parseman 0.34.0 → 0.36.0

package/dist/index.js

@@ -335,28 +335,28 @@ function rollbackBufList(b, keyMulti, keySingle, len) {
     else if (len === 1) {
       b[keySingle] = arr[0];
       b[keyMulti] = void 0;
-    } else arr.length = len;
+    } else if (arr.length !== len) arr.length = len;
     return;
   }
   if (len === 0) b[keySingle] = void 0;
 }
 function rollbackCstCapture(ctx, mark) {
-  if (ctx._errors && mark.errors !== void 0) ctx._errors.length = mark.errors;
+  if (ctx._errors && mark.errors !== void 0 && ctx._errors.length !== mark.errors) ctx._errors.length = mark.errors;
   const b = ctx._cstBuf;
   if (b) {
     rollbackBufList(b, "raw", "rawSingle", mark.raw);
     rollbackBufList(b, "ch", "single", mark.leaves);
     if (b.tl) {
       if (mark.tlog === 0) b.tl = void 0;
-      else b.tl.length = mark.tlog;
+      else if (b.tl.length !== mark.tlog) b.tl.length = mark.tlog;
     }
-    if (ctx._fields) ctx._fields.length = mark.fields;
+    if (ctx._fields && ctx._fields.length !== mark.fields) ctx._fields.length = mark.fields;
     return;
   }
-  if (ctx._cstRawChildren) ctx._cstRawChildren.length = mark.raw;
-  if (ctx._cstTriviaLog) ctx._cstTriviaLog.length = mark.tlog;
-  if (ctx._cstLeaves) ctx._cstLeaves.length = mark.leaves;
-  if (ctx._fields) ctx._fields.length = mark.fields;
+  if (ctx._cstRawChildren && ctx._cstRawChildren.length !== mark.raw) ctx._cstRawChildren.length = mark.raw;
+  if (ctx._cstTriviaLog && ctx._cstTriviaLog.length !== mark.tlog) ctx._cstTriviaLog.length = mark.tlog;
+  if (ctx._cstLeaves && ctx._cstLeaves.length !== mark.leaves) ctx._cstLeaves.length = mark.leaves;
+  if (ctx._fields && ctx._fields.length !== mark.fields) ctx._fields.length = mark.fields;
 }
 function pushCstTriviaEntry(ctx, start, end, kindIndex) {
   const insertIdx = cstRawLen(ctx);
@@ -1049,6 +1049,10 @@ var REC = { scan: recoverScan, at: matchesAt, or: orSentinel, sentinel: firstSet
 
 // src/combinators/expect.ts
 function deriveExpected(c) {
+  return [...new Set(derive(c, /* @__PURE__ */ new Set()))];
+}
+function derive(c, seen) {
+  const deriveExpected2 = (p) => derive(p, seen);
   const def = c._def;
   switch (def.tag) {
     case "literal":
@@ -1060,20 +1064,32 @@ function deriveExpected(c) {
     case "label":
       return [def.label];
     case "choice":
-      return def.parsers.flatMap(deriveExpected);
-    case "sequence":
-      return def.parsers.length > 0 ? deriveExpected(def.parsers[0]) : [];
+      return def.parsers.flatMap(deriveExpected2);
+    // 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
+    // never actually required. A leading `not(…)` is zero-width and contributes
+    // nothing (its own expectation is about what must NOT be here), but it is
+    // nullable, so keep scanning past it. Stop at the first term that must match.
+    case "sequence": {
+      const out = [];
+      for (const term of def.parsers) {
+        if (!isZeroWidthAssertion(term)) out.push(...deriveExpected2(term));
+        if (!matchesEmpty(term)) break;
+      }
+      return out;
+    }
     case "attempt":
-      return deriveExpected(def.parser);
+      return deriveExpected2(def.parser);
     // Delegating wrappers whose first-token failure is their inner parser's — must
     // pass through so a start-failure (and the first-set-miss fast-path in attempt/
     // node, which reads this) names the real expected token, not the wrapper tag.
     case "field":
     case "withCtx":
     case "recover":
-      return deriveExpected(def.parser);
+      return deriveExpected2(def.parser);
     case "skip":
-      return deriveExpected(def.main);
+      return deriveExpected2(def.main);
     case "expect":
       return def.expected;
     case "node":
@@ -1087,13 +1103,17 @@ function deriveExpected(c) {
     case "transform":
     case "not":
     case "peek":
-      return deriveExpected(def.parser);
+      return deriveExpected2(def.parser);
     case "lazy": {
       const name = c._ruleName;
+      if (seen.has(c)) return name ? [name] : [];
+      seen.add(c);
       try {
-        return deriveExpected(def.thunk());
+        return deriveExpected2(def.thunk());
       } catch {
         return name ? [name] : [];
+      } finally {
+        seen.delete(c);
       }
     }
     default:
@@ -1224,14 +1244,14 @@ function choice(...args) {
         const result = parsers[i].parse(input, pos, ctx);
         if (!result.ok) {
           rollbackCstCapture(ctx, mark);
-          if (logLen !== void 0 && ctx._triviaLog) ctx._triviaLog.length = logLen;
+          if (logLen !== void 0 && ctx._triviaLog && ctx._triviaLog.length !== logLen) ctx._triviaLog.length = logLen;
           expected.push(...result.expected);
           continue;
         }
         const checks = autoNot[i];
         if (checks && autoNotFires(input, result.span.end, checks)) {
           rollbackCstCapture(ctx, mark);
-          if (logLen !== void 0 && ctx._triviaLog) ctx._triviaLog.length = logLen;
+          if (logLen !== void 0 && ctx._triviaLog && ctx._triviaLog.length !== logLen) ctx._triviaLog.length = logLen;
           continue;
         }
         return result;
@@ -1558,7 +1578,7 @@ function saveTriviaMark(ctx) {
 }
 function rollbackTrivia(ctx, mark) {
   rollbackCstCapture(ctx, { raw: mark.raw, tlog: mark.tlog, leaves: mark.leaves, fields: mark.fields, errors: mark.errors });
-  if (ctx._triviaLog) ctx._triviaLog.length = mark.log;
+  if (ctx._triviaLog && ctx._triviaLog.length !== mark.log) ctx._triviaLog.length = mark.log;
 }
 function saveLookaheadMark(ctx) {
   return { ...saveTriviaMark(ctx), probeBest: ctx._probe ? ctx._probe.best : null };
@@ -1926,7 +1946,7 @@ function attempt(parser2) {
       const result = parser2.parse(input, pos, ctx);
       if (result.ok) return result;
       rollbackCstCapture(ctx, mark);
-      if (log !== void 0 && ctx._triviaLog) ctx._triviaLog.length = log;
+      if (log !== void 0 && ctx._triviaLog && ctx._triviaLog.length !== log) ctx._triviaLog.length = log;
       return { ok: false, expected: result.expected, span: { start: pos, end: pos } };
     }
   };
@@ -3000,7 +3020,7 @@ function leaf(root, fn) {
 }
 
 // src/version.ts
-var PARSEMAN_VERSION = "0.34.0";
+var PARSEMAN_VERSION = "0.36.0";
 
 // src/analysis/gating.ts
 var isAny = (fs) => fs.kind === "any";
@@ -5191,9 +5211,9 @@ function planDisjointDispatch(parsers) {
   return total >= SWITCH_MIN_CASES ? { kind: "switch", cases } : { kind: "if" };
 }
 function captureRestoreBody(mL, mR, mTl, mLg, mF = null) {
-  const fieldRestore = mF ? `; if (_ctx._fields) _ctx._fields.length = ${mF}` : "";
-  const base = `if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${mL}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${mR}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${mTl}${fieldRestore}`;
-  return mLg ? `${base}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${mLg}` : base;
+  const fieldRestore = mF ? `; if (_ctx._fields && _ctx._fields.length !== ${mF}) _ctx._fields.length = ${mF}` : "";
+  const base = `if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${mL}) _ctx._cstLeaves.length = ${mL}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${mR}) _ctx._cstRawChildren.length = ${mR}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${mTl}) _ctx._cstTriviaLog.length = ${mTl}${fieldRestore}`;
+  return mLg ? `${base}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${mLg}) _ctx._triviaLog.length = ${mLg}` : base;
 }
 function mayLeavePartialCapture(p, seen = /* @__PURE__ */ new Set()) {
   if (seen.has(p)) return false;
@@ -5555,7 +5575,7 @@ function emitSeqValues(def, ctx, pos) {
         const endAfterV = v(ctx, "_sea");
         stmts.push(
           `${ind(ctx)}const ${endAfterV} = ${r2.endVar}`,
-          `${ind(ctx)}if (${endAfterV} > ${scanEndV}) { ${curV} = ${endAfterV} } else { if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${markLog}; }`
+          `${ind(ctx)}if (${endAfterV} > ${scanEndV}) { ${curV} = ${endAfterV} } else { if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${markV}) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${markTl}) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${markLog}) _ctx._triviaLog.length = ${markLog}; }`
         );
         valueVars.push(r2.valueVar);
         continue;
@@ -6042,7 +6062,7 @@ function emitMany(def, ctx, pos) {
         `${ind(ctx)}const ${npV} = ${capFn}(input, ${curV}, _ctx, 1)`
       );
       itemPos = npV;
-      rollback = `if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${markLog}; `;
+      rollback = `if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${markV}) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${markTl}) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${markLog}) _ctx._triviaLog.length = ${markLog}; `;
     } else {
       const trivFn = ensureTriviaFn(ctx);
       const npV = v(ctx, "_np");
@@ -6105,7 +6125,7 @@ function emitAttempt(p, def, ctx, pos) {
   const log = v(ctx, "_atg");
   const fields = v(ctx, "_atf");
   const errors = v(ctx, "_ate");
-  const rollback = `if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${leaves}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${raw}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${trivia2}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${log}; if (_ctx._fields) _ctx._fields.length = ${fields}; if (_ctx._errors) _ctx._errors.length = ${errors}`;
+  const rollback = `if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${leaves}) _ctx._cstLeaves.length = ${leaves}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${raw}) _ctx._cstRawChildren.length = ${raw}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${trivia2}) _ctx._cstTriviaLog.length = ${trivia2}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${log}) _ctx._triviaLog.length = ${log}; if (_ctx._fields && _ctx._fields.length !== ${fields}) _ctx._fields.length = ${fields}; if (_ctx._errors && _ctx._errors.length !== ${errors}) _ctx._errors.length = ${errors}`;
   const traceId = ctx.coverage?.plan.attempts.get(p);
   const traceRollback = traceId === void 0 ? "" : ` _ctx._grammarTrace?.write({ id: ${JSON.stringify(traceId)}, phase: 'rollback', offset: ${pos} });`;
   const preGuard = [];
@@ -6168,7 +6188,7 @@ function emitSepBy(_p, def, ctx, pos) {
         `${ind(ctx)}const ${tl} = _ctx._cstTriviaLog ? _ctx._cstTriviaLog.length : 0`,
         `${ind(ctx)}const ${lg} = _ctx._triviaLog ? _ctx._triviaLog.length : 0`
       ],
-      rb: `if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${lv}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${rw}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${tl}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${lg}; `
+      rb: `if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${lv}) _ctx._cstLeaves.length = ${lv}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${rw}) _ctx._cstRawChildren.length = ${rw}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${tl}) _ctx._cstTriviaLog.length = ${tl}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${lg}) _ctx._triviaLog.length = ${lg}; `
     };
   };
   if (rec) {
@@ -6214,7 +6234,7 @@ function emitSepBy(_p, def, ctx, pos) {
         `${ind(ctx)}const ${spV} = ${capFn}(input, ${curV}, _ctx, 1)`
       );
       sepAtPos = spV;
-      const rollbackToSep = `if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${markLv}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${markLog}; `;
+      const rollbackToSep = `if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${markLv}) _ctx._cstLeaves.length = ${markLv}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${markV}) _ctx._cstRawChildren.length = ${markV}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${markTl}) _ctx._cstTriviaLog.length = ${markTl}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${markLog}) _ctx._triviaLog.length = ${markLog}; `;
       const { stmts: sepStmts, okVar: sepOk, endVar: sepEnd } = emitFallible(def.separator, ctx, sepAtPos, true);
       stmts.push(...sepStmts, `${ind(ctx)}if (!${sepOk}) { ${rollbackToSep}break }`);
       const post = postSepMarks();
@@ -6256,7 +6276,7 @@ function emitSepBy(_p, def, ctx, pos) {
     }
     const { stmts: sepStmts, okVar: sepOk, endVar: sepEnd } = emitFallible(def.separator, ctx, sepAtPos, true);
     stmts.push(...sepStmts, `${ind(ctx)}if (!${sepOk}) break`);
-    const nextRb = markLv ? `if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${markLv}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${markRw}; ` : "";
+    const nextRb = markLv ? `if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${markLv}) _ctx._cstLeaves.length = ${markLv}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${markRw}) _ctx._cstRawChildren.length = ${markRw}; ` : "";
     const post = postSepMarks();
     stmts.push(...post.decl);
     const { stmts: nextStmts, okVar: nextOk, valVar: nextVal, endVar: nextEnd } = emitFallible(def.parser, ctx, sepEnd, true);
@@ -6341,7 +6361,7 @@ function emitNot(def, ctx, pos) {
       ] : [],
       ...stmts,
       ...sinksLive ? [
-        `${ind(ctx)}if (_ctx._cstLeaves) _ctx._cstLeaves.length = ${leaves}; if (_ctx._cstRawChildren) _ctx._cstRawChildren.length = ${raw}; if (_ctx._cstTriviaLog) _ctx._cstTriviaLog.length = ${tl}; if (_ctx._triviaLog) _ctx._triviaLog.length = ${log}; if (_ctx._fields) _ctx._fields.length = ${fields}; if (_ctx._errors) _ctx._errors.length = ${errors}`
+        `${ind(ctx)}if (_ctx._cstLeaves && _ctx._cstLeaves.length !== ${leaves}) _ctx._cstLeaves.length = ${leaves}; if (_ctx._cstRawChildren && _ctx._cstRawChildren.length !== ${raw}) _ctx._cstRawChildren.length = ${raw}; if (_ctx._cstTriviaLog && _ctx._cstTriviaLog.length !== ${tl}) _ctx._cstTriviaLog.length = ${tl}; if (_ctx._triviaLog && _ctx._triviaLog.length !== ${log}) _ctx._triviaLog.length = ${log}; if (_ctx._fields && _ctx._fields.length !== ${fields}) _ctx._fields.length = ${fields}; if (_ctx._errors && _ctx._errors.length !== ${errors}) _ctx._errors.length = ${errors}`
       ] : [],
       ...emitIfFail(ctx, okVar, failBody(ctx, label2, pos))
     ],