Files
ww/lib/bufio/bufiotest.ww
Hojun-Cho 036b2c851f lib+test: graduate bufio to Hare scanner on io.stream
Scanner subset: newscanner, finish, scanbyte, scanline, scantok,
overflow (named-void). Buffer caller-owned; EOF_DISCARD default.
Skips Hare's scanner-as-io.stream embed pending task #6 (chained
N_DOT-of-N_DOT miscompile).
2026-05-13 20:38:55 +09:00

414 lines
11 KiB
Plaintext

// bufiotest — exercises lib/bufio. Run with `out/bin/ww run lib/bufio/bufiotest.ww`.
//
// Each @test enumerates parallel `[N]T` arrays of inputs and
// expectations, then iterates one body across them. Parallel arrays
// (rather than `[N]struct{...}`) match the same cgen workaround
// memiotest leans on.
use bufio;
use bytes;
use io;
use memio;
// Direct exit(2) binding rather than `use os;` — os exports
// read/write/close, which collide with io.read/write/close under
// the driver's flat-scope concat.
@symbol("rt_syscall") fn syscall1ww(num: i64, a: i64) i64;
fn doexit(code: i32) void = {
syscall1ww(60i64, code: i64);
};
// signalled — bumped by main before each test so a failing exit code
// pinpoints the offending case.
let signalled: i32 = 0;
fn fail() void = { doexit(signalled + 10); };
// putstr — copy the bytes of `s` into `into` starting at `off`,
// returning the new offset.
fn putstr(s: str, into: []u8, off: i32) i32 = {
let i: i32 = 0;
for (i < s.len) {
into[off + i] = s[i];
i += 1;
};
return off + s.len;
};
// ---- A closed-on-read stream for the io.closed surfacing test ---------
fn closedread(s: *io.stream, buf: []u8) (i32 | io.eof | io.closed) = {
let e: io.closed; return e;
};
fn closedwrite(s: *io.stream, buf: []u8) (i32 | io.closed) = {
let e: io.closed; return e;
};
fn closedclose(s: *io.stream) (void | io.closed) = { return; };
fn closedstream(s: *io.stream) void = {
s.ctx = nil;
s.read = closedread;
s.write = closedwrite;
s.close = closedclose;
};
// ---- scanbyte: drain four bytes through a 2-byte window ---------------
// Reading buf is smaller than the stream contents (2 vs 4) — the
// scanner has to refill twice. Row 4 is one past EOF, row 5 stays at
// EOF (idempotent).
@test fn scanbytecases() void = {
let raw: [4]u8;
raw[0] = 11u8; raw[1] = 22u8; raw[2] = 33u8; raw[3] = 44u8;
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, raw[0:4]);
let buf: [2]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:2]);
// (wantbyte, weof)
let want: [6]u8;
let weof: [6]i32;
want[0]=11u8; weof[0]=0;
want[1]=22u8; weof[1]=0;
want[2]=33u8; weof[2]=0;
want[3]=44u8; weof[3]=0;
want[4]=0u8; weof[4]=1; // past end → eof
want[5]=0u8; weof[5]=1; // stay at eof
let i: i32 = 0;
for (i < 6) {
let r: (u8 | io.eof | io.closed) = bufio.scanbyte(&sc);
match (r) {
case let b: u8 => {
if (weof[i] != 0) { fail(); };
if (b != want[i]) { fail(); };
};
case io.eof => { if (weof[i] == 0) { fail(); }; };
case io.closed => fail();
};
i += 1;
};
bufio.finish(&sc);
};
// ---- scanline: 3 lines + tail fragment + extra calls stay at EOF ------
@test fn scanlinecases() void = {
let src: [32]u8;
let n: i32 = putstr("foo\nbar\nbaz\ntrailing", src[0:32], 0);
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, src[0:n]);
let buf: [32]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:32]);
// (wantlen, wantfirst, weof, wovr)
// Row 0..2: terminated lines.
// Row 3: tail "trailing" — EOF_DISCARD drops it, returns eof.
// Row 4: subsequent call stays at eof.
let wl: [5]i32;
let wf: [5]u8;
let weof: [5]i32;
let wovr: [5]i32;
wl[0]=3; wf[0]=102u8; weof[0]=0; wovr[0]=0; // foo
wl[1]=3; wf[1]=98u8; weof[1]=0; wovr[1]=0; // bar
wl[2]=3; wf[2]=98u8; weof[2]=0; wovr[2]=0; // baz
wl[3]=0; wf[3]=0u8; weof[3]=1; wovr[3]=0; // trailing → eof
wl[4]=0; wf[4]=0u8; weof[4]=1; wovr[4]=0; // stay eof
let i: i32 = 0;
for (i < 5) {
let r: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc);
match (r) {
case let v: str => {
if (weof[i] != 0) { fail(); };
if (wovr[i] != 0) { fail(); };
if (v.len != wl[i]) { fail(); };
if (v.len > 0 && v[0] != wf[i]) { fail(); };
};
case io.eof => { if (weof[i] == 0) { fail(); }; };
case io.closed => fail();
case bufio.overflow => { if (wovr[i] == 0) { fail(); }; };
};
i += 1;
};
bufio.finish(&sc);
};
// ---- scanline: line longer than buffer → overflow --------------------
@test fn scanlineoverflow() void = {
let src: [32]u8;
let n: i32 = putstr("ABCDEFGHI\n", src[0:32], 0);
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, src[0:n]);
// Buffer too small for the 9-byte payload + delim. Caller
// owns the budget; overflow signals "raise it".
let buf: [4]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:4]);
let r: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc);
match (r) {
case let v: str => fail();
case io.eof => fail();
case io.closed => fail();
case bufio.overflow => { };
};
bufio.finish(&sc);
};
// ---- scantok: multi-delim hit, delim-at-start, EOF before delim ------
@test fn scantokcases() void = {
let src: [16]u8;
// ",a,,b,c" — leading delim, empty token, multi-hit, trailing fragment.
let n: i32 = putstr(",a,,b,c", src[0:16], 0);
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, src[0:n]);
let buf: [8]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:8]);
// (wantlen, wantfirst, weof)
// Row 0: leading "," → empty token.
// Row 1: "a" before next ",".
// Row 2: empty token between consecutive ",,".
// Row 3: "b".
// Row 4: "c" — no trailing delim → EOF_DISCARD drops + eof.
let wl: [5]i32;
let wf: [5]u8;
let weof: [5]i32;
wl[0]=0; wf[0]=0u8; weof[0]=0;
wl[1]=1; wf[1]=97u8; weof[1]=0; // 'a'
wl[2]=0; wf[2]=0u8; weof[2]=0;
wl[3]=1; wf[3]=98u8; weof[3]=0; // 'b'
wl[4]=0; wf[4]=0u8; weof[4]=1; // 'c' discarded
let i: i32 = 0;
for (i < 5) {
let r: ([]u8 | io.eof | io.closed | bufio.overflow) = bufio.scantok(&sc, 44u8); // ','
match (r) {
case let v: []u8 => {
if (weof[i] != 0) { fail(); };
if (v.len != wl[i]) { fail(); };
if (v.len > 0 && v[0] != wf[i]) { fail(); };
};
case io.eof => { if (weof[i] == 0) { fail(); }; };
case io.closed => fail();
case bufio.overflow => fail();
};
i += 1;
};
bufio.finish(&sc);
};
// ---- empty stream: every variant returns eof immediately --------------
@test fn emptystream() void = {
let raw: [1]u8;
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, raw[0:0]);
let buf: [8]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:8]);
let r1: (u8 | io.eof | io.closed) = bufio.scanbyte(&sc);
match (r1) {
case let b: u8 => fail();
case io.eof => { };
case io.closed => fail();
};
let r2: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc);
match (r2) {
case let v: str => fail();
case io.eof => { };
case io.closed => fail();
case bufio.overflow => fail();
};
let r3: ([]u8 | io.eof | io.closed | bufio.overflow) = bufio.scantok(&sc, 10u8);
match (r3) {
case let v: []u8 => fail();
case io.eof => { };
case io.closed => fail();
case bufio.overflow => fail();
};
bufio.finish(&sc);
};
// ---- io.closed surfacing on a stream that returns closed --------------
@test fn closedsource() void = {
let m: io.stream;
closedstream(&m);
let buf: [8]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:8]);
let r1: (u8 | io.eof | io.closed) = bufio.scanbyte(&sc);
match (r1) {
case let b: u8 => fail();
case io.eof => fail();
case io.closed => { };
};
let r2: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc);
match (r2) {
case let v: str => fail();
case io.eof => fail();
case io.closed => { };
case bufio.overflow => fail();
};
let r3: ([]u8 | io.eof | io.closed | bufio.overflow) = bufio.scantok(&sc, 32u8);
match (r3) {
case let v: []u8 => fail();
case io.eof => fail();
case io.closed => { };
case bufio.overflow => fail();
};
bufio.finish(&sc);
};
// ---- boundary: empty line + idempotent EOF on scantok ----------------
// scanline on a leading '\n' must return an empty (len=0) view, not
// skip the empty line and pretend it's "the first non-empty line".
// scantok after EOF_DISCARD must keep returning io.eof on every
// subsequent call — same idempotency guaranteed by scanbyte rows 4/5
// and scanline rows 3/4, but the scantok path filled at the bottom
// of `for (true)` is its own state machine. Worth pinning.
@test fn boundarycases() void = {
// ---- empty line via scanline -----------------------------------
let s1: [16]u8;
let n1: i32 = putstr("\nfoo\n", s1[0:16], 0);
let m1mem: memio.state;
let m1: io.stream;
memio.fixed(&m1mem, &m1, s1[0:n1]);
let b1: [8]u8;
let sc1: bufio.scanner;
bufio.newscanner(&sc1, &m1, b1[0:8]);
let r1: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc1);
match (r1) {
case let v: str => { if (v.len != 0) { fail(); }; };
case io.eof => fail();
case io.closed => fail();
case bufio.overflow => fail();
};
let r2: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc1);
match (r2) {
case let v: str => {
if (v.len != 3) { fail(); };
if (v[0] != 102u8) { fail(); }; // 'f'
};
case io.eof => fail();
case io.closed => fail();
case bufio.overflow => fail();
};
bufio.finish(&sc1);
// ---- scantok idempotent EOF -----------------------------------
// "ab" with delim ',' — no delim, EOF_DISCARD drops "ab", first
// call returns eof. The next call must too (state stays at eof).
let s2: [8]u8;
let n2: i32 = putstr("ab", s2[0:8], 0);
let m2mem: memio.state;
let m2: io.stream;
memio.fixed(&m2mem, &m2, s2[0:n2]);
let b2: [8]u8;
let sc2: bufio.scanner;
bufio.newscanner(&sc2, &m2, b2[0:8]);
let t1: ([]u8 | io.eof | io.closed | bufio.overflow) = bufio.scantok(&sc2, 44u8);
match (t1) {
case let v: []u8 => fail();
case io.eof => { };
case io.closed => fail();
case bufio.overflow => fail();
};
let t2: ([]u8 | io.eof | io.closed | bufio.overflow) = bufio.scantok(&sc2, 44u8);
match (t2) {
case let v: []u8 => fail();
case io.eof => { };
case io.closed => fail();
case bufio.overflow => fail();
};
bufio.finish(&sc2);
};
// ---- multi-fill: window smaller than longest token, shift path --------
// Buffer 4B, line "abcd\nxy\n". After reading "abcd", a delim search
// finds nothing, the buffer is full at start=0 → overflow. So bump
// to 5B (room for "abcd" + '\n'). Reads come in 5-byte chunks; the
// second line "xy" fits trivially and exercises the shift path.
@test fn multifill() void = {
let src: [32]u8;
let n: i32 = putstr("abcd\nxy\n", src[0:32], 0);
let mem: memio.state;
let m: io.stream;
memio.fixed(&mem, &m, src[0:n]);
let buf: [5]u8;
let sc: bufio.scanner;
bufio.newscanner(&sc, &m, buf[0:5]);
// (wantlen, wantfirst, weof)
let wl: [3]i32;
let wf: [3]u8;
let weof: [3]i32;
wl[0]=4; wf[0]=97u8; weof[0]=0; // abcd
wl[1]=2; wf[1]=120u8; weof[1]=0; // xy
wl[2]=0; wf[2]=0u8; weof[2]=1; // eof
let i: i32 = 0;
for (i < 3) {
let r: (str | io.eof | io.closed | bufio.overflow) = bufio.scanline(&sc);
match (r) {
case let v: str => {
if (weof[i] != 0) { fail(); };
if (v.len != wl[i]) { fail(); };
if (v.len > 0 && v[0] != wf[i]) { fail(); };
};
case io.eof => { if (weof[i] == 0) { fail(); }; };
case io.closed => fail();
case bufio.overflow => fail();
};
i += 1;
};
bufio.finish(&sc);
};
export fn main() i32 = {
signalled = 1; scanbytecases();
signalled = 2; scanlinecases();
signalled = 3; scanlineoverflow();
signalled = 4; scantokcases();
signalled = 5; emptystream();
signalled = 6; closedsource();
signalled = 7; boundarycases();
signalled = 8; multifill();
return 0;
};