ww/lib/fnmatch/fnmatch.ww

// fnmatch — shell wildcard pattern matching. Port of Hare's
// fnmatch:: (ref/hare/fnmatch/fnmatch.ha) using ww byte-indexed
// cursors in place of Hare's strings::iterator.
//
// Surface today:
//
//   fnmatch.flag           — enum i32 bitmask (NONE / PATHNAME /
//                            NOESCAPE / PERIOD)
//   fnmatch.fnmatch(pattern: str, string: str, flags: flag) bool
//
// Matching rules (Hare-spec):
//
//   - '?'          matches any single byte
//   - '*'          matches any byte sequence (including empty)
//   - '['…']'      bracket expression. Inside: byte ranges (`a-z`),
//                  POSIX character classes ([:alnum:], [:alpha:],
//                  …), leading '!' negates. Hare-style only: '^'
//                  is *not* accepted as a negation marker; it
//                  yields invalid.
//   - '\\<c>'      escape — `<c>` is taken literally. Disabled by
//                  flag.NOESCAPE, in which case '\\' is itself a
//                  literal.
//   - any other    matches itself.
//
// Flags (see [[flag]]):
//
//   - PATHNAME: '/' in the string is matched only by a literal '/'
//     in the pattern (segments delimited by '/' are matched
//     independently with PATHNAME stripped).
//   - NOESCAPE: '\\' loses its escape meaning.
//   - PERIOD:   a leading '.' (or — under PATHNAME — a '.'
//     immediately following a '/') must be matched by a literal
//     '.' in the pattern, not by '?' / '*' / '[…]'.
//
// Divergence from Hare:
//
//   - ww has no UTF-8 rune iteration on str; matching is byte-wise.
//     Pattern and string bytes compare directly, bracket ranges
//     `[a-z]` operate over byte values (ASCII-only meaningful;
//     multibyte UTF-8 only matches byte-identically, not
//     codepoint-equivalently). Same precedent as
//     [[strings.byteindex]]'s rune-needle arm (which matches a
//     single ASCII byte). When ww grows UTF-8 iteration, this
//     module graduates "in one go" (lib/CLAUDE.md) and the
//     byte-only caveat retires.
//
//   - Hare returns invalid up the public surface via `?` so callers
//     can distinguish bad patterns from non-matches. ww collapses
//     invalid → false at the public boundary, matching Hare's
//     fnmatch.ha:53 (`return b is bool && b: bool;`).
//
//   - Hare's algorithm uses strings::iterator (with `next` / `prev`
//     / `riter` / `slice` / `iterstr`). ww has none of these
//     yet; the port models a cursor as a bare `i32` byte index
//     into the source `str` and threads `(pat, *i32)` through
//     pat_next / match_bracket / match_ctype. The tail-match phase
//     uses a fresh forward cursor at `string.len - cnt` instead of
//     a reverse iterator — produces the same byte sequence.
//
//   - Hare's match_ctype builds a `[name, *fn(rune) bool]` table
//     and dispatches via fn-pointer call; ww uses a sequence of
//     `streq` comparisons routing to the matching [[ascii.is*]]
//     predicate. Equivalent, narrower lowering.
//
// Owning model: pure function — no allocation, no mutation of
// caller-owned data. Pattern and string are read-only byte sources;
// the returned `bool` is the only outward effect.
//
//   if (fnmatch.fnmatch("*.c", "foo.c", fnmatch.flag.NONE)) { … };
//   if (fnmatch.fnmatch("a/*.c", "a/x.c",
//        fnmatch.flag.PATHNAME)) { … };

package fnmatch;

import ascii;
import strings;

// flag — bitmask altering match semantics. Stored as `enum i32`
// to match [[os.flag]] / [[temp.mode]]; the four named values are
// small unsigned bit positions where i32-vs-u32 makes no
// observable difference. Bitwise AND/OR composition lives in the
// public boundary, which casts to i32 once and back.
export type flag = enum i32 {
	NONE     = 0,
	PATHNAME = 1,
	NOESCAPE = 2,
	PERIOD   = 4,
};

// invalid — internal sentinel for a malformed pattern (bad bracket,
// unterminated escape, unknown POSIX class, …). Module-private:
// Hare exposes this to callers via `?`, ww collapses it to `false`
// at the public boundary (mirrors Hare's `fnmatch` line 53). Once
// callers need pattern-validity diagnostics, a `try`-shaped public
// entry can be added without churning the existing surface.
type invalid = !void;

// Token tags returned by [[pat_next]]. void aliases so they can sit
// alongside `u8` (a literal pattern byte) and `invalid` in one
// tagged-union return. Same shape as [[io.eof]] / [[io.closed]].
type star     = void;
type question = void;
type bracket  = void;
type endt     = void;

// pat_next — consume one pattern element starting at `*pos`.
// Returns the matching token tag for metacharacters (`*` / `?` /
// `[`), `endt` at exhaustion, the literal byte otherwise. Under
// flag.NOESCAPE, `\\` is itself a literal byte; otherwise `\\<c>`
// returns `<c>` and an unterminated `\\` returns `invalid`.
// Mirrors Hare's pat_next (fnmatch.ha:313).
fn pat_next(pat: str, pos: *i32, fl: flag) (u8 | star | question | bracket | endt | invalid) = {
	if (*pos >= pat.len) { let e: endt; return e; };
	let c: u8 = pat[*pos];
	*pos += 1;
	if (c == 42u8) { let r: star;     return r; };	// '*'
	if (c == 63u8) { let r: question; return r; };	// '?'
	if (c == 91u8) { let r: bracket;  return r; };	// '['
	if (c == 92u8) {					// '\\'
		if (((fl as i32) & (flag.NOESCAPE as i32)) == 0) {
			if (*pos >= pat.len) { let e: invalid; return e; };
			let c2: u8 = pat[*pos];
			*pos += 1;
			return c2;
		};
		return 92u8;
	};
	return c;
};

// advance_or_err — read one byte from `s` at `*pos`, advancing.
// Returns invalid on exhaustion. Mirrors Hare's advance_or_err
// (fnmatch.ha:336), narrowed to bytes.
fn advance_or_err(s: str, pos: *i32) (u8 | invalid) = {
	if (*pos >= s.len) { let e: invalid; return e; };
	let c: u8 = s[*pos];
	*pos += 1;
	return c;
};

// streq — byte-equal compare. ascii is the only character set we
// fold against; the POSIX class names are pure ASCII, so a flat
// length-checked memcmp matches Hare's behavior here.
fn streq(a: str, b: str) bool = {
	if (a.len != b.len) { return false; };
	let i: i32 = 0;
	for (i < a.len) {
		if (a[i] != b[i]) { return false; };
		i += 1;
	};
	return true;
};

// match_ctype — parse a "[:name:]" character class. `*pos` points
// just past the leading colon on entry; the function consumes
// through the closing "]" and returns whether `c` is a member of
// the named class. Mirrors Hare's match_ctype (fnmatch.ha:284),
// modulo: ascii.is* are dispatched by direct name compare instead
// of through a fn-pointer table.
fn match_ctype(pat: str, pos: *i32, c: u8) (bool | invalid) = {
	let nameoff: i32 = *pos;
	let i: i32 = 0;
	let r: u8 = 0u8;
	for (r != 58u8) {				// ':'
		let rr = advance_or_err(pat, pos);
		match (rr) {
		case let e: invalid => return e;
		case let f: u8 => { r = f; };
		};
		if (!ascii.valid(r: rune)) {
			let e: invalid; return e;
		};
		i += 1;
	};
	let rr = advance_or_err(pat, pos);
	match (rr) {
	case let e: invalid => return e;
	case let f: u8 => {
		if (f != 93u8) { let e: invalid; return e; };	// ']'
	};
	};
	let name: str;
	name.ptr = pat.ptr + (nameoff: u64);
	name.len = i - 1;
	let cr: rune = c: rune;
	if (streq(name, "alnum"))  { return ascii.isalnum(cr); };
	if (streq(name, "alpha"))  { return ascii.isalpha(cr); };
	if (streq(name, "blank"))  { return ascii.isblank(cr); };
	if (streq(name, "cntrl"))  { return ascii.iscntrl(cr); };
	if (streq(name, "digit"))  { return ascii.isdigit(cr); };
	if (streq(name, "graph"))  { return ascii.isgraph(cr); };
	if (streq(name, "lower"))  { return ascii.islower(cr); };
	if (streq(name, "print"))  { return ascii.isprint(cr); };
	if (streq(name, "punct"))  { return ascii.ispunct(cr); };
	if (streq(name, "space"))  { return ascii.isspace(cr); };
	if (streq(name, "upper"))  { return ascii.isupper(cr); };
	if (streq(name, "xdigit")) { return ascii.isxdigit(cr); };
	let e: invalid; return e;
};

// match_bracket — consume one `[…]` group at `*pos` and decide
// whether `c` is a member. Mirrors Hare's match_bracket
// (fnmatch.ha:220).
//
// On entry `*pos` is positioned just past the leading '['; on
// successful return it sits just past the closing ']'.
fn match_bracket(pat: str, pos: *i32, c: u8) (bool | invalid) = {
	let oldpos: i32 = *pos;
	let r0 = advance_or_err(pat, pos);
	let first: u8 = 0u8;
	match (r0) {
	case let e: invalid => return e;
	case let f: u8 => { first = f; };
	};
	if (first == 94u8) { let e: invalid; return e; };	// '^' is Hare-invalid
	let inv: bool = false;
	if (first == 33u8) {					// '!'
		inv = true;
		let r1 = advance_or_err(pat, pos);
		match (r1) {
		case let e: invalid => return e;
		case let f: u8 => { first = f; };
		};
	};
	let found: bool = (first != 91u8) && (first == c);
	let havelast: bool = true;
	let last: u8 = first;
	if (first == 93u8) {					// ']' as first member
		let r2 = advance_or_err(pat, pos);
		match (r2) {
		case let e: invalid => return e;
		case let f: u8 => { first = f; };
		};
	};
	let r: u8 = first;
	let loop: bool = true;
	for (loop) {
		if (r == 93u8) {				// ']' closes
			loop = false;
		} else { if (r == 45u8) {			// '-' range
			let er = advance_or_err(pat, pos);
			let endv: u8 = 0u8;
			match (er) {
			case let e: invalid => return e;
			case let f: u8 => { endv = f; };
			};
			if (endv == 93u8) {
				// Trailing '-' matches itself. Un-eat ']'
				// so the next iteration sees it.
				*pos -= 1;
				last = 45u8;
				havelast = true;
				if (c == 45u8) { found = true; };
			} else {
				if (!havelast) { let e: invalid; return e; };
				if ((last: u32) <= (c: u32) && (c: u32) <= (endv: u32)) {
					found = true;
				};
				havelast = false;	// forbid `a-f-n`
			};
		} else { if (r == 91u8) {			// '['
			let nx = advance_or_err(pat, pos);
			let nxv: u8 = 0u8;
			match (nx) {
			case let e: invalid => return e;
			case let f: u8 => { nxv = f; };
			};
			if (nxv == 61u8) { let e: invalid; return e; };	// '='
			if (nxv == 46u8) { let e: invalid; return e; };	// '.'
			if (nxv == 58u8) {				// ':' class
				let t = match_ctype(pat, pos, c);
				match (t) {
				case let e: invalid => return e;
				case let v: bool => {
					if (v) { found = true; };
				};
				};
			} else {
				// Not a class — un-eat the byte after '['.
				*pos -= 1;
				if (c == 91u8) { found = true; };
			};
			last = 91u8;
			havelast = true;
		} else {					// literal byte
			if (c == r) { found = true; };
			last = r;
			havelast = true;
		}; }; };
		if (loop) {
			let rr = advance_or_err(pat, pos);
			match (rr) {
			case let e: invalid => return e;
			case let f: u8 => { r = f; };
			};
		};
	};
	// Hare's degeneracy check: `[xyx]` where the bracket spans
	// exactly `[X..X]` for X in `=`, `.`, `:` is invalid (would
	// have been an empty collating/equivalence/class).
	let cnt: i32 = *pos - oldpos;
	if (havelast && first == last && cnt >= 4) {
		if (first == 61u8) { let e: invalid; return e; };
		if (first == 46u8) { let e: invalid; return e; };
		if (first == 58u8) { let e: invalid; return e; };
	};
	if (inv) { return !found; };
	return found;
};

// fnmatch_internal — core "sea of stars" matcher. Drives the same
// three-phase algorithm as Hare (fnmatch.ha:101):
//
//   1. Exact match against pattern up to the first '*' (prefix).
//   2. Exact match of the remaining pattern after the last '*'
//      against the corresponding tail of the string.
//   3. Greedy left-to-right match of each pattern segment between
//      stars against the leftover middle of the string.
//
// Step (3) has no exponential corner because each star anchors a
// "match found" at increasing string positions.
fn fnmatch_internal(pattern: str, string: str, fl: flag) (bool | invalid) = {
	if (((fl as i32) & (flag.PERIOD as i32)) != 0) {
		if (strings.hasprefix(string, ".") && !strings.hasprefix(pattern, ".")) {
			return false;
		};
	};
	let pp: i32 = 0;
	let sp: i32 = 0;

	// ---- prefix: match up to the first '*' ----------------------
	let sawstar: bool = false;
	for (!sawstar) {
		let scur: i32 = sp;
		let isdone: bool = (scur >= string.len);
		let curs: u8 = 0u8;
		if (!isdone) { curs = string[scur]; scur += 1; };

		let t = pat_next(pattern, &pp, fl);
		let advance: bool = true;
		match (t) {
		case let e: invalid => return e;
		case star => { sawstar = true; advance = false; };
		case endt => { return isdone; };
		case question => {
			if (isdone) { return false; };
		};
		case bracket => {
			if (isdone) { return false; };
			let mr = match_bracket(pattern, &pp, curs);
			match (mr) {
			case let e: invalid => return e;
			case let b: bool => { if (!b) { return false; }; };
			};
		};
		case let r: u8 => {
			if (isdone) { return false; };
			if (curs != r) { return false; };
		};
		};
		if (advance) { sp = scur; };
	};

	// ---- find the tail (token count after the last '*') ---------
	let pp_copy: i32 = pp;
	let pp_last: i32 = pp;
	let pp_last_cnt: i32 = 0;
	let cnt: i32 = 0;
	let pdone: bool = false;
	for (!pdone) {
		let t = pat_next(pattern, &pp, fl);
		match (t) {
		case let e: invalid => return e;
		case endt => { pdone = true; };
		case star => {
			pp_last = pp;
			pp_last_cnt = cnt + 1;
			cnt += 1;
		};
		case bracket => {
			let mr = match_bracket(pattern, &pp, 0u8);
			match (mr) {
			case let e: invalid => return e;
			case let _b: bool => { };
			};
			cnt += 1;
		};
		case question => { cnt += 1; };
		case let r: u8 => { cnt += 1; };
		};
	};
	pp = pp_last;
	let tail_cnt: i32 = cnt - pp_last_cnt;

	// String has to leave room for the post-star tail; the prefix
	// already consumed up to sp.
	let sp_copy: i32 = sp;
	let tail_pos: i32 = string.len - tail_cnt;
	if (tail_pos < sp_copy) { return false; };

	// ---- tail: match (from pp) against string[tail_pos..] -------
	let ts: i32 = tail_pos;
	let tdone: bool = false;
	for (!tdone) {
		let t = pat_next(pattern, &pp, fl);
		let havec: bool = (ts < string.len);
		let ch: u8 = 0u8;
		if (havec) { ch = string[ts]; };
		match (t) {
		case let e: invalid => return e;
		case endt => {
			if (havec) { return false; };
			tdone = true;
		};
		case star => {
			// Unreachable — pp starts after pp_last (past the
			// last star), so pat_next can't yield another star.
			// Treat defensively as invalid pattern.
			let e: invalid; return e;
		};
		case question => {
			if (!havec) { return false; };
			ts += 1;
		};
		case bracket => {
			if (!havec) { return false; };
			let mr = match_bracket(pattern, &pp, ch);
			match (mr) {
			case let e: invalid => return e;
			case let b: bool => { if (!b) { return false; }; };
			};
			ts += 1;
		};
		case let r: u8 => {
			if (!havec) { return false; };
			if (ch != r) { return false; };
			ts += 1;
		};
		};
	};

	// ---- middle: greedy match of each star-delimited subpattern --
	let mid_pat: str;
	mid_pat.ptr = pattern.ptr + (pp_copy: u64);
	mid_pat.len = pp_last - pp_copy;
	let mid_str: str;
	mid_str.ptr = string.ptr + (sp_copy: u64);
	mid_str.len = tail_pos - sp_copy;

	let mpp_copy: i32 = 0;
	let msp_copy: i32 = 0;

	let outer: bool = true;
	for (outer) {
		let mpp: i32 = mpp_copy;
		if (mpp >= mid_pat.len) { return true; };
		let msp: i32 = msp_copy;
		let inner: bool = true;
		let sawstarm: bool = false;
		for (inner) {
			let scur: i32 = msp;
			let isdone: bool = (scur >= mid_str.len);
			let cur: u8 = 0u8;
			if (!isdone) { cur = mid_str[scur]; scur += 1; };

			let t = pat_next(mid_pat, &mpp, fl);
			let matched: bool = false;
			let advance: bool = true;
			match (t) {
			case let e: invalid => return e;
			case endt => {
				// mid_pat by construction never ends with end
				// here (the last token was a star); treat as
				// invalid defensively.
				let e: invalid; return e;
			};
			case question => { matched = !isdone; };
			case bracket => {
				if (isdone) { matched = false; }
				else {
					let mr = match_bracket(mid_pat, &mpp, cur);
					match (mr) {
					case let e: invalid => return e;
					case let b: bool => { matched = b; };
					};
				};
			};
			case let r: u8 => {
				matched = (!isdone) && (cur == r);
			};
			case star => {
				mpp_copy = mpp;
				msp_copy = msp;
				inner = false;
				sawstarm = true;
				advance = false;
			};
			};
			if (advance) {
				if (!matched) {
					inner = false;
				} else {
					msp = scur;
				};
			};
		};
		if (!sawstarm) {
			// Inner failed — advance the outer string anchor by
			// one byte and retry from the top of mid_pat.
			if (msp_copy >= mid_str.len) { return false; };
			msp_copy += 1;
		};
	};
	return false;
};

// fnmatch_pathname — PATHNAME variant: split pattern and string on
// '/' and match each segment independently with [[fnmatch_internal]].
// Mirrors Hare's fnmatch_pathname (fnmatch.ha:58).
fn fnmatch_pathname(pattern: str, string: str, fl: flag) (bool | invalid) = {
	let pp: i32 = 0;
	let segstart_pat: i32 = 0;
	let sp: i32 = 0;
	let donetokens: bool = false;
	let final: bool = false;
	for (!final) {
		segstart_pat = pp;
		// Walk pattern until the next '/' (segment break) or end.
		let inner: bool = true;
		let kind: i32 = 0;	// 0 = saw '/', 1 = saw end
		for (inner) {
			let t = pat_next(pattern, &pp, fl);
			match (t) {
			case let e: invalid => return e;
			case endt => { kind = 1; inner = false; };
			case let r: u8 => {
				if (r == 47u8) { kind = 0; inner = false; };
			};
			case bracket => {
				let mr = match_bracket(pattern, &pp, 0u8);
				match (mr) {
				case let e: invalid => return e;
				case let _b: bool => { };
				};
			};
			case question => { };
			case star => { };
			};
		};
		// Pull the next string token (Hare's strings::next_token
		// on `/`).
		if (donetokens) { return false; };
		let segstart_s: i32 = sp;
		for (sp < string.len) {
			if (string[sp] == 47u8) { break; };
			sp += 1;
		};
		let seg: str;
		seg.ptr = string.ptr + (segstart_s: u64);
		seg.len = sp - segstart_s;
		let hadslash: bool = (sp < string.len);
		if (hadslash) { sp += 1; }
		else { donetokens = true; };

		let p_slice: str;
		if (kind == 0) {
			// pp is just past the '/'; slice excludes it.
			p_slice.ptr = pattern.ptr + (segstart_pat: u64);
			p_slice.len = (pp - 1) - segstart_pat;
		} else {
			// kind == 1: end of pattern; this is the final
			// segment-pattern. Whole tail of pattern from
			// segstart_pat is the slice.
			p_slice.ptr = pattern.ptr + (segstart_pat: u64);
			p_slice.len = pp - segstart_pat;
			final = true;
		};
		let r = fnmatch_internal(p_slice, seg, fl);
		match (r) {
		case let e: invalid => return e;
		case let b: bool => {
			if (!b) { return false; };
		};
		};
		if (final) {
			// Final segment: require string also exhausted
			// (Hare's `&& strings::next_token(&tok) is done`).
			if (!donetokens) { return false; };
			return true;
		};
	};
	return false;
};

// fnmatch — public entry. Selects PATHNAME-split or core matcher
// based on `flags`; invalid pattern → false (Hare's collapsing
// semantic, fnmatch.ha:53).
export fn fnmatch(pattern: str, string: str, flags: flag) bool = {
	let r: (bool | invalid);
	if (((flags as i32) & (flag.PATHNAME as i32)) != 0) {
		r = fnmatch_pathname(pattern, string, flags);
	} else {
		r = fnmatch_internal(pattern, string, flags);
	};
	match (r) {
	case invalid => return false;
	case let b: bool => return b;
	};
};