parseman 0.28.0 → 0.29.0

package/dist/index.js

@@ -4140,6 +4140,61 @@ function tryInlineDestructureTransform(src, valueVars) {
 }
 
 // src/compiler/codegen.ts
+function leadingFirstSetRecipe(p, seen = /* @__PURE__ */ new Set()) {
+  if (seen.has(p)) return { concrete: empty(), refs: [] };
+  seen.add(p);
+  const d = p._def;
+  const rec = (c) => leadingFirstSetRecipe(c, seen);
+  const merge2 = (a, b) => ({ concrete: union(a.concrete, b.concrete), refs: [...a.refs, ...b.refs] });
+  switch (d.tag) {
+    case "lazy": {
+      const name = p._ruleName;
+      if (name !== void 0) return { concrete: empty(), refs: [name] };
+      try {
+        return rec(d.thunk());
+      } catch {
+        return { concrete: any(), refs: [] };
+      }
+    }
+    case "literal":
+    case "regex":
+    case "keywords":
+      return { concrete: p._meta.firstSet, refs: [] };
+    case "choice": {
+      let out = { concrete: empty(), refs: [] };
+      for (const arm of d.parsers) out = merge2(out, rec(arm));
+      return out;
+    }
+    case "sequence": {
+      let out = { concrete: empty(), refs: [] };
+      for (const term of d.parsers) {
+        out = merge2(out, rec(term));
+        if (!matchesEmpty(term)) return out;
+      }
+      return out;
+    }
+    case "oneOrMore":
+    case "many":
+    case "optional":
+    case "transform":
+    case "label":
+    case "field":
+    case "trivia":
+    case "token":
+    case "leaf":
+    case "node":
+    case "grammar":
+    case "expect":
+      return rec(d.parser);
+    case "sepBy":
+      return rec(d.parser);
+    case "skip":
+      return rec(d.main);
+    // not / scanTo / guard / withCtx / recover / unknown → shallow set (safe).
+    default:
+      return { concrete: p._meta.firstSet, refs: [] };
+  }
+}
 var HOST_READS_DECL = "const _hostReads = (b, n) => { if (b === undefined) return false; let s; try { s = Function.prototype.toString.call(b) } catch (e) { return true } if (/\\barguments\\b/.test(s)) return true; const m = /^[^(]*\\(([\\s\\S]*?)\\)/.exec(s); if (m && /\\.\\.\\.|=/.test(m[1])) return true; return b.length > n }";
 var _loweringSink = null;
 function v(ctx, prefix = "_v") {
@@ -5186,6 +5241,13 @@ function emitMany(def, ctx, pos) {
       itemPos = npV;
     }
   }
+  if (!ctx.recovery && !ctx.coverage && !failsAtStart(def.parser) && needsFirstSetGuard(def.parser)) {
+    const lcV = v(ctx, "_lc");
+    stmts.push(
+      `${ind(ctx)}const ${lcV} = ${itemPos} < input.length ? (input.codePointAt(${itemPos}) ?? -1) : -1`,
+      `${ind(ctx)}if (!(${firstSetCond(lcV, def.parser._meta.firstSet)})) { ${rollback}break }`
+    );
+  }
   const { stmts: iterStmts, okVar: iterOk, valVar: iterVal, endVar: iterEnd } = emitFallible(def.parser, ctx, itemPos, true);
   stmts.push(...iterStmts);
   if (ctx.recovery) {
@@ -5236,8 +5298,19 @@ function emitAttempt(p, def, ctx, pos) {
   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 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 = [];
+  if (!ctx.recovery && !ctx.coverage && needsFirstSetGuard(def.parser)) {
+    const gcV = v(ctx, "_agc");
+    const gExp = armStaticExpected(ctx, def.parser);
+    const gFail = ctx.failLabel ? `{ _ctx._fe = ${pos};${ctx.recordFail ? ` _ctx._fx = ${gExp};` : ""} break ${ctx.failLabel} }` : `{ _ctx._fe = ${pos}; ${failReturnArr(gExp, pos)} }`;
+    preGuard.push(
+      `${ind(ctx)}const ${gcV} = ${pos} < input.length ? (input.codePointAt(${pos}) ?? -1) : -1`,
+      `${ind(ctx)}if (!(${firstSetCond(gcV, def.parser._meta.firstSet)})) ${gFail}`
+    );
+  }
   return {
     stmts: [
+      ...preGuard,
       `${ind(ctx)}const ${leaves} = _ctx._cstLeaves?.length ?? 0, ${raw} = _ctx._cstRawChildren?.length ?? 0, ${trivia2} = _ctx._cstTriviaLog?.length ?? 0, ${log} = _ctx._triviaLog?.length ?? 0, ${fields} = _ctx._fields?.length ?? 0, ${errors} = _ctx._errors?.length ?? 0`,
       ...inner.stmts,
       ...emitIfFail(ctx, `!${inner.okVar}`, `{ ${rollback};${traceRollback} _ctx._fe = ${pos}; ${propagateFailBody(ctx)} }`)
@@ -5617,7 +5690,18 @@ ${i}const ${chV} = ${profileRecognizer} ? undefined : [], ${rawV} = ${profileRec
   const saveTrivia = needsTriviaFrame ? `, ${st} = _ctx.captureTrivia, ${stl} = _ctx._cstTriviaLog${smk ? `, ${smk} = _ctx._triviaCaptureMask` : ""}` : "";
   const installTrivia = needsTriviaFrame ? `; _ctx.captureTrivia = ${capTLv ?? "false"}; _ctx._cstTriviaLog = ${innerTl}${smk ? `; _ctx._triviaCaptureMask = ${capTLv} && _ctx.build !== undefined && _ctx.build._parsemanTriviaKinds !== undefined ? _ctx.build._parsemanTriviaKinds(${JSON.stringify(def.type)}) : ${smk}` : ""}` : "";
   const restoreTrivia = needsTriviaFrame ? `; _ctx.captureTrivia = ${st}; _ctx._cstTriviaLog = ${stl}${smk ? `; _ctx._triviaCaptureMask = ${smk}` : ""}` : "";
+  const preGuard = [];
+  if (!ctx.recovery && !ctx.coverage && (capturesChildren || structural) && needsFirstSetGuard(def.parser)) {
+    const gcV = v(ctx, "_ngc");
+    const gExp = armStaticExpected(ctx, def.parser);
+    const gFail = ctx.failLabel ? ctx.recordFail ? `{ _ctx._fe = ${pos}; _ctx._fx = ${gExp}; break ${ctx.failLabel} }` : `break ${ctx.failLabel}` : failReturnArr(gExp, pos);
+    preGuard.push(
+      `${i}const ${gcV} = ${pos} < input.length ? (input.codePointAt(${pos}) ?? -1) : -1`,
+      `${i}if (!(${firstSetCond(gcV, def.parser._meta.firstSet)})) ${gFail}`
+    );
+  }
   const stmts = [
+    ...preGuard,
     profHoist,
     allocStmt,
     `${i}const ${sc} = _ctx._cstChildren, ${sl} = _ctx._cstLeaves, ${sr} = _ctx._cstRawChildren${saveTrivia}${sf ? `, ${sf} = _ctx._fields` : ""}`,
@@ -6534,6 +6618,7 @@ function compileLinkable(ruleMapArg, ns, opts) {
     ].join("\n")));
   }
   const firstSets = /* @__PURE__ */ new Map();
+  const firstSetRecipes = /* @__PURE__ */ new Map();
   for (const [key, rule] of ruleMap) {
     let resolved = rule;
     const d = rule._def;
@@ -6545,6 +6630,7 @@ function compileLinkable(ruleMapArg, ns, opts) {
       }
     }
     firstSets.set(key, canMatchEmptyAtStart(resolved) ? { kind: "any" } : resolved._meta.firstSet);
+    firstSetRecipes.set(key, canMatchEmptyAtStart(resolved) ? { concrete: any(), refs: [] } : leadingFirstSetRecipe(resolved));
   }
   return {
     ns,
@@ -6553,6 +6639,7 @@ function compileLinkable(ruleMapArg, ns, opts) {
     ruleFns,
     wrappers,
     firstSets,
+    firstSetRecipes,
     deps: ruleDependencies(ruleMap),
     needsEmptyTl: !!ctx.needsEmptyTl,
     needsHostReads: !!ctx.needsHostReads,
@@ -7188,10 +7275,30 @@ function fusedBody(pieces) {
   ];
   const wrapperEntries = [...winner].map(([k, p]) => `${JSON.stringify(k)}: ${p.wrappers.get(k)}`);
   const rawBody = [...lines, "return {", wrapperEntries.join(",\n"), "}"].join("\n");
-  const finalFS = /* @__PURE__ */ new Map();
+  const ANY2 = { kind: "any" };
+  const recipes = /* @__PURE__ */ new Map();
+  const shallow = /* @__PURE__ */ new Map();
   for (const [k, p] of winner) {
+    const r = p.firstSetRecipes?.get(k);
+    if (r) recipes.set(k, r);
     const fs = p.firstSets?.get(k);
-    if (fs) finalFS.set(k, fs);
+    if (fs) shallow.set(k, fs);
+  }
+  const finalFS = /* @__PURE__ */ new Map();
+  for (const [k] of winner) finalFS.set(k, recipes.get(k)?.concrete ?? shallow.get(k) ?? ANY2);
+  for (let iter = 0; iter <= recipes.size; iter++) {
+    let changed = false;
+    for (const [k, r] of recipes) {
+      let fs = r.concrete;
+      for (const ref2 of r.refs) {
+        fs = union(fs, recipes.has(ref2) || shallow.has(ref2) ? finalFS.get(ref2) ?? ANY2 : ANY2);
+      }
+      if (JSON.stringify(finalFS.get(k)) !== JSON.stringify(fs)) {
+        finalFS.set(k, fs);
+        changed = true;
+      }
+    }
+    if (!changed) break;
   }
   const body = rawBody.replace(
     // Rule name + code-point var are both JS identifiers (rule names are validated