parseman 0.26.2 → 0.27.0

package/dist/plugin/index.js

@@ -2217,6 +2217,55 @@ function parserHasOwnFields(p, seen = /* @__PURE__ */ new Set()) {
       return false;
   }
 }
+function parserHasTriviaSite(p, seen = /* @__PURE__ */ new Set()) {
+  if (seen.has(p)) return true;
+  seen.add(p);
+  const d = p._def;
+  switch (d.tag) {
+    // Trivia is skipped between/around elements or iterations → a site.
+    case "sequence":
+    case "many":
+    case "oneOrMore":
+    case "sepBy":
+    case "scanTo":
+    case "recover":
+    case "skip":
+      return true;
+    // Nested node manages its own trivia frame; none logs here.
+    case "node":
+      return false;
+    // Pure terminals + trivia-suppressing token: no site.
+    case "regex":
+    case "literal":
+    case "keywords":
+    case "guard":
+    case "token":
+      return false;
+    // Transparent single-child wrappers: recurse.
+    case "optional":
+    case "transform":
+    case "trivia":
+    case "label":
+    case "expect":
+    case "withCtx":
+    case "not":
+    case "field":
+      return parserHasTriviaSite(d.parser, seen);
+    case "choice":
+      return d.parsers.some((x) => parserHasTriviaSite(x, seen));
+    case "grammar":
+      return parserHasTriviaSite(d.parser, seen);
+    case "lazy": {
+      try {
+        return parserHasTriviaSite(d.thunk(), seen);
+      } catch {
+        return true;
+      }
+    }
+    default:
+      return true;
+  }
+}
 function buildFieldMap(captures) {
   if (!captures?.length) return void 0;
   const out = {};
@@ -5191,32 +5240,49 @@ function emitNode(def, ctx, pos) {
   const smk = structural ? v(ctx, "_smk") : null;
   if (structural) ctx.needsHostReads = true;
   const hostTriviaGate = `_ctx.build !== undefined && (_ctx.build._parsemanCaptureTrivia !== undefined ? _ctx.build._parsemanCaptureTrivia(${JSON.stringify(def.type)}) : (_ctx._pmCapTL ??= _hostReads(_ctx.build, 5)))`;
-  const allocStmt = structural ? `${i}const ${capTLv} = ${hostTriviaGate}, ${capSTv} = _ctx._pmCapST ??= (_ctx.build === undefined || _hostReads(_ctx.build, 6))${capFv ? `, ${capFv} = _ctx.build !== undefined && _hostReads(_ctx.build, 2)` : ""}
-${i}const ${chV} = [], ${rawV} = [], ${tlV} = ${capTLv} ? [] : _EMPTY_TL` : capturesTrivia ? `${i}const ${chV} = [], ${rawV} = [], ${tlV} = []` : `${i}const ${chV} = [], ${rawV} = []`;
+  const pmV = v(ctx, "_pm"), recV = v(ctx, "_rec"), capV = v(ctx, "_cap");
+  const profHoist = `${i}const ${pmV} = _ctx._pmProfile, ${recV} = ${pmV}?.phase === 'recognizer', ${capV} = ${pmV}?.phase === 'capture'`;
+  const profileRecognizer = recV;
+  const profileCapture = capV;
+  const allocStmt = structural ? `${i}const ${capTLv} = !(${profileRecognizer}) && (${profileCapture} || ${hostTriviaGate}), ${capSTv} = !(${profileRecognizer} || ${profileCapture}) && (_ctx._pmCapST ??= (_ctx.build === undefined || _hostReads(_ctx.build, 6)))${capFv ? `, ${capFv} = !(${profileRecognizer}) && (${profileCapture} || (_ctx.build !== undefined && _hostReads(_ctx.build, 2)))` : ""}
+${i}const ${chV} = ${profileRecognizer} ? undefined : [], ${rawV} = ${profileRecognizer} ? undefined : [], ${tlV} = ${capTLv} ? [] : _EMPTY_TL` : capturesTrivia ? `${i}const ${chV} = ${profileRecognizer} ? undefined : [], ${rawV} = ${profileRecognizer} ? undefined : [], ${tlV} = ${profileRecognizer} ? undefined : []` : `${i}const ${chV} = ${profileRecognizer} ? undefined : [], ${rawV} = ${profileRecognizer} ? undefined : []`;
   const innerTl = structural ? `${capTLv} ? ${tlV} : undefined` : capturesTrivia ? tlV : "undefined";
-  const fieldsOn = structural ? capFv ?? "false" : capturesFields ? "true" : "false";
+  const fieldsOn = structural ? capFv ?? "false" : `${profileRecognizer} ? false : ${capturesFields ? "true" : "false"}`;
   const sf = hasFields ? v(ctx, "_sf") : null;
   const fArr = hasFields ? v(ctx, "_fa") : null;
   const fObj = hasFields ? v(ctx, "_fields") : "undefined";
+  const needsTriviaFrame = parserHasTriviaSite(def.parser);
+  const saveTrivia = needsTriviaFrame ? `, ${st} = _ctx.captureTrivia, ${stl} = _ctx._cstTriviaLog${smk ? `, ${smk} = _ctx._triviaCaptureMask` : ""}` : "";
+  const installTrivia = needsTriviaFrame ? `; _ctx.captureTrivia = true; _ctx._cstTriviaLog = ${innerTl}${smk ? `; _ctx._triviaCaptureMask = ${capTLv} && _ctx.build._parsemanTriviaKinds !== undefined ? _ctx.build._parsemanTriviaKinds(${JSON.stringify(def.type)}) : ${smk}` : ""}` : "";
+  const restoreTrivia = needsTriviaFrame ? `; _ctx.captureTrivia = ${st}; _ctx._cstTriviaLog = ${stl}${smk ? `; _ctx._triviaCaptureMask = ${smk}` : ""}` : "";
   const stmts = [
+    profHoist,
     allocStmt,
-    `${i}const ${sc} = _ctx._cstChildren, ${sl} = _ctx._cstLeaves, ${sr} = _ctx._cstRawChildren, ${st} = _ctx.captureTrivia, ${stl} = _ctx._cstTriviaLog${smk ? `, ${smk} = _ctx._triviaCaptureMask` : ""}${sf ? `, ${sf} = _ctx._fields` : ""}`,
-    `${i}_ctx._cstChildren = ${chV}; _ctx._cstLeaves = ${chV}; _ctx._cstRawChildren = ${rawV}; _ctx.captureTrivia = true; _ctx._cstTriviaLog = ${innerTl}${smk ? `; _ctx._triviaCaptureMask = ${capTLv} && _ctx.build._parsemanTriviaKinds !== undefined ? _ctx.build._parsemanTriviaKinds(${JSON.stringify(def.type)}) : ${smk}` : ""}${sf ? `; _ctx._fields = ${fieldsOn} ? [] : undefined` : ""}`
+    `${i}const ${sc} = _ctx._cstChildren, ${sl} = _ctx._cstLeaves, ${sr} = _ctx._cstRawChildren${saveTrivia}${sf ? `, ${sf} = _ctx._fields` : ""}`,
+    `${i}_ctx._cstChildren = ${chV}; _ctx._cstLeaves = ${chV}; _ctx._cstRawChildren = ${rawV}${installTrivia}${sf ? `; _ctx._fields = ${fieldsOn} ? [] : undefined` : ""}`
   ];
   const { stmts: innerStmts, okVar, endVar } = emitFallible(def.parser, ctx, pos);
   stmts.push(...innerStmts);
   if (sf && fArr) {
     stmts.push(`${i}const ${fArr} = _ctx._fields`);
   }
-  stmts.push(`${i}_ctx._cstChildren = ${sc}; _ctx._cstLeaves = ${sl}; _ctx._cstRawChildren = ${sr}; _ctx.captureTrivia = ${st}; _ctx._cstTriviaLog = ${stl}${smk ? `; _ctx._triviaCaptureMask = ${smk}` : ""}${sf ? `; _ctx._fields = ${sf}` : ""}`);
+  stmts.push(`${i}_ctx._cstChildren = ${sc}; _ctx._cstLeaves = ${sl}; _ctx._cstRawChildren = ${sr}${restoreTrivia}${sf ? `; _ctx._fields = ${sf}` : ""}`);
   stmts.push(...emitIfFail(ctx, `!${okVar}`, propagateFailBody(ctx)));
+  stmts.push(
+    `${i}if (_ctx._pmProfile) {`,
+    `${i}  _ctx._pmProfile.nodes++`,
+    `${i}  if (${profileCapture}) {`,
+    `${i}    _ctx._pmProfile.childSlots += ${chV}.length; _ctx._pmProfile.rawSlots += ${rawV}.length; _ctx._pmProfile.triviaSlots += ${tlV}.length${fArr ? `; _ctx._pmProfile.fieldSlots += ${fArr} ? ${fArr}.length : 0` : ""}`,
+    `${i}  }`,
+    `${i}}`
+  );
   let stV = "undefined";
   if (structural) {
     stV = v(ctx, "_nst");
     stmts.push(`${i}const ${stV} = ${capSTv} && _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
   } else if (clonesState) {
     stV = v(ctx, "_nst");
-    stmts.push(`${i}const ${stV} = _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
+    stmts.push(`${i}const ${stV} = !(${profileRecognizer} || ${profileCapture}) && _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
   }
   if (sf && fArr) {
     const fe = v(ctx, "_fe");
@@ -5236,15 +5302,17 @@ ${i}const ${chV} = [], ${rawV} = [], ${tlV} = ${capTLv} ? [] : _EMPTY_TL` : capt
   }
   const ndV = v(ctx, "_nd");
   const buildExpr = structural ? `{ _tag: 'node', type: ${JSON.stringify(def.type)}, span: { start: ${pos}, end: ${endVar} }, state: ${stV} ?? null, children: ${chV} }` : mkType ? emitInlineMkNodeExpr(mkType, chV, rawV, pos, endVar, tlV) : `${buildRef(ctx)}[${buildIdx}](${chV}, ${fObj}, { start: ${pos}, end: ${endVar} }, ${rawV}, ${tlV}, ${stV})`;
-  const ndExpr = ctx.ns || structural ? `_ctx.build !== undefined ? _ctx.build(${JSON.stringify(def.type)}, ${chV}, ${fObj}, { start: ${pos}, end: ${endVar} }, ${rawV}, ${tlV}, ${stV}) : (${buildExpr})` : buildExpr;
+  const ndExpr = ctx.ns || structural ? `_ctx.build !== undefined ? (_ctx._pmProfile?.phase === 'host' && _ctx._pmProfile.hostCalls++, _ctx.build(${JSON.stringify(def.type)}, ${chV}, ${fObj}, { start: ${pos}, end: ${endVar} }, ${rawV}, ${tlV}, ${stV})) : (${buildExpr})` : buildExpr;
   const hostCollapseExpr = ctx.ns || structural ? `_ctx.build !== undefined && _ctx.build._parsemanCstCollapse !== undefined && ${chV}.length === 1 && ${rawV}.length === 1 && _ctx.build._parsemanCstCollapse(${JSON.stringify(def.type)}, ${chV}[0], ${chV}, ${rawV}) ? ${chV}[0] : (${ndExpr})` : ndExpr;
   const unwrapExpr = `${chV}.length === 1 ? (${chV}[0] !== null && typeof ${chV}[0] === 'object' && ${chV}[0]._tag === 'leaf' ? ${chV}[0].value : ${chV}[0]) : (${ndExpr})`;
   const collapseExpr = `${chV}.length === 1 ? ${chV}[0] : (${ndExpr})`;
   const finalExpr = def.unwrap ? unwrapExpr : def.collapse ? collapseExpr : hostCollapseExpr;
   stmts.push(
-    `${i}const ${ndV} = ${finalExpr}`,
-    `${i}if (${sc}) ${sc}.push(${ndV})`,
-    `${i}if (${sr}) ${sr}.push((typeof ${ndV} === 'object' && ${ndV} !== null && (${ndV}._tag === 'node' || ${ndV}._tag === 'leaf' || ${ndV}._tag === 'parseError')) ? ${ndV} : { _tag: 'leaf', value: typeof ${ndV} === 'string' ? ${ndV} : '', span: { start: ${pos}, end: ${endVar} } })`
+    `${i}const ${ndV} = (${profileRecognizer} || ${profileCapture}) ? undefined : (${finalExpr})`,
+    `${i}if (!(${profileRecognizer})) {`,
+    `${i}  if (${sc}) ${sc}.push(${ndV})`,
+    `${i}  if (${sr}) ${sr}.push((typeof ${ndV} === 'object' && ${ndV} !== null && (${ndV}._tag === 'node' || ${ndV}._tag === 'leaf' || ${ndV}._tag === 'parseError')) ? ${ndV} : { _tag: 'leaf', value: typeof ${ndV} === 'string' ? ${ndV} : '', span: { start: ${pos}, end: ${endVar} } })`,
+    `${i}}`
   );
   return { stmts, valueVar: ndV, endVar };
 }
@@ -5530,7 +5598,7 @@ function emitDispatch(p, ctx, pos) {
         ctx.indent = 1;
         ctx.failLabel = "_pfail";
         ctx.recordFail = true;
-        const innerR = emit(innerParser, ctx, "_pos");
+        const innerR = emitDispatch(innerParser, ctx, "_pos");
         ctx.indent = savedIndent;
         ctx.failLabel = savedFailLabel;
         ctx.recordFail = savedRecord;
@@ -6247,6 +6315,23 @@ function escapeClassChar(ch) {
   return ch.replace(/[\\\]^-]/g, "\\$&");
 }
 
+// src/combinators/withCtx.ts
+function withCtx(extra, combinator) {
+  const meta = {
+    firstSet: combinator._meta.firstSet,
+    canMatchNewline: combinator._meta.canMatchNewline,
+    isTrivia: false
+  };
+  return {
+    _tag: "withCtx",
+    _meta: meta,
+    _def: { tag: "withCtx", extra, parser: combinator },
+    parse(input, pos, ctx) {
+      return combinator.parse(input, pos, { ...ctx, state: extra });
+    }
+  };
+}
+
 // src/compiler/ir-serialize.ts
 function childrenOf2(def) {
   switch (def.tag) {
@@ -6263,6 +6348,7 @@ function childrenOf2(def) {
     case "field":
     case "not":
     case "node":
+    case "withCtx":
     case "expect":
       return [def.parser];
     case "grammar":
@@ -6278,7 +6364,6 @@ function childrenOf2(def) {
     case "regex":
     case "keywords":
     case "guard":
-    case "withCtx":
     case "recover":
     case "unknown":
       return [];
@@ -6353,6 +6438,17 @@ function evalRuleMapIR(ir) {
     c._def.gateSrcs = gateSrcs;
     return c;
   };
+  const _wc = (src, inner) => {
+    let extra;
+    try {
+      extra = (0, eval)(`(${src})`);
+    } catch {
+      extra = void 0;
+    }
+    const w = withCtx(extra, inner);
+    w._def.extraSrc = src;
+    return w;
+  };
   const fn = new Function(
     "rules",
     "ref",
@@ -6379,6 +6475,7 @@ function evalRuleMapIR(ir) {
     "_tf",
     "_nd",
     "_gch",
+    "_wc",
     `return (${ir})`
   );
   const map = fn(
@@ -6406,7 +6503,8 @@ function evalRuleMapIR(ir) {
     expect,
     _tf,
     _nd,
-    _gch
+    _gch,
+    _wc
   );
   return Object.entries(map);
 }
@@ -6462,7 +6560,7 @@ ${body}
       this.analyze(target, active);
       return;
     }
-    if (def.tag === "guard" || def.tag === "withCtx" || def.tag === "recover" || def.tag === "unknown") {
+    if (def.tag === "guard" || def.tag === "recover" || def.tag === "unknown") {
       throw new Unserializable(`unsupported tag ${def.tag}`);
     }
     this.counts.set(c, (this.counts.get(c) ?? 0) + 1);
@@ -6571,8 +6669,11 @@ ${body}
         const trivia2 = def.clearTrivia ? "null" : def.triviaParser ? kid(def.triviaParser) : "undefined";
         return `parser({ trivia: ${trivia2}${def.trackLines ? ", trackLines: true" : ""} }, ${kid(def.parser)})`;
       }
+      case "withCtx": {
+        if (def.extraSrc === void 0) throw new Unserializable("withCtx without extraSrc");
+        return `_wc(${JSON.stringify(def.extraSrc)}, ${kid(def.parser)})`;
+      }
       case "guard":
-      case "withCtx":
       case "recover":
       case "unknown":
         throw new Unserializable(`unsupported tag ${def.tag}`);
@@ -6684,23 +6785,6 @@ function guard(predicate) {
   };
 }
 
-// src/combinators/withCtx.ts
-function withCtx(extra, combinator) {
-  const meta = {
-    firstSet: combinator._meta.firstSet,
-    canMatchNewline: combinator._meta.canMatchNewline,
-    isTrivia: false
-  };
-  return {
-    _tag: "withCtx",
-    _meta: meta,
-    _def: { tag: "withCtx", extra, parser: combinator },
-    parse(input, pos, ctx) {
-      return combinator.parse(input, pos, { ...ctx, state: extra });
-    }
-  };
-}
-
 // src/cst/offset-model.ts
 var { isFinite } = Number;