parseman 0.50.7 → 0.51.0

package/dist/index.js

@@ -3314,7 +3314,7 @@ 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":
@@ -4342,7 +4342,12 @@ 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":
@@ -4863,7 +4868,7 @@ 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":
@@ -5246,7 +5251,12 @@ function childrenOf2(d) {
     case "recover":
       return [d.parser, d.sentinel];
     case "scanTo":
-      return [d.sentinel, ...d.skip];
+      return [
+        d.sentinel,
+        ...d.skip,
+        ...d.recoverAt === void 0 ? [] : [d.recoverAt],
+        ...d.stopAt === void 0 ? [] : [d.stopAt]
+      ];
     case "grammar":
       return d.triviaParser ? [d.parser, d.triviaParser] : [d.parser];
     case "routed":
@@ -5893,7 +5903,7 @@ function hasNodeDef(p, seen = /* @__PURE__ */ new Set()) {
     case "sepBy":
       return hasNodeDef(d.parser, seen) || hasNodeDef(d.separator, seen);
     case "scanTo":
-      return hasNodeDef(d.sentinel, seen) || d.skip.some((x) => hasNodeDef(x, seen));
+      return hasNodeDef(d.sentinel, seen) || d.skip.some((x) => hasNodeDef(x, seen)) || d.recoverAt !== void 0 && hasNodeDef(d.recoverAt, seen) || d.stopAt !== void 0 && hasNodeDef(d.stopAt, seen);
     case "recover":
       return hasNodeDef(d.parser, seen) || hasNodeDef(d.sentinel, seen);
     case "expect":
@@ -5942,7 +5952,7 @@ function hasDirectBuildDef(p, seen = /* @__PURE__ */ new Set()) {
     case "sepBy":
       return hasDirectBuildDef(d.parser, seen) || hasDirectBuildDef(d.separator, seen);
     case "scanTo":
-      return hasDirectBuildDef(d.sentinel, seen) || d.skip.some((x) => hasDirectBuildDef(x, seen));
+      return hasDirectBuildDef(d.sentinel, seen) || d.skip.some((x) => hasDirectBuildDef(x, seen)) || d.recoverAt !== void 0 && hasDirectBuildDef(d.recoverAt, seen) || d.stopAt !== void 0 && hasDirectBuildDef(d.stopAt, seen);
     case "recover":
       return hasDirectBuildDef(d.parser, seen) || hasDirectBuildDef(d.sentinel, seen);
     case "routed":
@@ -5979,7 +5989,7 @@ function mayCommitFailure(p, seen = /* @__PURE__ */ new Set(), resolve) {
     case "withCtx":
       return mayCommitFailure(d.parser, seen, resolve);
     case "scanTo":
-      return mayCommitFailure(d.sentinel, seen, resolve) || d.skip.some((x) => mayCommitFailure(x, seen, resolve));
+      return mayCommitFailure(d.sentinel, seen, resolve) || d.skip.some((x) => mayCommitFailure(x, seen, resolve)) || d.recoverAt !== void 0 && mayCommitFailure(d.recoverAt, seen, resolve) || d.stopAt !== void 0 && mayCommitFailure(d.stopAt, seen, resolve);
     case "lazy": {
       try {
         return mayCommitFailure(d.thunk(), seen, resolve);
@@ -7885,6 +7895,8 @@ var Encoder = class {
         flags: (d.raw ? 1 : 0) | (d.orEOF ? 2 : 0),
         skip: d.skip.map((c) => this.subtree(c)),
         sentinel: this.subtree(d.sentinel),
+        ...d.recoverAt === void 0 ? {} : { recoverAt: this.subtree(d.recoverAt) },
+        ...d.stopAt === void 0 ? {} : { stopAt: this.subtree(d.stopAt) },
         sent: sentDef.tag === "literal" ? sentDef.value : null
       }));
     }
@@ -8318,6 +8330,8 @@ var Encoder = class {
     const refs = [];
     for (const s of this.scans) {
       if (s.sentinel !== void 0) refs.push(s.sentinel);
+      if (s.recoverAt !== void 0) refs.push(s.recoverAt);
+      if (s.stopAt !== void 0) refs.push(s.stopAt);
       refs.push(...s.skip);
     }
     for (const set of this.scanSkipSets) refs.push(...set);
@@ -8337,7 +8351,9 @@ var Encoder = class {
     this.scans = this.scans.map((s) => ({
       ...s,
       skip: s.skip.map(res),
-      ...s.sentinel === void 0 ? {} : { sentinel: res(s.sentinel) }
+      ...s.sentinel === void 0 ? {} : { sentinel: res(s.sentinel) },
+      ...s.recoverAt === void 0 ? {} : { recoverAt: res(s.recoverAt) },
+      ...s.stopAt === void 0 ? {} : { stopAt: res(s.stopAt) }
     }));
     this.scanSkipSets = this.scanSkipSets.map((set) => set.map(res));
   }
@@ -11584,7 +11600,7 @@ 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,
@@ -11594,7 +11610,15 @@ 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);
@@ -11606,28 +11630,130 @@ 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) cur++;
+        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;
+          }
+        }
+        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"];
@@ -11928,6 +12054,8 @@ function assemble(t, prog, cfg) {
   for (const s of prog.scans ?? []) {
     for (const r of s.skip) labelExtraIps.push(r[0]);
     if (s.sentinel !== void 0) labelExtraIps.push(s.sentinel[0]);
+    if (s.recoverAt !== void 0) labelExtraIps.push(s.recoverAt[0]);
+    if (s.stopAt !== void 0) labelExtraIps.push(s.stopAt[0]);
   }
   for (const set of prog.scanSkip ?? []) for (const r of set) labelExtraIps.push(r[0]);
   const closureLabels = computeSiteLabels(
@@ -14613,6 +14741,8 @@ function assemble(t, prog, cfg) {
     for (const s of prog.scans ?? []) {
       for (const r of s.skip) extraIps.push(r[0]);
       if (s.sentinel !== void 0) extraIps.push(s.sentinel[0]);
+      if (s.recoverAt !== void 0) extraIps.push(s.recoverAt[0]);
+      if (s.stopAt !== void 0) extraIps.push(s.stopAt[0]);
     }
     for (const set of prog.scanSkip ?? []) for (const r of set) extraIps.push(r[0]);
     const roots = [...Object.values(prog.rules), ...extraIps];
@@ -14824,7 +14954,13 @@ function assemble(t, prog, cfg) {
       continue;
     }
     const sentDef = typeof s.sent === "string" ? { tag: "literal", value: s.sent, caseInsensitive: false } : void 0;
-    scansArr.push(scanTo(subtreeComb(s.sentinel, sentDef), { skip, raw, orEOF: (s.flags & 2) !== 0 }));
+    scansArr.push(scanTo(subtreeComb(s.sentinel, sentDef), {
+      skip,
+      raw,
+      orEOF: (s.flags & 2) !== 0,
+      ...s.recoverAt === void 0 ? {} : { recoverAt: subtreeComb(s.recoverAt) },
+      ...s.stopAt === void 0 ? {} : { stopAt: subtreeComb(s.stopAt) }
+    }));
   }
   const scans = scansArr;
   const scanSkip = (prog.scanSkip ?? []).map((set) => set.map((r) => subtreeComb(r)));
@@ -15041,6 +15177,8 @@ function emitRef(r) {
 function emitScanSpec(s) {
   const parts = [`kind:${s.kind}`, `flags:${s.flags}`, `skip:[${s.skip.map(emitRef).join(",")}]`];
   if (s.sentinel !== void 0) parts.push(`sentinel:${emitRef(s.sentinel)}`);
+  if (s.recoverAt !== void 0) parts.push(`recoverAt:${emitRef(s.recoverAt)}`);
+  if (s.stopAt !== void 0) parts.push(`stopAt:${emitRef(s.stopAt)}`);
   if (s.sent !== void 0) parts.push(`sent:${s.sent === null ? "null" : jsString(s.sent)}`);
   if (s.open !== void 0) parts.push(`open:${jsString(s.open)}`);
   if (s.close !== void 0) parts.push(`close:${jsString(s.close)}`);
@@ -15067,6 +15205,8 @@ function emitAssemblies(prog, cfgs) {
   for (const s of prog.scans ?? []) {
     for (const r of s.skip) extraIps.push(r[0]);
     if (s.sentinel !== void 0) extraIps.push(s.sentinel[0]);
+    if (s.recoverAt !== void 0) extraIps.push(s.recoverAt[0]);
+    if (s.stopAt !== void 0) extraIps.push(s.stopAt[0]);
   }
   for (const set of prog.scanSkip ?? []) for (const r of set) extraIps.push(r[0]);
   const out = [];
@@ -15209,7 +15349,12 @@ function children(def, winners) {
     case "sepBy":
       return [def.parser, def.separator];
     case "scanTo":
-      return [def.sentinel, ...def.skip];
+      return [
+        def.sentinel,
+        ...def.skip,
+        ...def.recoverAt === void 0 ? [] : [def.recoverAt],
+        ...def.stopAt === void 0 ? [] : [def.stopAt]
+      ];
     case "lazy": {
       let resolved;
       try {
@@ -15322,7 +15467,7 @@ function payloadKey(p, d) {
     case "expect":
       return `expect\0${d.label ?? ""}\0${d.expected.join("
")}`;
     case "scanTo":
-      return `scanTo\0${d.raw}\0${d.orEOF}`;
+      return `scanTo\0${d.raw}\0${d.orEOF}\0${d.recoverAt !== void 0}\0${d.stopAt !== void 0}`;
     case "guard":
       return `guard\0${fnKey(d.predSrc, d.predicate)}`;
     case "adjacency":
@@ -15437,8 +15582,12 @@ function segment(d, index) {
       return index === 0 ? "sepBy.item" : "sepBy.sep";
     case "recover":
       return index === 0 ? "recover.body" : "recover.sentinel";
-    case "scanTo":
-      return index === 0 ? "scanTo.sentinel" : `scanTo.skip[${index - 1}]`;
+    case "scanTo": {
+      if (index === 0) return "scanTo.sentinel";
+      if (index <= d.skip.length) return `scanTo.skip[${index - 1}]`;
+      if (d.recoverAt !== void 0 && index === d.skip.length + 1) return "scanTo.recoverAt";
+      return "scanTo.stopAt";
+    }
     default:
       return d.tag;
   }
@@ -16714,7 +16863,12 @@ function childrenOf3(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 "lazy":
@@ -17208,7 +17362,7 @@ ${outerDecls.map((d) => "  " + d).join("\n")}
       case "expect":
         return `expect(${kid(def.parser)}${def.label !== void 0 ? `, ${JSON.stringify(def.label)}` : ""})`;
       case "scanTo":
-        return `scanTo(${kid(def.sentinel)}, { skip: [${def.skip.map(kid).join(", ")}]${def.raw ? ", raw: true" : ""}, orEOF: ${def.orEOF} })`;
+        return `scanTo(${kid(def.sentinel)}, { skip: [${def.skip.map(kid).join(", ")}]${def.raw ? ", raw: true" : ""}, orEOF: ${def.orEOF}${def.recoverAt === void 0 ? "" : `, recoverAt: ${kid(def.recoverAt)}`}${def.stopAt === void 0 ? "" : `, stopAt: ${kid(def.stopAt)}`} })`;
       case "transform": {
         if (def.fnSrc === void 0) throw new Unserializable("transform without fnSrc");
         return `_tf(${kid(def.parser)}, ${JSON.stringify(def.fnSrc)}${def.recognitionOnly ? ", true" : ""})`;
@@ -17388,7 +17542,7 @@ function compileRuleMap(ruleMap, opts = {}) {
 }
 
 // src/version.ts
-var PARSEMAN_VERSION = "0.50.7";
+var PARSEMAN_VERSION = "0.51.0";
 
 // src/compiler/compile-linkable-table.ts
 function isLocal(rule) {
@@ -18454,8 +18608,12 @@ function slotLabel(d, index) {
       return index === 0 ? "sepBy.item" : "sepBy.sep";
     case "recover":
       return index === 0 ? "recover.parser" : "recover.sentinel";
-    case "scanTo":
-      return index === 0 ? "scanTo.sentinel" : `scanTo.skip[${index - 1}]`;
+    case "scanTo": {
+      if (index === 0) return "scanTo.sentinel";
+      if (index <= d.skip.length) return `scanTo.skip[${index - 1}]`;
+      if (d.recoverAt !== void 0 && index === d.skip.length + 1) return "scanTo.recoverAt";
+      return "scanTo.stopAt";
+    }
     case "grammar":
       return index === 0 ? "grammar" : "grammar.trivia";
     case "node":
@@ -20121,7 +20279,13 @@ function coverageEntry(entry, collector, maps, trace) {
         case "expect":
           return expect(build(def.parser), def.label);
         case "scanTo":
-          return scanTo(build(def.sentinel), { skip: def.skip.map(build), orEOF: def.orEOF });
+          return scanTo(build(def.sentinel), {
+            skip: def.skip.map(build),
+            raw: def.raw,
+            orEOF: def.orEOF,
+            ...def.recoverAt === void 0 ? {} : { recoverAt: build(def.recoverAt) },
+            ...def.stopAt === void 0 ? {} : { stopAt: build(def.stopAt) }
+          });
         // Zero-width assertion with no coverage-bearing child; reuse it verbatim so
         // the rebuilt sequence still recognises the boundary marker.
         case "adjacency":