parseman 0.50.6 → 0.51.0

package/dist/spec/index.js

@@ -31,7 +31,7 @@ function intersects(a, b) {
   return false;
 }
 function fromChar(code) {
-  return { kind: "ranges", ranges: [{ lo: code, hi: code }] };
+  return fromRange(code, code);
 }
 function startsFirstSet(combinator, input, pos) {
   const fs = combinator._meta.firstSet;
@@ -124,11 +124,10 @@ function matchesEmptyBody(p, seen, resolve) {
   }
 }
 function isZeroWidthAssertion(p) {
-  const tag = p._def.tag;
-  return tag === "not" || tag === "adjacency";
+  return p._tag === "not" || p._tag === "adjacency";
 }
 function isPositiveLookahead(p) {
-  return p._def.tag === "peek";
+  return p._tag === "peek";
 }
 function narrowBy(acc, constraint) {
   if (constraint.kind === "any") return acc;
@@ -144,34 +143,42 @@ function narrowBy(acc, constraint) {
   return ranges.length === 0 ? { kind: "empty" } : { kind: "ranges", ranges };
 }
 function applyAssertion(fs, assertion) {
-  if (assertion === null) return fs;
-  if (fs.kind === "empty") return assertion;
-  return narrowBy(fs, assertion) ?? fs;
+  return assertion === null ? fs : fs.kind === "empty" ? assertion : narrowBy(fs, assertion);
 }
-function sequenceFirstSet(parsers) {
-  let fs = empty();
-  let assertion = null;
-  for (const p of parsers) {
+function sequenceSet(parsers, fs, empties, trivia2) {
+  const afterBoundary = (i) => {
+    const rest = at(i);
+    return trivia2 ? union(rest, fs(trivia2)) : rest;
+  };
+  const at = (i) => {
+    const p = parsers[i];
     if (isPositiveLookahead(p)) {
-      assertion = narrowBy(assertion, p._meta.firstSet);
-      continue;
+      const constraint = fs(p);
+      return i + 1 === parsers.length ? constraint : applyAssertion(afterBoundary(i + 1), constraint);
     }
-    if (!isZeroWidthAssertion(p)) fs = union(fs, p._meta.firstSet);
-    if (!matchesEmpty(p)) return applyAssertion(fs, assertion);
-  }
-  return applyAssertion(fs, assertion);
+    if (isZeroWidthAssertion(p)) {
+      return i + 1 === parsers.length ? empty() : afterBoundary(i + 1);
+    }
+    const first = fs(p);
+    if (!empties(p) || i + 1 === parsers.length) return first;
+    return union(first, afterBoundary(i + 1));
+  };
+  return at(0);
 }
-function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve) {
+function sequenceFirstSet(parsers) {
+  return sequenceSet(parsers, (p) => p._meta.firstSet, matchesEmpty);
+}
+function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve, activeTrivia) {
   if (seen.has(p)) return any();
   seen.add(p);
   try {
-    return firstSetBody(p, seen, resolve);
+    return firstSetBody(p, seen, resolve, activeTrivia);
   } finally {
     seen.delete(p);
   }
 }
-function firstSetBody(p, seen, resolve) {
-  const fs = (c) => firstSetOf(c, seen, resolve);
+function firstSetBody(p, seen, resolve, activeTrivia) {
+  const fs = (c) => firstSetOf(c, seen, resolve, activeTrivia);
   const empties = (c) => matchesEmpty(c, /* @__PURE__ */ new Set(), resolve);
   const d = p._def;
   switch (d.tag) {
@@ -197,17 +204,7 @@ function firstSetBody(p, seen, resolve) {
       return out;
     }
     case "sequence": {
-      let out = empty();
-      let assertion = null;
-      for (const term of d.parsers) {
-        if (isPositiveLookahead(term)) {
-          assertion = narrowBy(assertion, fs(term));
-          continue;
-        }
-        if (!isZeroWidthAssertion(term)) out = union(out, fs(term));
-        if (!empties(term)) return applyAssertion(out, assertion);
-      }
-      return applyAssertion(out, assertion);
+      return sequenceSet(d.parsers, fs, empties, activeTrivia);
     }
     case "peek": {
       const inner = d.parser;
@@ -222,12 +219,16 @@ function firstSetBody(p, seen, resolve) {
     case "token":
     case "leaf":
     case "node":
-    case "grammar":
-    case "expect":
-      return fs(d.parser);
     case "sepBy":
+    case "expect":
       return fs(d.parser);
-    // both min 0 and min 1 start with the item
+    case "grammar":
+      return firstSetOf(
+        d.parser,
+        seen,
+        resolve,
+        d.clearTrivia ? null : d.triviaParser ?? activeTrivia
+      );
     default:
       return p._meta.firstSet;
   }
@@ -1705,7 +1706,12 @@ function childrenOf(def) {
     case "sepBy":
       return [def.parser, def.separator];
     case "scanTo":
-      return [def.sentinel, ...def.skip];
+      return [
+        def.sentinel,
+        ...def.skip,
+        ...def.recoverAt === void 0 ? [] : [def.recoverAt],
+        ...def.stopAt === void 0 ? [] : [def.stopAt]
+      ];
     case "routed":
       return def.fallback ? [def.fallback] : [];
     case "literal":
@@ -3750,7 +3756,7 @@ function parserUsesRouted(parser2, seen = /* @__PURE__ */ new Set()) {
     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));
+      return parserUsesRouted(def.sentinel, seen) || def.skip.some((entry) => parserUsesRouted(entry, seen)) || def.recoverAt !== void 0 && parserUsesRouted(def.recoverAt, seen) || def.stopAt !== void 0 && parserUsesRouted(def.stopAt, seen);
     case "recover":
       return parserUsesRouted(def.parser, seen) || parserUsesRouted(def.sentinel, seen);
     case "lazy":
@@ -4657,7 +4663,7 @@ function parserHasOwnFields(p, seen = /* @__PURE__ */ new Set()) {
     case "grammar":
       return parserHasOwnFields(d.parser, seen) || (d.triviaParser ? parserHasOwnFields(d.triviaParser, seen) : false);
     case "scanTo":
-      return parserHasOwnFields(d.sentinel, seen) || d.skip.some((x) => parserHasOwnFields(x, seen));
+      return parserHasOwnFields(d.sentinel, seen) || d.skip.some((x) => parserHasOwnFields(x, seen)) || d.recoverAt !== void 0 && parserHasOwnFields(d.recoverAt, seen) || d.stopAt !== void 0 && parserHasOwnFields(d.stopAt, seen);
     case "recover":
       return parserHasOwnFields(d.parser, seen) || parserHasOwnFields(d.sentinel, seen);
     case "routed":
@@ -5034,7 +5040,7 @@ function resolveScanSkip(explicitSkip, raw, ctx) {
   out.push(...explicitSkip);
   return out;
 }
-function scanTo(sentinel, { skip, raw = false, orEOF = false } = {}) {
+function scanTo(sentinel, { skip, raw = false, orEOF = false, recoverAt, stopAt } = {}) {
   const meta = {
     firstSet: any(),
     canMatchNewline: true,
@@ -5044,7 +5050,15 @@ function scanTo(sentinel, { skip, raw = false, orEOF = false } = {}) {
   return {
     _tag: "scanTo",
     _meta: meta,
-    _def: { tag: "scanTo", sentinel, skip: explicitSkip, raw, orEOF },
+    _def: {
+      tag: "scanTo",
+      sentinel,
+      skip: explicitSkip,
+      raw,
+      orEOF,
+      ...recoverAt === void 0 ? {} : { recoverAt },
+      ...stopAt === void 0 ? {} : { stopAt }
+    },
     parse(input, pos, ctx) {
       let cur = pos;
       const skip2 = resolveScanSkip(explicitSkip, raw, ctx);
@@ -5056,28 +5070,130 @@ function scanTo(sentinel, { skip, raw = false, orEOF = false } = {}) {
           pushCstLeaf(ctx, leaf2);
         }
       };
-      while (cur < input.length) {
-        const s = sentinel.parse(input, cur, probeCtx);
-        if (s.ok) {
+      if (recoverAt === void 0 && stopAt === void 0) {
+        while (cur < input.length) {
+          const s = sentinel.parse(input, cur, probeCtx);
+          if (s.ok) {
+            if (ctx.trackLines) recordLineRangeFromContext(ctx, input, pos, cur);
+            emit(cur);
+            return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
+          }
+          let advanced = false;
+          for (const skipper of skip2) {
+            const r = skipper.parse(input, cur, probeCtx);
+            if (r.ok && r.span.end > cur) {
+              cur = r.span.end;
+              advanced = true;
+              break;
+            }
+          }
+          if (!advanced) cur++;
+        }
+        if (orEOF) {
           if (ctx.trackLines) recordLineRangeFromContext(ctx, input, pos, cur);
           emit(cur);
           return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
         }
+        const sentDef2 = sentinel._def;
+        const expected2 = sentDef2.tag === "literal" ? [JSON.stringify(sentDef2.value)] : ["sentinel"];
+        return { ok: false, expected: expected2, span: { start: pos, end: cur } };
+      }
+      let recovery = -1;
+      let firstSentinel = -1;
+      let sentinelCount = 0;
+      let recoveryErrorCount = -1;
+      let sentinelErrorCount = -1;
+      const succeed = (end) => {
+        if (ctx.trackLines) recordLineRangeFromContext(ctx, input, pos, end);
+        emit(end);
+        return { ok: true, value: input.slice(pos, end), span: { start: pos, end } };
+      };
+      const recoveredEnd = () => {
+        const acceptSentinel = firstSentinel >= 0 && (sentinelCount & 1) === 1;
+        const errors = ctx._errors;
+        if (errors !== void 0) {
+          const keep = acceptSentinel ? sentinelErrorCount : recoveryErrorCount;
+          if (keep >= 0 && errors.length > keep) errors.length = keep;
+        }
+        return acceptSentinel ? firstSentinel : recovery;
+      };
+      while (cur < input.length) {
+        const sentinelProbeErrorCount = ctx._errors?.length ?? -1;
+        const s = sentinel.parse(input, cur, probeCtx);
+        const sentinelProbeErrors = ctx._errors;
+        if (sentinelProbeErrors !== void 0 && sentinelProbeErrorCount >= 0 && sentinelProbeErrors.length > sentinelProbeErrorCount) {
+          sentinelProbeErrors.length = sentinelProbeErrorCount;
+        }
+        if (s.ok) {
+          if (recovery < 0) return succeed(cur);
+          if (s.span.end <= cur) {
+            throw new TypeError("scanTo recovery requires a sentinel that consumes input");
+          }
+          if (firstSentinel < 0) {
+            firstSentinel = cur;
+            sentinelErrorCount = sentinelProbeErrorCount;
+          }
+          sentinelCount++;
+          cur = s.span.end;
+          continue;
+        }
+        if (stopAt !== void 0) {
+          const errorCount = ctx._errors?.length ?? -1;
+          const stop = stopAt.parse(input, cur, probeCtx);
+          const errors = ctx._errors;
+          if (errors !== void 0 && errorCount >= 0 && errors.length > errorCount) {
+            errors.length = errorCount;
+          }
+          if (stop.ok) {
+            if (recovery < 0) {
+              recovery = cur;
+              recoveryErrorCount = errorCount;
+            }
+            return succeed(recoveredEnd());
+          }
+        }
         let advanced = false;
         for (const skipper of skip2) {
+          const errorCount = ctx._errors?.length ?? -1;
           const r = skipper.parse(input, cur, probeCtx);
           if (r.ok && r.span.end > cur) {
             cur = r.span.end;
             advanced = true;
             break;
           }
+          const errors = ctx._errors;
+          if (errors !== void 0 && errorCount >= 0 && errors.length > errorCount) {
+            errors.length = errorCount;
+          }
+        }
+        if (advanced) continue;
+        if (recoverAt !== void 0) {
+          const errorCount = ctx._errors?.length ?? -1;
+          const recover = recoverAt.parse(input, cur, probeCtx);
+          const errors = ctx._errors;
+          if (errors !== void 0 && errorCount >= 0 && errors.length > errorCount) {
+            errors.length = errorCount;
+          }
+          if (recover.ok) {
+            if (recover.span.end <= cur) {
+              throw new TypeError("scanTo recoverAt must consume input");
+            }
+            if (recovery < 0) {
+              recovery = cur;
+              recoveryErrorCount = errorCount;
+            }
+            if (firstSentinel >= 0 && (sentinelCount & 1) === 1) {
+              return succeed(recoveredEnd());
+            }
+            cur = recover.span.end;
+            continue;
+          }
         }
-        if (!advanced) cur++;
+        cur++;
       }
+      if (recovery >= 0) return succeed(recoveredEnd());
       if (orEOF) {
-        if (ctx.trackLines) recordLineRangeFromContext(ctx, input, pos, cur);
-        emit(cur);
-        return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
+        return succeed(cur);
       }
       const sentDef = sentinel._def;
       const expected = sentDef.tag === "literal" ? [JSON.stringify(sentDef.value)] : ["sentinel"];