parseman 0.38.0 → 0.39.1

package/dist/spec/index.js

@@ -48,6 +48,8 @@ function matchesEmpty(p, seen = /* @__PURE__ */ new Set(), resolve) {
       return d.value.length === 0;
     case "keywords":
       return false;
+    case "routed":
+      return false;
     case "regex":
       try {
         const m = new RegExp(d.source).exec("");
@@ -74,6 +76,8 @@ function matchesEmpty(p, seen = /* @__PURE__ */ new Set(), resolve) {
       return d.parsers.every(me);
     case "choice":
       return d.parsers.some(me);
+    case "dispatch":
+      return me(d.selector);
     case "transform":
     case "label":
     case "trivia":
@@ -147,6 +151,8 @@ function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve) {
     case "keywords":
       return p._meta.firstSet;
     // terminals: no refs, cached set is exact
+    case "dispatch":
+      return fs(d.selector);
     case "lazy":
       try {
         return fs(d.thunk());
@@ -279,6 +285,12 @@ function markUnusedValues(root) {
       case "choice":
         for (const p of def.parsers) visit(p, consumed);
         return;
+      case "dispatch":
+        visit(def.selector, consumed);
+        for (const entry of def.cases) visit(entry.parser, consumed);
+        if (def.matchers) for (const entry of def.matchers) visit(entry.parser, consumed);
+        if (def.otherwise) visit(def.otherwise, consumed);
+        return;
       case "node":
         visit(def.parser, false);
         return;
@@ -689,6 +701,8 @@ function derive(c, seen) {
       return [def.label];
     case "choice":
       return def.parsers.flatMap(deriveExpected2);
+    case "dispatch":
+      return deriveExpected2(def.selector);
     // Union through the NULLABLE prefix, mirroring `sequenceFirstSet`. A leading
     // `optional(…)`/`many(…)` can match nothing, so the term after it is equally
     // able to fail first — deriving from term 0 alone named a token the parse
@@ -728,6 +742,8 @@ function derive(c, seen) {
     case "not":
     case "peek":
       return deriveExpected2(def.parser);
+    case "routed":
+      return ["routed()"];
     case "lazy": {
       const name = c._ruleName;
       if (seen.has(c)) return name ? [name] : [];
@@ -827,6 +843,7 @@ function choice(...args) {
           const result = parsers[idx].parse(input, pos, ctx);
           if (result.ok) return result;
           expected.push(...result.expected);
+          if (result.committed) return { ok: false, expected, span: { start: pos, end: pos }, committed: true };
           return { ok: false, expected, span: { start: pos, end: pos } };
         }
         return {
@@ -867,6 +884,7 @@ function choice(...args) {
           rollbackCstCapture(ctx, mark);
           if (logLen !== void 0 && ctx._triviaLog && ctx._triviaLog.length !== logLen) ctx._triviaLog.length = logLen;
           expected.push(...result.expected);
+          if (result.committed) return { ok: false, expected, span: result.span, committed: true };
           continue;
         }
         const checks = autoNot[i];
@@ -1817,6 +1835,7 @@ function repItem(combinator, input, cur, ctx, guardable) {
   const result = combinator.parse(input, pos, ctx);
   if (!result.ok) {
     rollbackTrivia(ctx, mark);
+    if (result.committed) return { fail: result, failPos: pos };
     return { fail: result, failPos: pos };
   }
   if (result.span.end === pos) {
@@ -1860,6 +1879,7 @@ function many(combinator, opts = {}) {
         const item = repItem(combinator, input, cur, ctx, guardable);
         if (item === "stop") break;
         if ("fail" in item) {
+          if (!item.fail.ok && item.fail.committed) return item.fail;
           const sync = ctx._tolerant ? ctx._sync : void 0;
           if (sync === void 0) break;
           if (matchesAt(sync, input, item.failPos, ctx)) break;
@@ -1904,6 +1924,7 @@ function atLeast(combinator, min, max) {
       while (count < min) {
         const item = repItem(combinator, input, cur, ctx, guardable);
         if (item === "stop" || "fail" in item) {
+          if (item !== "stop" && !item.fail.ok && item.fail.committed) return item.fail;
           expected ??= deriveExpected(combinator);
           return { ok: false, expected: expected.length > 0 ? expected : [combinator._tag], span: { start: cur, end: cur } };
         }
@@ -1916,6 +1937,7 @@ function atLeast(combinator, min, max) {
         const item = repItem(combinator, input, cur, ctx, guardable);
         if (item === "stop") break;
         if ("fail" in item) {
+          if (!item.fail.ok && item.fail.committed) return item.fail;
           const sync = ctx._tolerant ? ctx._sync : void 0;
           if (sync === void 0) break;
           if (matchesAt(sync, input, item.failPos, ctx)) break;
@@ -1954,6 +1976,7 @@ function optional(combinator) {
       const result = combinator.parse(input, pos, ctx);
       if (result.ok) return result;
       rollbackTrivia(ctx, mark);
+      if (result.committed) return result;
       return { ok: true, value: null, span: { start: pos, end: pos } };
     }
   };
@@ -1990,6 +2013,7 @@ function sepBy(combinator, separator, opts = {}) {
         values.push(first.value);
         cur = first.span.end;
       } else {
+        if (first.committed) return first;
         const term = ctx._tolerant ? ctx._sync : void 0;
         if (term === void 0 || matchesAt(term, input, pos, ctx)) {
           if (min >= 1) return failAt2(pos);
@@ -2017,6 +2041,7 @@ function sepBy(combinator, separator, opts = {}) {
         const sep = separator.parse(input, sepPos, ctx);
         if (!sep.ok) {
           rollbackTrivia(ctx, loopMark);
+          if (sep.committed) return sep;
           break;
         }
         const sepMark = trailing === "forbid" ? void 0 : saveTriviaMark(ctx);
@@ -2032,6 +2057,7 @@ function sepBy(combinator, separator, opts = {}) {
         }
         const next = combinator.parse(input, nextPos, ctx);
         if (!next.ok) {
+          if (next.committed) return next;
           const term = ctx._tolerant ? ctx._sync : void 0;
           if (term !== void 0 && !matchesAt(term, input, nextPos, ctx)) {
             expected ??= deriveExpected(combinator);
@@ -2265,6 +2291,8 @@ function parserHasOwnFields(p, seen = /* @__PURE__ */ new Set()) {
     case "sequence":
     case "choice":
       return d.parsers.some((x) => parserHasOwnFields(x, seen));
+    case "dispatch":
+      return parserHasOwnFields(d.selector, seen) || d.cases.some((x) => parserHasOwnFields(x.parser, seen)) || (d.matchers ? d.matchers.some((entry) => parserHasOwnFields(entry.parser, seen)) : false) || (d.otherwise ? parserHasOwnFields(d.otherwise, seen) : false);
     case "sepBy":
       return parserHasOwnFields(d.parser, seen) || parserHasOwnFields(d.separator, seen);
     case "skip":
@@ -2379,6 +2407,245 @@ function keywords(words, opts = {}) {
   };
 }
 
+// src/combinators/dispatch.ts
+function parserUsesRouted(parser2, seen = /* @__PURE__ */ new Set()) {
+  if (seen.has(parser2)) return false;
+  seen.add(parser2);
+  const def = parser2._def;
+  switch (def.tag) {
+    case "routed":
+      return true;
+    case "sequence":
+    case "choice":
+      return def.parsers.some((entry) => parserUsesRouted(entry, seen));
+    case "dispatch":
+      return false;
+    case "sepBy":
+      return parserUsesRouted(def.parser, seen) || parserUsesRouted(def.separator, seen);
+    case "skip":
+      return parserUsesRouted(def.main, seen) || parserUsesRouted(def.skipped, seen);
+    case "grammar":
+      return parserUsesRouted(def.parser, seen) || (def.triviaParser ? parserUsesRouted(def.triviaParser, seen) : false);
+    case "scanTo":
+      return parserUsesRouted(def.sentinel, seen) || def.skip.some((entry) => parserUsesRouted(entry, seen));
+    case "recover":
+      return parserUsesRouted(def.parser, seen) || parserUsesRouted(def.sentinel, seen);
+    case "lazy":
+      try {
+        return parserUsesRouted(def.thunk(), seen);
+      } catch {
+        return false;
+      }
+    case "many":
+    case "oneOrMore":
+    case "optional":
+    case "attempt":
+    case "transform":
+    case "trivia":
+    case "token":
+    case "leaf":
+    case "label":
+    case "field":
+    case "not":
+    case "peek":
+    case "node":
+    case "withCtx":
+    case "expect":
+      return parserUsesRouted(def.parser, seen);
+    default:
+      return false;
+  }
+}
+function branchUsesRouted(branch) {
+  return branch.usesRouted === true || parserUsesRouted(branch.parser);
+}
+function asciiFoldKey(key) {
+  let out = "";
+  for (let i = 0; i < key.length; i++) {
+    const c = key.charCodeAt(i);
+    out += String.fromCharCode(c >= 65 && c <= 90 ? c + 32 : c);
+  }
+  return out;
+}
+function matchesDispatchMatcher(value, matcher) {
+  const candidate = matcher.caseInsensitive ? asciiFoldKey(value) : value;
+  const expected = matcher.caseInsensitive ? asciiFoldKey(matcher.value) : matcher.value;
+  switch (matcher.kind) {
+    case "startsWith":
+      return candidate.startsWith(expected);
+    case "endsWith":
+      return candidate.endsWith(expected);
+    case "matches": {
+      const flags = matcher.caseInsensitive && !matcher.flags?.includes("i") ? `${matcher.flags ?? ""}i` : matcher.flags ?? "";
+      return new RegExp(matcher.value, flags).test(value);
+    }
+  }
+}
+function isStringMatcher(value) {
+  if (typeof value !== "object" || value === null) return false;
+  const kind = value.kind;
+  return kind === "startsWith" || kind === "endsWith" || kind === "matches";
+}
+function fixedMatcher(kind, value) {
+  if (typeof value !== "string") throw new TypeError(`parseman: ${kind}() value must be a string`);
+  if (value.length === 0) throw new RangeError(`parseman: ${kind}() requires a non-empty value`);
+  return { kind, value };
+}
+function startsWith(prefix) {
+  return fixedMatcher("startsWith", prefix);
+}
+function endsWith(suffix) {
+  return fixedMatcher("endsWith", suffix);
+}
+function matches(pattern) {
+  if (!(pattern instanceof RegExp)) throw new TypeError("parseman: matches() requires a RegExp");
+  if (pattern.flags.includes("g") || pattern.flags.includes("y")) {
+    throw new TypeError("parseman: matches() does not accept global or sticky regex flags");
+  }
+  return { kind: "matches", value: pattern.source, flags: pattern.flags };
+}
+function when(key, parser2, opts = {}) {
+  const usesRouted = parserUsesRouted(parser2);
+  if (isStringMatcher(key)) {
+    return { kind: "whenMatcher", matcher: key, parser: parser2, caseInsensitive: opts.caseInsensitive ?? false, usesRouted };
+  }
+  const keys = Array.isArray(key) ? [...key] : [key];
+  if (keys.length === 0) throw new RangeError("parseman: when() requires at least one key");
+  for (const item of keys) {
+    if (typeof item !== "string") throw new TypeError("parseman: when() keys must be strings");
+  }
+  return { kind: "when", keys, parser: parser2, caseInsensitive: opts.caseInsensitive ?? false, usesRouted };
+}
+function otherwise(parser2) {
+  return { kind: "otherwise", parser: parser2, usesRouted: parserUsesRouted(parser2) };
+}
+function routed() {
+  return {
+    _tag: "routed",
+    _meta: { firstSet: { kind: "any" }, canMatchNewline: false, isTrivia: false },
+    _def: { tag: "routed" },
+    parse(_input, pos, ctx) {
+      const item = ctx._routed;
+      if (item === void 0) {
+        return { ok: false, expected: ["routed()"], span: { start: pos, end: pos } };
+      }
+      if (pos !== item.span.start) {
+        return { ok: false, expected: ["routed()"], span: { start: pos, end: pos } };
+      }
+      if (cstCaptureActive(ctx)) {
+        pushCstLeaf(ctx, { _tag: "leaf", value: item.value, span: item.span });
+      }
+      return { ok: true, value: item.value, span: item.span };
+    }
+  };
+}
+function dispatch(selector, ...arms) {
+  if (parserUsesRouted(selector)) {
+    throw new Error("parseman: routed() can only appear inside a dispatch() branch");
+  }
+  let fallback;
+  let fallbackUsesRouted = false;
+  const matchers = [];
+  const cases = [];
+  const seen = [];
+  for (let i = 0; i < arms.length; i++) {
+    const arm = arms[i];
+    if (arm.kind === "otherwise") {
+      if (fallback !== void 0) throw new Error("parseman: dispatch() accepts at most one otherwise() arm");
+      if (i !== arms.length - 1) throw new Error("parseman: otherwise() must be the last dispatch() arm");
+      fallback = arm.parser;
+      fallbackUsesRouted = arm.usesRouted === true;
+      continue;
+    }
+    if (arm.kind === "whenMatcher") {
+      const matcher = arm.matcher;
+      matchers.push({
+        kind: matcher.kind,
+        value: matcher.value,
+        ...matcher.kind === "matches" ? { flags: matcher.flags } : {},
+        parser: arm.parser,
+        caseInsensitive: arm.caseInsensitive,
+        ...arm.usesRouted === true ? { usesRouted: true } : {}
+      });
+      continue;
+    }
+    for (const key of arm.keys) {
+      const folded = asciiFoldKey(key);
+      for (const prior of seen) {
+        const overlaps = arm.caseInsensitive || prior.caseInsensitive ? folded === prior.folded : key === prior.raw;
+        if (overlaps) throw new Error(`parseman: duplicate dispatch key ${JSON.stringify(key)}`);
+      }
+      seen.push({ raw: key, folded, caseInsensitive: arm.caseInsensitive });
+    }
+    cases.push({
+      keys: arm.keys,
+      parser: arm.parser,
+      caseInsensitive: arm.caseInsensitive,
+      ...arm.usesRouted === true ? { usesRouted: true } : {}
+    });
+  }
+  const meta = {
+    firstSet: selector._meta.firstSet,
+    canMatchNewline: selector._meta.canMatchNewline || cases.some((entry) => entry.parser._meta.canMatchNewline) || matchers.some((entry) => entry.parser._meta.canMatchNewline) || (fallback?._meta.canMatchNewline ?? false),
+    isTrivia: false
+  };
+  const byKey = /* @__PURE__ */ new Map();
+  const byFoldedKey = /* @__PURE__ */ new Map();
+  for (const entry of cases) {
+    for (const key of entry.keys) {
+      if (entry.caseInsensitive) byFoldedKey.set(asciiFoldKey(key), entry);
+      else byKey.set(key, entry);
+    }
+  }
+  return {
+    _tag: "dispatch",
+    _meta: meta,
+    _def: {
+      tag: "dispatch",
+      selector,
+      cases,
+      ...matchers.length === 0 ? {} : { matchers },
+      ...fallback === void 0 ? {} : { otherwise: fallback, ...fallbackUsesRouted ? { otherwiseUsesRouted: true } : {} }
+    },
+    parse(input, pos, ctx) {
+      if (parserUsesRouted(selector)) {
+        throw new Error("parseman: routed() can only appear inside a dispatch() branch");
+      }
+      const selectorMark = saveTriviaMark(ctx);
+      const selected = selector.parse(input, pos, ctx);
+      if (!selected.ok) return selected;
+      const branch = byKey.get(selected.value) ?? (byFoldedKey.size === 0 ? void 0 : byFoldedKey.get(asciiFoldKey(selected.value))) ?? matchers.find((matcher) => matchesDispatchMatcher(selected.value, matcher)) ?? (fallback === void 0 ? void 0 : { keys: [], parser: fallback, caseInsensitive: false, usesRouted: fallbackUsesRouted });
+      if (branch === void 0) {
+        const expected = cases.flatMap((entry) => entry.keys.map((key) => JSON.stringify(key)));
+        return { ok: false, expected, span: { start: selected.span.end, end: selected.span.end } };
+      }
+      const savedRouted = ctx._routed;
+      const usesRouted = branchUsesRouted(branch);
+      let mark = saveTriviaMark(ctx);
+      if (usesRouted) {
+        rollbackTrivia(ctx, selectorMark);
+        mark = saveTriviaMark(ctx);
+        ctx._routed = { value: selected.value, span: selected.span };
+      }
+      let result;
+      try {
+        result = branch.parser.parse(input, usesRouted ? pos : selected.span.end, ctx);
+      } finally {
+        if (usesRouted) ctx._routed = savedRouted;
+      }
+      if (!result.ok) {
+        rollbackTrivia(ctx, mark);
+        return { ...result, committed: true };
+      }
+      return {
+        ok: true,
+        value: [selected.value, result.value],
+        span: { start: pos, end: result.span.end }
+      };
+    }
+  };
+}
+
 // src/combinators/attempt.ts
 function attempt(parser2) {
   const guardable = parser2._meta.firstSet.kind !== "any" && !matchesEmpty(parser2);
@@ -2401,6 +2668,7 @@ function attempt(parser2) {
       if (result.ok) return result;
       rollbackCstCapture(ctx, mark);
       if (log !== void 0 && ctx._triviaLog && ctx._triviaLog.length !== log) ctx._triviaLog.length = log;
+      if (result.committed) return result;
       return { ok: false, expected: result.expected, span: { start: pos, end: pos } };
     }
   };
@@ -2810,6 +3078,13 @@ function evalRuleMapIR(ir) {
     "keywords",
     "sequence",
     "choice",
+    "dispatch",
+    "when",
+    "startsWith",
+    "endsWith",
+    "matches",
+    "otherwise",
+    "routed",
     "attempt",
     "many",
     "oneOrMore",
@@ -2843,6 +3118,13 @@ function evalRuleMapIR(ir) {
     keywords,
     sequence,
     choice,
+    dispatch,
+    when,
+    startsWith,
+    endsWith,
+    matches,
+    otherwise,
+    routed,
     attempt,
     many,
     oneOrMore,
@@ -3037,6 +3319,15 @@ var Builder = class {
         const items = def.parsers.map((p) => this.walk(p)).filter(nonEmpty);
         return flattenChoice(items);
       }
+      case "dispatch": {
+        const selector = this.walk(def.selector);
+        const tails = [
+          ...def.cases.map((entry) => this.walk(entry.parser)),
+          ...def.matchers ? def.matchers.map((entry) => this.walk(entry.parser)) : [],
+          ...def.otherwise ? [this.walk(def.otherwise)] : []
+        ];
+        return flattenSeq([selector, flattenChoice(tails)]);
+      }
       // `max`/`min` are carried through so a BOUNDED repeat does not render as an
       // unbounded one — the spec is generated to be read as the truth about what
       // parses, and `many(x, { min: 3, max: 8 })` is not `x+`.
@@ -3066,6 +3357,8 @@ var Builder = class {
         return { kind: "not", item: this.walk(def.parser) };
       case "peek":
         return { kind: "peek", item: this.walk(def.parser) };
+      case "routed":
+        return { kind: "empty" };
       // Transparent semantic wrappers — render the inner syntax.
       case "transform":
       case "token":
@@ -3223,7 +3516,7 @@ function render(node2) {
     case "choice": {
       const text = node2.items.map((it) => {
         const r = render(it);
-        return wrap(r.text, r.level, ALT);
+        return r.text.length === 0 ? "/* empty */" : wrap(r.text, r.level, ALT);
       }).join(" | ");
       return { text, level: ALT };
     }