Compare commits

...

21 Commits

Author SHA1 Message Date
7545b046af progress: render a single download with the same table as several
block() no longer special-cases one download with a [bracketed] one-liner;
it always draws the summary line, the column header, and one row per
download. The non-TTY path emits the same summary for one or many. This
drops lineOne and the dual rendering path it created, so a column or state
change is made in one place, not two that could drift.
2026-06-22 08:42:30 +09:00
cf9082ff72 all: cut bloated HTTP-client and tuning options
This is a personal torrent client, not a web client or daemon, so the
rarely-used HTTP-client knobs and redundant tuning flags are gone — flag,
config field, implementation, and tests removed for each. The option
surface drops from 54 to 28.

Web-client cluster removed: --all-proxy (HTTP_PROXY env still works),
load/save-cookies, --referer, -U/--user-agent, --http-user/--http-passwd,
--conditional-get, --remote-time, --check-certificate, --ca-certificate,
--lowest-speed-limit. Pass --header for a one-off auth/cookie/UA header.

Redundant knobs removed: -x and -k (folded into -s), --connect-timeout,
--retry-wait, per-item --max-download-limit/-u, --bt-max-peers,
--bt-stop-timeout, --disable-ipv6, --auto-save-interval, --human-readable,
--stop, -T (a positional .torrent works). follow-torrent and human-readable
are now unconditional defaults.
2026-06-22 08:30:12 +09:00
2afcb861e4 cli: parse each option value once
validate() parsed every value to bounds-check it, then the typed getters
re-parsed the same string at read time (with a comment admitting the
failure was unreachable). A single parseValue() now validates and converts,
finalize() runs it once per option after layering, and the getters just
read the stored typed value.
2026-06-22 08:29:25 +09:00
b98c36acb7 main: fold the two infohash-dedup paths into one
dedupeTorrents and markInfoHash were the same map-of-infohash logic, one
batch and one incremental, with the same hazard explained in prose four
times. Both now go through a single seenInfoHash helper.
2026-06-22 08:27:34 +09:00
a40cc67c38 progress: redesign the multi-download table
Drop the stacked progress bars for an aligned column table under a
header. Middle-elide names so downloads sharing a long prefix stay
distinct (a leading truncation rendered them identically), drop the
per-row DL:/UL: labels, and break the summary into downloading/seeding/
queued counts. Add SIZE (completed/total) and CN/SD (peer/seeder)
columns and show the seed share ratio in place of an idle upload rate.

Also fix the in-place redraw: the block could fill the pane's bottom row
and scroll it out from under the cursor, stranding a summary line each
tick. Reserve that row so the table falls back to the summary line
instead.
2026-06-22 04:13:58 +09:00
b632f37e3d download: free the -j slot once a torrent starts seeding 2026-06-22 01:59:23 +09:00
6e77cde4cc all: reword comments to state behavior directly
Comments (and one test comment) now describe the rules on their own terms
instead of by comparison to another downloader. No behavior change.
2026-06-21 16:10:13 +09:00
6c045999f2 cli: make bool parsing case-sensitive
Completes the boolean fix: only literal "true"/"false" are accepted, with no
case folding, so "True"/"TRUE" are rejected. Drop strings.ToLower. Keep
TrimSpace so a config value with surrounding whitespace still parses; only the
case folding was invented leniency.
2026-06-21 15:52:42 +09:00
10672f3816 all: cut bloat found in the pike audit
Dead, vestigial, and over-built code with no behavior change (except the
bool-vocabulary trim, a deliberate behavior tightening):

- httpdl: drop the vestigial sort+copy in validSegs (work-stealing was
  removed in 270812d, so segments always tile [0,total) ascending now);
  validate order in place, which also rejects a scrambled sidecar. Inline
  the endOff() accessor to s.end.
- cli: shrink the boolean vocabulary to true/false, which is all that's
  accepted. The invented yes/no/1/0/on/off spellings are gone, so e.g.
  --enable-dht=yes (or yes in got.conf) is now a usage error.
- download: remove the dead, never-called Stat.Done() method.
- progress: drop speed(), a pure alias of humanSize; call humanSize directly.
- main: replace the jobsHolder mutex box with atomic.Pointer[[]job] for the
  forced-exit session save; the capability stays, the lock and type go.
- bt: collapse isAddrInUse's X||X (missinggo.IsAddrInUse is that exact
  string match); drop the now-unused missinggo import (tidy -> indirect).
- option: note that prealloc and falloc are identical here.
2026-06-21 15:45:56 +09:00
3eb2d5ffcb cli: trim basic help, drop invented flag, parallelize by default
Surface area, not the machinery, was the problem: 16 flags crowded the
bare --help. Re-tag so basic shows only the 10 outcome-changing flags;
every other flag still works behind --help=<group>.

- cut --bt-metadata-timeout, an invented flag with no real counterpart. Keep
  its 60s magnet-metadata stall as a fixed internal default (bt.go), not a
  user knob.
- fix the parallelism footgun: a bare `got URL` used min(split,x)=1
  connection. When -x is unset, let --split drive per-host connections
  (capped at 16), so `got URL` gets 5 and `-s16` gets 16. An explicit -x
  is honored verbatim, including -x1. Locked by TestHTTPConnDefault.
- -s is the per-download speed dial now, so it moves to basic; -x (a
  per-server cap) moves to the http group. -j help says "files at once,
  not connections within one file".
2026-06-21 15:10:57 +09:00
2d99c39604 cli: simplify option model per pike review
- appendList: one home for the List newline-join rule (was duplicated
  across accumulate and set)
- fold the redundant Int64 getter into Int; drop the truthy helper for
  the shared boolWord vocabulary
- add Opt.Metavar so help derives the value placeholder from data
  instead of branching on the option name (seed-time)

No behavior change; --help output is byte-identical.
2026-06-21 14:34:14 +09:00
508708b039 httpdl/cli: drop --auto-split; classify errors by type, not message text
Two Pike cleanups.

--auto-split was a third connection-count policy (beside -x and -s) that added
an extra knob rather than keeping the model minimal. Removing it also retires the
now-dead autoConns/maxAutoConns and the per-host transport cap that existed only
to scale for it; min(split, M*-x) is the sole policy again.

main's exit-code mapping fell back to substring-matching third-party error text
(strings.Contains "connection refused"/"timeout"/...), which rots when a
dependency rewords a message. The typed checks (errors.Is on the syscall errno,
*net.DNSError, net.Error.Timeout, context.DeadlineExceeded) already cover the
real stdlib errors -- verified end to end: refused -> 6, bad host -> 19. The two
cases that genuinely needed the text match are our own errors, now typed
sentinels: httpdl.ErrTimeout (idle timeout) and the bt metadata timeout wrapping
context.DeadlineExceeded.
2026-06-20 23:55:39 +09:00
ebc630b231 httpdl: drop work-stealing segment pool for one worker per segment
The pool let an idle worker steal the back half of a slower in-flight
segment, behind a mutex that also serialized every advance. At the default
-x 1 it never fired, and it carried the trickiest invariant in the tree
(minSplit >= readBuf keeps an owner's in-flight write out of the stolen tail).

makeSegments already caps the segment count at the connection count, so the
segments map one-to-one onto workers: give each its own goroutine, advance
written with a plain atomic add, and snapshot the fixed slice with no lock.
pool/newPool/acquire/steal/advance and the seg.owned field all go away.

Net -156 lines. Behaviour at the user's flags is unchanged, except a slow
mirror's tail segment is no longer rebalanced onto idle connections.
2026-06-20 23:43:15 +09:00
e3505855c1 httpdl: fix progress-reporter race on name/out; harden mirror & resume validation
Findings from a Rob-Pike-lens review (bugs/races/network), each verified
against the code before fixing:

- httpdl: name/out are now atomic.Pointer[string] -- the engine publishes a
  download to the reporter before Run resolves the name, so Stat raced the
  write (bt already did this)
- httpdl: a malformed --proxy fails loudly instead of silently bypassing it
- httpdl: a 206 must carry a matching Content-Range; a 200 in segmented mode is
  fatal so it fails over instead of burning the retry budget
- httpdl: single-stream mirror failover validates the range before appending;
  ErrTooSlow only when the error is a real ctx cancellation
- httpdl: idle guard tracks progress by timestamp (no Reset/Stop race, no
  sticky fired flag)
- httpdl/control: reject a resume file whose segments don't tile [0,total)
- bt: clamp the listen-port range; verify on-disk data before choosing pieces
  under --check-integrity
- cli: reject size overflow; show --seed-time=MIN; clamp --select-file range
- progress/engine/main: clamp ETA against int64 overflow; show queued
  downloads as waiting; join the reporter on exit instead of a 20ms sleep
2026-06-20 23:28:26 +09:00
c4c03dde60 gitignore: ignore downloaded media so an in-tree run can't commit content
running got in the repo root drops downloaded files next to the source;
torrents/control files were already ignored, add common media (mkv/mp4/
avi/mov/webm/iso/mp3/flac) so anime/ISO downloads don't show up as
untracked and can't be committed by accident.
2026-06-20 21:54:29 +09:00
2cce0fcde1 cli: show -x/-s/-k in basic help
max-connection-per-server, split and min-split-size were tagged HTTP, so
they only appeared under --help=all. Tag them Basic so they show in the
default --help, where users expect the core connection knobs.
2026-06-20 21:53:33 +09:00
456ae3545b httpdl: unify completed counter and dedupe mirror loop; engine: harden untrack
From a Rob Pike pass over the tree:
- singleOnce adds deltas to d.completed like pump does, instead of
  storing absolute offsets, so the counter has one write discipline.
- one overMirrors helper replaces the three copy-pasted mirror-fallover
  loops in probeRetry, fetchSeg and single.
- failedDownload becomes a pointer, the only value-type Download impl, so
  Engine.untrack's == is always pointer identity and can't compare the
  fields of a value (keying on Stat().ID is out — a torrent's ID changes
  from source to infohash once metadata loads).
2026-06-20 21:12:07 +09:00
bb49578892 bt: filter anacrolix logs to errors so warnings don't corrupt the readout
webseed/peer/tracker warnings (the 403/429 chatter) go through slog, the
rest through the legacy analog logger; both default to Warning, which
interleaves with and corrupts the live progress block. Filter both to
Error and above so only genuine errors reach stderr. quietSlogger is a
small testable helper; anacrolix/log becomes a direct dependency.
2026-06-20 21:11:57 +09:00
3fc29e3a86 progress: show multiple downloads as a bar table
lineMany crammed every download onto one line as name+percent, dropping
the speed column that answers "which one is stalled" and producing ugly
double brackets for names like "[Gecko]". render N>1 as a redrawn block:
one row each (name, bar, percent, DL speed, ETA/stalled), a header with
the active count and totals, falling back to a single summary line when
the table would overflow the window. one download keeps its dashboard.
2026-06-20 19:45:49 +09:00
7336f9c95e main: dedupe followed .torrent files by infohash
a .torrent fetched over http is added to the shared client in a second
pass (followUps); two that name the same torrent would share one torrent
on the refcount-less client, and the first to finish would Drop it out
from under the other, stranding it at 0%. dedupe the followed files by
infohash, the guard build() already applies to pass-1 sources.
2026-06-20 19:28:32 +09:00
554beb6b48 httpdl: add segment work-stealing and --auto-split
an idle worker now steals the back half of the busiest in-flight segment
instead of exiting, so one slow segment no longer drains alone over a single
connection. control file persists stolen tails; resume re-tiles.

add --auto-split (opt-in): derive the connection count from file size (one
per min-split-size, up to 16), overriding -x/-s. default stays at 1
connection for a single server.
2026-06-19 14:01:59 +09:00
31 changed files with 1513 additions and 1651 deletions

10
.gitignore vendored
View File

@@ -7,3 +7,13 @@ ref/
bin/ bin/
.torrent.db* .torrent.db*
*.part *.part
# media downloaded by running got in the repo tree (never commit content)
*.mkv
*.mp4
*.avi
*.mov
*.webm
*.iso
*.mp3
*.flac

Binary file not shown.

158
README.md
View File

@@ -1,13 +1,10 @@
# got # got
A small, Rob-Pike-style command-line download tool in Go: one program that A command-line download tool in Go. It fetches HTTP(S) files over multiple
downloads HTTP(S) URLs over several connections and BitTorrent connections and downloads BitTorrent magnets and `.torrent` files — one binary,
magnets/torrents, with a focused set of options. a small set of flags. It is a downloader, not a web client: there are no
cookie/proxy/auth flags — pass `--header` for a one-off header (auth token,
The design leans on Go's runtime instead of a hand-rolled reactor. There is no cookie, user-agent) when a download needs one.
single-threaded event-poll loop: one goroutine per download blocks on real I/O,
a buffered channel of slots bounds concurrency, and `context` carries shutdown.
What remains is the data — and, per Pike, the data is the design.
## Build ## Build
@@ -15,148 +12,63 @@ What remains is the data — and, per Pike, the data is the design.
go build -o got . go build -o got .
``` ```
Requires Go 1.25+. The BitTorrent engine is Needs Go 1.25+.
[`anacrolix/torrent`](https://github.com/anacrolix/torrent); everything else is
the standard library plus `golang.org/x/{term,time}`.
## Use ## Usage
```sh ```sh
# segmented HTTP download, 16 connections, into ./downloads # HTTP download: parallel out of the box (5 connections); -s turns it up
got -x16 -s16 -d downloads https://example.com/big.iso got https://example.com/big.iso
got -s16 https://example.com/big.iso
# resume an interrupted download # resume an interrupted download
got -c -x16 https://example.com/big.iso got -c https://example.com/big.iso
# the same file from several mirrors at once (one file, connections spread # one file from several mirrors (segments spread across the servers,
# across servers, a dead mirror falls over); -Z downloads them separately instead # a dead mirror falls over); -Z downloads them as separate files instead.
got -x4 https://a.example/big.iso https://b.example/big.iso got -s16 https://a.example/big.iso https://b.example/big.iso
# a torrent or magnet; seed for 30 minutes after finishing # a torrent or magnet; seed for 30 minutes after finishing
got --seed-time=30 ubuntu.torrent got --seed-time=30 ubuntu.torrent
got 'magnet:?xt=urn:btih:...' got 'magnet:?xt=urn:btih:...'
# list the files in a torrent without downloading # many downloads, two at a time
got -S -T some.torrent
# several downloads at once, two running in parallel
got -j2 -i urls.txt got -j2 -i urls.txt
``` ```
Run `got --help` (or `--help=http`, `--help=bittorrent`, ...) for the full Run `got --help` (or `--help=all`) for every option.
option list.
### Options ## Common options
A focused subset of options, with familiar short flags and sensible defaults.
| flag | meaning | default | | flag | meaning | default |
|------|---------|---------| |------|---------|---------|
| `-d, --dir` | output directory | `.` | | `-d, --dir` | output directory | `.` |
| `-o, --out` | output filename (single download) | from URL | | `-o, --out` | output filename | from URL |
| `-i, --input-file` | read downloads line by line (TAB-separated URLs = mirrors; `-` = stdin) | |
| `-c, --continue` | resume a partial download | false | | `-c, --continue` | resume a partial download | false |
| `-x, --max-connection-per-server` | connections to one server (116) | 1 | | `-s, --split` | connections per download (the speed dial) | 5 |
| `-s, --split` | split a download into N connections | 5 | | `-j, --max-concurrent-downloads` | files downloaded at once | 5 |
| `-Z, --force-sequential` | download each command-line URL separately, not as mirrors | false |
| `-k, --min-split-size` | do not split below this size | 20M |
| `-j, --max-concurrent-downloads` | parallel downloads | 5 |
| `--max-overall-download-limit` | global speed cap | 0 (off) | | `--max-overall-download-limit` | global speed cap | 0 (off) |
| `--max-download-limit` | per-download speed cap | 0 (off) | | `--checksum` | verify the finished file: `TYPE=DIGEST` (sha-256, sha-1, …) | |
| `--retry-wait` | seconds to wait between retries | 0 | | `--seed-time` | minutes to seed after a torrent finishes | by ratio |
| `--load-cookies` / `--save-cookies` | read/write a Netscape `cookies.txt` | |
| `--conditional-get` | skip the download if the local file is up to date | false |
| `--remote-time` | set the file's mtime from the server | false |
| `--checksum` | verify the finished file: `TYPE=DIGEST` (sha-256, sha-1, md5, …) | |
| `--dry-run` | check the file is available but do not download it | false |
| `--ca-certificate` | verify HTTPS against the CA certificates in FILE (PEM) | |
| `--stop` | stop the program after N seconds | 0 (off) |
| `--disable-ipv6` | force IPv4-only connections | false |
| `-T, --torrent-file` | a `.torrent` file | |
| `--seed-time` | minutes to seed after finishing | seed by ratio |
| `--seed-ratio` | stop seeding at this ratio | 1.0 | | `--seed-ratio` | stop seeding at this ratio | 1.0 |
| `--listen-port` | port (range) for incoming peers | 6881-6999 |
| `--enable-dht` | use the BitTorrent DHT | true |
| `--select-file` | fetch only these file indexes (`1,3-5`) | all |
| `-S, --show-files` | list torrent files and exit | | | `-S, --show-files` | list torrent files and exit | |
| `--bt-stop-timeout` | give up if no download progress for N s | 0 (off) | | `-q, --quiet` | no progress output | false |
| `--bt-metadata-timeout` | give up if a magnet can't fetch metadata in N s | 60 |
| `--save-session` | on exit, write unfinished downloads to FILE | |
| `-q, --quiet` | no progress readout | false |
### Resuming across runs ## Resume
There is no daemon — resume is a plain file. Point HTTP downloads resume from a small `<file>.got` sidecar — just re-run with `-c`.
`--save-session` and `-i` at the same file and add `-c`: unfinished downloads Torrents resume from the data already on disk. To resume across runs, point
are loaded at start and the still-unfinished ones are written back on exit. `--save-session` and `-i` at the same file:
```sh ```sh
got -c --save-session=got.session -i got.session <new urls/magnets...> got -c --save-session=got.session -i got.session <urls/magnets...>
``` ```
On the first run the session file need not exist. HTTP downloads resume from ## Notes
their `.got` control file; torrents resume from the data already on disk
(re-checked by the engine). A magnet with no reachable peers no longer hangs:
`--bt-metadata-timeout` bounds the metadata fetch and defaults to 60s (set it to
0 to wait forever).
## Design - BitTorrent uses [`anacrolix/torrent`](https://github.com/anacrolix/torrent);
HTTP uses the standard library.
``` - Not implemented: FTP/SFTP, Metalink, and the JSON-RPC server.
main classify URIs, wire options, run, report exit code - Speed caps are global: `--max-overall-download-limit` /
└─ download the Download interface + the scheduler (-j semaphore, ctx) `--max-overall-upload-limit` apply across the whole run.
├─ httpdl []Segment + worker pool + WriteAt + JSON resume sidecar - Standard `HTTP_PROXY` / `HTTPS_PROXY` environment variables are honored.
└─ bt anacrolix/torrent behind the Download interface
├─ cli one flat option table -> hand parser -> layered config
└─ progress one ticker goroutine, pull-snapshot, \r line redraw
```
The contract is one small interface:
```go
type Download interface {
Name() string
Run(ctx context.Context) error // blocks in its own goroutine
Stat() Stat // a value snapshot, lock-free to read
}
```
- **HTTP** splits the file into byte-range `Segment`s handed to a pool of
workers. Each worker streams its range straight to one shared file with
`os.File.WriteAt` — safe for concurrent non-overlapping writes, so there is no
shared seek offset and no mutex. Resume is a small JSON sidecar
(`<file>.got`) holding per-segment progress plus validators (length +
ETag/Last-Modified) so a stale file is never trusted.
- **BitTorrent** is `anacrolix/torrent` configured from the CLI options and
driven to completion, then seeded for `--seed-time` minutes or up to
`--seed-ratio`.
- **Options** are one flat `[]Opt` table — the single source of truth for
parsing, validation, defaults and `--help`. Layers apply in order:
built-in defaults, config file, proxy environment, command line.
- **Progress** is a single goroutine that ticks once a second, pulls a snapshot
of the running downloads, derives speeds from the change since the last tick,
and redraws one line with `\r` + erase-to-end-of-line.
## Status
Implemented: segmented HTTP(S) with resume — including a foreign/browser-started
partial file (`-c`) — multi-mirror downloads (several URLs, or TAB-grouped `-i`
lines, fetch one file with connections spread across servers and fallover; `-Z`
to download them separately), retries, rate limits, auto-renaming,
`Content-Disposition` naming, conditional GET, cookies
(`--load-cookies`/`--save-cookies`), HTTP basic auth, `--remote-time`, whole-file
checksum verification (`--checksum`), and a `--dry-run` availability check;
BitTorrent download + seeding (magnet and `.torrent`,
DHT, trackers, file selection, show-files, metadata-fetch timeout), seeding that
stops on whichever of `--seed-time`/`--seed-ratio` comes first, and a listen port
chosen from the whole range so a busy port no longer disables BitTorrent;
a command-line interface with documented exit codes, config file, live progress,
and session save/reload for resume across runs.
Deferred (not implemented yet): FTP/SFTP, Metalink, and the
JSON-RPC server. The architecture leaves room for
each — a new protocol is just another `Download`. One known limitation: the
BitTorrent engine (`anacrolix/torrent`) applies download/upload rate limits
client-wide, so `--max-upload-limit`/`--max-download-limit` act per-run rather
than per-torrent when several torrents run at once (`--max-overall-*` are exact);
DHT/PEX likewise cannot yet be disabled per private torrent.

141
bt/bt.go
View File

@@ -9,13 +9,16 @@ package bt
import ( import (
"context" "context"
"fmt" "fmt"
"io"
"log/slog"
"os"
"path/filepath" "path/filepath"
"strconv" "strconv"
"strings" "strings"
"sync/atomic" "sync/atomic"
"time" "time"
missinggo "github.com/anacrolix/missinggo/v2" alog "github.com/anacrolix/log"
"github.com/anacrolix/torrent" "github.com/anacrolix/torrent"
"github.com/anacrolix/torrent/metainfo" "github.com/anacrolix/torrent/metainfo"
"github.com/hanbok/got/download" "github.com/hanbok/got/download"
@@ -26,14 +29,9 @@ import (
type ClientConfig struct { type ClientConfig struct {
Dir string Dir string
ListenPortSpec string // raw --listen-port spec ("6881-6999,7000"); empty lets the OS choose. ListenPortSpec string // raw --listen-port spec ("6881-6999,7000"); empty lets the OS choose.
DHT bool DHT bool // --enable-dht
MaxPeers int // max peer connections per torrent (0 = library default)
OverallDown *rate.Limiter // global down limit (may be nil) OverallDown *rate.Limiter // global down limit (may be nil)
OverallUp *rate.Limiter // global up limit (may be nil) OverallUp *rate.Limiter // global up limit (may be nil)
DownLimit int64 // per-run bytes/s, 0 = unlimited (used only without a global limiter)
UpLimit int64
UserAgent string
DisableIPv6 bool // force IPv4-only (--disable-ipv6)
} }
// ParsePorts expands a --listen-port spec into a flat list of candidate ports. // ParsePorts expands a --listen-port spec into a flat list of candidate ports.
@@ -64,57 +62,54 @@ func ParsePorts(spec string) []int {
if err != nil { if err != nil {
continue continue
} }
for i := a; i <= b; i++ { // Clamp to the valid port space before iterating: a parseable but oversized
if i >= 1 && i <= 65535 { // spec like "1-9999999999" would otherwise spin for ~10^10 iterations and
ports = append(ports, i) // hang startup, even though only [1,65535] can ever be appended.
if a < 1 {
a = 1
} }
if b > 65535 {
b = 65535
}
for i := a; i <= b; i++ {
ports = append(ports, i)
} }
} }
return ports return ports
} }
// quietSlogger filters anacrolix's structured logging down to Error and above,
// so the routine webseed/peer/tracker warning chatter (the 403/429 noise) no
// longer interleaves with — and corrupts — the live progress display. Genuine
// errors still reach w.
func quietSlogger(w io.Writer) *slog.Logger {
return slog.New(slog.NewTextHandler(w, &slog.HandlerOptions{Level: slog.LevelError}))
}
// NewClient builds the shared anacrolix client from the run-wide config. When a // NewClient builds the shared anacrolix client from the run-wide config. When a
// listen-port spec is given it tries each candidate port in turn, advancing past // listen-port spec is given it tries each candidate port in turn, advancing past
// any that is already in use, and falls back to an OS-chosen port if none bind, // any that is already in use, and falls back to an OS-chosen port if none bind,
// so a single busy port does not disable BitTorrent entirely. // so a single busy port does not disable BitTorrent entirely. The rate limits
// // and the DHT toggle are client-wide (anacrolix has no per-torrent equivalent),
// Two per-torrent features have no equivalent in anacrolix/torrent v1.61.0, // which is why they live here on the shared client rather than per download.
// which exposes only client-wide knobs; both must therefore be configured run-wide
// here rather than attached to an individual Torrent:
//
// 1. No per-Torrent rate limiter. Throttles are read from the client-wide
// DownloadRateLimiter/UploadRateLimiter (no TorrentSpec rate field exists), so
// per-torrent --max-download-limit / --max-upload-limit cannot be
// honored individually; the effective cap is the shared limiter set below.
// 2. No per-Torrent DHT/PEX/LSD toggle for BEP27 private torrents. Those are
// client-wide only (NoDHT, DisablePEX, etc.), so a private torrent's leak
// prevention must come from a run-wide setting (e.g. --enable-dht=false).
func NewClient(cfg ClientConfig) (*torrent.Client, error) { func NewClient(cfg ClientConfig) (*torrent.Client, error) {
c := torrent.NewDefaultClientConfig() c := torrent.NewDefaultClientConfig()
// Keep the library's own logging out of the readout. Webseed/peer warnings go
// through slog, the rest through the legacy analog logger, so filter both to
// Error and above (the analog default is Warning, which is what leaks today).
c.Slogger = quietSlogger(os.Stderr)
c.Logger = alog.Default.FilterLevel(alog.Error)
c.DataDir = cfg.Dir c.DataDir = cfg.Dir
if c.DataDir == "" { if c.DataDir == "" {
c.DataDir = "." c.DataDir = "."
} }
c.Seed = true c.Seed = true
c.NoDHT = !cfg.DHT c.NoDHT = !cfg.DHT
if cfg.MaxPeers > 0 { if cfg.OverallDown != nil {
c.EstablishedConnsPerTorrent = cfg.MaxPeers
}
if cfg.UserAgent != "" {
c.HTTPUserAgent = cfg.UserAgent
}
c.DisableIPv6 = cfg.DisableIPv6
switch {
case cfg.OverallDown != nil:
c.DownloadRateLimiter = cfg.OverallDown c.DownloadRateLimiter = cfg.OverallDown
case cfg.DownLimit > 0:
c.DownloadRateLimiter = rate.NewLimiter(rate.Limit(cfg.DownLimit), download.LimiterBurst(cfg.DownLimit))
} }
switch { if cfg.OverallUp != nil {
case cfg.OverallUp != nil:
c.UploadRateLimiter = cfg.OverallUp c.UploadRateLimiter = cfg.OverallUp
case cfg.UpLimit > 0:
c.UploadRateLimiter = rate.NewLimiter(rate.Limit(cfg.UpLimit), download.LimiterBurst(cfg.UpLimit))
} }
// Candidate ports from the --listen-port spec. // Candidate ports from the --listen-port spec.
@@ -148,10 +143,11 @@ func NewClient(cfg ClientConfig) (*torrent.Client, error) {
return cl, nil return cl, nil
} }
// isAddrInUse reports whether err is an "address already in use" bind failure, // isAddrInUse reports whether err is an "address already in use" bind failure.
// matching either missinggo's helper or the raw string the library wraps. // (anacrolix/missinggo's IsAddrInUse is this exact string match, so we just do
// it directly rather than carry the import for it.)
func isAddrInUse(err error) bool { func isAddrInUse(err error) bool {
return err != nil && (missinggo.IsAddrInUse(err) || strings.Contains(err.Error(), "address already in use")) return err != nil && strings.Contains(err.Error(), "address already in use")
} }
// Options are the per-download settings. // Options are the per-download settings.
@@ -159,8 +155,6 @@ type Options struct {
SeedTimeSet bool // was --seed-time given? SeedTimeSet bool // was --seed-time given?
SeedTime time.Duration // how long to seed (0 with SeedTimeSet means no seeding) SeedTime time.Duration // how long to seed (0 with SeedTimeSet means no seeding)
SeedRatio float64 // stop seeding at this ratio; checked alongside SeedTime SeedRatio float64 // stop seeding at this ratio; checked alongside SeedTime
StopTimeout time.Duration // abort if no download progress for this long (0 = off)
MetaTimeout time.Duration // abort a magnet that can't fetch metadata in this long (0 = off)
SelectFiles map[int]bool // 1-based file indexes to fetch; empty = all SelectFiles map[int]bool // 1-based file indexes to fetch; empty = all
CheckIntegrity bool // re-verify data against piece hashes before downloading CheckIntegrity bool // re-verify data against piece hashes before downloading
DryRun bool // fetch metadata only, then stop without downloading (--dry-run) DryRun bool // fetch metadata only, then stop without downloading (--dry-run)
@@ -180,12 +174,17 @@ type Download struct {
// t is published once Run adds the torrent to the client. It is atomic so // t is published once Run adds the torrent to the client. It is atomic so
// Stat stays lock-free, exactly like httpdl.Stat and bt's own name field. // Stat stays lock-free, exactly like httpdl.Stat and bt's own name field.
t atomic.Pointer[torrent.Torrent] t atomic.Pointer[torrent.Torrent]
// seeding is closed once Run finishes downloading and transitions to
// seed-only, so the engine can drop this torrent from the -j concurrency
// count. Created in New, closed exactly once in Run.
seeding chan struct{}
} }
// New builds a torrent download on the shared client. source is a .torrent path // New builds a torrent download on the shared client. source is a .torrent path
// when isFile is true, otherwise a magnet URI. // when isFile is true, otherwise a magnet URI.
func New(client *torrent.Client, source string, isFile bool, opts Options) *Download { func New(client *torrent.Client, source string, isFile bool, opts Options) *Download {
d := &Download{client: client, source: source, isFile: isFile, opts: opts} d := &Download{client: client, source: source, isFile: isFile, opts: opts, seeding: make(chan struct{})}
if isFile { if isFile {
d.setName(strings.TrimSuffix(filepath.Base(source), ".torrent")) d.setName(strings.TrimSuffix(filepath.Base(source), ".torrent"))
} else { } else {
@@ -210,6 +209,12 @@ func (d *Download) setStatus(s download.Status) {
atomic.StoreInt32(&d.status, int32(s)) atomic.StoreInt32(&d.status, int32(s))
} }
// SeedingStarted is closed when this torrent finishes downloading and switches
// to seed-only. The engine watches it to release the -j concurrency slot at that
// moment: seeding is not downloading, so a seed-only torrent must not hold a slot
// that a queued download could use. Implements download.Seeder.
func (d *Download) SeedingStarted() <-chan struct{} { return d.seeding }
// id is the process-unique identity for this download: the infohash once known, // id is the process-unique identity for this download: the infohash once known,
// otherwise the source string (a pre-metadata magnet has no infohash yet, and // otherwise the source string (a pre-metadata magnet has no infohash yet, and
// two such magnets would otherwise collide on Name). // two such magnets would otherwise collide on Name).
@@ -289,20 +294,24 @@ func (d *Download) Run(ctx context.Context) (err error) {
return nil return nil
} }
if err := d.choose(t); err != nil { // --check-integrity: re-hash the existing on-disk data BEFORE arming the
return err // request loop, so already-good pieces are not re-requested from peers
} // (verify first, then fetch only what is missing).
if d.opts.CheckIntegrity { if d.opts.CheckIntegrity {
if err := t.VerifyDataContext(ctx); err != nil { if err := t.VerifyDataContext(ctx); err != nil {
return err return err
} }
} }
if err := d.choose(t); err != nil {
return err
}
if err := d.wait(ctx, t); err != nil { if err := d.wait(ctx, t); err != nil {
return err return err
} }
d.setStatus(download.Seeding) d.setStatus(download.Seeding)
close(d.seeding) // free the -j slot: a seed-only torrent no longer downloads
d.seed(ctx, t) d.seed(ctx, t)
d.setStatus(download.Complete) d.setStatus(download.Complete)
return nil return nil
@@ -347,36 +356,35 @@ func (d *Download) choose(t *torrent.Torrent) error {
return nil return nil
} }
// awaitInfo blocks until the torrent metadata arrives. It gives up after // metadataTimeout bounds the magnet metadata-fetch phase. A magnet with no
// --bt-metadata-timeout (default 60s) so a magnet with no reachable peers — a // reachable peers — a dead link, or UDP trackers behind a firewall with no DHT —
// dead link, or UDP trackers behind a firewall with no DHT — fails fast instead // would otherwise hang forever, so awaitInfo gives up after this fixed window.
// of hanging forever. Setting --bt-metadata-timeout=0 disables the timeout and // It is not a user-facing flag — 60s is a generous ceiling that a healthy swarm
// waits indefinitely. // clears in well under a second.
const metadataTimeout = 60 * time.Second
// awaitInfo blocks until the torrent metadata arrives, giving up after
// metadataTimeout so a peerless magnet fails fast instead of hanging forever.
func (d *Download) awaitInfo(ctx context.Context, t *torrent.Torrent) error { func (d *Download) awaitInfo(ctx context.Context, t *torrent.Torrent) error {
var deadline <-chan time.Time tm := time.NewTimer(metadataTimeout)
if d.opts.MetaTimeout > 0 {
tm := time.NewTimer(d.opts.MetaTimeout)
defer tm.Stop() defer tm.Stop()
deadline = tm.C
}
select { select {
case <-t.GotInfo(): case <-t.GotInfo():
return nil return nil
case <-ctx.Done(): case <-ctx.Done():
return ctx.Err() return ctx.Err()
case <-deadline: case <-tm.C:
return fmt.Errorf("timed out fetching metadata after %s", d.opts.MetaTimeout) // Wrap DeadlineExceeded so the exit-code mapping classifies it as a timeout
// (2) via errors.Is, rather than matching on the message text.
return fmt.Errorf("timed out fetching metadata after %s: %w", metadataTimeout, context.DeadlineExceeded)
} }
} }
// wait blocks until the wanted data is complete, ctx is cancelled, or the // wait blocks until the wanted data is complete or ctx is cancelled, polling
// download stalls past --bt-stop-timeout. // completion a couple of times a second.
func (d *Download) wait(ctx context.Context, t *torrent.Torrent) error { func (d *Download) wait(ctx context.Context, t *torrent.Torrent) error {
tick := time.NewTicker(500 * time.Millisecond) tick := time.NewTicker(500 * time.Millisecond)
defer tick.Stop() defer tick.Stop()
last := t.BytesCompleted()
lastChange := time.Now()
for { for {
if d.complete(t) { if d.complete(t) {
return nil return nil
@@ -385,11 +393,6 @@ func (d *Download) wait(ctx context.Context, t *torrent.Torrent) error {
case <-ctx.Done(): case <-ctx.Done():
return ctx.Err() return ctx.Err()
case <-tick.C: case <-tick.C:
if n := t.BytesCompleted(); n != last {
last, lastChange = n, time.Now()
} else if d.opts.StopTimeout > 0 && time.Since(lastChange) > d.opts.StopTimeout {
return fmt.Errorf("no progress for %s", d.opts.StopTimeout)
}
} }
} }
} }

View File

@@ -1,10 +1,38 @@
package bt package bt
import ( import (
"bytes"
"context"
"log/slog"
"reflect" "reflect"
"strings"
"testing" "testing"
) )
// TestQuietSloggerDropsWarnings: the library's warning chatter (the webseed
// 403/429 noise) is filtered out so it can't corrupt the live display, while a
// genuine error still gets through.
func TestQuietSloggerDropsWarnings(t *testing.T) {
var buf bytes.Buffer
lg := quietSlogger(&buf)
lg.Warn("webseed request error", "err", "429 Too Many Requests")
if buf.Len() != 0 {
t.Errorf("warning leaked into output: %q", buf.String())
}
// anacrolix's webseed code logs via Log(ctx, level, ...); confirm that path is
// filtered too, not just the Warn helper.
lg.Log(context.Background(), slog.LevelWarn, "another warning")
if buf.Len() != 0 {
t.Errorf("Log-at-Warn leaked: %q", buf.String())
}
lg.Error("real failure")
if !strings.Contains(buf.String(), "real failure") {
t.Errorf("error was suppressed, want it kept: %q", buf.String())
}
}
func TestParsePorts(t *testing.T) { func TestParsePorts(t *testing.T) {
tests := []struct { tests := []struct {
spec string spec string

View File

@@ -89,19 +89,31 @@ func flagLabel(o *Opt) string {
b.WriteString(" ") b.WriteString(" ")
} }
b.WriteString("--" + o.Long) b.WriteString("--" + o.Long)
switch o.Kind { if m := metavar(o); m != "" {
case Bool: b.WriteString("=" + m)
// no argument shown
case Size:
b.WriteString("=SIZE")
case Int:
b.WriteString("=N")
case Float:
b.WriteString("=RATIO")
case Enum:
b.WriteString("=" + strings.Join(o.Enum, "|"))
default:
b.WriteString("=VAL")
} }
return b.String() return b.String()
} }
// metavar is the placeholder shown after a value-taking flag, e.g. the "N" in
// "--split=N". An explicit Opt.Metavar wins; otherwise it derives from the Kind.
// Bool takes no value, so its metavar is empty and no "=ARG" is printed.
func metavar(o *Opt) string {
if o.Metavar != "" {
return o.Metavar
}
switch o.Kind {
case Bool:
return ""
case Size:
return "SIZE"
case Int:
return "N"
case Float:
return "RATIO"
case Enum:
return strings.Join(o.Enum, "|")
default:
return "VAL"
}
}

View File

@@ -45,11 +45,15 @@ func (t Tag) String() string {
// Opt describes one option: its names, value kind, default and help text. // Opt describes one option: its names, value kind, default and help text.
type Opt struct { type Opt struct {
Long string // long name without dashes, e.g. "max-connection-per-server" Long string // long name without dashes, e.g. "max-concurrent-downloads"
Short byte // short flag byte, e.g. 'x'; 0 if none Short byte // short flag byte, e.g. 'x'; 0 if none
Kind Kind Kind Kind
Default string Default string
Help string Help string
// Metavar overrides the "=PLACEHOLDER" shown after the flag in --help. Empty
// means derive it from Kind (N, SIZE, RATIO, ...). Set it only when the
// Kind-derived default would mislead, e.g. seed-time is minutes, not a ratio.
Metavar string
Tag Tag Tag Tag
Enum []string // valid values when Kind == Enum Enum []string // valid values when Kind == Enum
// Min is the inclusive lower Int/Size bound. The zero value enforces a // Min is the inclusive lower Int/Size bound. The zero value enforces a
@@ -60,68 +64,44 @@ type Opt struct {
Max int64 Max int64
} }
// options is the single source of truth: a focused, practical subset. // options is the single source of truth: a focused, practical subset. The Tag
// decides which --help group an option shows under; the Basic group is the bare
// `got --help` and is kept to the handful of flags that change WHAT happens, not
// how fast. Every other flag still works — it just lives behind --help=<group>.
var options = []Opt{ var options = []Opt{
// --- basic --- // --- basic --- the outcome-changing core shown by a bare `got --help`.
{Long: "dir", Short: 'd', Kind: Str, Help: "directory to store downloaded files", Tag: Basic}, {Long: "dir", Short: 'd', Kind: Str, Help: "directory to store downloaded files", Tag: Basic},
{Long: "out", Short: 'o', Kind: Str, Help: "output filename for the download", Tag: Basic}, {Long: "out", Short: 'o', Kind: Str, Help: "output filename for the download", Tag: Basic},
{Long: "input-file", Short: 'i', Kind: Str, Help: "read URIs line by line from FILE (- for stdin)", Tag: Basic}, {Long: "input-file", Short: 'i', Kind: Str, Help: "read URIs line by line from FILE (- for stdin)", Tag: Basic},
{Long: "max-concurrent-downloads", Short: 'j', Kind: Int, Default: "5", Min: 1, Help: "how many files to download at once (not connections within one file)", Tag: Basic},
{Long: "continue", Short: 'c', Kind: Bool, Default: "false", Help: "resume a partially downloaded file", Tag: Basic}, {Long: "continue", Short: 'c', Kind: Bool, Default: "false", Help: "resume a partially downloaded file", Tag: Basic},
{Long: "max-concurrent-downloads", Short: 'j', Kind: Int, Default: "5", Min: 1, Help: "max number of parallel downloads", Tag: Basic}, {Long: "split", Short: 's', Kind: Int, Default: "5", Min: 1, Help: "connections per download — the speed dial (default 5; e.g. -s16 for 16)", Tag: Basic},
{Long: "check-integrity", Short: 'V', Kind: Bool, Default: "false", Help: "re-verify torrent data against piece hashes (BitTorrent)", Tag: Basic},
{Long: "show-files", Short: 'S', Kind: Bool, Default: "false", Help: "list files in a torrent and exit", Tag: Basic},
{Long: "allow-overwrite", Kind: Bool, Default: "false", Help: "overwrite an existing file", Tag: Basic},
{Long: "auto-file-renaming", Kind: Bool, Default: "true", Help: "rename file (.1, .2, ...) if it already exists", Tag: Basic},
{Long: "max-overall-download-limit", Kind: Size, Default: "0", Help: "global download speed limit (0 = unlimited)", Tag: Basic}, {Long: "max-overall-download-limit", Kind: Size, Default: "0", Help: "global download speed limit (0 = unlimited)", Tag: Basic},
{Long: "max-download-limit", Kind: Size, Default: "0", Help: "per-download speed limit (0 = unlimited)", Tag: Basic}, {Long: "checksum", Kind: Str, Help: "verify the finished file: TYPE=DIGEST, e.g. sha-256=<hex> (md5, sha-1, sha-224, sha-256, sha-384, sha-512)", Tag: Basic},
{Long: "show-files", Short: 'S', Kind: Bool, Default: "false", Help: "list files in a torrent and exit", Tag: Basic},
{Long: "quiet", Short: 'q', Kind: Bool, Default: "false", Help: "suppress the progress readout", Tag: Basic}, {Long: "quiet", Short: 'q', Kind: Bool, Default: "false", Help: "suppress the progress readout", Tag: Basic},
{Long: "human-readable", Kind: Bool, Default: "true", Help: "show sizes as Ki/Mi/Gi", Tag: Basic},
// --- http --- // --- http ---
{Long: "max-connection-per-server", Short: 'x', Kind: Int, Default: "1", Min: 1, Max: 16, Help: "max connections to one server (1-16)", Tag: HTTP},
{Long: "split", Short: 's', Kind: Int, Default: "5", Min: 1, Help: "split a download into N connections; actual connections are min(max-connection-per-server, split), and -x defaults to 1", Tag: HTTP},
{Long: "min-split-size", Short: 'k', Kind: Size, Default: "20M", Min: 1 << 20, Max: 1 << 30, Help: "do not split a piece smaller than SIZE (1M-1024M)", Tag: HTTP},
{Long: "force-sequential", Short: 'Z', Kind: Bool, Default: "false", Help: "download each command-line URI as its own file instead of mirroring them", Tag: HTTP}, {Long: "force-sequential", Short: 'Z', Kind: Bool, Default: "false", Help: "download each command-line URI as its own file instead of mirroring them", Tag: HTTP},
{Long: "max-tries", Short: 'm', Kind: Int, Default: "5", Min: 0, Help: "max retries per segment (0 = unlimited)", Tag: HTTP}, {Long: "max-tries", Short: 'm', Kind: Int, Default: "5", Min: 0, Help: "max retries per segment (0 = unlimited)", Tag: HTTP},
{Long: "timeout", Short: 't', Kind: Int, Default: "60", Min: 1, Help: "connection timeout in seconds", Tag: HTTP}, {Long: "timeout", Short: 't', Kind: Int, Default: "60", Min: 1, Help: "connection/idle timeout in seconds", Tag: HTTP},
{Long: "header", Kind: List, Help: "append an extra HTTP header (repeatable)", Tag: HTTP}, {Long: "header", Kind: List, Help: "append an extra HTTP header (repeatable), e.g. 'Authorization: Bearer …'", Tag: HTTP},
{Long: "user-agent", Short: 'U', Kind: Str, Default: "got/1.0", Help: "HTTP User-Agent", Tag: HTTP},
{Long: "referer", Kind: Str, Help: "HTTP Referer header", Tag: HTTP},
{Long: "all-proxy", Kind: Str, Help: "proxy for all protocols (host:port)", Tag: HTTP},
{Long: "check-certificate", Kind: Bool, Default: "true", Help: "verify the server's TLS certificate", Tag: HTTP},
{Long: "ca-certificate", Kind: Str, Help: "verify HTTPS servers against the CA certificates in FILE (PEM)", Tag: HTTP},
{Long: "retry-wait", Kind: Int, Default: "0", Min: 0, Max: 600, Help: "seconds to wait between retries (0 = no wait)", Tag: HTTP},
{Long: "connect-timeout", Kind: Int, Default: "60", Min: 1, Help: "connection establishment timeout in seconds", Tag: HTTP},
{Long: "load-cookies", Kind: Str, Help: "load Cookies from FILE (Netscape/Mozilla format)", Tag: HTTP},
{Long: "save-cookies", Kind: Str, Help: "save Cookies to FILE on exit", Tag: HTTP},
{Long: "conditional-get", Kind: Bool, Default: "false", Help: "download only if the remote file is newer than the local one", Tag: HTTP},
{Long: "remote-time", Kind: Bool, Default: "false", Help: "set the local file's mtime from the server", Tag: HTTP},
{Long: "http-user", Kind: Str, Help: "HTTP basic-auth user", Tag: HTTP},
{Long: "http-passwd", Kind: Str, Help: "HTTP basic-auth password", Tag: HTTP},
{Long: "lowest-speed-limit", Kind: Size, Default: "0", Help: "abort if speed stays below SIZE bytes/sec (0 = off)", Tag: HTTP},
{Long: "checksum", Kind: Str, Help: "verify the finished file: TYPE=DIGEST, e.g. sha-256=<hex> (md5, sha-1, sha-224, sha-256, sha-384, sha-512)", Tag: HTTP},
// --- bittorrent --- // --- bittorrent ---
{Long: "torrent-file", Short: 'T', Kind: Str, Help: "path to a .torrent file", Tag: BitTorrent}, {Long: "check-integrity", Short: 'V', Kind: Bool, Default: "false", Help: "re-verify torrent data against piece hashes", Tag: BitTorrent},
{Long: "seed-time", Kind: Float, Help: "minutes to seed after completion (0 = no seeding)", Tag: BitTorrent}, {Long: "seed-time", Kind: Float, Metavar: "MIN", Help: "minutes to seed after completion (0 = no seeding)", Tag: BitTorrent},
{Long: "seed-ratio", Kind: Float, Default: "1.0", Help: "stop seeding at this share ratio (0 = unlimited)", Tag: BitTorrent}, {Long: "seed-ratio", Kind: Float, Default: "1.0", Help: "stop seeding at this share ratio (0 = unlimited)", Tag: BitTorrent},
{Long: "bt-stop-timeout", Kind: Int, Default: "0", Min: 0, Help: "stop a torrent with no download progress for N seconds (0 = off)", Tag: BitTorrent},
{Long: "bt-metadata-timeout", Kind: Int, Default: "60", Min: 0, Help: "stop a magnet that can't fetch metadata in N seconds (0 = off)", Tag: BitTorrent},
{Long: "listen-port", Kind: Str, Default: "6881-6999", Help: "TCP port (range) for incoming peers", Tag: BitTorrent}, {Long: "listen-port", Kind: Str, Default: "6881-6999", Help: "TCP port (range) for incoming peers", Tag: BitTorrent},
{Long: "enable-dht", Kind: Bool, Default: "true", Help: "enable the BitTorrent DHT", Tag: BitTorrent}, {Long: "enable-dht", Kind: Bool, Default: "true", Help: "enable the BitTorrent DHT (turn off for private torrents)", Tag: BitTorrent},
{Long: "bt-max-peers", Kind: Int, Default: "55", Min: 0, Help: "max peers per torrent (0 = unlimited)", Tag: BitTorrent},
{Long: "max-overall-upload-limit", Kind: Size, Default: "0", Help: "global upload speed limit (0 = unlimited)", Tag: BitTorrent}, {Long: "max-overall-upload-limit", Kind: Size, Default: "0", Help: "global upload speed limit (0 = unlimited)", Tag: BitTorrent},
{Long: "max-upload-limit", Short: 'u', Kind: Size, Default: "0", Help: "per-torrent upload speed limit", Tag: BitTorrent},
{Long: "select-file", Kind: Str, Help: "download only these file indexes, e.g. 1,3-5", Tag: BitTorrent}, {Long: "select-file", Kind: Str, Help: "download only these file indexes, e.g. 1,3-5", Tag: BitTorrent},
{Long: "follow-torrent", Kind: Bool, Default: "true", Help: "after fetching a .torrent over HTTP, download its content", Tag: BitTorrent},
// --- advanced --- // --- advanced ---
{Long: "file-allocation", Short: 'a', Kind: Enum, Default: "prealloc", Enum: []string{"none", "prealloc", "trunc", "falloc"}, Help: "how to allocate disk space", Tag: Advanced}, {Long: "allow-overwrite", Kind: Bool, Default: "false", Help: "overwrite an existing file instead of renaming the new one", Tag: Advanced},
{Long: "auto-file-renaming", Kind: Bool, Default: "true", Help: "rename file (.1, .2, ...) if it already exists", Tag: Advanced},
{Long: "file-allocation", Short: 'a', Kind: Enum, Default: "prealloc", Enum: []string{"none", "prealloc", "trunc", "falloc"}, Help: "how to reserve disk space (prealloc and falloc are identical here)", Tag: Advanced},
{Long: "dry-run", Kind: Bool, Default: "false", Help: "check that the file is available but do not download it", Tag: Advanced}, {Long: "dry-run", Kind: Bool, Default: "false", Help: "check that the file is available but do not download it", Tag: Advanced},
{Long: "stop", Kind: Int, Default: "0", Min: 0, Help: "stop the program after N seconds (0 = off)", Tag: Advanced},
{Long: "disable-ipv6", Kind: Bool, Default: "false", Help: "disable IPv6 (force IPv4-only connections)", Tag: Advanced},
{Long: "save-session", Kind: Str, Help: "on exit, write unfinished downloads to FILE (reload with -i)", Tag: Advanced}, {Long: "save-session", Kind: Str, Help: "on exit, write unfinished downloads to FILE (reload with -i)", Tag: Advanced},
{Long: "auto-save-interval", Kind: Int, Default: "60", Min: 0, Max: 600, Help: "save the session file every N seconds (0 = only on exit)", Tag: Advanced},
{Long: "conf-path", Kind: Str, Help: "path to the config file", Tag: Advanced}, {Long: "conf-path", Kind: Str, Help: "path to the config file", Tag: Advanced},
{Long: "no-conf", Kind: Bool, Default: "false", Help: "do not read a config file", Tag: Advanced}, {Long: "no-conf", Kind: Bool, Default: "false", Help: "do not read a config file", Tag: Advanced},
} }

View File

@@ -2,21 +2,24 @@ package cli
import ( import (
"fmt" "fmt"
"math"
"strconv" "strconv"
"strings" "strings"
) )
// Options is the resolved configuration: a flat name->value map plus a record // Options is the resolved configuration: a flat name->value map plus a record
// of which names were explicitly set (so callers can tell a default from a // of which names were explicitly set (so callers can tell a default from a
// chosen value). Values are kept as strings end-to-end and converted only at // chosen value). The raw strings carry the layered override merge; finalize()
// the point of use by the typed getters below. // then parses each one exactly once into typed, and the numeric/bool getters
// read that — no value is parsed twice.
type Options struct { type Options struct {
vals map[string]string vals map[string]string
set map[string]bool set map[string]bool
typed map[string]any // parsed value per option, filled once by finalize()
} }
func newOptions() *Options { func newOptions() *Options {
return &Options{vals: map[string]string{}, set: map[string]bool{}} return &Options{vals: map[string]string{}, set: map[string]bool{}, typed: map[string]any{}}
} }
// IsSet reports whether name was given on the command line or in the config // IsSet reports whether name was given on the command line or in the config
@@ -26,47 +29,30 @@ func (o *Options) IsSet(name string) bool { return o.set[name] }
// Str returns the raw string value (empty if unset and no default). // Str returns the raw string value (empty if unset and no default).
func (o *Options) Str(name string) string { return o.vals[name] } func (o *Options) Str(name string) string { return o.vals[name] }
// boolWords is the single accepted vocabulary for boolean options, mapping each // boolWords is the accepted vocabulary for boolean options: exactly true and
// recognised spelling to its truth value. The Bool reader, truthy, and // false, nothing else. boolWord is the single consult point — the Bool reader,
// validate() all consult it, so exactly the words that validate are honoured // validate(), and the no-conf bootstrap all go through it so exactly the words
// (no "accepted by the reader but rejected by validate" surprises like --x=on). // that validate are honoured.
var boolWords = map[string]bool{ var boolWords = map[string]bool{
"true": true, "yes": true, "1": true, "on": true, "true": true,
"false": false, "no": false, "0": false, "off": false, "false": false,
} }
// boolWord reports a value's truth and whether it is a recognised boolean word. // boolWord reports a value's truth and whether it is a recognised boolean word.
// Whitespace is trimmed, but case is significant: "True"/"TRUE" are rejected, so
// the match is against exactly "true"/"false".
func boolWord(s string) (val, ok bool) { func boolWord(s string) (val, ok bool) {
val, ok = boolWords[strings.ToLower(strings.TrimSpace(s))] val, ok = boolWords[strings.TrimSpace(s)]
return val, ok return val, ok
} }
// Bool reports whether the value is a truthy boolean word. // Bool, Int, Float and Size read the value finalize() already parsed and stored
func (o *Options) Bool(name string) bool { // in typed; an unset option (no entry) reads as the zero value. The conversion
val, _ := boolWord(o.vals[name]) // happened once, at finalize, so these never re-parse a string.
return val func (o *Options) Bool(name string) bool { v, _ := o.typed[name].(bool); return v }
} func (o *Options) Int(name string) int { v, _ := o.typed[name].(int64); return int(v) }
func (o *Options) Float(name string) float64 { v, _ := o.typed[name].(float64); return v }
// Int returns the value as an int, or 0 if empty/invalid. func (o *Options) Size(name string) int64 { v, _ := o.typed[name].(int64); return v }
func (o *Options) Int(name string) int { return int(o.Int64(name)) }
// Int64 returns the value as an int64, or 0 if empty/invalid.
func (o *Options) Int64(name string) int64 {
n, _ := strconv.ParseInt(strings.TrimSpace(o.vals[name]), 10, 64)
return n
}
// Float returns the value as a float64, or 0 if empty/invalid.
func (o *Options) Float(name string) float64 {
f, _ := strconv.ParseFloat(strings.TrimSpace(o.vals[name]), 64)
return f
}
// Size returns a byte count parsed from a value like "20M" or "512K".
func (o *Options) Size(name string) int64 {
n, _ := parseSize(o.vals[name])
return n
}
// List returns a repeatable option's accumulated values. // List returns a repeatable option's accumulated values.
func (o *Options) List(name string) []string { func (o *Options) List(name string) []string {
@@ -105,5 +91,10 @@ func parseSize(s string) (int64, error) {
if n < 0 { if n < 0 {
return 0, fmt.Errorf("negative size %q", s) return 0, fmt.Errorf("negative size %q", s)
} }
// Reject a value whose unit multiply would overflow int64 and silently wrap to
// a bogus (positive or negative) byte count.
if mult > 1 && n > math.MaxInt64/mult {
return 0, fmt.Errorf("size %q too large", s)
}
return n * mult, nil return n * mult, nil
} }

View File

@@ -27,8 +27,8 @@ type Result struct {
} }
// Parse resolves args into options and URIs, applying the layered override // Parse resolves args into options and URIs, applying the layered override
// order: built-in defaults, then the config file, then proxy // order: built-in defaults, then the config file, then the command line
// environment variables, then the command line (last wins). // (last wins).
func Parse(args []string) (*Result, error) { func Parse(args []string) (*Result, error) {
cmdVals, uris, action, helpTag, err := parseArgs(args) cmdVals, uris, action, helpTag, err := parseArgs(args)
if err != nil { if err != nil {
@@ -41,23 +41,47 @@ func Parse(args []string) (*Result, error) {
opts := newOptions() opts := newOptions()
applyDefaults(opts) applyDefaults(opts)
if !truthy(cmdVals["no-conf"]) { if noConf, _ := boolWord(cmdVals["no-conf"]); !noConf {
explicit := cmdVals["conf-path"] explicit := cmdVals["conf-path"]
path, fromUser := confPath(explicit) path, fromUser := confPath(explicit)
if err := applyConfig(opts, path, fromUser); err != nil { if err := applyConfig(opts, path, fromUser); err != nil {
return nil, err return nil, err
} }
} }
applyEnv(opts)
for name, val := range cmdVals { for name, val := range cmdVals {
if err := set(opts, name, val); err != nil { if err := set(opts, name, val); err != nil {
return nil, err return nil, err
} }
} }
if err := opts.finalize(); err != nil {
return nil, err
}
return &Result{Action: Run, Options: opts, URIs: uris}, nil return &Result{Action: Run, Options: opts, URIs: uris}, nil
} }
// finalize parses every resolved value once into its typed form, validating it
// in the same pass; the numeric/bool getters then just read the result. It runs
// after all layering, so each option is converted exactly once, from its final
// (last-wins) string — not once per layer and again per read.
func (o *Options) finalize() error {
for i := range options {
opt := &options[i]
raw, ok := o.vals[opt.Long]
if !ok {
continue
}
v, err := parseValue(opt, raw)
if err != nil {
return err
}
if v != nil {
o.typed[opt.Long] = v
}
}
return nil
}
// parseArgs walks the argument list once. It recognises --long, --long=val, // parseArgs walks the argument list once. It recognises --long, --long=val,
// -x, -xval, -x val and short bool bundles like -cq. Like getopt_long, it // -x, -xval, -x val and short bool bundles like -cq. Like getopt_long, it
// permutes: a non-flag token becomes a URI but scanning continues so flags may // permutes: a non-flag token becomes a URI but scanning continues so flags may
@@ -138,14 +162,23 @@ func parseArgs(args []string) (vals map[string]string, uris []string, action Act
// accumulate stores a value, joining repeatable List options with newlines. // accumulate stores a value, joining repeatable List options with newlines.
func accumulate(vals map[string]string, o *Opt, val string) { func accumulate(vals map[string]string, o *Opt, val string) {
if o.Kind == List { if o.Kind == List {
if prev, ok := vals[o.Long]; ok { vals[o.Long] = appendList(vals[o.Long], val)
vals[o.Long] = prev + "\n" + val
return return
} }
}
vals[o.Long] = val vals[o.Long] = val
} }
// appendList is the single home of the List newline-join rule, shared by the
// raw-parse layer (accumulate) and the resolved layer (set). An empty prev (the
// option not yet seen) yields val unchanged, so the first value carries no
// leading newline.
func appendList(prev, val string) string {
if prev == "" {
return val
}
return prev + "\n" + val
}
func applyDefaults(o *Options) { func applyDefaults(o *Options) {
for i := range options { for i := range options {
opt := &options[i] opt := &options[i]
@@ -191,86 +224,73 @@ func applyConfig(o *Options, path string, explicit bool) error {
return sc.Err() return sc.Err()
} }
// applyEnv folds the conventional proxy environment variables into all-proxy // set stores a raw value, marking the option as explicitly set. Validation and
// unless it was already set explicitly. // conversion happen later, once, in finalize(); set only records the string.
func applyEnv(o *Options) {
if o.set["all-proxy"] {
return
}
for _, key := range []string{"all_proxy", "https_proxy", "http_proxy"} {
if v := os.Getenv(key); v != "" {
o.vals["all-proxy"] = v
return
}
}
}
// set validates a value against its option descriptor and stores it, marking
// the option as explicitly set.
func set(o *Options, name, val string) error { func set(o *Options, name, val string) error {
opt := byLong[name] opt := byLong[name]
if opt == nil { if opt == nil {
return fmt.Errorf("unknown option %q", name) return fmt.Errorf("unknown option %q", name)
} }
if err := validate(opt, val); err != nil { if opt.Kind == List && o.set[name] {
return err val = appendList(o.vals[name], val)
}
if opt.Kind == List {
if prev, ok := o.vals[name]; ok && o.set[name] {
val = prev + "\n" + val
}
} }
o.vals[name] = val o.vals[name] = val
o.set[name] = true o.set[name] = true
return nil return nil
} }
func validate(o *Opt, val string) error { // parseValue converts a raw string to the typed value for o.Kind, validating it
// in the process: it is the single place a value is parsed. Bool/Int/Size/Float
// return their typed value (read back by the getters); Str/List/Enum need no
// stored conversion and return nil, with Enum still checked for membership.
func parseValue(o *Opt, val string) (any, error) {
switch o.Kind { switch o.Kind {
case Bool: case Bool:
// Booleans accept true/false plus the yes/no/1/0/on/off spellings, but // Booleans accept exactly true/false (the boolWords vocabulary the Bool
// nothing else, so e.g. --enable-dht=flase is a parse error. The accepted // getter honours); --enable-dht=flase is a parse error.
// set is boolWords, the same vocabulary the readers honour. v, ok := boolWord(val)
if _, ok := boolWord(val); !ok { if !ok {
return fmt.Errorf("--%s: %q is not a boolean (true/false)", o.Long, val) return nil, fmt.Errorf("--%s: %q is not a boolean (true/false)", o.Long, val)
} }
return v, nil
case Int: case Int:
n, err := strconv.ParseInt(strings.TrimSpace(val), 10, 64) n, err := strconv.ParseInt(strings.TrimSpace(val), 10, 64)
if err != nil { if err != nil {
return fmt.Errorf("--%s: %q is not an integer", o.Long, val) return nil, fmt.Errorf("--%s: %q is not an integer", o.Long, val)
} }
if err := checkBounds(o, n); err != nil { if err := checkBounds(o, n); err != nil {
return err return nil, err
} }
return n, nil
case Size: case Size:
n, err := parseSize(val) n, err := parseSize(val)
if err != nil { if err != nil {
return fmt.Errorf("--%s: %v", o.Long, err) return nil, fmt.Errorf("--%s: %v", o.Long, err)
} }
if err := checkBounds(o, n); err != nil { if err := checkBounds(o, n); err != nil {
return err return nil, err
} }
return n, nil
case Float: case Float:
// seed-time and seed-ratio are both rates/durations that must be // seed-time and seed-ratio must be non-negative: a negative SeedTime would
// non-negative; a negative SeedTime would fire the seed timer immediately // fire the seed timer immediately and a negative SeedRatio is meaningless.
// and a negative SeedRatio is silently ignored. Reject negatives here.
// Opt.Min/Max are int64, so we only enforce the floor, not full bounds.
f, err := strconv.ParseFloat(strings.TrimSpace(val), 64) f, err := strconv.ParseFloat(strings.TrimSpace(val), 64)
if err != nil { if err != nil {
return fmt.Errorf("--%s: %q is not a number", o.Long, val) return nil, fmt.Errorf("--%s: %q is not a number", o.Long, val)
} }
if f < 0 { if f < 0 {
return fmt.Errorf("--%s: %v below minimum 0", o.Long, f) return nil, fmt.Errorf("--%s: %v below minimum 0", o.Long, f)
} }
return f, nil
case Enum: case Enum:
for _, e := range o.Enum { for _, e := range o.Enum {
if val == e { if val == e {
return nil return nil, nil
} }
} }
return fmt.Errorf("--%s: %q not one of %s", o.Long, val, strings.Join(o.Enum, "|")) return nil, fmt.Errorf("--%s: %q not one of %s", o.Long, val, strings.Join(o.Enum, "|"))
} }
return nil return nil, nil
} }
// checkBounds enforces an Int/Size option's numeric range. Min defaults to a // checkBounds enforces an Int/Size option's numeric range. Min defaults to a
@@ -304,8 +324,3 @@ func confPath(explicit string) (path string, fromUser bool) {
} }
return filepath.Join(home, ".config", "got", "got.conf"), false return filepath.Join(home, ".config", "got", "got.conf"), false
} }
func truthy(s string) bool {
val, _ := boolWord(s)
return val
}

View File

@@ -13,7 +13,7 @@ func TestParseArgs(t *testing.T) {
wantVal map[string]string wantVal map[string]string
uris []string uris []string
}{ }{
{"short attached", []string{"-x16", "https://e.com"}, Run, map[string]string{"max-connection-per-server": "16"}, []string{"https://e.com"}}, {"short attached", []string{"-m3", "https://e.com"}, Run, map[string]string{"max-tries": "3"}, []string{"https://e.com"}},
{"long equals + bool", []string{"--split=8", "-c", "u"}, Run, map[string]string{"split": "8", "continue": "true"}, []string{"u"}}, {"long equals + bool", []string{"--split=8", "-c", "u"}, Run, map[string]string{"split": "8", "continue": "true"}, []string{"u"}},
{"long separate value", []string{"--out", "f.bin", "u"}, Run, map[string]string{"out": "f.bin"}, []string{"u"}}, {"long separate value", []string{"--out", "f.bin", "u"}, Run, map[string]string{"out": "f.bin"}, []string{"u"}},
{"bool bundle", []string{"-cq", "u"}, Run, map[string]string{"continue": "true", "quiet": "true"}, []string{"u"}}, {"bool bundle", []string{"-cq", "u"}, Run, map[string]string{"continue": "true", "quiet": "true"}, []string{"u"}},
@@ -22,7 +22,7 @@ func TestParseArgs(t *testing.T) {
{"permute flag after uri", []string{"u", "--split=8"}, Run, map[string]string{"split": "8"}, []string{"u"}}, {"permute flag after uri", []string{"u", "--split=8"}, Run, map[string]string{"split": "8"}, []string{"u"}},
{"permute interleaved", []string{"a", "-c", "b", "--quiet"}, Run, map[string]string{"continue": "true", "quiet": "true"}, []string{"a", "b"}}, {"permute interleaved", []string{"a", "-c", "b", "--quiet"}, Run, map[string]string{"continue": "true", "quiet": "true"}, []string{"a", "b"}},
{"dash dash terminator", []string{"--split=8", "--", "--not-a-flag", "u"}, Run, map[string]string{"split": "8"}, []string{"--not-a-flag", "u"}}, {"dash dash terminator", []string{"--split=8", "--", "--not-a-flag", "u"}, Run, map[string]string{"split": "8"}, []string{"--not-a-flag", "u"}},
{"bt metadata timeout", []string{"--bt-metadata-timeout=120", "magnet:x"}, Run, map[string]string{"bt-metadata-timeout": "120"}, []string{"magnet:x"}}, {"long int + magnet uri", []string{"--timeout=120", "magnet:x"}, Run, map[string]string{"timeout": "120"}, []string{"magnet:x"}},
{"help", []string{"--help"}, ShowHelp, nil, nil}, {"help", []string{"--help"}, ShowHelp, nil, nil},
{"help short", []string{"-h"}, ShowHelp, nil, nil}, {"help short", []string{"-h"}, ShowHelp, nil, nil},
{"version", []string{"-v"}, ShowVersion, nil, nil}, {"version", []string{"-v"}, ShowVersion, nil, nil},
@@ -68,21 +68,6 @@ func TestParseArgsErrors(t *testing.T) {
} }
} }
// got bounds the magnet metadata fetch by default rather than waiting forever;
// guard the default value and its zero floor.
func TestBTMetadataTimeoutDefault(t *testing.T) {
o, ok := byLong["bt-metadata-timeout"]
if !ok {
t.Fatal("bt-metadata-timeout option is missing")
}
if o.Default != "60" {
t.Errorf("bt-metadata-timeout default = %q, want %q", o.Default, "60")
}
if o.Min != 0 {
t.Errorf("bt-metadata-timeout Min = %d, want 0 (reject negatives, allow 0=off)", o.Min)
}
}
func TestParseSize(t *testing.T) { func TestParseSize(t *testing.T) {
tests := []struct { tests := []struct {
in string in string
@@ -109,73 +94,84 @@ func TestParseSize(t *testing.T) {
} }
} }
func TestValidateRange(t *testing.T) { func TestParseValueRange(t *testing.T) {
x := byLong["max-connection-per-server"] // A synthetic bounded Int exercises the Min/Max range check independent of
if err := validate(x, "16"); err != nil { // which options happen to carry bounds.
t.Errorf("validate 16: %v", err) o := &Opt{Long: "n", Kind: Int, Min: 1, Max: 16}
if _, err := parseValue(o, "16"); err != nil {
t.Errorf("parseValue 16: %v", err)
} }
if err := validate(x, "99"); err == nil { if _, err := parseValue(o, "99"); err == nil {
t.Errorf("validate 99: want out-of-range error") t.Errorf("parseValue 99: want out-of-range error")
}
if _, err := parseValue(o, "0"); err == nil {
t.Errorf("parseValue 0: want below-minimum error")
} }
a := byLong["file-allocation"] a := byLong["file-allocation"]
if err := validate(a, "bogus"); err == nil { if _, err := parseValue(a, "bogus"); err == nil {
t.Errorf("validate enum bogus: want error") t.Errorf("parseValue enum bogus: want error")
} }
} }
func TestValidateIntFloor(t *testing.T) { func TestParseValueIntFloor(t *testing.T) {
// Min:0 now enforces a floor of 0: negatives are rejected. // Min:0 enforces a floor of 0: negatives are rejected.
m := byLong["max-tries"] // Min 0, "0 = unlimited" runtime meaning preserved m := byLong["max-tries"] // Min 0, "0 = unlimited" runtime meaning preserved
if err := validate(m, "0"); err != nil { if _, err := parseValue(m, "0"); err != nil {
t.Errorf("validate max-tries 0: %v, want nil (0 stays a valid input)", err) t.Errorf("parseValue max-tries 0: %v, want nil (0 stays a valid input)", err)
} }
if err := validate(m, "-1"); err == nil { if _, err := parseValue(m, "-1"); err == nil {
t.Errorf("validate max-tries -1: want below-minimum error") t.Errorf("parseValue max-tries -1: want below-minimum error")
} }
} }
func TestValidateFloatNonNegative(t *testing.T) { func TestParseValueFloatNonNegative(t *testing.T) {
// Float options (seed-time, seed-ratio) reject negatives: a negative // Float options (seed-time, seed-ratio) reject negatives: a negative
// SeedTime would fire the seed timer immediately and a negative SeedRatio // SeedTime would fire the seed timer immediately and a negative SeedRatio
// is silently ignored, so neither is a meaningful input. // is silently ignored, so neither is a meaningful input.
for _, name := range []string{"seed-ratio", "seed-time"} { for _, name := range []string{"seed-ratio", "seed-time"} {
o := byLong[name] o := byLong[name]
if err := validate(o, "-1"); err == nil { if _, err := parseValue(o, "-1"); err == nil {
t.Errorf("validate %s -1: want below-minimum error", name) t.Errorf("parseValue %s -1: want below-minimum error", name)
} }
if err := validate(o, "0"); err != nil { if _, err := parseValue(o, "0"); err != nil {
t.Errorf("validate %s 0: %v, want nil", name, err) t.Errorf("parseValue %s 0: %v, want nil", name, err)
} }
if err := validate(o, "1.5"); err != nil { if _, err := parseValue(o, "1.5"); err != nil {
t.Errorf("validate %s 1.5: %v, want nil", name, err) t.Errorf("parseValue %s 1.5: %v, want nil", name, err)
} }
} }
} }
func TestValidateBool(t *testing.T) { func TestParseValueBool(t *testing.T) {
b := byLong["enable-dht"] b := byLong["enable-dht"]
for _, ok := range []string{"true", "false", "yes", "no", "1", "0", "on", "off"} { // Only literal true/false are accepted (case-sensitive), with surrounding
if err := validate(b, ok); err != nil { // whitespace trimmed.
t.Errorf("validate bool %q: %v", ok, err) for _, ok := range []string{"true", "false", " true ", "false\t"} {
if _, err := parseValue(b, ok); err != nil {
t.Errorf("parseValue bool %q: %v", ok, err)
} }
} }
if err := validate(b, "flase"); err == nil { // The invented spellings and any case variant must fail.
t.Errorf("validate bool flase: want error") for _, bad := range []string{"yes", "no", "1", "0", "on", "off", "flase",
"True", "TRUE", "tRuE", "False", "FALSE"} {
if _, err := parseValue(b, bad); err == nil {
t.Errorf("parseValue bool %q: want error (only true/false accepted)", bad)
}
} }
} }
func TestValidateSizeBounds(t *testing.T) { func TestParseValueSizeBounds(t *testing.T) {
k := byLong["min-split-size"] // 1M..1024M (no G unit) k := &Opt{Long: "sz", Kind: Size, Min: 1 << 20, Max: 1 << 30} // 1M..1024M
if err := validate(k, "20M"); err != nil { if _, err := parseValue(k, "20M"); err != nil {
t.Errorf("validate min-split-size 20M: %v", err) t.Errorf("parseValue 20M: %v", err)
} }
if err := validate(k, "512K"); err == nil { if _, err := parseValue(k, "512K"); err == nil {
t.Errorf("validate min-split-size 512K: want below-minimum error") t.Errorf("parseValue 512K: want below-minimum error")
} }
// 1025M exceeds the 1<<30 cap and still parses, so it tests the bounds path // 1025M exceeds the 1<<30 cap and still parses, so it tests the bounds path
// (where "2G" would have failed earlier as an unknown unit). // (where "2G" would have failed earlier as an unknown unit).
if err := validate(k, "1025M"); err == nil { if _, err := parseValue(k, "1025M"); err == nil {
t.Errorf("validate min-split-size 1025M: want above-maximum error") t.Errorf("parseValue 1025M: want above-maximum error")
} }
} }
@@ -213,14 +209,17 @@ func TestExplicitConfMissingFatal(t *testing.T) {
func TestOptionsGetters(t *testing.T) { func TestOptionsGetters(t *testing.T) {
o := newOptions() o := newOptions()
o.vals["split"] = "8" o.vals["split"] = "8"
o.vals["min-split-size"] = "20M" o.vals["max-overall-download-limit"] = "20M"
o.vals["continue"] = "true" o.vals["continue"] = "true"
o.vals["seed-ratio"] = "1.5" o.vals["seed-ratio"] = "1.5"
if err := o.finalize(); err != nil { // parse the raw values once, as Parse does
t.Fatalf("finalize: %v", err)
}
if o.Int("split") != 8 { if o.Int("split") != 8 {
t.Errorf("Int split = %d", o.Int("split")) t.Errorf("Int split = %d", o.Int("split"))
} }
if o.Size("min-split-size") != 20<<20 { if o.Size("max-overall-download-limit") != 20<<20 {
t.Errorf("Size min-split-size = %d", o.Size("min-split-size")) t.Errorf("Size max-overall-download-limit = %d", o.Size("max-overall-download-limit"))
} }
if !o.Bool("continue") { if !o.Bool("continue") {
t.Errorf("Bool continue = false") t.Errorf("Bool continue = false")

View File

@@ -55,9 +55,6 @@ type Stat struct {
Seeders int // connected seeders (BitTorrent only) Seeders int // connected seeders (BitTorrent only)
} }
// Done reports whether the download has reached a terminal state.
func (s Stat) Done() bool { return s.Status == Complete || s.Status == Errored }
// A Download is one logical job: a URL, a torrent, a magnet. Run blocks in its // A Download is one logical job: a URL, a torrent, a magnet. Run blocks in its
// own goroutine until the work finishes, fails, or ctx is cancelled. Stat may // own goroutine until the work finishes, fails, or ctx is cancelled. Stat may
// be called concurrently at any time and must not block; implementations back // be called concurrently at any time and must not block; implementations back
@@ -67,3 +64,14 @@ type Download interface {
Run(ctx context.Context) error Run(ctx context.Context) error
Stat() Stat Stat() Stat
} }
// Seeder is an optional capability of a Download whose work continues after its
// content is fully downloaded — a torrent that keeps running to seed.
// SeedingStarted is closed at that download->seed transition. The engine uses it
// to free the concurrency slot the moment a download stops downloading: a
// seed-only torrent is no longer a download, so it must not count against
// --max-concurrent-downloads (-j) and block a queued download. HTTP downloads do
// not implement this, so they hold their slot until they finish.
type Seeder interface {
SeedingStarted() <-chan struct{}
}

View File

@@ -46,6 +46,12 @@ func (e *Engine) Run(ctx context.Context, downloads []Download) []Result {
go func(i int, d Download) { go func(i int, d Download) {
defer wg.Done() defer wg.Done()
// Track immediately so a download still queued for a concurrency slot
// is visible to the progress reporter as Waiting, rather than hidden
// until it starts running.
e.track(d)
defer e.untrack(d)
// Acquire a concurrency slot, or bail if we're shutting down // Acquire a concurrency slot, or bail if we're shutting down
// before this download ever started. // before this download ever started.
select { select {
@@ -54,10 +60,24 @@ func (e *Engine) Run(ctx context.Context, downloads []Download) []Result {
results <- Result{Name: d.Name(), Index: i, Err: ctx.Err(), Stat: d.Stat()} results <- Result{Name: d.Name(), Index: i, Err: ctx.Err(), Stat: d.Stat()}
return return
} }
defer func() { <-slots }() // Release the slot exactly once — whether the download finishes or, for
// a torrent, first detaches into seed-only. A seeding torrent is no
e.track(d) // longer downloading, so it stops counting against -j and lets a queued
defer e.untrack(d) // download start; this goroutine lives on to seed.
var once sync.Once
release := func() { once.Do(func() { <-slots }) }
defer release()
if s, ok := d.(Seeder); ok {
done := make(chan struct{})
defer close(done)
go func() {
select {
case <-s.SeedingStarted():
release()
case <-done:
}
}()
}
err := d.Run(ctx) err := d.Run(ctx)
results <- Result{Name: d.Name(), Index: i, Err: err, Stat: d.Stat()} results <- Result{Name: d.Name(), Index: i, Err: err, Stat: d.Stat()}
@@ -74,7 +94,8 @@ func (e *Engine) Run(ctx context.Context, downloads []Download) []Result {
return out return out
} }
// Snapshot returns a Stat for every currently running download. // Snapshot returns a Stat for every tracked download — those running plus those
// still queued for a concurrency slot (reported as Waiting).
func (e *Engine) Snapshot() []Stat { func (e *Engine) Snapshot() []Stat {
e.mu.Lock() e.mu.Lock()
defer e.mu.Unlock() defer e.mu.Unlock()
@@ -91,6 +112,10 @@ func (e *Engine) track(d Download) {
e.mu.Unlock() e.mu.Unlock()
} }
// untrack removes d from the active set. Every Download implementation is a
// pointer type, so the == compares identities — never the fields of a value,
// which could panic on a non-comparable one. (Keying on Stat().ID is not an
// option: a torrent's ID changes from source to infohash once metadata loads.)
func (e *Engine) untrack(d Download) { func (e *Engine) untrack(d Download) {
e.mu.Lock() e.mu.Lock()
defer e.mu.Unlock() defer e.mu.Unlock()

107
download/engine_test.go Normal file
View File

@@ -0,0 +1,107 @@
package download
import (
"context"
"testing"
"time"
)
// fakeDownload is a Download that blocks in Run until released, and (optionally)
// announces a download->seed transition so the engine can free its -j slot.
type fakeDownload struct {
name string
started chan struct{} // closed when Run begins
seeding chan struct{} // SeedingStarted signal
release chan struct{} // Run returns once this is closed
detach bool // if set, Run signals seeding right after starting
}
func newFake(name string, detach bool) *fakeDownload {
return &fakeDownload{
name: name,
started: make(chan struct{}),
seeding: make(chan struct{}),
release: make(chan struct{}),
detach: detach,
}
}
func (f *fakeDownload) Name() string { return f.name }
func (f *fakeDownload) Stat() Stat { return Stat{Name: f.name} }
func (f *fakeDownload) SeedingStarted() <-chan struct{} { return f.seeding }
func (f *fakeDownload) Run(ctx context.Context) error {
close(f.started)
if f.detach {
close(f.seeding)
}
select {
case <-f.release:
return nil
case <-ctx.Done():
return ctx.Err()
}
}
func started(f *fakeDownload, d time.Duration) bool {
select {
case <-f.started:
return true
case <-time.After(d):
return false
}
}
// A torrent that has finished downloading and is now seeding must not keep
// occupying a -j slot: with -j 1, a seeding download has to let a queued one run.
func TestEngineSeedingFreesSlot(t *testing.T) {
eng := NewEngine(1)
a := newFake("a", true)
b := newFake("b", true)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
done := make(chan []Result, 1)
go func() { done <- eng.Run(ctx, []Download{a, b}) }()
// Neither is ever released, yet both must start: each frees the single slot
// the instant it detaches into seeding.
if !started(a, time.Second) || !started(b, time.Second) {
t.Fatal("both downloads should start once seeding frees the -j slot")
}
close(a.release)
close(b.release)
<-done
}
// Control: without a seeding transition, -j 1 still serializes — the second
// download waits for the first to finish before it starts.
func TestEngineLimitsConcurrentDownloads(t *testing.T) {
eng := NewEngine(1)
a := newFake("a", false)
b := newFake("b", false)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
done := make(chan []Result, 1)
go func() { done <- eng.Run(ctx, []Download{a, b}) }()
var first, second *fakeDownload
select {
case <-a.started:
first, second = a, b
case <-b.started:
first, second = b, a
case <-time.After(time.Second):
t.Fatal("one download should start")
}
if started(second, 100*time.Millisecond) {
t.Fatal("second download must not start while the only slot is held")
}
close(first.release) // first finishes, freeing the slot
if !started(second, time.Second) {
t.Fatal("second download should start after the first frees the slot")
}
close(second.release)
<-done
}

4
go.mod
View File

@@ -3,7 +3,7 @@ module github.com/hanbok/got
go 1.25 go 1.25
require ( require (
github.com/anacrolix/missinggo/v2 v2.10.0 github.com/anacrolix/log v0.17.1-0.20251118025802-918f1157b7bb
github.com/anacrolix/torrent v1.61.0 github.com/anacrolix/torrent v1.61.0
golang.org/x/sys v0.38.0 golang.org/x/sys v0.38.0
golang.org/x/term v0.37.0 golang.org/x/term v0.37.0
@@ -19,9 +19,9 @@ require (
github.com/anacrolix/envpprof v1.4.0 // indirect github.com/anacrolix/envpprof v1.4.0 // indirect
github.com/anacrolix/generics v0.1.1-0.20251125230353-15d98d46693b // indirect github.com/anacrolix/generics v0.1.1-0.20251125230353-15d98d46693b // indirect
github.com/anacrolix/go-libutp v1.3.2 // indirect github.com/anacrolix/go-libutp v1.3.2 // indirect
github.com/anacrolix/log v0.17.1-0.20251118025802-918f1157b7bb // indirect
github.com/anacrolix/missinggo v1.3.0 // indirect github.com/anacrolix/missinggo v1.3.0 // indirect
github.com/anacrolix/missinggo/perf v1.0.0 // indirect github.com/anacrolix/missinggo/perf v1.0.0 // indirect
github.com/anacrolix/missinggo/v2 v2.10.0 // indirect
github.com/anacrolix/mmsg v1.0.1 // indirect github.com/anacrolix/mmsg v1.0.1 // indirect
github.com/anacrolix/multiless v0.4.0 // indirect github.com/anacrolix/multiless v0.4.0 // indirect
github.com/anacrolix/stm v0.5.0 // indirect github.com/anacrolix/stm v0.5.0 // indirect

View File

@@ -1,22 +0,0 @@
package httpdl
import (
"os"
"path/filepath"
"testing"
)
func TestLoadCAPool(t *testing.T) {
// A missing file is an error, not a silent fall-back to the system roots.
if _, err := loadCAPool(filepath.Join(t.TempDir(), "absent.pem")); err == nil {
t.Errorf("loadCAPool(missing): want error")
}
// A file with no PEM certificates is an error (AppendCertsFromPEM found none).
bad := filepath.Join(t.TempDir(), "bad.pem")
if err := os.WriteFile(bad, []byte("not a certificate\n"), 0o644); err != nil {
t.Fatal(err)
}
if _, err := loadCAPool(bad); err == nil {
t.Errorf("loadCAPool(garbage): want error (no certificates)")
}
}

View File

@@ -1,97 +0,0 @@
package httpdl
import (
"context"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"testing"
"github.com/hanbok/got/download"
)
// TestConditionalGet304 checks that --conditional-get sends If-Modified-Since
// for an existing local file and treats a 304 as success without rewriting the
// file.
func TestConditionalGet304(t *testing.T) {
dir := t.TempDir()
out := filepath.Join(dir, "file.bin")
const existing = "already here"
if err := os.WriteFile(out, []byte(existing), 0o644); err != nil {
t.Fatal(err)
}
gotIfModSince := false
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.Header.Get("If-Modified-Since") != "" {
gotIfModSince = true
}
w.WriteHeader(http.StatusNotModified)
}))
defer srv.Close()
d := New([]string{srv.URL + "/file.bin"}, Config{
Dir: dir,
Out: "file.bin",
MaxConnPerServer: 1,
ConditionalGet: true,
AllowOverwrite: true,
UserAgent: "got-test",
})
if err := d.Run(context.Background()); err != nil {
t.Fatalf("conditional-get 304 should succeed, got %v", err)
}
if !gotIfModSince {
t.Fatal("expected an If-Modified-Since header on the request")
}
if d.Stat().Status != download.Complete {
t.Fatalf("status = %v, want Complete", d.Stat().Status)
}
// The file must be untouched: 304 means "up to date".
b, err := os.ReadFile(out)
if err != nil {
t.Fatal(err)
}
if string(b) != existing {
t.Fatalf("file was rewritten: %q, want %q", b, existing)
}
}
// TestConditionalGetSkippedWithControlFile verifies the rule that
// --conditional-get is ignored when a .got control file exists, so an
// in-progress resume is never short-circuited by a 304.
func TestConditionalGetSkippedWithControlFile(t *testing.T) {
dir := t.TempDir()
out := filepath.Join(dir, "file.bin")
if err := os.WriteFile(out, []byte("partial"), 0o644); err != nil {
t.Fatal(err)
}
// A control file present means a resume is in flight.
if err := os.WriteFile(controlPath(out), []byte("{}"), 0o644); err != nil {
t.Fatal(err)
}
sawIfModSince := false
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.Header.Get("If-Modified-Since") != "" {
sawIfModSince = true
}
// Serve a small body so the run can complete.
http.Error(w, "no range", http.StatusOK)
}))
defer srv.Close()
d := New([]string{srv.URL + "/file.bin"}, Config{
Dir: dir,
Out: "file.bin",
MaxConnPerServer: 1,
ConditionalGet: true,
AllowOverwrite: true,
UserAgent: "got-test",
})
_ = d.Run(context.Background())
if sawIfModSince {
t.Fatal("If-Modified-Since must not be sent when a control file exists")
}
}

View File

@@ -27,7 +27,9 @@ type segState struct {
func controlPath(out string) string { return out + ".got" } func controlPath(out string) string { return out + ".got" }
// snapshot builds a control record from the live segments. // snapshot builds a control record from the live segments. The segment slice is
// fixed once the file is divided and each segment's frontier is owned by one
// worker, so reading written atomically gives a consistent record with no lock.
func snapshot(url string, total int64, etag, lastmod string, segs []seg) control { func snapshot(url string, total int64, etag, lastmod string, segs []seg) control {
c := control{URL: url, Total: total, ETag: etag, LastModified: lastmod, Segs: make([]segState, len(segs))} c := control{URL: url, Total: total, ETag: etag, LastModified: lastmod, Segs: make([]segState, len(segs))}
for i := range segs { for i := range segs {
@@ -73,7 +75,34 @@ func loadControl(out, url string, total int64, etag, lastmod string) *control {
if (lastmod != "" || c.LastModified != "") && c.LastModified != lastmod { if (lastmod != "" || c.LastModified != "") && c.LastModified != lastmod {
return nil return nil
} }
// Reject a control file whose segments do not exactly tile [0,total): a
// truncated/corrupted/hand-edited sidecar that still parses as JSON could
// otherwise mark a segment done() without its bytes on disk (inflated Written)
// or leave an un-downloaded hole, both of which would be reported as a complete
// file. We restart cleanly instead of trusting it.
if !validSegs(c.Segs, c.Total) {
return nil
}
return &c return &c
} }
// validSegs reports whether segs cover [0,total) with no gap or overlap and a
// sane written count for each. got always writes segments in ascending start
// order, so coverage is checked in place; an out-of-order sidecar (only possible
// from a hand-edited or corrupted file) is rejected, which restarts cleanly.
func validSegs(segs []segState, total int64) bool {
if total <= 0 || len(segs) == 0 {
return false
}
var next int64
for _, s := range segs {
length := s.End - s.Start + 1
if s.Start != next || s.End < s.Start || s.Written < 0 || s.Written > length {
return false
}
next = s.End + 1
}
return next == total
}
func removeControl(out string) { os.Remove(controlPath(out)) } func removeControl(out string) { os.Remove(controlPath(out)) }

View File

@@ -1,177 +0,0 @@
package httpdl
import (
"bufio"
"fmt"
"math"
"net/http"
"net/http/cookiejar"
"net/url"
"os"
"strconv"
"strings"
"time"
)
// Cookie handling implements the --load-cookies / --save-cookies options, which
// read and write the Netscape/Mozilla cookies.txt format (one tab-separated
// cookie per line). We back it with the standard net/http/cookiejar so the
// http.Client applies and tracks Set-Cookie headers automatically; the file
// format is just the on-disk representation of that jar.
// loadCookieJar builds a cookie jar seeded from a Netscape cookies.txt file. A
// missing file yields an empty jar (not an error): there are simply no cookies
// to load. A malformed line is skipped, giving per-line tolerance.
func loadCookieJar(path string) (*cookiejar.Jar, error) {
jar, err := cookiejar.New(nil)
if err != nil {
return nil, err
}
if path == "" {
return jar, nil
}
f, err := os.Open(path)
if err != nil {
if os.IsNotExist(err) {
return jar, nil // nothing to load yet
}
return nil, fmt.Errorf("load-cookies %s: %w", path, err)
}
defer f.Close()
// Group cookies by the URL their domain implies, since SetCookies keys on a
// URL. We synthesise a representative URL per (domain, secure, path).
byURL := map[string][]*http.Cookie{}
urls := map[string]*url.URL{}
sc := bufio.NewScanner(f)
for sc.Scan() {
line := sc.Text()
c, u := parseNetscapeCookie(line)
if c == nil {
continue
}
key := u.String()
if _, ok := urls[key]; !ok {
urls[key] = u
}
byURL[key] = append(byURL[key], c)
}
if err := sc.Err(); err != nil {
return nil, fmt.Errorf("load-cookies %s: %w", path, err)
}
for key, cs := range byURL {
jar.SetCookies(urls[key], cs)
}
return jar, nil
}
// parseNetscapeCookie parses one cookies.txt line into a cookie plus the URL it
// belongs to, or (nil, nil) for a comment, blank or malformed line. The format
// is seven tab-separated fields:
//
// domain includeSubdomains path secure expiry name value
//
// Curl marks HttpOnly cookies with a "#HttpOnly_" line prefix rather than a
// real comment, so we recognise that prefix instead of dropping the line.
// Honouring HttpOnly_ rather than skipping every "#" line never loses a cookie.
func parseNetscapeCookie(line string) (*http.Cookie, *url.URL) {
httpOnly := false
if strings.HasPrefix(line, "#HttpOnly_") {
httpOnly = true
line = strings.TrimPrefix(line, "#HttpOnly_")
} else if strings.HasPrefix(line, "#") {
return nil, nil
}
if strings.TrimSpace(line) == "" {
return nil, nil
}
f := strings.Split(line, "\t")
if len(f) < 6 {
return nil, nil
}
domain := f[0]
hostOnly := !strings.EqualFold(f[1], "TRUE")
path := f[2]
secure := strings.EqualFold(f[3], "TRUE")
name := f[5]
value := ""
if len(f) >= 7 {
value = f[6]
}
if name == "" || domain == "" {
return nil, nil
}
c := &http.Cookie{
Name: name,
Value: value,
Path: path,
Secure: secure,
HttpOnly: httpOnly,
}
// A non-host-only cookie (includeSubdomains TRUE) carries a Domain attribute
// so the jar applies it to subdomains; a host-only one leaves Domain empty so
// the jar binds it to the exact request host.
if !hostOnly {
c.Domain = domain
}
// expiry 0 means a session cookie; otherwise it is a Unix timestamp. The jar
// drops already-expired cookies on load.
if exp, err := strconv.ParseInt(strings.TrimSpace(f[4]), 10, 64); err == nil && exp > 0 {
c.Expires = time.Unix(exp, 0)
}
// The jar needs a URL whose host and scheme match the cookie. A leading dot
// denotes a domain cookie; strip it for the host but keep the cookie Domain
// so the jar treats it as domain-wide.
host := strings.TrimPrefix(domain, ".")
scheme := "http"
if secure {
scheme = "https"
}
u := &url.URL{Scheme: scheme, Host: host, Path: path}
return c, u
}
// saveCookieJar writes every cookie the jar holds (for the hosts we have
// contacted) back to path in Netscape format, implementing --save-cookies on
// exit. Because net/http/cookiejar exposes cookies only per URL, we ask it for
// the cookies of each host we recorded a request against.
func saveCookieJar(jar *cookiejar.Jar, urls []*url.URL, path string) error {
if path == "" || jar == nil {
return nil
}
var b strings.Builder
b.WriteString("# Netscape HTTP Cookie File\n")
seen := map[string]bool{}
for _, u := range urls {
for _, c := range jar.Cookies(u) {
// Cookies() drops the domain/path/expiry, so reconstruct sensible
// values: host-only domain, root path, far-future expiry. This is the
// best a host-only readback can do; it round-trips the name/value pair
// that authenticated downloads actually need.
key := u.Host + "\x00" + c.Name
if seen[key] {
continue
}
seen[key] = true
expiry := int64(math.MaxInt32) // session-stable far future
line := strings.Join([]string{
u.Host, "FALSE", "/",
boolToTRUE(u.Scheme == "https"),
strconv.FormatInt(expiry, 10),
c.Name, c.Value,
}, "\t")
b.WriteString(line)
b.WriteByte('\n')
}
}
return os.WriteFile(path, []byte(b.String()), 0o600)
}
func boolToTRUE(b bool) string {
if b {
return "TRUE"
}
return "FALSE"
}

View File

@@ -1,126 +0,0 @@
package httpdl
import (
"net/http"
"os"
"path/filepath"
"strconv"
"testing"
"time"
)
func TestParseNetscapeCookie(t *testing.T) {
future := time.Now().Add(24 * time.Hour).Unix()
tests := []struct {
name string
line string
wantNil bool
wantName string
wantValue string
wantDom string // expected http.Cookie.Domain (empty for host-only)
wantHTTP bool
}{
{
name: "domain cookie",
line: ".example.com\tTRUE\t/\tFALSE\t" + itoa(future) + "\tSID\tabc123",
wantName: "SID",
wantValue: "abc123",
wantDom: ".example.com",
},
{
name: "host-only cookie",
line: "example.com\tFALSE\t/\tTRUE\t" + itoa(future) + "\ttoken\txyz",
wantName: "token",
wantValue: "xyz",
wantDom: "", // host-only: no Domain attribute
},
{
name: "httponly prefix kept",
line: "#HttpOnly_example.com\tFALSE\t/\tFALSE\t" + itoa(future) + "\tHO\tsecret",
wantName: "HO",
wantValue: "secret",
wantHTTP: true,
},
{
name: "empty value field",
line: "example.com\tFALSE\t/\tFALSE\t0\tflag\t",
wantName: "flag",
wantValue: "",
},
{
name: "comment line",
line: "# this is a comment",
wantNil: true,
},
{
name: "blank line",
line: " ",
wantNil: true,
},
{
name: "too few fields",
line: "example.com\tFALSE\t/\tFALSE\t0",
wantNil: true,
},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
c, u := parseNetscapeCookie(tc.line)
if tc.wantNil {
if c != nil {
t.Fatalf("expected nil cookie, got %+v", c)
}
return
}
if c == nil {
t.Fatal("expected a cookie, got nil")
}
if c.Name != tc.wantName || c.Value != tc.wantValue {
t.Fatalf("name/value = %q/%q, want %q/%q", c.Name, c.Value, tc.wantName, tc.wantValue)
}
if c.Domain != tc.wantDom {
t.Fatalf("domain = %q, want %q", c.Domain, tc.wantDom)
}
if c.HttpOnly != tc.wantHTTP {
t.Fatalf("httponly = %v, want %v", c.HttpOnly, tc.wantHTTP)
}
if u == nil {
t.Fatal("expected a URL, got nil")
}
})
}
}
func itoa(n int64) string { return strconv.FormatInt(n, 10) }
func itoa64(n int64) string { return strconv.FormatInt(n, 10) }
func TestLoadCookieJarMissingFile(t *testing.T) {
// A missing cookies file is not an error: there are simply no cookies to load.
jar, err := loadCookieJar(filepath.Join(t.TempDir(), "nope.txt"))
if err != nil {
t.Fatalf("missing file should not error: %v", err)
}
if jar == nil {
t.Fatal("expected an empty jar, got nil")
}
}
func TestLoadCookieJarAppliesCookie(t *testing.T) {
dir := t.TempDir()
path := filepath.Join(dir, "cookies.txt")
future := time.Now().Add(24 * time.Hour).Unix()
data := "# Netscape HTTP Cookie File\n" +
".example.com\tTRUE\t/\tFALSE\t" + itoa64(future) + "\tSID\tabc123\n"
if err := os.WriteFile(path, []byte(data), 0o600); err != nil {
t.Fatal(err)
}
jar, err := loadCookieJar(path)
if err != nil {
t.Fatal(err)
}
req, _ := http.NewRequest(http.MethodGet, "http://www.example.com/file", nil)
got := jar.Cookies(req.URL)
if len(got) != 1 || got[0].Name != "SID" || got[0].Value != "abc123" {
t.Fatalf("jar did not apply the domain cookie to a subdomain: %+v", got)
}
}

View File

@@ -1,6 +1,6 @@
// Package httpdl downloads a single HTTP(S) resource over one or more // Package httpdl downloads a single HTTP(S) resource over one or more
// connections. The model is deliberately flat: split the file into byte-range // connections. The model is deliberately flat: split the file into byte-range
// segments, hand them to a pool of worker goroutines, and have each worker // segments, give each segment its own worker goroutine, and have each worker
// stream its range straight to disk with WriteAt (safe for concurrent, // stream its range straight to disk with WriteAt (safe for concurrent,
// non-overlapping writes — no shared seek, no mutex). Goroutines blocking on // non-overlapping writes — no shared seek, no mutex). Goroutines blocking on
// real I/O keep the model simple: no segment manager, no piece storage, no // real I/O keep the model simple: no segment manager, no piece storage, no
@@ -9,8 +9,6 @@ package httpdl
import ( import (
"context" "context"
"crypto/tls"
"crypto/x509"
"errors" "errors"
"fmt" "fmt"
"hash" "hash"
@@ -18,7 +16,6 @@ import (
"mime" "mime"
"net" "net"
"net/http" "net/http"
"net/http/cookiejar"
"net/url" "net/url"
"os" "os"
"path" "path"
@@ -35,6 +32,15 @@ import (
const readBuf = 32 * 1024 const readBuf = 32 * 1024
// minSplitSize is the smallest piece a file is split into: a segment shorter
// than this is not worth its own connection, so makeSegments stops dividing once
// the pieces would fall below it.
const minSplitSize = 20 << 20
// userAgent is sent on every request; there is no flag to change it. A site that
// needs a different one can be given a User-Agent header via --header.
const userAgent = "got/1.0"
// ErrOutputExists reports that the resolved output file already exists and // ErrOutputExists reports that the resolved output file already exists and
// neither --allow-overwrite nor -c was given. main maps it to a distinct exit // neither --allow-overwrite nor -c was given. main maps it to a distinct exit
// code via errors.Is, so it must stay on the error chain (CONTRACT). // code via errors.Is, so it must stay on the error chain (CONTRACT).
@@ -45,60 +51,21 @@ var ErrOutputExists = errors.New("output file already exists")
// (CONTRACT). // (CONTRACT).
var ErrNotFound = errors.New("not found") var ErrNotFound = errors.New("not found")
// ErrTooSlow reports a transfer aborted by --lowest-speed-limit. main maps it to
// exit code 5 via errors.Is, so it must stay on the error chain (CONTRACT).
var ErrTooSlow = errors.New("download speed too low")
// Config holds everything one HTTP download needs. It is assembled by main from // Config holds everything one HTTP download needs. It is assembled by main from
// the resolved CLI options. // the resolved CLI options.
type Config struct { type Config struct {
Dir string Dir string
Out string Out string
Split int // --split: total connections across all mirrors Split int // --split: total connections across all mirrors
MaxConnPerServer int // --max-connection-per-server: per-host connection cap MinSplit int64 // smallest piece to split into; 0 = minSplitSize default (internal, no flag)
MinSplit int64 Tries int // --max-tries: 0 = unlimited
Tries int // 0 = unlimited
Timeout time.Duration Timeout time.Duration
FileAlloc string // none | prealloc | trunc | falloc FileAlloc string // none | prealloc | trunc | falloc
Continue bool Continue bool
AllowOverwrite bool AllowOverwrite bool
AutoRename bool AutoRename bool
Headers []string Headers []string // --header: extra request headers
UserAgent string
Referer string
Proxy string
CheckCert bool
Limit int64 // per-download bytes/s, 0 = unlimited
OverallLimiter *rate.Limiter // shared across all downloads, may be nil OverallLimiter *rate.Limiter // shared across all downloads, may be nil
// RetryWait is the pause between retries of a transient failure
// (--retry-wait). 0 means retry immediately with no sleep.
RetryWait time.Duration
// AutoSaveInterval controls how often the .got control file is rewritten
// (--auto-save-interval). <=0 falls back to 60s.
AutoSaveInterval time.Duration
// ConnectTimeout is the time allowed to establish the TCP/TLS connection
// (--connect-timeout, default 60s), distinct from Timeout which
// covers stalls once data is flowing. <=0 falls back to Timeout.
ConnectTimeout time.Duration
// LoadCookies / SaveCookies are paths to a Netscape cookies.txt read into
// the client's jar before the download and written back on exit.
LoadCookies string
SaveCookies string
// ConditionalGet sends If-Modified-Since (from the existing local file's
// mtime) so a 304 reports success without re-downloading
// (--conditional-get).
ConditionalGet bool
// RemoteTime sets the finished file's mtime from the server's Last-Modified
// header (-R / --remote-time).
RemoteTime bool
// HTTPUser / HTTPPasswd enable HTTP Basic auth (--http-user/passwd).
HTTPUser string
HTTPPasswd string
// LowestSpeedLimit aborts a download whose speed stays at or below this many
// bytes/sec for a sustained window (--lowest-speed-limit). 0 = off.
LowestSpeedLimit int64
// Checksum is a "<type>=<digest>" spec (e.g. "sha-256=ab…"); when set, // Checksum is a "<type>=<digest>" spec (e.g. "sha-256=ab…"); when set,
// the finished file is hashed and compared, failing with ErrChecksum on a // the finished file is hashed and compared, failing with ErrChecksum on a
// mismatch (--checksum). Empty means no verification. // mismatch (--checksum). Empty means no verification.
@@ -106,11 +73,6 @@ type Config struct {
// DryRun probes the resource — proving it exists and its size — but downloads // DryRun probes the resource — proving it exists and its size — but downloads
// nothing, then reports success (--dry-run). // nothing, then reports success (--dry-run).
DryRun bool DryRun bool
// CACert is a path to a PEM bundle of CA certificates used to verify HTTPS
// servers (--ca-certificate). Empty uses the system roots.
CACert string
// DisableIPv6 forces IPv4-only dialing (--disable-ipv6).
DisableIPv6 bool
} }
// Download implements download.Download for one HTTP(S) file, fetched from one // Download implements download.Download for one HTTP(S) file, fetched from one
@@ -119,26 +81,21 @@ type Config struct {
// connections. // connections.
type Download struct { type Download struct {
uris []string // mirror list; uris[0] is the primary (naming + resume key) uris []string // mirror list; uris[0] is the primary (naming + resume key)
maxConns int // segment workers = min(split, len(uris)*max-connection-per-server) maxConns int // segment-worker cap = min(split, len(uris))
cfg Config cfg Config
client *http.Client client *http.Client
limit []*rate.Limiter limit []*rate.Limiter
jar *cookiejar.Jar // non-nil when cookies are in use, for save-cookies
// name and out are resolved in Run after the probe response is known (the // name and out are resolved in Run after the probe response is known (the
// final post-redirect URL and any Content-Disposition can change them), then // final post-redirect URL and any Content-Disposition can change them). The
// never mutated again, so Stat can read name from another goroutine without a // engine publishes a Download to the progress reporter before Run starts, so
// lock once Run has set it. They are seeded in New so a Stat before Run still // Stat reads these from the reporter goroutine while resolveName writes them;
// has a sensible best-effort name. // they are atomic.Pointer[string] (exactly like bt.Download.name) to keep that
name string // read/write race-free. Seeded in New so a Stat before Run has a best-effort name.
out string name atomic.Pointer[string]
out atomic.Pointer[string]
initErr error initErr error
// mu guards seenURLs, which records every URL we sent a request to so
// SaveCookies can ask the jar for the cookies of each contacted host.
mu sync.Mutex
seenURLs []*url.URL
// live counters, read by Stat from any goroutine // live counters, read by Stat from any goroutine
total int64 // -1 until known total int64 // -1 until known
completed int64 completed int64
@@ -146,21 +103,6 @@ type Download struct {
conns int32 // active connections conns int32 // active connections
} }
// loadCAPool reads a PEM bundle of CA certificates for verifying HTTPS servers
// (--ca-certificate). An unreadable file or one containing no
// certificates is an error rather than a silent fall-back to the system roots.
func loadCAPool(path string) (*x509.CertPool, error) {
pem, err := os.ReadFile(path)
if err != nil {
return nil, err
}
pool := x509.NewCertPool()
if !pool.AppendCertsFromPEM(pem) {
return nil, fmt.Errorf("ca-certificate %s: no certificates found", path)
}
return pool, nil
}
// primary is the canonical URL — used for naming and the resume sidecar so they // primary is the canonical URL — used for naming and the resume sidecar so they
// stay stable regardless of which mirror served the bytes. // stay stable regardless of which mirror served the bytes.
func (d *Download) primary() string { return d.uris[0] } func (d *Download) primary() string { return d.uris[0] }
@@ -173,82 +115,27 @@ func (d *Download) mirror(n int) string { return d.uris[n%len(d.uris)] }
// New builds an HTTP download for uris, which must all serve identical content // New builds an HTTP download for uris, which must all serve identical content
// (mirrors). uris[0] is the primary. // (mirrors). uris[0] is the primary.
func New(uris []string, cfg Config) *Download { func New(uris []string, cfg Config) *Download {
// --connect-timeout bounds only connection establishment; once data flows
// --timeout (the idle guard) takes over. The dial and the TLS handshake each
// get the full connectTimeout (Go applies it per phase), so an HTTPS connect
// can take up to ~2x connectTimeout in the worst case — there is no single
// shared connect budget.
connectTimeout := cfg.ConnectTimeout
if connectTimeout <= 0 {
connectTimeout = cfg.Timeout
}
var initErr error var initErr error
if cfg.MinSplit <= 0 {
// --ca-certificate: verify HTTPS servers against a custom CA bundle instead of cfg.MinSplit = minSplitSize
// the system roots. A bad bundle becomes an initErr so Run fails loudly.
tlsCfg := &tls.Config{InsecureSkipVerify: !cfg.CheckCert}
if cfg.CACert != "" {
if pool, err := loadCAPool(cfg.CACert); err != nil {
initErr = err
} else {
tlsCfg.RootCAs = pool
}
} }
// --disable-ipv6: force IPv4 by dialing "tcp4"; a plain "tcp" would let the // One connect budget: --timeout bounds both establishing the connection and
// resolver hand back a v6 address. // idle stalls once data flows. Proxies still come from the standard
dialer := &net.Dialer{Timeout: connectTimeout} // HTTP_PROXY/HTTPS_PROXY environment, and TLS uses the system roots.
dialContext := dialer.DialContext
if cfg.DisableIPv6 {
dialContext = func(ctx context.Context, network, addr string) (net.Conn, error) {
if network == "tcp" {
network = "tcp4"
}
return dialer.DialContext(ctx, network, addr)
}
}
// MaxConnsPerHost is the per-host cap (--max-connection-per-server): with
// mirrors the total worker count can exceed it, but no single host does.
perHost := cfg.MaxConnPerServer
if perHost < 1 {
perHost = 1
}
tr := &http.Transport{ tr := &http.Transport{
Proxy: http.ProxyFromEnvironment, Proxy: http.ProxyFromEnvironment,
MaxConnsPerHost: perHost, MaxConnsPerHost: cfg.Split,
MaxIdleConnsPerHost: perHost, MaxIdleConnsPerHost: cfg.Split,
ResponseHeaderTimeout: cfg.Timeout, ResponseHeaderTimeout: cfg.Timeout,
TLSHandshakeTimeout: connectTimeout, TLSHandshakeTimeout: cfg.Timeout,
DialContext: dialContext, DialContext: (&net.Dialer{Timeout: cfg.Timeout}).DialContext,
TLSClientConfig: tlsCfg,
}
if cfg.Proxy != "" {
if pu, err := url.Parse(cfg.Proxy); err == nil {
tr.Proxy = http.ProxyURL(pu)
}
}
// Seed the cookie jar from --load-cookies; a load error becomes an initErr so
// Run surfaces it rather than silently downloading uncredentialed.
var jar *cookiejar.Jar
if cfg.LoadCookies != "" || cfg.SaveCookies != "" {
if j, err := loadCookieJar(cfg.LoadCookies); err != nil {
if initErr == nil {
initErr = err
}
} else {
jar = j
}
} }
var limit []*rate.Limiter var limit []*rate.Limiter
if cfg.OverallLimiter != nil { if cfg.OverallLimiter != nil {
limit = append(limit, cfg.OverallLimiter) limit = append(limit, cfg.OverallLimiter)
} }
if cfg.Limit > 0 {
limit = append(limit, rate.NewLimiter(rate.Limit(cfg.Limit), download.LimiterBurst(cfg.Limit)))
}
// Seed a best-effort name from -o or the primary URL. The real name is // Seed a best-effort name from -o or the primary URL. The real name is
// finalised in Run after probe(), where Content-Disposition and the // finalised in Run after probe(), where Content-Disposition and the
@@ -268,47 +155,54 @@ func New(uris []string, cfg Config) *Download {
if len(uris) == 0 && initErr == nil { if len(uris) == 0 && initErr == nil {
initErr = errors.New("no URL to download") initErr = errors.New("no URL to download")
} }
// Worker count: --split connections are spread across the mirrors, capped // Worker count is --split: the file is cut into that many byte-range segments,
// at --max-connection-per-server per host, so the ceiling is // each its own connection, whether served by one host or spread across mirrors.
// min(split, M*max-conn-per-server). One mirror reduces to min(split, x).
maxConns := cfg.Split maxConns := cfg.Split
if m := max(1, len(uris)) * perHost; maxConns > m {
maxConns = m
}
if maxConns < 1 { if maxConns < 1 {
maxConns = 1 maxConns = 1
} }
// Only attach the jar when one was built: a typed-nil *cookiejar.Jar stored d := &Download{
// in the http.Client.Jar interface is non-nil and would panic on use.
client := &http.Client{Transport: tr}
if jar != nil {
client.Jar = jar
}
return &Download{
uris: uris, uris: uris,
maxConns: maxConns, maxConns: maxConns,
cfg: cfg, cfg: cfg,
client: client, client: &http.Client{Transport: tr},
limit: limit, limit: limit,
jar: jar,
name: name,
out: out,
total: -1, total: -1,
initErr: initErr, initErr: initErr,
} }
d.setName(name)
d.setOut(out)
return d
} }
func (d *Download) Name() string { return d.name } func (d *Download) Name() string { return d.loadName() }
// Path returns the resolved output file path (used by --follow-torrent). // Path returns the resolved output file path, so a fetched .torrent can be
func (d *Download) Path() string { return d.out } // followed into a BitTorrent download of its content.
func (d *Download) Path() string { return d.loadOut() }
// name and out are read by Stat from the reporter goroutine while Run resolves
// them, so they are stored atomically (mirroring bt.Download.name).
func (d *Download) setName(s string) { d.name.Store(&s) }
func (d *Download) setOut(s string) { d.out.Store(&s) }
func (d *Download) loadName() string {
if p := d.name.Load(); p != nil {
return *p
}
return ""
}
func (d *Download) loadOut() string {
if p := d.out.Load(); p != nil {
return *p
}
return ""
}
func (d *Download) Stat() download.Stat { func (d *Download) Stat() download.Stat {
return download.Stat{ return download.Stat{
Name: d.name, Name: d.loadName(),
ID: d.out, // resolved output path; stable per process (CONTRACT) ID: d.loadOut(), // resolved output path; stable per process (CONTRACT)
IsBT: false, IsBT: false,
Status: download.Status(atomic.LoadInt32(&d.status)), Status: download.Status(atomic.LoadInt32(&d.status)),
Total: atomic.LoadInt64(&d.total), Total: atomic.LoadInt64(&d.total),
@@ -332,7 +226,6 @@ type probeResult struct {
lastmod string lastmod string
cdisp string // raw Content-Disposition header cdisp string // raw Content-Disposition header
finalURL *url.URL finalURL *url.URL
notModified bool // server answered 304 to a conditional GET
} }
// Run probes the resource, finalises the output name, decides between a single // Run probes the resource, finalises the output name, decides between a single
@@ -355,12 +248,6 @@ func (d *Download) Run(ctx context.Context) (err error) {
if err != nil { if err != nil {
return err return err
} }
// --conditional-get: the server said our local copy is current, so there is
// nothing to download. Report success without touching the file.
if pr.notModified {
d.setStatus(download.Complete)
return nil
}
atomic.StoreInt64(&d.total, pr.total) atomic.StoreInt64(&d.total, pr.total)
// Finalise name/out now that the post-redirect URL and Content-Disposition // Finalise name/out now that the post-redirect URL and Content-Disposition
@@ -388,36 +275,22 @@ func (d *Download) Run(ctx context.Context) (err error) {
return nil return nil
} }
// --lowest-speed-limit: cancel the transfer if its speed stays at or below out := d.loadOut()
// the limit past a startup grace. The watcher cancels xferCtx and sets a
// flag so we can return the distinct "too slow" error rather than a bare
// cancellation.
xferCtx, stopGuard, tooSlow := d.speedGuard(ctx)
if !pr.ranged || pr.total <= 0 { if !pr.ranged || pr.total <= 0 {
err = d.single(xferCtx, d.out) err = d.single(ctx, out)
} else { } else {
err = d.segmented(xferCtx, d.out, pr.total, pr.etag, pr.lastmod) err = d.segmented(ctx, out, pr.total, pr.etag, pr.lastmod)
} }
stopGuard()
if err != nil { if err != nil {
if tooSlow() {
return fmt.Errorf("%s: %w (<= %d bytes/sec)", d.primary(), ErrTooSlow, d.cfg.LowestSpeedLimit)
}
return err return err
} }
// --checksum: verify the finished file before declaring success, so a // --checksum: verify the finished file before declaring success, so a
// corrupted or tampered download fails (--checksum, exit code 32). // corrupted or tampered download fails (--checksum, exit code 32).
if newSum != nil { if newSum != nil {
if err = verifyFile(d.out, newSum, sumWant); err != nil { if err = verifyFile(out, newSum, sumWant); err != nil {
return err return err
} }
} }
// -R / --remote-time: stamp the finished file with the server's mtime.
if d.cfg.RemoteTime && pr.lastmod != "" {
if t, err := http.ParseTime(pr.lastmod); err == nil {
os.Chtimes(d.out, t, t)
}
}
d.setStatus(download.Complete) d.setStatus(download.Complete)
return nil return nil
} }
@@ -427,6 +300,7 @@ func (d *Download) Run(ctx context.Context) (err error) {
// the original URL. Overwrite / auto-rename is then resolved against the final // the original URL. Overwrite / auto-rename is then resolved against the final
// path, honouring -c so a resumable file is never renamed (items [2],[8],[13]). // path, honouring -c so a resumable file is never renamed (items [2],[8],[13]).
func (d *Download) resolveName(pr probeResult) error { func (d *Download) resolveName(pr probeResult) error {
out := d.loadOut()
if d.cfg.Out == "" { if d.cfg.Out == "" {
name := "" name := ""
if cd := pr.cdisp; cd != "" { if cd := pr.cdisp; cd != "" {
@@ -438,8 +312,9 @@ func (d *Download) resolveName(pr probeResult) error {
if name == "" { if name == "" {
name = nameFromURL(d.primary()) name = nameFromURL(d.primary())
} }
d.name = name out = filepath.Join(d.cfg.Dir, name)
d.out = filepath.Join(d.cfg.Dir, name) d.setName(name)
d.setOut(out)
} }
// --dry-run writes nothing, so skip overwrite/auto-rename resolution; the name // --dry-run writes nothing, so skip overwrite/auto-rename resolution; the name
@@ -450,38 +325,24 @@ func (d *Download) resolveName(pr probeResult) error {
// -c resumes either our own .got sidecar or a foreign/browser partial file, // -c resumes either our own .got sidecar or a foreign/browser partial file,
// so an existing file is never auto-renamed under -c (item [2]). // so an existing file is never auto-renamed under -c (item [2]).
if d.cfg.Continue && fileExists(d.out) { if d.cfg.Continue && fileExists(out) {
return nil return nil
} }
if fileExists(d.out) && !d.cfg.AllowOverwrite { if fileExists(out) && !d.cfg.AllowOverwrite {
if d.cfg.AutoRename { if d.cfg.AutoRename {
u, err := uniqueName(d.out) u, err := uniqueName(out)
if err != nil { if err != nil {
return err return err
} }
d.out = u d.setOut(u)
d.name = filepath.Base(u) d.setName(filepath.Base(u))
return nil return nil
} }
return fmt.Errorf("%s: %w (use --allow-overwrite or -c)", d.out, ErrOutputExists) return fmt.Errorf("%s: %w (use --allow-overwrite or -c)", out, ErrOutputExists)
} }
return nil return nil
} }
// backoff waits RetryWait between attempts, but never after the final attempt,
// and returns ctx.Err() if cancelled while waiting (item [20]).
func backoff(ctx context.Context, wait time.Duration, tries, attempt int) error {
if wait <= 0 || (tries != 0 && attempt >= tries-1) {
return nil
}
select {
case <-ctx.Done():
return ctx.Err()
case <-time.After(wait):
return nil
}
}
// retriesExhausted wraps the final cause once the retry budget is spent, so the // retriesExhausted wraps the final cause once the retry budget is spent, so the
// caller — and the exit-code mapping in main — can still see whether it was DNS, // caller — and the exit-code mapping in main — can still see whether it was DNS,
// a refused connection, or a timeout, rather than an opaque "too many retries". // a refused connection, or a timeout, rather than an opaque "too many retries".
@@ -499,12 +360,11 @@ func retriesExhausted(what string, tries int, cause error) error {
} }
// withRetries runs attempt up to d.cfg.Tries times (0 = unlimited), the shared // withRetries runs attempt up to d.cfg.Tries times (0 = unlimited), the shared
// tries/backoff/fatal policy that probe, the single stream, and each segment all // tries/fatal policy that probe, the single stream, and each segment all use so
// use so one transient DNS/connect/5xx does not fail the whole download (item // one transient DNS/connect/5xx does not fail the whole download (item [18]). It
// [18]). It returns immediately on success (nil), on context cancellation, or on // returns immediately on success (nil), on context cancellation, or on a fatal{}
// a fatal{} error (retrying cannot help), sleeps RetryWait between attempts, and // error (retrying cannot help), and wraps the final cause with retriesExhausted
// wraps the final cause with retriesExhausted(what, ...) once the budget is // once the budget is spent so the underlying cause stays inspectable.
// spent so the underlying cause stays inspectable.
func (d *Download) withRetries(ctx context.Context, what string, attempt func() error) error { func (d *Download) withRetries(ctx context.Context, what string, attempt func() error) error {
tries := d.cfg.Tries tries := d.cfg.Tries
var lastErr error var lastErr error
@@ -521,36 +381,47 @@ func (d *Download) withRetries(ctx context.Context, what string, attempt func()
if errors.As(err, &fe) { if errors.As(err, &fe) {
return err // permanent: retrying cannot help return err // permanent: retrying cannot help
} }
if err := backoff(ctx, d.cfg.RetryWait, tries, n); err != nil {
return err
}
} }
return retriesExhausted(what, tries, lastErr) return retriesExhausted(what, tries, lastErr)
} }
// overMirrors calls fn against each mirror URL in turn, starting at offset start
// so callers can spread work across mirrors (segment index) and rotate on retry.
// It returns nil on the first success, ctx.Err() if cancelled between mirrors, or
// the last error once every mirror is spent.
func (d *Download) overMirrors(ctx context.Context, start int, fn func(uri string) error) error {
var lastErr error
for i := range d.uris {
if ctx.Err() != nil {
return ctx.Err()
}
if err := fn(d.mirror(start + i)); err != nil {
lastErr = err
continue
}
return nil
}
return lastErr
}
// probeRetry runs probe within the shared retry policy (item [18]). The probe // probeRetry runs probe within the shared retry policy (item [18]). The probe
// result is captured through pr because withRetries only carries an error. // result is captured through pr because withRetries only carries an error.
func (d *Download) probeRetry(ctx context.Context) (probeResult, error) { func (d *Download) probeRetry(ctx context.Context) (probeResult, error) {
// Probe mirrors in order (primary first); the first that answers anchors the // Probe mirrors in order (primary first); the first that answers anchors the
// size and validators for the whole download. A dead/404 primary falls over // size and validators for the whole download. A dead/404 primary falls over
// to the next mirror. // to the next mirror.
var lastErr error
for _, u := range d.uris {
if ctx.Err() != nil {
return probeResult{}, ctx.Err()
}
var pr probeResult var pr probeResult
err := d.withRetries(ctx, u, func() error { err := d.overMirrors(ctx, 0, func(uri string) error {
return d.withRetries(ctx, uri, func() error {
var perr error var perr error
pr, perr = d.probe(ctx, u) pr, perr = d.probe(ctx, uri)
return perr return perr
}) })
if err == nil { })
if err != nil {
return probeResult{}, err
}
return pr, nil return pr, nil
}
lastErr = err
}
return probeResult{}, lastErr
} }
// probe issues a one-byte ranged GET to learn the total size, whether the // probe issues a one-byte ranged GET to learn the total size, whether the
@@ -561,15 +432,6 @@ func (d *Download) probe(ctx context.Context, uri string) (probeResult, error) {
if err != nil { if err != nil {
return probeResult{}, err return probeResult{}, err
} }
// --conditional-get: ask the server to skip the transfer when our local file
// is already current. Only done when there is no .got control file (an
// in-progress resume must not be short-circuited), using the local file's
// mtime for If-Modified-Since.
if d.cfg.ConditionalGet && !fileExists(controlPath(d.out)) {
if fi, err := os.Stat(d.out); err == nil {
req.Header.Set("If-Modified-Since", fi.ModTime().UTC().Format(http.TimeFormat))
}
}
resp, err := d.client.Do(req) resp, err := d.client.Do(req)
if err != nil { if err != nil {
return probeResult{}, err return probeResult{}, err
@@ -586,11 +448,6 @@ func (d *Download) probe(ctx context.Context, uri string) (probeResult, error) {
pr.finalURL = resp.Request.URL pr.finalURL = resp.Request.URL
} }
if resp.StatusCode == http.StatusNotModified {
pr.notModified = true
return pr, nil
}
switch resp.StatusCode { switch resp.StatusCode {
case http.StatusPartialContent: case http.StatusPartialContent:
// Content-Range: bytes 0-0/12345 // Content-Range: bytes 0-0/12345
@@ -652,37 +509,23 @@ func (d *Download) segmented(ctx context.Context, out string, total int64, etag,
saveDone := make(chan struct{}) saveDone := make(chan struct{})
go d.saveLoop(ctx, out, total, etag, lastmod, segs, saveDone) go d.saveLoop(ctx, out, total, etag, lastmod, segs, saveDone)
jobs := make(chan *seg) // One worker per segment. Segments are non-overlapping byte ranges written
go func() { // with WriteAt, so the workers share no cursor and need no coordination — a
defer close(jobs) // finished worker simply exits. The division (makeSegments / a resumed
for i := range segs { // control) fixes the parallelism up front; there is no rebalancing.
if segs[i].done() {
continue
}
select {
case jobs <- &segs[i]:
case <-ctx.Done(): // stop feeding once we're tearing down
return
}
}
}()
var ( var (
wg sync.WaitGroup wg sync.WaitGroup
errOnce sync.Once errOnce sync.Once
runErr error runErr error
) )
for w := 0; w < d.maxConns; w++ { for i := range segs {
wg.Add(1) wg.Add(1)
go func() { go func(s *seg) {
defer wg.Done() defer wg.Done()
for s := range jobs {
if err := d.fetchSeg(ctx, f, s, total); err != nil { if err := d.fetchSeg(ctx, f, s, total); err != nil {
errOnce.Do(func() { runErr = err; cancel() }) errOnce.Do(func() { runErr = err; cancel() })
return
} }
} }(&segs[i])
}()
} }
wg.Wait() wg.Wait()
cancel() cancel()
@@ -706,24 +549,14 @@ func (d *Download) fetchSeg(ctx context.Context, f *os.File, s *seg, total int64
// The segment fails only once every mirror is spent. Starting at s.index // The segment fails only once every mirror is spent. Starting at s.index
// spreads the segments across mirrors round-robin, and each attempt resumes // spreads the segments across mirrors round-robin, and each attempt resumes
// from s.offset(). // from s.offset().
var lastErr error return d.overMirrors(ctx, s.index, func(uri string) error {
for i := range d.uris { return d.withRetries(ctx, fmt.Sprintf("segment %d", s.index), func() error {
if ctx.Err() != nil {
return ctx.Err()
}
uri := d.mirror(s.index + i)
err := d.withRetries(ctx, fmt.Sprintf("segment %d", s.index), func() error {
if s.done() { if s.done() {
return nil return nil
} }
return d.fetchOnce(ctx, f, s, uri, total) return d.fetchOnce(ctx, f, s, uri, total)
}) })
if err == nil { })
return nil
}
lastErr = err
}
return lastErr
} }
func (d *Download) fetchOnce(ctx context.Context, f *os.File, s *seg, uri string, total int64) error { func (d *Download) fetchOnce(ctx context.Context, f *os.File, s *seg, uri string, total int64) error {
@@ -744,129 +577,95 @@ func (d *Download) fetchOnce(ctx context.Context, f *os.File, s *seg, uri string
atomic.AddInt32(&d.conns, -1) atomic.AddInt32(&d.conns, -1)
}() }()
if resp.StatusCode != http.StatusPartialContent { if resp.StatusCode != http.StatusPartialContent {
return statusError(fmt.Sprintf("segment %d: expected 206, got %s", s.index, resp.Status), resp.StatusCode) err := statusError(fmt.Sprintf("segment %d: expected 206, got %s", s.index, resp.Status), resp.StatusCode)
// A 200 means this mirror ignores Range entirely; retrying it only burns the
// budget, so make it fatal and let fetchSeg fail over to the next mirror.
if resp.StatusCode == http.StatusOK {
err = fatal{err}
}
return err
} }
// A mirror must serve the same file as the primary, or its bytes written at // A mirror must serve the same file as the primary, or its bytes written at
// this offset would corrupt the output. Reject a 206 whose total length // this offset would corrupt the output. Require a Content-Range that confirms
// disagrees with the probe (validate the total length); fatal so withRetries // both the offset we asked for and the total length the probe saw; a missing,
// stops and fetchSeg falls over to the next mirror instead of writing it. // unparseable, or divergent range is fatal so withRetries stops and fetchSeg
if _, _, t, ok := parseContentRange(resp.Header.Get("Content-Range")); ok && t != total { // falls over to the next mirror instead of writing unverified bytes.
return fatal{fmt.Errorf("segment %d: %s reports length %d, want %d", s.index, uri, t, total)} if start, _, t, ok := parseContentRange(resp.Header.Get("Content-Range")); !ok || t != total || start != s.offset() {
return fatal{fmt.Errorf("segment %d: %s returned an unverifiable range %q (want start %d, length %d)",
s.index, uri, resp.Header.Get("Content-Range"), s.offset(), total)}
} }
body, stop := d.idleGuard(resp.Body, cancel) body, stop := d.idleGuard(resp.Body, cancel)
defer stop() defer stop()
return d.pump(ctx, f, s, body) return d.pump(ctx, f, s, body)
} }
// errIdleTimeout marks a read that stalled past the idle window, so callers can // ErrTimeout marks a transfer that stalled past the idle window (--timeout): a
// report a distinct "timed out" instead of the bare "context canceled" that the // distinct "timed out" instead of the bare "context canceled" the cancelled
// cancelled request would otherwise surface (item [39]). // request would otherwise surface (item [39]). main maps it to exit code 2 via
var errIdleTimeout = errors.New("idle timeout: no data received") // errors.Is, so it must stay on the error chain (CONTRACT).
var ErrTimeout = errors.New("idle timeout: no data received")
// idleReader resets a watchdog timer on every read; if a read stalls longer // idleReader records the time of the last byte received; a background watcher
// than the idle window the timer fires and cancels the request, so the blocked // (idleGuard) cancels the request once a full idle window passes with no
// Read returns an error instead of hanging forever on a wedged connection. // progress. Tracking progress by timestamp — rather than arming a per-read timer
// — means a slow-but-advancing transfer is never aborted, and there is no
// Reset/Stop race that could poison a connection that was still delivering data.
type idleReader struct { type idleReader struct {
r io.Reader r io.Reader
timer *time.Timer
idle time.Duration idle time.Duration
fired atomic.Bool // set when the watchdog cancelled the request last atomic.Int64 // UnixNano of the last byte received
timedOut atomic.Bool // set when the watcher cancelled the request
} }
func (ir *idleReader) Read(p []byte) (int, error) { func (ir *idleReader) Read(p []byte) (int, error) {
ir.timer.Reset(ir.idle)
n, err := ir.r.Read(p) n, err := ir.r.Read(p)
ir.timer.Stop() // only the Read itself counts as idle, not write/throttle gaps if n > 0 {
if err != nil && ir.fired.Load() { ir.last.Store(time.Now().UnixNano())
// The cancellation came from our watchdog, not the caller's ctx, so }
if err != nil && ir.timedOut.Load() {
// The cancellation came from our watcher, not the caller's ctx, so
// translate the read error into the distinct idle-timeout sentinel. // translate the read error into the distinct idle-timeout sentinel.
return n, fmt.Errorf("%w (after %s)", errIdleTimeout, ir.idle) return n, fmt.Errorf("%w (after %s)", ErrTimeout, ir.idle)
} }
return n, err return n, err
} }
// idleGuard wraps body with an idle timeout of d.cfg.Timeout, cancelling the // watch cancels the request once idle passes with no byte received. It polls a
// request if no data arrives in that window. It returns the reader to use and a // few times per window so a stall is caught within ~idle of the last byte,
// stop func to call when the transfer is done. // without a shared timer that Read would have to Reset/Stop.
func (ir *idleReader) watch(stop <-chan struct{}, cancel context.CancelFunc) {
tick := ir.idle / 4
if tick <= 0 {
tick = ir.idle
}
t := time.NewTicker(tick)
defer t.Stop()
for {
select {
case <-stop:
return
case now := <-t.C:
if now.UnixNano()-ir.last.Load() >= int64(ir.idle) {
ir.timedOut.Store(true)
cancel()
return
}
}
}
}
// idleGuard wraps body with an idle timeout of d.cfg.Timeout: a watcher goroutine
// cancels the request when no byte arrives for that long. It returns the reader to
// use and a stop func to call (exactly once) when the transfer is done.
func (d *Download) idleGuard(body io.Reader, cancel context.CancelFunc) (io.Reader, func()) { func (d *Download) idleGuard(body io.Reader, cancel context.CancelFunc) (io.Reader, func()) {
if d.cfg.Timeout <= 0 { if d.cfg.Timeout <= 0 {
return body, func() {} return body, func() {}
} }
ir := &idleReader{r: body, idle: d.cfg.Timeout} ir := &idleReader{r: body, idle: d.cfg.Timeout}
ir.timer = time.AfterFunc(d.cfg.Timeout, func() { ir.last.Store(time.Now().UnixNano())
ir.fired.Store(true) stop := make(chan struct{})
cancel() go ir.watch(stop, cancel)
}) return ir, func() { close(stop) }
return ir, func() { ir.timer.Stop() }
}
// speedStartupGrace is how long a download may stay slow before the
// lowest-speed-limit watchdog starts checking it: a startup idle time of
// 10 seconds.
const speedStartupGrace = 10 * time.Second
// belowSpeed reports whether the bytes transferred between two cumulative
// d.completed samples (prev -> now over one window) put the download at or
// below limit. A negative delta is not a slow second: d.completed is not
// strictly monotonic — singleOnce rewrites it with an absolute StoreInt64 and a
// resumed Range answered by a 200 resets the offset to 0, so the counter can
// jump backward on a retry. Treating that backward jump as "below limit" would
// false-abort a healthy download, so a reset (delta < 0) reports false; the
// caller has already advanced its baseline, so the next full window measures the
// real speed. Only a genuine non-negative delta at or under the limit aborts.
func belowSpeed(prev, now, limit int64) bool {
delta := now - prev
if delta < 0 {
return false // counter reset, not a slow window
}
return delta <= limit
}
// speedGuard implements --lowest-speed-limit. It returns a child context that
// the watcher cancels (setting a flag) once the measured download speed has
// stayed at or below the limit past the startup grace, plus a stop func and a
// predicate reporting whether the watcher fired. When the limit is 0 it is a
// no-op passthrough.
//
// Speed is sampled from d.completed over a one-second window: a single slow
// second after the grace aborts the download as too slow.
func (d *Download) speedGuard(ctx context.Context) (context.Context, func(), func() bool) {
if d.cfg.LowestSpeedLimit <= 0 {
return ctx, func() {}, func() bool { return false }
}
child, cancel := context.WithCancel(ctx)
var fired atomic.Bool
done := make(chan struct{})
go func() {
defer close(done)
start := time.Now()
ticker := time.NewTicker(time.Second)
defer ticker.Stop()
last := atomic.LoadInt64(&d.completed)
for {
select {
case <-child.Done():
return
case <-ticker.C:
now := atomic.LoadInt64(&d.completed)
prev := last
last = now
if time.Since(start) < speedStartupGrace {
continue
}
if belowSpeed(prev, now, d.cfg.LowestSpeedLimit) {
fired.Store(true)
cancel()
return
}
}
}
}()
stop := func() {
cancel()
<-done
}
return child, stop, func() bool { return fired.Load() }
} }
// fatal marks an error that retrying cannot fix. Unwrap exposes the wrapped // fatal marks an error that retrying cannot fix. Unwrap exposes the wrapped
@@ -901,7 +700,9 @@ func statusError(ctxMsg string, code int) error {
} }
// pump copies the response body into the file at the segment's running offset, // pump copies the response body into the file at the segment's running offset,
// stopping at the segment end and respecting rate limits. // stopping at the segment end and respecting rate limits. Only this worker writes
// s.written, so the advance is a plain atomic add (Stat and the snapshot read it
// atomically); the loop ends when the range is filled.
func (d *Download) pump(ctx context.Context, f *os.File, s *seg, body io.Reader) error { func (d *Download) pump(ctx context.Context, f *os.File, s *seg, body io.Reader) error {
buf := make([]byte, readBuf) buf := make([]byte, readBuf)
for s.remaining() > 0 { for s.remaining() > 0 {
@@ -914,7 +715,7 @@ func (d *Download) pump(ctx context.Context, f *os.File, s *seg, body io.Reader)
if _, werr := f.WriteAt(buf[:rd], s.offset()); werr != nil { if _, werr := f.WriteAt(buf[:rd], s.offset()); werr != nil {
return werr return werr
} }
s.advance(int64(rd)) s.addWritten(int64(rd))
atomic.AddInt64(&d.completed, int64(rd)) atomic.AddInt64(&d.completed, int64(rd))
d.throttle(ctx, rd) d.throttle(ctx, rd)
} }
@@ -953,23 +754,13 @@ func (d *Download) single(ctx context.Context, out string) error {
first := true first := true
// One connection at a time, but try each mirror: a mirror that fails its // One connection at a time, but try each mirror: a mirror that fails its
// retry budget falls over to the next (resuming from what's already on disk). // retry budget falls over to the next (resuming from what's already on disk).
var lastErr error return d.overMirrors(ctx, 0, func(uri string) error {
for i := range d.uris { return d.withRetries(ctx, uri, func() error {
if ctx.Err() != nil {
return ctx.Err()
}
uri := d.mirror(i)
err := d.withRetries(ctx, uri, func() error {
resumeFromDisk := !first || foreignResume resumeFromDisk := !first || foreignResume
first = false first = false
return d.singleOnce(ctx, out, uri, resumeFromDisk) return d.singleOnce(ctx, out, uri, resumeFromDisk)
}) })
if err == nil { })
return nil
}
lastErr = err
}
return lastErr
} }
// singleOnce performs one single-stream transfer attempt. When resumeFromDisk is // singleOnce performs one single-stream transfer attempt. When resumeFromDisk is
@@ -1005,6 +796,19 @@ func (d *Download) singleOnce(ctx context.Context, out, uri string, resumeFromDi
if resp.StatusCode != http.StatusOK && resp.StatusCode != http.StatusPartialContent { if resp.StatusCode != http.StatusOK && resp.StatusCode != http.StatusPartialContent {
return statusError(fmt.Sprintf("%s: %s", uri, resp.Status), resp.StatusCode) return statusError(fmt.Sprintf("%s: %s", uri, resp.Status), resp.StatusCode)
} }
// On failover, resumeAt is the prefix a previous mirror wrote. If this mirror
// honours the range, make sure it serves the same file before we append to that
// prefix — a divergent total or start means different content, and writing it
// past the existing bytes would corrupt the output. Mirror the segmented guard
// (fatal, so we fall over rather than append unverified bytes). A same-mirror
// retry compares against its own earlier total, so it never false-trips.
if resp.StatusCode == http.StatusPartialContent && resumeAt > 0 {
if prev := atomic.LoadInt64(&d.total); prev > 0 {
if start, _, t, ok := parseContentRange(resp.Header.Get("Content-Range")); ok && (t != prev || start != resumeAt) {
return fatal{fmt.Errorf("%s: returned a divergent range %q (want start %d, length %d)", uri, resp.Header.Get("Content-Range"), resumeAt, prev)}
}
}
}
// If we asked to resume but the server replied 200 (no range honoured), start // If we asked to resume but the server replied 200 (no range honoured), start
// over from the top rather than appending past the existing bytes. // over from the top rather than appending past the existing bytes.
if resp.StatusCode == http.StatusOK { if resp.StatusCode == http.StatusOK {
@@ -1047,7 +851,7 @@ func (d *Download) singleOnce(ctx context.Context, out, uri string, resumeFromDi
} }
off += int64(rd) off += int64(rd)
got += int64(rd) got += int64(rd)
atomic.StoreInt64(&d.completed, off) atomic.AddInt64(&d.completed, int64(rd)) // add deltas, as pump does
d.throttle(ctx, rd) d.throttle(ctx, rd)
} }
if err == io.EOF { if err == io.EOF {
@@ -1066,13 +870,13 @@ func (d *Download) singleOnce(ctx context.Context, out, uri string, resumeFromDi
} }
} }
// saveInterval is how often the .got control file is rewritten so a crash can
// resume; on a clean finish it is removed.
const saveInterval = 60 * time.Second
func (d *Download) saveLoop(ctx context.Context, out string, total int64, etag, lastmod string, segs []seg, done chan<- struct{}) { func (d *Download) saveLoop(ctx context.Context, out string, total int64, etag, lastmod string, segs []seg, done chan<- struct{}) {
defer close(done) defer close(done)
interval := d.cfg.AutoSaveInterval t := time.NewTicker(saveInterval)
if interval <= 0 {
interval = 60 * time.Second // --auto-save-interval default
}
t := time.NewTicker(interval)
defer t.Stop() defer t.Stop()
for { for {
select { select {
@@ -1103,13 +907,9 @@ func (d *Download) request(ctx context.Context, uri, rangeHdr string) (*http.Req
if err != nil { if err != nil {
return nil, err return nil, err
} }
req.Header.Set("User-Agent", d.cfg.UserAgent) req.Header.Set("User-Agent", userAgent)
if d.cfg.Referer != "" { // --header sets arbitrary request headers, applied last so they can override
req.Header.Set("Referer", d.cfg.Referer) // the User-Agent (or add a Referer, Authorization, Cookie, ...) when needed.
}
if d.cfg.HTTPUser != "" {
req.SetBasicAuth(d.cfg.HTTPUser, d.cfg.HTTPPasswd)
}
for _, h := range d.cfg.Headers { for _, h := range d.cfg.Headers {
if k, v, ok := strings.Cut(h, ":"); ok { if k, v, ok := strings.Cut(h, ":"); ok {
req.Header.Set(strings.TrimSpace(k), strings.TrimSpace(v)) req.Header.Set(strings.TrimSpace(k), strings.TrimSpace(v))
@@ -1118,33 +918,9 @@ func (d *Download) request(ctx context.Context, uri, rangeHdr string) (*http.Req
if rangeHdr != "" { if rangeHdr != "" {
req.Header.Set("Range", rangeHdr) req.Header.Set("Range", rangeHdr)
} }
if d.jar != nil {
d.recordURL(req.URL)
}
return req, nil return req, nil
} }
// recordURL remembers a contacted URL so SaveCookies can later read back the
// jar's cookies for that host. Duplicates are tolerated; saveCookieJar dedups.
func (d *Download) recordURL(u *url.URL) {
d.mu.Lock()
d.seenURLs = append(d.seenURLs, u)
d.mu.Unlock()
}
// SaveCookies writes the client's cookie jar back to the --save-cookies file,
// if one was configured. It is a no-op when cookies are not in use. main calls
// it once per HTTP download on exit (cookies are saved at session end).
func (d *Download) SaveCookies() error {
if d.cfg.SaveCookies == "" || d.jar == nil {
return nil
}
d.mu.Lock()
urls := append([]*url.URL{}, d.seenURLs...)
d.mu.Unlock()
return saveCookieJar(d.jar, urls, d.cfg.SaveCookies)
}
// seedSegmentsFromPrefix marks the first prefix bytes of the file as already // seedSegmentsFromPrefix marks the first prefix bytes of the file as already
// downloaded, segment by segment. A foreign partial file has no per-segment // downloaded, segment by segment. A foreign partial file has no per-segment
// metadata, so the only thing we can trust is that bytes [0,prefix) are present; // metadata, so the only thing we can trust is that bytes [0,prefix) are present;

View File

@@ -30,25 +30,24 @@ func TestMirrorPick(t *testing.T) {
} }
} }
// TestMaxConns checks the connection ceiling min(split, mirrors*x), floored // TestMaxConns checks the connection ceiling: --split segment workers, floored
// at 1, since that decides how many segment workers fan out. // at 1. Mirrors are alternate sources, not extra connections, so they do not
// raise the count.
func TestMaxConns(t *testing.T) { func TestMaxConns(t *testing.T) {
cases := []struct{ split, x, mirrors, want int }{ cases := []struct{ split, mirrors, want int }{
{5, 1, 1, 1}, // default -x1 -s5: one connection {5, 1, 5}, // split is the connection count on one server
{5, 16, 1, 5}, // split is the ceiling on one server {16, 1, 16},
{16, 16, 1, 16}, {5, 3, 5}, // 3 mirrors, still split connections
{5, 1, 3, 3}, // 3 mirrors * 1 per host < split {0, 1, 1}, // floor at 1
{16, 2, 3, 6}, // 3 mirrors * 2 per host < split
{0, 1, 1, 1}, // floor at 1
} }
for _, c := range cases { for _, c := range cases {
uris := make([]string, c.mirrors) uris := make([]string, c.mirrors)
for i := range uris { for i := range uris {
uris[i] = "http://h/x" uris[i] = "http://h/x"
} }
d := New(uris, Config{Split: c.split, MaxConnPerServer: c.x}) d := New(uris, Config{Split: c.split})
if d.maxConns != c.want { if d.maxConns != c.want {
t.Errorf("split=%d x=%d mirrors=%d: maxConns=%d, want %d", c.split, c.x, c.mirrors, d.maxConns, c.want) t.Errorf("split=%d mirrors=%d: maxConns=%d, want %d", c.split, c.mirrors, d.maxConns, c.want)
} }
} }
} }
@@ -77,8 +76,7 @@ func TestMirrorFallback(t *testing.T) {
// Primary is healthy (so the probe anchors size/validators), bad is a mirror // Primary is healthy (so the probe anchors size/validators), bad is a mirror
// that 404s every segment it's handed. // that 404s every segment it's handed.
d := New([]string{good.URL + "/f.bin", bad.URL + "/f.bin"}, Config{ d := New([]string{good.URL + "/f.bin", bad.URL + "/f.bin"}, Config{
Dir: dir, Out: "f.bin", Split: 4, MaxConnPerServer: 2, MinSplit: 1, Dir: dir, Out: "f.bin", Split: 4, MinSplit: 1, Tries: 1,
Tries: 1, UserAgent: "got-test",
}) })
if err := d.Run(context.Background()); err != nil { if err := d.Run(context.Background()); err != nil {
t.Fatalf("Run with a 404 mirror should still complete via the healthy one: %v", err) t.Fatalf("Run with a 404 mirror should still complete via the healthy one: %v", err)
@@ -121,8 +119,7 @@ func TestMirrorDivergentRejected(t *testing.T) {
dir := t.TempDir() dir := t.TempDir()
out := filepath.Join(dir, "f.bin") out := filepath.Join(dir, "f.bin")
d := New([]string{primary.URL + "/f.bin", div.URL + "/f.bin"}, Config{ d := New([]string{primary.URL + "/f.bin", div.URL + "/f.bin"}, Config{
Dir: dir, Out: "f.bin", Split: 4, MaxConnPerServer: 2, MinSplit: 1, Dir: dir, Out: "f.bin", Split: 4, MinSplit: 1, Tries: 1,
Tries: 1, UserAgent: "got-test",
}) })
if err := d.Run(context.Background()); err != nil { if err := d.Run(context.Background()); err != nil {
t.Fatalf("Run: %v", err) t.Fatalf("Run: %v", err)

View File

@@ -5,8 +5,9 @@ import "sync/atomic"
// seg is one contiguous byte range of the output file, downloaded by a single // seg is one contiguous byte range of the output file, downloaded by a single
// ranged GET. A segment IS a byte range — a plain HTTP downloader does not need // ranged GET. A segment IS a byte range — a plain HTTP downloader does not need
// the Piece/Segment/block layering that exists only to share code with // the Piece/Segment/block layering that exists only to share code with
// BitTorrent. written is updated with atomic ops so Stat() can read it while a // BitTorrent. start and end are fixed when the file is divided; written is
// worker advances it. // advanced (atomically) by the segment's one worker, so Stat() and the resume
// snapshot can read its frontier while it downloads.
type seg struct { type seg struct {
index int index int
start int64 // first byte offset, inclusive start int64 // first byte offset, inclusive
@@ -16,14 +17,16 @@ type seg struct {
func (s *seg) length() int64 { return s.end - s.start + 1 } func (s *seg) length() int64 { return s.end - s.start + 1 }
func (s *seg) done() bool { return atomic.LoadInt64(&s.written) >= s.length() } func (s *seg) done() bool { return atomic.LoadInt64(&s.written) >= s.length() }
func (s *seg) advance(n int64) { atomic.AddInt64(&s.written, n) } func (s *seg) addWritten(n int64) { atomic.AddInt64(&s.written, n) }
func (s *seg) progress() int64 { return atomic.LoadInt64(&s.written) } func (s *seg) progress() int64 { return atomic.LoadInt64(&s.written) }
func (s *seg) offset() int64 { return s.start + atomic.LoadInt64(&s.written) } func (s *seg) offset() int64 { return s.start + atomic.LoadInt64(&s.written) }
func (s *seg) remaining() int64 { return s.length() - atomic.LoadInt64(&s.written) } func (s *seg) remaining() int64 { return s.length() - atomic.LoadInt64(&s.written) }
// makeSegments divides a file of total bytes into contiguous segments, using at // makeSegments divides a file of total bytes into contiguous segments, using at
// most conns of them and never splitting below minSplit. The remainder lands in // most conns of them and never splitting below minSplit. The remainder lands in
// the last segment. With conns==1 (the default) this yields one segment. // the last segment. With conns==1 (the default) this yields one segment. Each
// segment gets its own worker, so the division here fixes the parallelism: there
// is no later rebalancing.
func makeSegments(total, minSplit int64, conns int) []seg { func makeSegments(total, minSplit int64, conns int) []seg {
if conns < 1 { if conns < 1 {
conns = 1 conns = 1

View File

@@ -3,6 +3,9 @@ package httpdl
import ( import (
"os" "os"
"path/filepath" "path/filepath"
"sort"
"sync"
"sync/atomic"
"testing" "testing"
) )
@@ -52,14 +55,14 @@ func TestSegProgress(t *testing.T) {
if s.done() { if s.done() {
t.Fatalf("new segment should not be done") t.Fatalf("new segment should not be done")
} }
s.advance(60) s.addWritten(60)
if s.offset() != 160 { if s.offset() != 160 {
t.Errorf("offset = %d, want 160", s.offset()) t.Errorf("offset = %d, want 160", s.offset())
} }
if s.remaining() != 40 { if s.remaining() != 40 {
t.Errorf("remaining = %d, want 40", s.remaining()) t.Errorf("remaining = %d, want 40", s.remaining())
} }
s.advance(40) s.addWritten(40)
if !s.done() { if !s.done() {
t.Errorf("segment should be done after writing full length") t.Errorf("segment should be done after writing full length")
} }
@@ -186,3 +189,107 @@ func TestParseContentRange(t *testing.T) {
t.Errorf("parseContentRange(garbage) ok = true, want false") t.Errorf("parseContentRange(garbage) ok = true, want false")
} }
} }
// tilesCover checks that segs cover [0,total) exactly: sorted by start they must
// be contiguous with no gap and no overlap.
func tilesCover(t *testing.T, segs []seg, total int64) {
t.Helper()
sorted := append([]seg(nil), segs...)
sort.Slice(sorted, func(i, j int) bool { return sorted[i].start < sorted[j].start })
var next int64
for _, s := range sorted {
if s.start != next {
t.Fatalf("segment gap/overlap: next byte %d, got start %d", next, s.start)
}
next = s.end + 1
}
if next != total {
t.Fatalf("segments cover %d bytes, want %d", next, total)
}
}
// TestSegmentsConcurrentCoverage drives the segments the way real workers do —
// one worker per segment, each copying its range in small chunks and committing
// with addWritten. The coverage array proves the core invariant: every byte is
// written exactly once, with no gap or overlap between adjacent segments.
func TestSegmentsConcurrentCoverage(t *testing.T) {
const (
total = int64(64 * readBuf) // 2 MiB
minSplit = int64(readBuf)
chunk = int64(readBuf / 4) // one "read"
conns = 8
)
segs := makeSegments(total, minSplit, conns)
if len(segs) < 2 {
t.Fatalf("expected several segments, got %d", len(segs))
}
cover := make([]uint32, total)
var wg sync.WaitGroup
for i := range segs {
wg.Add(1)
go func(s *seg) {
defer wg.Done()
for s.remaining() > 0 {
n := chunk
if s.remaining() < n {
n = s.remaining()
}
off := s.offset()
for j := off; j < off+n; j++ {
atomic.AddUint32(&cover[j], 1)
}
s.addWritten(n)
}
}(&segs[i])
}
wg.Wait()
for i, c := range cover {
if c != 1 {
t.Fatalf("byte %d written %d times, want exactly 1", i, c)
}
}
tilesCover(t, segs, total)
}
// TestSnapshotRoundTrips checks the persistence path an interrupt-and-resume
// relies on: a snapshot of partially-written segments records each frontier, and
// segsFromControl rebuilds a set that still tiles the file and keeps the bytes
// already written.
func TestSnapshotRoundTrips(t *testing.T) {
const (
total = int64(8 * readBuf)
minSplit = int64(readBuf)
)
segs := makeSegments(total, minSplit, 4)
if len(segs) < 2 {
t.Fatalf("expected several segments, got %d", len(segs))
}
// Each segment makes some progress, so the snapshot has a real frontier to
// carry across the round trip.
var want int64
for i := range segs {
n := int64(readBuf)
if n > segs[i].length() {
n = segs[i].length()
}
segs[i].addWritten(n)
want += n
}
c := snapshot("http://x", total, "", "", segs)
if len(c.Segs) != len(segs) {
t.Fatalf("snapshot has %d segments, want %d", len(c.Segs), len(segs))
}
rebuilt := segsFromControl(&c)
tilesCover(t, rebuilt, total)
var written int64
for _, s := range rebuilt {
written += s.progress()
}
if written != want {
t.Errorf("rebuilt written = %d, want %d (bytes must survive the round trip)", written, want)
}
}

View File

@@ -45,9 +45,7 @@ func TestSingleShortBody(t *testing.T) {
d := New([]string{srv.URL + "/file.bin"}, Config{ d := New([]string{srv.URL + "/file.bin"}, Config{
Dir: dir, Dir: dir,
Out: "file.bin", Out: "file.bin",
MaxConnPerServer: 1,
Tries: 1, // a single attempt so the retryable short read surfaces, not loops Tries: 1, // a single attempt so the retryable short read surfaces, not loops
UserAgent: "got-test",
}) })
err := d.single(context.Background(), out) err := d.single(context.Background(), out)
@@ -77,9 +75,7 @@ func TestSingleCompleteFullBody(t *testing.T) {
d := New([]string{srv.URL + "/file.bin"}, Config{ d := New([]string{srv.URL + "/file.bin"}, Config{
Dir: dir, Dir: dir,
Out: "file.bin", Out: "file.bin",
MaxConnPerServer: 1,
Tries: 1, Tries: 1,
UserAgent: "got-test",
}) })
if err := d.single(context.Background(), out); err != nil { if err := d.single(context.Background(), out); err != nil {

View File

@@ -1,32 +0,0 @@
package httpdl
import "testing"
// TestBelowSpeed covers the speedGuard decision, including the counter-reset
// case (item [1]): d.completed can jump backward when singleOnce rewrites it
// with an absolute StoreInt64 and a resumed Range is answered by a 200 that
// resets the offset to 0. A backward jump must NOT be read as a slow window, or
// a healthy download would be false-aborted.
func TestBelowSpeed(t *testing.T) {
const limit = 1000
cases := []struct {
name string
prev, now int64
want bool
}{
{"fast", 0, 5000, false},
{"exactly at limit aborts", 1000, 2000, true},
{"just under limit aborts", 1000, 1999, true},
{"just over limit ok", 1000, 2001, false},
{"stalled aborts", 4242, 4242, true},
{"counter reset is not slow", 1_000_000, 0, false},
{"small backward jump is not slow", 5000, 4500, false},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
if got := belowSpeed(c.prev, c.now, limit); got != c.want {
t.Errorf("belowSpeed(%d, %d, %d) = %v; want %v", c.prev, c.now, limit, got, c.want)
}
})
}
}

263
main.go
View File

@@ -17,7 +17,7 @@ import (
"sort" "sort"
"strconv" "strconv"
"strings" "strings"
"sync" "sync/atomic"
"syscall" "syscall"
"time" "time"
@@ -88,30 +88,33 @@ func run(args []string) int {
sessionFile := opts.Str("save-session") sessionFile := opts.Str("save-session")
ctx, cancel := context.WithCancel(context.Background()) ctx, cancel := context.WithCancel(context.Background())
defer cancel() defer cancel()
// --stop: cancel everything after N seconds, exactly like a Ctrl-C, so an var jobsRef atomic.Pointer[[]job]
// unattended run can't hang forever (0 = off). // publish snapshots the slice header (j is a by-value copy) and stores a
if stop := opts.Int("stop"); stop > 0 { // pointer to it, so the signal handler can read the current jobs lock-free
t := time.AfterFunc(time.Duration(stop)*time.Second, cancel) // without racing the `jobs = append(...)` reassignment in the follow pass.
defer t.Stop() publish := func(j []job) { jobsRef.Store(&j) }
} go handleSignals(cancel, sessionFile, &jobsRef)
jobsRef := &jobsHolder{}
go handleSignals(cancel, sessionFile, jobsRef)
eng := download.NewEngine(opts.Int("max-concurrent-downloads")) eng := download.NewEngine(opts.Int("max-concurrent-downloads"))
uiCtx, uiCancel := context.WithCancel(context.Background()) uiCtx, uiCancel := context.WithCancel(context.Background())
uiDone := make(chan struct{})
if !opts.Bool("quiet") && !opts.Bool("dry-run") { if !opts.Bool("quiet") && !opts.Bool("dry-run") {
rep := progress.New(eng.Snapshot, opts.Bool("human-readable")) rep := progress.New(eng.Snapshot, true)
go rep.Run(uiCtx) go func() {
rep.Run(uiCtx)
close(uiDone)
}()
} else {
close(uiDone)
} }
jobsRef.set(jobs) publish(jobs)
results := runJobs(ctx, eng, jobs) results := runJobs(ctx, eng, jobs)
// --follow-torrent: a .torrent fetched over HTTP becomes a BitTorrent // A .torrent fetched over HTTP becomes a BitTorrent download of its content,
// download of its content (on by default). Done as a second pass so the // done as a second pass so the engine and scheduler stay simple.
// engine and scheduler stay simple. if ctx.Err() == nil && !opts.Bool("dry-run") {
if opts.Bool("follow-torrent") && ctx.Err() == nil && !opts.Bool("dry-run") {
if follow := followUps(jobs, btClient, btOptions(opts)); len(follow) > 0 { if follow := followUps(jobs, btClient, btOptions(opts)); len(follow) > 0 {
// The engine numbers each Result by its position in the slice it was // The engine numbers each Result by its position in the slice it was
// passed, so the follow pass restarts at Index 0. Offset those by the // passed, so the follow pass restarts at Index 0. Offset those by the
@@ -119,7 +122,7 @@ func run(args []string) int {
// Index space, so the sort below restores input order deterministically. // Index space, so the sort below restores input order deterministically.
nInitial := len(jobs) nInitial := len(jobs)
jobs = append(jobs, follow...) jobs = append(jobs, follow...)
jobsRef.set(jobs) publish(jobs)
followResults := runJobs(ctx, eng, follow) followResults := runJobs(ctx, eng, follow)
for i := range followResults { for i := range followResults {
followResults[i].Index += nInitial followResults[i].Index += nInitial
@@ -129,9 +132,7 @@ func run(args []string) int {
} }
uiCancel() uiCancel()
time.Sleep(20 * time.Millisecond) // let the renderer clear its line <-uiDone // wait for the renderer's final frame before printing the OK/FAIL lines
saveCookies(jobs)
if sessionFile != "" { if sessionFile != "" {
if err := saveSession(sessionFile, jobs); err != nil { if err := saveSession(sessionFile, jobs); err != nil {
@@ -147,25 +148,6 @@ func run(args []string) int {
return report(results) return report(results)
} }
// jobsHolder lets the signal handler reach the latest job list for a
// best-effort session save on a forced (second-Ctrl-C) exit.
type jobsHolder struct {
mu sync.Mutex
jobs []job
}
func (h *jobsHolder) set(jobs []job) {
h.mu.Lock()
h.jobs = jobs
h.mu.Unlock()
}
func (h *jobsHolder) get() []job {
h.mu.Lock()
defer h.mu.Unlock()
return h.jobs
}
// job pairs a download with the source string that produced it, so we can write // job pairs a download with the source string that produced it, so we can write
// the still-unfinished ones back to a session file. // the still-unfinished ones back to a session file.
type job struct { type job struct {
@@ -190,35 +172,43 @@ type failedDownload struct {
err error err error
} }
func (f failedDownload) Name() string { return f.name } func (f *failedDownload) Name() string { return f.name }
func (f failedDownload) Run(context.Context) error { return f.err } func (f *failedDownload) Run(context.Context) error { return f.err }
func (f failedDownload) Stat() download.Stat { func (f *failedDownload) Stat() download.Stat {
return download.Stat{Name: f.name, Status: download.Errored, Total: -1} return download.Stat{Name: f.name, Status: download.Errored, Total: -1}
} }
// dedupeTorrents finds torrent/magnet sources that name the same torrent as an // seenInfoHash records source under its torrent infohash in seen and reports
// earlier source (same infohash), returning each duplicate URI mapped to the // whether that infohash was already recorded, returning the source that first
// earlier source it repeats. HTTP(S) URLs and any source whose infohash can't be // claimed it. The shared client keys torrents by infohash and does not refcount,
// read without a network fetch are never duplicates here. The shared client keys // so two sources naming one torrent would have the first to finish Drop it out
// torrents by infohash and does not refcount, so build() turns each duplicate // from under the other; callers fail the duplicate instead. A source whose
// into a failed download (duplicate infohash) rather than letting two jobs // infohash can't be known without a network fetch (an HTTP URL), or that is
// share — and Drop — one torrent. // v2-only or unreadable, is never a duplicate: it isn't recorded, so it runs.
func dedupeTorrents(uris []string, follow bool) map[string]string { func seenInfoHash(seen map[metainfo.Hash]string, source string, isFile bool) (first string, dup bool) {
h, ok := bt.SourceInfoHash(source, isFile)
if !ok {
return "", false
}
if first, dup = seen[h]; dup {
return first, true
}
seen[h] = source
return "", false
}
// dedupeTorrents finds torrent/magnet sources that repeat an earlier source's
// infohash, returning each duplicate URI mapped to the source it repeats.
func dedupeTorrents(uris []string) map[string]string {
seen := map[metainfo.Hash]string{} seen := map[metainfo.Hash]string{}
dup := map[string]string{} dup := map[string]string{}
for _, u := range uris { for _, u := range uris {
k := uriKind(u, follow) k := uriKind(u)
if k != kindMagnet && k != kindTorrent { if k != kindMagnet && k != kindTorrent {
continue continue
} }
h, ok := bt.SourceInfoHash(u, k == kindTorrent) if first, isDup := seenInfoHash(seen, u, k == kindTorrent); isDup {
if !ok {
continue
}
if first, isDup := seen[h]; isDup {
dup[u] = first dup[u] = first
} else {
seen[h] = u
} }
} }
return dup return dup
@@ -231,7 +221,7 @@ func build(opts *cli.Options, targets []target) ([]job, *torrent.Client, error)
// finish would Drop it out from under the other. The duplicate is reported as // finish would Drop it out from under the other. The duplicate is reported as
// a failed download (duplicate infohash, exit 12); the loop below does this, // a failed download (duplicate infohash, exit 12); the loop below does this,
// so the duplicate never obtains or Drops a shared torrent. // so the duplicate never obtains or Drops a shared torrent.
dups := dedupeTorrents(flat, opts.Bool("follow-torrent")) dups := dedupeTorrents(flat)
overallDL := makeLimiter(opts.Size("max-overall-download-limit")) overallDL := makeLimiter(opts.Size("max-overall-download-limit"))
overallUL := makeLimiter(opts.Size("max-overall-upload-limit")) overallUL := makeLimiter(opts.Size("max-overall-upload-limit"))
@@ -249,7 +239,7 @@ func build(opts *cli.Options, targets []target) ([]job, *torrent.Client, error)
// it (e.g. the listen port is taken) disables BitTorrent but still lets HTTP // it (e.g. the listen port is taken) disables BitTorrent but still lets HTTP
// downloads run; those torrents then report the error. // downloads run; those torrents then report the error.
var client *torrent.Client var client *torrent.Client
if needsBT(opts, flat) { if needsBT(flat) {
c, err := bt.NewClient(btClientConfig(opts, overallDL, overallUL)) c, err := bt.NewClient(btClientConfig(opts, overallDL, overallUL))
if err != nil { if err != nil {
fmt.Fprintln(os.Stderr, "got: bittorrent disabled:", err) fmt.Fprintln(os.Stderr, "got: bittorrent disabled:", err)
@@ -263,14 +253,14 @@ func build(opts *cli.Options, targets []target) ([]job, *torrent.Client, error)
u := t.uris[0] // primary: classifies the target and names the session source u := t.uris[0] // primary: classifies the target and names the session source
src := strings.Join(t.uris, "\t") src := strings.Join(t.uris, "\t")
if first, isDup := dups[u]; isDup { if first, isDup := dups[u]; isDup {
out = append(out, job{source: src, dl: failedDownload{ out = append(out, job{source: src, dl: &failedDownload{
name: u, name: u,
err: fmt.Errorf("%w: same torrent as %s", errDuplicate, first), err: fmt.Errorf("%w: same torrent as %s", errDuplicate, first),
}}) }})
continue continue
} }
var dl download.Download var dl download.Download
switch uriKind(u, opts.Bool("follow-torrent")) { switch uriKind(u) {
case kindMagnet: case kindMagnet:
dl = bt.New(client, u, false, bo) dl = bt.New(client, u, false, bo)
case kindHTTP, kindFollow: case kindHTTP, kindFollow:
@@ -289,16 +279,15 @@ func build(opts *cli.Options, targets []target) ([]job, *torrent.Client, error)
} }
// uriKind classifies a URI by scheme into the taxonomy build() and needsBT() // uriKind classifies a URI by scheme into the taxonomy build() and needsBT()
// both switch on. Scheme wins over suffix: an http URL ending in .torrent is a // both switch on. Scheme wins over suffix: an http URL ending in .torrent is
// file to fetch over HTTP, and with --follow-torrent it becomes a torrent only // fetched over HTTP first, then followed as a torrent (kindFollow).
// after the fetch (kindFollow), so it still downloads over HTTP first. func uriKind(u string) kind {
func uriKind(u string, follow bool) kind {
lu := strings.ToLower(u) lu := strings.ToLower(u)
switch { switch {
case strings.HasPrefix(u, "magnet:"): case strings.HasPrefix(u, "magnet:"):
return kindMagnet return kindMagnet
case strings.HasPrefix(u, "http://"), strings.HasPrefix(u, "https://"): case strings.HasPrefix(u, "http://"), strings.HasPrefix(u, "https://"):
if follow && strings.HasSuffix(lu, ".torrent") { if strings.HasSuffix(lu, ".torrent") {
return kindFollow return kindFollow
} }
return kindHTTP return kindHTTP
@@ -320,10 +309,9 @@ const (
// needsBT reports whether the run will involve any torrent, so we only start a // needsBT reports whether the run will involve any torrent, so we only start a
// client when one is actually needed. // client when one is actually needed.
func needsBT(opts *cli.Options, uris []string) bool { func needsBT(uris []string) bool {
follow := opts.Bool("follow-torrent")
for _, u := range uris { for _, u := range uris {
switch uriKind(u, follow) { switch uriKind(u) {
case kindMagnet, kindTorrent, kindFollow: case kindMagnet, kindTorrent, kindFollow:
return true return true
} }
@@ -358,6 +346,7 @@ func runJobs(ctx context.Context, eng *download.Engine, jobs []job) []download.R
// so a regular download that merely ends in ".torrent" is left alone. // so a regular download that merely ends in ".torrent" is left alone.
func followUps(jobs []job, client *torrent.Client, bo bt.Options) []job { func followUps(jobs []job, client *torrent.Client, bo bt.Options) []job {
var out []job var out []job
seen := map[metainfo.Hash]string{}
for _, j := range jobs { for _, j := range jobs {
h, ok := j.dl.(*httpdl.Download) h, ok := j.dl.(*httpdl.Download)
if !ok || j.dl.Stat().Status != download.Complete { if !ok || j.dl.Stat().Status != download.Complete {
@@ -373,24 +362,20 @@ func followUps(jobs []job, client *torrent.Client, bo bt.Options) []job {
if _, err := bt.Files(path); err != nil { if _, err := bt.Files(path); err != nil {
continue // downloaded file is not a valid torrent continue // downloaded file is not a valid torrent
} }
// build() dedupes pass-1 sources but cannot reach a not-yet-fetched HTTP
// .torrent, so dedupe the fetched ones here too against the same hazard.
if first, dup := seenInfoHash(seen, path, true); dup {
out = append(out, job{source: path, dl: &failedDownload{
name: path,
err: fmt.Errorf("%w: same torrent as %s", errDuplicate, first),
}})
continue
}
out = append(out, job{source: path, dl: bt.New(client, path, true, bo)}) out = append(out, job{source: path, dl: bt.New(client, path, true, bo)})
} }
return out return out
} }
// saveCookies asks each HTTP download to write its cookie jar back to the
// --save-cookies file, persisting cookies at session end. A download with no
// --save-cookies configured is a no-op.
func saveCookies(jobs []job) {
for _, j := range jobs {
if h, ok := j.dl.(*httpdl.Download); ok {
if err := h.SaveCookies(); err != nil {
fmt.Fprintln(os.Stderr, "got:", err)
}
}
}
}
// target is one download: a single torrent/magnet/URL, or several HTTP mirror // target is one download: a single torrent/magnet/URL, or several HTTP mirror
// URLs that serve identical bytes (uris[0] is the primary). // URLs that serve identical bytes (uris[0] is the primary).
type target struct{ uris []string } type target struct{ uris []string }
@@ -405,34 +390,30 @@ func allURIs(ts []target) []string {
return u return u
} }
// gatherURIs collects download targets from the command line, --torrent-file, // gatherURIs collects download targets from the command line and --input-file.
// and --input-file. Plain command-line http/https URIs are grouped into ONE // Plain command-line http/https URIs are grouped into ONE mirrored download by
// mirrored download by default (every command-line URI names the same file); // default (every command-line URI names the same file); -Z/--force-sequential
// -Z/--force-sequential makes each its own download. Torrents and magnets // makes each its own download. Torrents, magnets, and .torrent URLs (fetched
// are always one target each; --input-file groups TAB-separated URIs per line. // over HTTP then followed) are always one target each, since distinct torrents
// are not mirrors of one another; --input-file groups TAB-separated URIs per line.
func gatherURIs(opts *cli.Options, cmdURIs []string) []target { func gatherURIs(opts *cli.Options, cmdURIs []string) []target {
var ts []target var ts []target
follow := opts.Bool("follow-torrent")
seq := opts.Bool("force-sequential") seq := opts.Bool("force-sequential")
var httpGroup []string var httpGroup []string
for _, u := range cmdURIs { for _, u := range cmdURIs {
switch uriKind(u, follow) { // Only plain HTTP URLs collapse into one mirror download (alternate sources
case kindHTTP, kindFollow: // for the same file); -Z opts out. A .torrent URL (kindFollow), like a
if seq { // torrent or magnet, is its own download — distinct torrents are never
ts = append(ts, target{uris: []string{u}}) // mirrors of one another.
} else { if uriKind(u) == kindHTTP && !seq {
httpGroup = append(httpGroup, u) httpGroup = append(httpGroup, u)
} } else {
default:
ts = append(ts, target{uris: []string{u}}) ts = append(ts, target{uris: []string{u}})
} }
} }
if len(httpGroup) > 0 { if len(httpGroup) > 0 {
ts = append(ts, target{uris: httpGroup}) ts = append(ts, target{uris: httpGroup})
} }
if tf := opts.Str("torrent-file"); tf != "" {
ts = append(ts, target{uris: []string{tf}})
}
if in := opts.Str("input-file"); in != "" { if in := opts.Str("input-file"); in != "" {
ts = append(ts, readInputFile(in)...) ts = append(ts, readInputFile(in)...)
} }
@@ -491,8 +472,6 @@ func httpConfig(opts *cli.Options, single bool, overallDL *rate.Limiter) httpdl.
Dir: opts.Str("dir"), Dir: opts.Str("dir"),
Out: out, Out: out,
Split: opts.Int("split"), Split: opts.Int("split"),
MaxConnPerServer: opts.Int("max-connection-per-server"),
MinSplit: opts.Size("min-split-size"),
Tries: opts.Int("max-tries"), Tries: opts.Int("max-tries"),
Timeout: time.Duration(opts.Int("timeout")) * time.Second, Timeout: time.Duration(opts.Int("timeout")) * time.Second,
FileAlloc: opts.Str("file-allocation"), FileAlloc: opts.Str("file-allocation"),
@@ -500,26 +479,9 @@ func httpConfig(opts *cli.Options, single bool, overallDL *rate.Limiter) httpdl.
AllowOverwrite: opts.Bool("allow-overwrite"), AllowOverwrite: opts.Bool("allow-overwrite"),
AutoRename: opts.Bool("auto-file-renaming"), AutoRename: opts.Bool("auto-file-renaming"),
Headers: opts.List("header"), Headers: opts.List("header"),
UserAgent: opts.Str("user-agent"),
Referer: opts.Str("referer"),
Proxy: opts.Str("all-proxy"),
CheckCert: opts.Bool("check-certificate"),
Limit: opts.Size("max-download-limit"),
OverallLimiter: overallDL, OverallLimiter: overallDL,
RetryWait: time.Duration(opts.Int("retry-wait")) * time.Second,
AutoSaveInterval: time.Duration(opts.Int("auto-save-interval")) * time.Second,
ConnectTimeout: time.Duration(opts.Int("connect-timeout")) * time.Second,
LoadCookies: opts.Str("load-cookies"),
SaveCookies: opts.Str("save-cookies"),
ConditionalGet: opts.Bool("conditional-get"),
RemoteTime: opts.Bool("remote-time"),
HTTPUser: opts.Str("http-user"),
HTTPPasswd: opts.Str("http-passwd"),
LowestSpeedLimit: opts.Size("lowest-speed-limit"),
Checksum: opts.Str("checksum"), Checksum: opts.Str("checksum"),
DryRun: opts.Bool("dry-run"), DryRun: opts.Bool("dry-run"),
CACert: opts.Str("ca-certificate"),
DisableIPv6: opts.Bool("disable-ipv6"),
} }
} }
@@ -529,13 +491,8 @@ func btClientConfig(opts *cli.Options, overallDL, overallUL *rate.Limiter) bt.Cl
Dir: opts.Str("dir"), Dir: opts.Str("dir"),
ListenPortSpec: opts.Str("listen-port"), ListenPortSpec: opts.Str("listen-port"),
DHT: opts.Bool("enable-dht"), DHT: opts.Bool("enable-dht"),
MaxPeers: opts.Int("bt-max-peers"),
OverallDown: overallDL, OverallDown: overallDL,
OverallUp: overallUL, OverallUp: overallUL,
DownLimit: opts.Size("max-download-limit"),
UpLimit: opts.Size("max-upload-limit"),
UserAgent: opts.Str("user-agent"),
DisableIPv6: opts.Bool("disable-ipv6"),
} }
} }
@@ -545,8 +502,6 @@ func btOptions(opts *cli.Options) bt.Options {
SeedTimeSet: opts.IsSet("seed-time"), SeedTimeSet: opts.IsSet("seed-time"),
SeedTime: time.Duration(opts.Float("seed-time") * float64(time.Minute)), SeedTime: time.Duration(opts.Float("seed-time") * float64(time.Minute)),
SeedRatio: opts.Float("seed-ratio"), SeedRatio: opts.Float("seed-ratio"),
StopTimeout: time.Duration(opts.Int("bt-stop-timeout")) * time.Second,
MetaTimeout: time.Duration(opts.Int("bt-metadata-timeout")) * time.Second,
SelectFiles: parseSelect(opts.Str("select-file")), SelectFiles: parseSelect(opts.Str("select-file")),
CheckIntegrity: opts.Bool("check-integrity"), CheckIntegrity: opts.Bool("check-integrity"),
DryRun: opts.Bool("dry-run"), DryRun: opts.Bool("dry-run"),
@@ -667,8 +622,6 @@ func exitCode(err error) int {
return 6 return 6
case errors.Is(err, httpdl.ErrNotFound): case errors.Is(err, httpdl.ErrNotFound):
return 3 return 3
case errors.Is(err, httpdl.ErrTooSlow):
return 5
case errors.Is(err, httpdl.ErrChecksum): case errors.Is(err, httpdl.ErrChecksum):
return 32 return 32
case errors.Is(err, errDuplicate): case errors.Is(err, errDuplicate):
@@ -679,42 +632,29 @@ func exitCode(err error) int {
// isNameResolution reports whether err is (or wraps) a DNS lookup failure — // isNameResolution reports whether err is (or wraps) a DNS lookup failure —
// "name resolution failed" (19), which usually points at the user's own // "name resolution failed" (19), which usually points at the user's own
// network or resolver rather than a dead remote. // network or resolver rather than a dead remote. The stdlib reports every such
// failure (no-such-host, server-misbehaving) as *net.DNSError.
func isNameResolution(err error) bool { func isNameResolution(err error) bool {
var de *net.DNSError var de *net.DNSError
if errors.As(err, &de) { return errors.As(err, &de)
return true
}
msg := strings.ToLower(err.Error())
return strings.Contains(msg, "no such host") || strings.Contains(msg, "server misbehaving")
} }
// isNetwork reports whether err is a connection-level failure — refused, reset, // isNetwork reports whether err is a connection-level failure — refused, reset,
// or an unreachable host/network — a "network problem" (6). // or an unreachable host/network — a "network problem" (6). A dial error from
// net/http wraps the syscall errno, so the typed check sees through it.
func isNetwork(err error) bool { func isNetwork(err error) bool {
if errors.Is(err, syscall.ECONNREFUSED) || errors.Is(err, syscall.ECONNRESET) || return errors.Is(err, syscall.ECONNREFUSED) || errors.Is(err, syscall.ECONNRESET) ||
errors.Is(err, syscall.ENETUNREACH) || errors.Is(err, syscall.EHOSTUNREACH) { errors.Is(err, syscall.ENETUNREACH) || errors.Is(err, syscall.EHOSTUNREACH)
return true
}
msg := strings.ToLower(err.Error())
return strings.Contains(msg, "connection refused") ||
strings.Contains(msg, "connection reset") ||
strings.Contains(msg, "network is unreachable") ||
strings.Contains(msg, "no route to host")
} }
// isTimeout reports whether err is (or wraps) a timeout: a deadline exceeded, a // isTimeout reports whether err is (or wraps) a timeout: a deadline exceeded, a
// net.Error that timed out, or an idle-read timeout reported as such. // net.Error that timed out, or our own idle-read timeout (httpdl.ErrTimeout).
func isTimeout(err error) bool { func isTimeout(err error) bool {
if errors.Is(err, context.DeadlineExceeded) { if errors.Is(err, context.DeadlineExceeded) || errors.Is(err, httpdl.ErrTimeout) {
return true return true
} }
var ne net.Error var ne net.Error
if errors.As(err, &ne) && ne.Timeout() { return errors.As(err, &ne) && ne.Timeout()
return true
}
msg := strings.ToLower(err.Error())
return strings.Contains(msg, "timeout") || strings.Contains(msg, "timed out")
} }
// saveSession writes the sources of every download that did not finish to path, // saveSession writes the sources of every download that did not finish to path,
@@ -777,14 +717,18 @@ func progressNote(s download.Stat) string {
// TERM) cancels the graceful context so run() can wind down and save normally; // TERM) cancels the graceful context so run() can wind down and save normally;
// a second forces exit, but first writes the session best-effort so the resume // a second forces exit, but first writes the session best-effort so the resume
// list survives a force-quit. // list survives a force-quit.
func handleSignals(cancel context.CancelFunc, sessionFile string, jobs *jobsHolder) { func handleSignals(cancel context.CancelFunc, sessionFile string, jobs *atomic.Pointer[[]job]) {
sig := make(chan os.Signal, 2) sig := make(chan os.Signal, 2)
signal.Notify(sig, os.Interrupt, syscall.SIGTERM) signal.Notify(sig, os.Interrupt, syscall.SIGTERM)
<-sig // first: cancel the graceful context <-sig // first: cancel the graceful context
cancel() cancel()
<-sig // second: force exit <-sig // second: force exit
if sessionFile != "" { if sessionFile != "" {
if err := saveSession(sessionFile, jobs.get()); err != nil { var js []job // nil until the first publish; saveSession handles an empty list
if p := jobs.Load(); p != nil {
js = *p
}
if err := saveSession(sessionFile, js); err != nil {
fmt.Fprintln(os.Stderr, "got:", err) fmt.Fprintln(os.Stderr, "got:", err)
} }
} }
@@ -799,6 +743,11 @@ func makeLimiter(bps int64) *rate.Limiter {
return rate.NewLimiter(rate.Limit(bps), download.LimiterBurst(bps)) return rate.NewLimiter(rate.Limit(bps), download.LimiterBurst(bps))
} }
// maxSelectIndex bounds how far a --select-file range is expanded: a torrent has
// at most a few thousand files, so a degenerate spec like "1-9999999999" must not
// spin or balloon the map. Indexes past the real file count are ignored anyway.
const maxSelectIndex = 1 << 20
// parseSelect parses "1,3-5" into a set of 1-based file indexes. // parseSelect parses "1,3-5" into a set of 1-based file indexes.
func parseSelect(s string) map[int]bool { func parseSelect(s string) map[int]bool {
s = strings.TrimSpace(s) s = strings.TrimSpace(s)
@@ -816,10 +765,14 @@ func parseSelect(s string) map[int]bool {
continue continue
} }
b, _ := strconv.Atoi(strings.TrimSpace(hi)) b, _ := strconv.Atoi(strings.TrimSpace(hi))
for i := a; i <= b; i++ { if a < 1 {
if i > 0 { a = 1
out[i] = true
} }
if b > maxSelectIndex {
b = maxSelectIndex
}
for i := a; i <= b; i++ {
out[i] = true
} }
} }
return out return out

View File

@@ -11,6 +11,9 @@ import (
"syscall" "syscall"
"testing" "testing"
"github.com/anacrolix/torrent/bencode"
"github.com/anacrolix/torrent/metainfo"
"github.com/hanbok/got/bt"
"github.com/hanbok/got/cli" "github.com/hanbok/got/cli"
"github.com/hanbok/got/download" "github.com/hanbok/got/download"
"github.com/hanbok/got/httpdl" "github.com/hanbok/got/httpdl"
@@ -66,18 +69,15 @@ func TestExitCode(t *testing.T) {
err error err error
want int want int
}{ }{
{"idle timeout", errors.New("idle timeout: no data received"), 2}, {"idle timeout", fmt.Errorf("segment 0: %w (after 1m0s)", httpdl.ErrTimeout), 2},
{"metadata timed out", errors.New("timed out fetching metadata after 3s"), 2}, {"metadata timed out", fmt.Errorf("timed out fetching metadata after 3s: %w", context.DeadlineExceeded), 2},
{"deadline", context.DeadlineExceeded, 2}, {"deadline", context.DeadlineExceeded, 2},
{"file exists", fmt.Errorf("/tmp/x: %w (use --allow-overwrite or -c)", httpdl.ErrOutputExists), 13}, {"file exists", fmt.Errorf("/tmp/x: %w (use --allow-overwrite or -c)", httpdl.ErrOutputExists), 13},
{"dns typed", dnsErr, 19}, {"dns typed", dnsErr, 19},
{"dns wrapped through retries", fmt.Errorf("http://x: %w (after 5 tries)", dnsErr), 19}, {"dns wrapped through retries", fmt.Errorf("http://x: %w (after 5 tries)", dnsErr), 19},
{"dns by message", errors.New("dial tcp: lookup nope.invalid: no such host"), 19},
{"refused typed", refused, 6}, {"refused typed", refused, 6},
{"refused wrapped through retries", fmt.Errorf("http://x: %w (after 5 tries)", refused), 6}, {"refused wrapped through retries", fmt.Errorf("http://x: %w (after 5 tries)", refused), 6},
{"refused by message", errors.New("dial tcp 127.0.0.1:1: connect: connection refused"), 6},
{"not found", fmt.Errorf("%w: http://x/y: 404 Not Found", httpdl.ErrNotFound), 3}, {"not found", fmt.Errorf("%w: http://x/y: 404 Not Found", httpdl.ErrNotFound), 3},
{"too slow", fmt.Errorf("http://x: %w (<= %d bytes/sec)", httpdl.ErrTooSlow, 1024), 5},
{"checksum", fmt.Errorf("/tmp/x: %w (want a, got b)", httpdl.ErrChecksum), 32}, {"checksum", fmt.Errorf("/tmp/x: %w (want a, got b)", httpdl.ErrChecksum), 32},
{"duplicate", fmt.Errorf("%w: same torrent as y", errDuplicate), 12}, {"duplicate", fmt.Errorf("%w: same torrent as y", errDuplicate), 12},
{"unknown", errors.New("disk full"), 1}, {"unknown", errors.New("disk full"), 1},
@@ -103,7 +103,7 @@ func TestDedupeTorrents(t *testing.T) {
c := "magnet:?xt=urn:btih:1111111111111111111111111111111111111111&dn=other" c := "magnet:?xt=urn:btih:1111111111111111111111111111111111111111&dn=other"
web := "http://example.com/file.iso" web := "http://example.com/file.iso"
dup := dedupeTorrents([]string{a, web, b, c}, true) dup := dedupeTorrents([]string{a, web, b, c})
// b is the only duplicate (of a); a, web and c are not duplicates. // b is the only duplicate (of a); a, web and c are not duplicates.
if len(dup) != 1 || dup[b] != a { if len(dup) != 1 || dup[b] != a {
@@ -112,8 +112,8 @@ func TestDedupeTorrents(t *testing.T) {
} }
// TestGatherURIsGrouping checks the command-line grouping: plain http URIs // TestGatherURIsGrouping checks the command-line grouping: plain http URIs
// become one mirrored download by default, -Z splits them, and torrents/magnets // become one mirrored download by default, -Z splits them, and torrents,
// stay separate. // magnets, and .torrent URLs stay separate.
func TestGatherURIsGrouping(t *testing.T) { func TestGatherURIsGrouping(t *testing.T) {
parse := func(args ...string) []target { parse := func(args ...string) []target {
res, err := cli.Parse(append([]string{"--no-conf"}, args...)) res, err := cli.Parse(append([]string{"--no-conf"}, args...))
@@ -131,6 +131,43 @@ func TestGatherURIsGrouping(t *testing.T) {
if ts := parse("http://a/x", "magnet:?xt=urn:btih:abc"); len(ts) != 2 { if ts := parse("http://a/x", "magnet:?xt=urn:btih:abc"); len(ts) != 2 {
t.Errorf("http + magnet: got %d targets, want 2 (magnet never mirrors)", len(ts)) t.Errorf("http + magnet: got %d targets, want 2 (magnet never mirrors)", len(ts))
} }
// Distinct .torrent URLs are separate downloads, not mirrors of one file
// (follow-torrent defaults on, so each is fetched then followed).
if ts := parse("http://a/x.torrent", "http://b/y.torrent", "http://c/z.torrent"); len(ts) != 3 {
t.Errorf(".torrent URLs: got %d targets, want 3 separate downloads", len(ts))
}
// A plain http URL still mirror-groups even alongside a .torrent URL: one
// http group of 1 + one torrent target = 2.
if ts := parse("http://a/x", "http://b/y.torrent"); len(ts) != 2 {
t.Errorf("http + .torrent: got %d targets, want 2", len(ts))
}
}
// TestHTTPConnDefault locks the parallelism default: a bare `got URL` must use
// several connections, not one. --split is the single speed dial — it is the
// connection count directly, defaulting to 5.
func TestHTTPConnDefault(t *testing.T) {
cfg := func(args ...string) httpdl.Config {
res, err := cli.Parse(append([]string{"--no-conf"}, args...))
if err != nil {
t.Fatalf("parse %v: %v", args, err)
}
return httpConfig(res.Options, true, nil)
}
tests := []struct {
name string
args []string
want int
}{
{"bare URL uses split default", []string{"http://a/x"}, 5},
{"-s16 means 16", []string{"-s16", "http://a/x"}, 16},
{"-s1 stays single", []string{"-s1", "http://a/x"}, 1},
}
for _, tc := range tests {
if c := cfg(tc.args...); c.Split != tc.want {
t.Errorf("%s: Split = %d, want %d", tc.name, c.Split, tc.want)
}
}
} }
// TestReadInputFileTAB checks the input-file grouping: TAB-separated URIs on // TestReadInputFileTAB checks the input-file grouping: TAB-separated URIs on
@@ -152,10 +189,69 @@ func TestReadInputFileTAB(t *testing.T) {
} }
} }
// writeTorrent creates a .torrent file at dir/name whose info dict is built from
// a file of the given content and internal name. Two calls with the same
// internalName and content produce the same infohash regardless of the output
// filename, so the test can make two .torrent files that name one torrent.
func writeTorrent(t *testing.T, dir, name, internalName string, content []byte) string {
t.Helper()
data := filepath.Join(dir, internalName)
if err := os.WriteFile(data, content, 0o644); err != nil {
t.Fatal(err)
}
info := metainfo.Info{PieceLength: 32 * 1024}
if err := info.BuildFromFilePath(data); err != nil {
t.Fatal(err)
}
b, err := bencode.Marshal(info)
if err != nil {
t.Fatal(err)
}
tp := filepath.Join(dir, name)
f, err := os.Create(tp)
if err != nil {
t.Fatal(err)
}
defer f.Close()
if err := (&metainfo.MetaInfo{InfoBytes: b}).Write(f); err != nil {
t.Fatal(err)
}
return tp
}
// TestFollowUpInfoHashDedup covers the fix for two fetched .torrent files that
// name the same torrent: the second must be flagged a duplicate so followUps
// fails it rather than letting both share — and Drop — one torrent on the client.
func TestFollowUpInfoHashDedup(t *testing.T) {
dir := t.TempDir()
a := writeTorrent(t, dir, "a.torrent", "same", []byte("AAAAAAAA"))
b := writeTorrent(t, dir, "b.torrent", "same", []byte("AAAAAAAA")) // same infohash as a
c := writeTorrent(t, dir, "c.torrent", "diff", []byte("BBBBBBBB")) // a distinct torrent
// Sanity-check the fixture: a and b really share an infohash, c differs.
ha, _ := bt.SourceInfoHash(a, true)
hb, _ := bt.SourceInfoHash(b, true)
hc, _ := bt.SourceInfoHash(c, true)
if ha != hb || ha == hc {
t.Fatalf("fixture: want a==b!=c infohashes, got a=%v b=%v c=%v", ha, hb, hc)
}
seen := map[metainfo.Hash]string{}
if first, dup := seenInfoHash(seen, a, true); dup {
t.Fatalf("a is the first occurrence, wrongly flagged a duplicate of %s", first)
}
if first, dup := seenInfoHash(seen, b, true); !dup || first != a {
t.Errorf("b should duplicate a: got first=%q dup=%v, want %q true", first, dup, a)
}
if _, dup := seenInfoHash(seen, c, true); dup {
t.Errorf("c is a distinct torrent, wrongly flagged a duplicate")
}
}
func TestReportExit(t *testing.T) { func TestReportExit(t *testing.T) {
canceled := download.Result{Name: "a", Err: context.Canceled} canceled := download.Result{Name: "a", Err: context.Canceled}
done := download.Result{Name: "b", Err: nil} done := download.Result{Name: "b", Err: nil}
timeout := download.Result{Name: "c", Err: errors.New("idle timeout")} timeout := download.Result{Name: "c", Err: httpdl.ErrTimeout}
tests := []struct { tests := []struct {
name string name string

View File

@@ -41,12 +41,6 @@ func humanSize(n int64, human bool) string {
return fmt.Sprintf("%.0f%ciB", val, suffix) return fmt.Sprintf("%.0f%ciB", val, suffix)
} }
// speed formats a bytes-per-second rate. The DL:/UL: fields show the bare
// abbreviated size with no "/s" suffix.
func speed(bytesPerSec int64, human bool) string {
return humanSize(bytesPerSec, human)
}
// secfmt renders a duration as "1h2m3s", appending each unit only when it is // secfmt renders a duration as "1h2m3s", appending each unit only when it is
// nonzero (but still showing seconds when the whole input is 0): // nonzero (but still showing seconds when the whole input is 0):
// 3600s->"1h", 120s->"2m", 3720s->"1h2m". A non-positive or absurd // 3600s->"1h", 120s->"2m", 3720s->"1h2m". A non-positive or absurd

View File

@@ -53,16 +53,6 @@ func TestSecfmt(t *testing.T) {
} }
} }
func TestSpeedNoSuffix(t *testing.T) {
// The DL:/UL: fields show the bare abbreviated size with no "/s".
if got := speed(1536, true); got != "1.5KiB" {
t.Errorf("speed(1536,true) = %q, want %q", got, "1.5KiB")
}
if got := speed(2048, false); got != "2048B" {
t.Errorf("speed(2048,false) = %q, want %q", got, "2048B")
}
}
func TestPercent(t *testing.T) { func TestPercent(t *testing.T) {
if got := percent(50, 100); got != 50 { if got := percent(50, 100); got != 50 {
t.Errorf("percent(50,100) = %d, want 50", got) t.Errorf("percent(50,100) = %d, want 50", got)

View File

@@ -1,7 +1,8 @@
// Package progress renders a live, single-line status display. One goroutine ticks // Package progress renders a live status display: a summary line over one
// once a second, pulls a snapshot of the running downloads, and derives speeds // aligned row per download, collapsing to the summary alone when the rows would
// from the change since the previous tick. There are no locks here: the data // not fit. One goroutine ticks once a second, pulls a snapshot of the running
// arrives by value through the snapshot function. // downloads, and derives speeds from the change since the previous tick. There
// are no locks here: the data arrives by value through the snapshot function.
package progress package progress
import ( import (
@@ -9,6 +10,7 @@ import (
"fmt" "fmt"
"io" "io"
"os" "os"
"strconv"
"strings" "strings"
"time" "time"
"unicode/utf8" "unicode/utf8"
@@ -25,8 +27,10 @@ type Reporter struct {
w io.Writer w io.Writer
tty bool tty bool
width int width int
height int
win map[string]*speedWindow // keyed by Stat.ID win map[string]*speedWindow // keyed by Stat.ID
prevLines int // lines drawn in the previous TTY frame, redrawn in place
lastLog time.Time // last non-TTY line emitted lastLog time.Time // last non-TTY line emitted
} }
@@ -62,6 +66,7 @@ func New(snap func() []download.Stat, human bool) *Reporter {
w: os.Stdout, w: os.Stdout,
tty: term.IsTerminal(int(os.Stdout.Fd())), tty: term.IsTerminal(int(os.Stdout.Fd())),
width: 80, width: 80,
height: 24,
win: map[string]*speedWindow{}, win: map[string]*speedWindow{},
} }
} }
@@ -103,89 +108,251 @@ func (r *Reporter) render(final bool) {
} }
} }
var line string
switch len(stats) {
case 0:
if !final && r.tty {
return // nothing active; leave the line as-is
}
case 1:
line = r.lineOne(stats[0])
default:
line = r.lineMany(stats)
}
if r.tty { if r.tty {
r.width = termWidth(r.width) r.width = termWidth(r.width)
line = clampRunes(line, r.width) r.height = termHeight(r.height)
fmt.Fprint(r.w, "\r"+line+"\x1b[K") r.redraw(r.block(stats), final)
if final {
fmt.Fprintln(r.w)
}
return return
} }
// Not a TTY: can't redraw one line, so print sparingly (and always the // Not a TTY: no cursor control, so emit the summary line sparingly (and always
// final line) to keep redirected output and logs readable. // the final one) to keep redirected output and logs readable.
var line string
if len(stats) > 0 {
line = r.summary(stats)
}
if line != "" && (final || now.Sub(r.lastLog) >= logEvery) { if line != "" && (final || now.Sub(r.lastLog) >= logEvery) {
fmt.Fprintln(r.w, line) fmt.Fprintln(r.w, line)
r.lastLog = now r.lastLog = now
} }
} }
func (r *Reporter) lineOne(s download.Stat) string { // nameW is the fixed rune width of the name column in the table, matching
dl, ul := r.rates(s) // elide's cap so a padded name lines the columns up.
seeding := s.Status == download.Seeding const nameW = 20
var b strings.Builder
fmt.Fprintf(&b, "[%s ", shortName(s.Name)) // block builds the lines of the live display: a summary line, the column header,
// While seeding, report the share ratio in place of the size block and drop // and one row per download — or just the one-line summary when the table would
// the DL: field; otherwise show completed/total (or bare completed). // not fit the window. A single download renders the same way as several.
if seeding { func (r *Reporter) block(stats []download.Stat) []string {
fmt.Fprintf(&b, "SEED(%.1f)", ratio(s.Uploaded, s.Completed)) if len(stats) == 0 {
} else if s.Total > 0 { return nil
fmt.Fprintf(&b, "%s/%s(%d%%)", humanSize(s.Completed, r.human), humanSize(s.Total, r.human), percent(s.Completed, s.Total))
} else {
b.WriteString(humanSize(s.Completed, r.human))
} }
// CN is always shown (including HTTP); SD is shown for any torrent. if len(stats)+3 > r.height {
fmt.Fprintf(&b, " CN:%d", s.Conns) // The block is repainted in place by moving the cursor up over it, which
if s.IsBT { // breaks if printing the last row reaches the screen's bottom row and
fmt.Fprintf(&b, " SD:%d", s.Seeders) // scrolls the frame out from under the cursor. term height counts the row
// the shell prompt occupies, so a summary + header + one row each
// (len+2 lines) must still leave the bottom row free: fall back to the
// single summary line once len+2 would fill to the bottom.
return []string{r.summary(stats)}
} }
if !seeding { out := make([]string, 0, len(stats)+2)
fmt.Fprintf(&b, " DL:%s", speed(dl, r.human)) out = append(out, r.summary(stats), tableHeader())
for _, s := range stats {
out = append(out, r.tableRow(s))
} }
if seeding || ul > 0 { return out
fmt.Fprintf(&b, " UL:%s", speed(ul, r.human))
}
if !seeding && s.Total > 0 && dl > 0 {
eta := time.Duration(float64(s.Total-s.Completed)/float64(dl)) * time.Second
fmt.Fprintf(&b, " ETA:%s", secfmt(eta))
}
b.WriteByte(']')
return b.String()
} }
func (r *Reporter) lineMany(stats []download.Stat) string { // redraw repaints the block in place: move to the top of the previous frame,
// clear it, and print the new lines clamped to the terminal width. A stray line
// from stderr that lands in the block is painted over on the next tick.
func (r *Reporter) redraw(lines []string, final bool) {
if r.prevLines > 0 {
if r.prevLines > 1 {
fmt.Fprintf(r.w, "\x1b[%dA", r.prevLines-1) // up to the first line
}
fmt.Fprint(r.w, "\r\x1b[J") // column 0, clear to end of screen
}
for i, ln := range lines {
if i > 0 {
fmt.Fprint(r.w, "\n")
}
fmt.Fprint(r.w, clampRunes(ln, r.width))
}
r.prevLines = len(lines)
if final && len(lines) > 0 {
fmt.Fprintln(r.w) // leave the cursor below the block for the OK/FAIL lines
r.prevLines = 0
}
}
// summary is the header/aggregate line: a count of downloads in each state, the
// total down/up speed, and a stalled count when any download has gone quiet. The
// state breakdown replaces a single "active" tally so seeding and queued
// downloads are not lumped in with the ones actually moving bytes.
func (r *Reporter) summary(stats []download.Stat) string {
if len(stats) == 0 {
return ""
}
var totDL, totUL int64 var totDL, totUL int64
var downloading, seeding, queued, stalled int
for _, s := range stats { for _, s := range stats {
dl, ul := r.rates(s) dl, ul := r.rates(s)
totDL += dl totDL += dl
totUL += ul totUL += ul
} switch s.Status {
var b strings.Builder case download.Waiting:
fmt.Fprintf(&b, "[%d active DL:%s UL:%s]", len(stats), speed(totDL, r.human), speed(totUL, r.human)) queued++
for i, s := range stats { case download.Seeding:
if i >= 4 { seeding++
fmt.Fprintf(&b, "[+%d]", len(stats)-i) default:
break downloading++
} if r.isStalled(s, dl) {
if s.Total > 0 { stalled++
fmt.Fprintf(&b, "[%s %d%%]", shortName(s.Name), percent(s.Completed, s.Total))
} else {
fmt.Fprintf(&b, "[%s %s]", shortName(s.Name), humanSize(s.Completed, r.human))
} }
} }
return b.String() }
parts := []string{fmt.Sprintf("%d downloading", downloading)}
if seeding > 0 {
parts = append(parts, fmt.Sprintf("%d seeding", seeding))
}
if queued > 0 {
parts = append(parts, fmt.Sprintf("%d queued", queued))
}
line := strings.Join(parts, " · ") + fmt.Sprintf(" DL %s UL %s", r.rateStr(totDL), r.rateStr(totUL))
if stalled > 0 {
line += fmt.Sprintf(" (%d stalled)", stalled)
}
return line
}
// tableHeader names the columns once so the rows can drop the per-field DL:/UL:
// labels. It uses tableRow's exact format string so the headings line up over
// their columns. SIZE is completed/total; CN is the connection/peer count and SD
// the connected seeders — the fields that tell a stalled download (no peers)
// from a slow one (peers, no speed).
func tableHeader() string {
return fmt.Sprintf("%s %4s %-13s %-8s %3s %3s %s", padRune("NAME", nameW), "%", "SIZE", "RATE", "CN", "SD", "ETA")
}
// tableRow renders one download as aligned columns under tableHeader:
// "name pct size rate cn sd tail". The name is middle-elided to keep its
// distinguishing tail; percent, rate and the peer counts are fixed-width so the
// digits scan straight down; size is completed/total; the rate column carries
// the download speed or a state word (seeding/queued/...), and the tail carries
// the ETA, the share ratio while seeding, or a stalled flag.
func (r *Reporter) tableRow(s download.Stat) string {
dl, ul := r.rates(s)
rate, tail := r.rowMetric(s, dl, ul)
cn, sd := peerFields(s)
row := fmt.Sprintf("%s %4s %-13s %-8s %3s %3s %s", padRune(elide(s.Name, nameW), nameW), pctField(s), sizeField(s, r.human), rate, cn, sd, tail)
// A queued row's trailing SIZE/CN/SD/ETA columns are blank; drop the trailing
// spaces so the line ends where its content does.
return strings.TrimRight(row, " ")
}
// sizeField is the absolute-progress column: completed/total (e.g.
// "950MiB/5.0GiB"), just the completed bytes when the total is unknown (an HTTP
// stream with no Content-Length, or a pre-metadata magnet), or blank for a
// download that has not started.
func sizeField(s download.Stat, human bool) string {
switch {
case s.Status == download.Waiting:
return ""
case s.Total > 0:
return humanSize(s.Completed, human) + "/" + humanSize(s.Total, human)
case s.Completed > 0:
return humanSize(s.Completed, human)
default:
return ""
}
}
// peerFields formats the connection (CN) and seeder (SD) columns. SD is blank
// for HTTP, which has no seeders, and both are blank for a download with no live
// connections (queued, done, failed) so the columns aren't noise where they
// carry no meaning.
func peerFields(s download.Stat) (cn, sd string) {
switch s.Status {
case download.Active, download.Seeding:
cn = strconv.Itoa(s.Conns)
if s.IsBT {
sd = strconv.Itoa(s.Seeders)
}
}
return cn, sd
}
// rowMetric returns a table row's rate column and its tail for the download's
// state: the rate is the download speed or the state word (seeding/queued/done/
// failed); the tail is an ETA, the share ratio while seeding (the seed-stop
// criterion, more telling than an idle upload rate), or a "stalled" flag.
func (r *Reporter) rowMetric(s download.Stat, dl, ul int64) (rate, tail string) {
switch s.Status {
case download.Waiting:
return "queued", ""
case download.Seeding:
return "seeding", fmt.Sprintf("r%.2f", ratio(s.Uploaded, s.Completed))
case download.Complete:
return "done", ""
case download.Errored:
return "failed", ""
default:
rate = r.rateStr(dl)
switch {
case dl > 0 && s.Total > 0:
tail = secfmt(etaDuration(s.Total-s.Completed, dl))
case r.isStalled(s, dl):
tail = "stalled"
}
return rate, tail
}
}
// rateStr formats a bytes/sec speed for display, e.g. "388KiB/s".
func (r *Reporter) rateStr(bps int64) string {
return humanSize(bps, r.human) + "/s"
}
// maxETASeconds caps an ETA before it becomes a Duration: a near-stalled download
// yields an astronomically large seconds value, and time.Duration(secs) *
// time.Second would overflow int64 and wrap to a bogus tiny ETA. secfmt renders
// anything past 99h as "--", so clamping to 100h loses nothing.
const maxETASeconds = 100 * 3600
func etaDuration(remaining, dl int64) time.Duration {
secs := float64(remaining) / float64(dl)
if secs > maxETASeconds {
secs = maxETASeconds
}
return time.Duration(secs) * time.Second
}
// isStalled reports an active download that has gone a while with no new bytes,
// so the table can flag it rather than leave the eye to read DL:0B. A download
// too young to have a rate yet is not stalled.
func (r *Reporter) isStalled(s download.Stat, dl int64) bool {
if s.Status != download.Active || dl > 0 {
return false
}
w := r.win[s.ID]
if w == nil || len(w.samples) < 2 {
return false
}
return w.samples[len(w.samples)-1].t.Sub(w.samples[0].t) >= 4*time.Second
}
// pctField is the percent column: a number, full for seeding/complete, or "--"
// when the total is unknown.
func pctField(s download.Stat) string {
switch {
case s.Status == download.Seeding || s.Status == download.Complete:
return "100%"
case s.Total > 0:
return fmt.Sprintf("%3d%%", percent(s.Completed, s.Total))
default:
return " --"
}
}
// padRune right-pads s with spaces to w runes (it is never longer, having been
// through shortName) so table columns line up on rune count.
func padRune(s string, w int) string {
if n := utf8.RuneCountInString(s); n < w {
return s + strings.Repeat(" ", w-n)
}
return s
} }
// add records a sample and drops any that have fallen out of the window. A // add records a sample and drops any that have fallen out of the window. A
@@ -244,14 +411,24 @@ func ratio(uploaded, completed int64) float64 {
return float64(uploaded) / float64(completed) return float64(uploaded) / float64(completed)
} }
func shortName(s string) string { // elide shortens s to at most w runes, keeping the head and the tail joined by a
const max = 20 // single "…". Download names often share a long prefix (a release group, a
// Truncate on runes, not bytes, so a multibyte name (CJK, emoji) is never // series) and differ only at the end (an episode or volume number), so dropping
// cut mid-rune. // the middle keeps what tells two downloads apart — where a leading-only
if utf8.RuneCountInString(s) > max { // truncation would collapse them to the same string. A name within w prints
return string([]rune(s)[:max-1]) + "~" // whole. It works on runes, never bytes, so a multibyte glyph (CJK, emoji) is
} // never split.
func elide(s string, w int) string {
r := []rune(s)
if len(r) <= w {
return s return s
}
if w <= 1 {
return string(r[:w])
}
head := w / 2 // the w-1 kept runes split around one "…",
tail := w - 1 - head // head taking the extra rune when w is even
return string(r[:head]) + "…" + string(r[len(r)-tail:])
} }
// clampRunes limits a line to width runes (not bytes) so multibyte glyphs are // clampRunes limits a line to width runes (not bytes) so multibyte glyphs are
@@ -272,3 +449,13 @@ func termWidth(prev int) int {
} }
return 80 return 80
} }
func termHeight(prev int) int {
if _, h, err := term.GetSize(int(os.Stdout.Fd())); err == nil && h > 0 {
return h
}
if prev > 0 {
return prev
}
return 24
}

View File

@@ -1,9 +1,11 @@
package progress package progress
import ( import (
"bytes"
"strings" "strings"
"testing" "testing"
"time" "time"
"unicode/utf8"
"github.com/hanbok/got/download" "github.com/hanbok/got/download"
) )
@@ -76,52 +78,148 @@ func TestRatesKeyedByID(t *testing.T) {
} }
} }
// TestLineOneSeeding: while seeding, render SEED(ratio), drop DL:, keep UL:. func TestPctField(t *testing.T) {
func TestLineOneSeeding(t *testing.T) { if got := pctField(download.Stat{Status: download.Active, Total: 100, Completed: 50}); got != " 50%" {
r := newReporter(true) t.Errorf("active 50%% = %q, want ' 50%%'", got)
s := download.Stat{
ID: "h", Name: "f", IsBT: true, Status: download.Seeding,
Total: 1000, Completed: 1000, Uploaded: 1500, Conns: 3, Seeders: 2,
} }
line := r.lineOne(s) if got := pctField(download.Stat{Status: download.Seeding}); got != "100%" {
if !strings.Contains(line, "SEED(1.5)") { t.Errorf("seeding = %q, want 100%%", got)
t.Errorf("missing SEED(1.5): %q", line)
} }
if strings.Contains(line, "DL:") { if got := pctField(download.Stat{Status: download.Active, Total: -1}); got != " --" {
t.Errorf("DL: should be dropped while seeding: %q", line) t.Errorf("unknown total = %q, want ' --'", got)
}
if !strings.Contains(line, "UL:") {
t.Errorf("UL: should be kept while seeding: %q", line)
} }
} }
// TestLineOneCNSD: CN always shown; SD shown for torrents (even with 0 seeders), // TestRowMetric: the rate column and trailing field per download state.
// absent for HTTP. func TestRowMetric(t *testing.T) {
func TestLineOneCNSD(t *testing.T) {
r := newReporter(true) r := newReporter(true)
bt := r.lineOne(download.Stat{ID: "h", Name: "f", IsBT: true, Total: 10, Completed: 1, Conns: 4, Seeders: 0}) if rate, tail := r.rowMetric(download.Stat{Status: download.Active, Total: 100, Completed: 50}, 10, 0); !strings.HasSuffix(rate, "/s") || tail == "" {
if !strings.Contains(bt, "CN:4") || !strings.Contains(bt, "SD:0") { t.Errorf("active downloading: rate=%q tail=%q, want a …/s rate + an ETA", rate, tail)
t.Errorf("torrent line missing CN/SD: %q", bt)
} }
http := r.lineOne(download.Stat{ID: "h2", Name: "f", IsBT: false, Total: 10, Completed: 1, Conns: 1}) if rate, tail := r.rowMetric(download.Stat{Status: download.Seeding, Completed: 1000, Uploaded: 1500}, 0, 0); rate != "seeding" || tail != "r1.50" {
if !strings.Contains(http, "CN:1") { t.Errorf("seeding: rate=%q tail=%q, want seeding + r1.50 (share ratio)", rate, tail)
t.Errorf("http line missing CN: %q", http)
} }
if strings.Contains(http, "SD:") { if rate, _ := r.rowMetric(download.Stat{Status: download.Complete}, 0, 0); rate != "done" {
t.Errorf("http line should not show SD: %q", http) t.Errorf("complete rate=%q, want done", rate)
} }
} }
// TestShortNameRune ensures CJK names are truncated on runes, not bytes. // TestPeerFields: CN (connections/peers) and SD (seeders) show for active and
func TestShortNameRune(t *testing.T) { // seeding downloads; SD is blank for HTTP; both blank when there are no live
// connections (queued).
func TestPeerFields(t *testing.T) {
if cn, sd := peerFields(download.Stat{Status: download.Active, IsBT: true, Conns: 18, Seeders: 4}); cn != "18" || sd != "4" {
t.Errorf("active torrent: cn=%q sd=%q, want 18 / 4", cn, sd)
}
if cn, sd := peerFields(download.Stat{Status: download.Active, IsBT: false, Conns: 5}); cn != "5" || sd != "" {
t.Errorf("active http: cn=%q sd=%q, want 5 / blank", cn, sd)
}
if cn, sd := peerFields(download.Stat{Status: download.Waiting, Conns: 0}); cn != "" || sd != "" {
t.Errorf("queued: cn=%q sd=%q, want blank / blank", cn, sd)
}
}
// TestSizeField: completed/total when the total is known, just the completed
// bytes when it is not, and blank for a queued download.
func TestSizeField(t *testing.T) {
if got := sizeField(download.Stat{Status: download.Active, Completed: 512, Total: 2048}, false); got != "512B/2048B" {
t.Errorf("known total = %q, want 512B/2048B", got)
}
if got := sizeField(download.Stat{Status: download.Active, Completed: 512, Total: -1}, false); got != "512B" {
t.Errorf("unknown total = %q, want 512B", got)
}
if got := sizeField(download.Stat{Status: download.Waiting, Total: 2048}, false); got != "" {
t.Errorf("queued = %q, want blank", got)
}
}
// TestBlockTableAndFallback: several downloads render as summary+header+rows when
// they fit, and collapse to a single summary line when the window is too short.
// The fallback must leave the terminal's bottom row free (the stacked-summary
// redraw fix): summary+header+2 rows is 4 lines, so it needs height >= 5.
func TestBlockTableAndFallback(t *testing.T) {
r := newReporter(true)
stats := []download.Stat{
{ID: "a", Name: "a", Status: download.Active, Total: 100, Completed: 50},
{ID: "b", Name: "b", Status: download.Active, Total: 100, Completed: 10},
}
r.height = 24
if got := r.block(stats); len(got) != 4 { // summary + header + 2 rows
t.Errorf("table block = %d lines, want 4:\n%v", len(got), got)
}
// A single download renders the same way as several: summary + header + 1 row.
if got := r.block(stats[:1]); len(got) != 3 {
t.Errorf("single-download block = %d lines, want 3 (summary+header+row):\n%v", len(got), got)
}
r.height = 5 // 4 lines fit with the bottom row left free
if got := r.block(stats); len(got) != 4 {
t.Errorf("just-fits block = %d lines, want 4:\n%v", len(got), got)
}
r.height = 4 // one short: would fill to the bottom row, so fall back
if got := r.block(stats); len(got) != 1 {
t.Errorf("overflow block = %d lines, want 1 summary:\n%v", len(got), got)
}
}
// TestRenderRedraw: the TTY path draws the block, then on the next frame moves
// the cursor up over the previous frame and clears it before repainting.
func TestRenderRedraw(t *testing.T) {
snap := []download.Stat{
{ID: "a", Name: "alpha", IsBT: true, Status: download.Active, Total: 100, Completed: 50},
{ID: "b", Name: "beta", IsBT: true, Status: download.Seeding, Total: 100, Completed: 100, Uploaded: 200},
}
var buf bytes.Buffer
r := &Reporter{
snap: func() []download.Stat { return snap },
human: true, w: &buf, tty: true, width: 200, height: 24,
win: map[string]*speedWindow{},
}
r.render(false)
first := buf.String()
for _, want := range []string{"downloading", "seeding", "alpha", "NAME"} {
if !strings.Contains(first, want) {
t.Fatalf("first frame missing %q:\n%q", want, first)
}
}
if strings.Contains(first, "|") {
t.Errorf("first frame should carry no progress bar:\n%q", first)
}
buf.Reset()
r.render(false)
second := buf.String()
if !strings.Contains(second, "\x1b[3A") { // up prevLines-1 = 4-1 (summary+header+2 rows)
t.Errorf("second frame should move cursor up 3 lines:\n%q", second)
}
if !strings.Contains(second, "\x1b[J") {
t.Errorf("second frame should clear to end of screen:\n%q", second)
}
}
// TestElideKeepsTail: names sharing a long prefix but differing at the end stay
// distinct — the failure the old leading truncation caused — and the result is
// capped at the column width on runes with the distinguishing tail preserved.
func TestElideKeepsTail(t *testing.T) {
a := elide("[Group] Long Common Series Name vol.1", nameW)
b := elide("[Group] Long Common Series Name vol.2", nameW)
if a == b {
t.Errorf("names differing only in the tail collapsed: %q == %q", a, b)
}
if n := utf8.RuneCountInString(a); n != nameW {
t.Errorf("elide rune count = %d, want %d (%q)", n, nameW, a)
}
if !strings.Contains(a, "…") {
t.Errorf("expected a middle ellipsis: %q", a)
}
if !strings.HasSuffix(a, "vol.1") {
t.Errorf("expected the distinguishing tail kept: %q", a)
}
}
// TestElideRune ensures CJK names are elided on runes, not bytes.
func TestElideRune(t *testing.T) {
name := strings.Repeat("あ", 30) // 30 runes, 90 bytes name := strings.Repeat("あ", 30) // 30 runes, 90 bytes
got := shortName(name) if n := utf8.RuneCountInString(elide(name, nameW)); n != nameW {
runes := []rune(got) t.Errorf("elide rune count = %d, want %d", n, nameW)
if len(runes) != 20 {
t.Errorf("shortName rune count = %d, want 20", len(runes))
}
if runes[len(runes)-1] != '~' {
t.Errorf("expected trailing ~, got %q", got)
} }
} }