parseman 0.42.1 → 0.43.0

package/dist/spec/index.cjs

@@ -1286,6 +1286,104 @@ function parser(opts, root) {
   };
 }
 
+// src/cst/reflection.ts
+var GRAMMAR_REFLECTION = /* @__PURE__ */ Symbol.for("parseman.grammarReflection");
+function childrenOf(def) {
+  switch (def.tag) {
+    case "sequence":
+    case "choice":
+      return def.parsers;
+    case "dispatch":
+      return [
+        def.selector,
+        ...def.cases.map((c) => c.parser),
+        ...def.matchers ? def.matchers.map((c) => c.parser) : [],
+        ...def.otherwise ? [def.otherwise] : []
+      ];
+    case "many":
+    case "oneOrMore":
+    case "optional":
+    case "attempt":
+    case "transform":
+    case "trivia":
+    case "token":
+    case "leaf":
+    case "label":
+    case "field":
+    case "not":
+    case "peek":
+    case "node":
+    case "withCtx":
+    case "recover":
+    case "expect":
+      return [def.parser];
+    case "grammar":
+      return def.triviaParser ? [def.parser, def.triviaParser] : [def.parser];
+    case "sepBy":
+      return [def.parser, def.separator];
+    case "skip":
+      return [def.main, def.skipped];
+    case "scanTo":
+      return [def.sentinel, ...def.skip];
+    case "literal":
+    case "regex":
+    case "keywords":
+    case "routed":
+    case "guard":
+    case "lazy":
+    case "unknown":
+      return [];
+  }
+}
+function addNode(out, def) {
+  if (def.type === void 0) return;
+  const existing = out.get(def.type);
+  if (existing === void 0) {
+    out.set(def.type, def.tags === void 0 ? [] : [...def.tags]);
+    return;
+  }
+  if (def.tags === void 0) return;
+  for (const tag of def.tags) {
+    if (!existing.includes(tag)) existing.push(tag);
+  }
+}
+function collectFromCombinator(parser2, out, seen, followLazy = true) {
+  if (seen.has(parser2)) return;
+  seen.add(parser2);
+  const def = parser2._def;
+  if (def.tag === "lazy") {
+    if (followLazy) {
+      try {
+        collectFromCombinator(def.thunk(), out, seen, followLazy);
+      } catch {
+      }
+    }
+    return;
+  }
+  if (def.tag === "node") addNode(out, def);
+  for (const child of childrenOf(def)) collectFromCombinator(child, out, seen, followLazy);
+}
+function collectGrammarReflection(ruleMap, opts) {
+  const byType = /* @__PURE__ */ new Map();
+  const followLazy = opts?.followLazy !== false;
+  for (const [, rule] of ruleMap) {
+    if (!followLazy && rule._def.tag === "lazy") {
+      try {
+        const resolved = rule._def.thunk();
+        if (resolved._def.tag !== "lazy") collectFromCombinator(resolved, byType, /* @__PURE__ */ new Set(), false);
+      } catch {
+      }
+      continue;
+    }
+    collectFromCombinator(rule, byType, /* @__PURE__ */ new Set(), followLazy);
+  }
+  return { nodes: [...byType].map(([type, tags]) => ({ type, tags })) };
+}
+function attachGrammarReflection(grammar, reflection) {
+  Object.defineProperty(grammar, GRAMMAR_REFLECTION, { value: reflection, enumerable: false, configurable: true });
+  return grammar;
+}
+
 // src/combinators/parser.ts
 var RULE_ORDER = "__parsemanRuleOrder";
 function tagRule(r, key) {
@@ -1384,6 +1482,7 @@ function rules(a, b) {
     enumerable: false,
     configurable: true
   });
+  attachGrammarReflection(cache, collectGrammarReflection(Object.keys(definitions).map((key) => [key, cache[key]])));
   return cache;
 }
 
@@ -3010,13 +3109,14 @@ function node(typeOrCombinator, combinatorOrBuild, buildOrOpts, maybeOpts) {
   const collapse = opts?.collapse === true;
   const captureTrivia = opts?.captureTrivia === true;
   const trailingTrivia = opts?.trailingTrivia === true;
+  const tags = opts?.tags;
   if (unwrap && collapse) {
     throw new Error("node() options cannot set both unwrap and collapse");
   }
   if (project !== void 0 && (unwrap || collapse)) {
     throw new Error("node() options cannot combine project with unwrap or collapse");
   }
-  const def = unwrap || collapse || project !== void 0 || captureTrivia || trailingTrivia ? { ...baseDef, ...unwrap ? { unwrap: true } : {}, ...collapse ? { collapse: true } : {}, ...project !== void 0 ? { project } : {}, ...captureTrivia ? { captureTrivia: true } : {}, ...trailingTrivia ? { trailingTrivia: true } : {} } : baseDef;
+  const def = unwrap || collapse || project !== void 0 || captureTrivia || trailingTrivia || tags !== void 0 && tags.length > 0 ? { ...baseDef, ...unwrap ? { unwrap: true } : {}, ...collapse ? { collapse: true } : {}, ...project !== void 0 ? { project } : {}, ...captureTrivia ? { captureTrivia: true } : {}, ...trailingTrivia ? { trailingTrivia: true } : {}, ...tags !== void 0 && tags.length > 0 ? { tags } : {} } : baseDef;
   const capturesTrivia = captureTrivia || trailingTrivia || (build ? buildReadsTrivia(def) : project === void 0);
   const clonesState = build ? buildReadsState(def) : project === void 0;
   const hasOwnFields = parserHasOwnFields(combinator);
@@ -3062,7 +3162,7 @@ function node(typeOrCombinator, combinatorOrBuild, buildOrOpts, maybeOpts) {
       const cstOutput = cstOutputHost(ctx.build);
       const span = ctx.trackLines ? annotateSpanFromLineContext(r.span, ctx) : r.span;
       const hostState = clonesState ? st : ctx.state !== void 0 ? Object.assign({}, ctx.state) : void 0;
-      const built = unwrap && children.length === 1 ? unwrapChild(children[0]) : collapse && children.length === 1 ? children[0] : !build && project === void 0 && ctx.build?._parsemanCstCollapse && children.length === 1 && rawChildren.length === 1 && ctx.build._parsemanCstCollapse(nodeType, children[0], children, rawChildren) ? children[0] : project !== void 0 ? cstOutput && ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, hostState) : projectChild(children, project, nodeType) : build ? cstOutput && ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, hostState) : build(children, fields, span, rawChildren, triviaLog, st) : ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, st) : { _tag: "node", type: nodeType, span: ctx.trackLines ? { ...span } : { start: r.span.start, end: r.span.end }, state: st ?? null, children };
+      const built = unwrap && children.length === 1 ? unwrapChild(children[0]) : collapse && children.length === 1 ? children[0] : !build && project === void 0 && ctx.build?._parsemanCstCollapse && children.length === 1 && rawChildren.length === 1 && ctx.build._parsemanCstCollapse(nodeType, children[0], children, rawChildren) ? children[0] : project !== void 0 ? cstOutput && ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, hostState, tags) : projectChild(children, project, nodeType) : build ? cstOutput && ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, hostState, tags) : build(children, fields, span, rawChildren, triviaLog, st) : ctx.build ? ctx.build(nodeType, children, fields, span, rawChildren, triviaLog, st, tags) : { _tag: "node", type: nodeType, span: ctx.trackLines ? { ...span } : { start: r.span.start, end: r.span.end }, state: st ?? null, children };
       const rawEntry = isCstChild(built) ? built : { _tag: "leaf", value: typeof built === "string" ? built : typeof built === "object" && built !== null ? input.slice(r.span.start, r.span.end) : "", span };
       if (saved.buf !== void 0 || saved.ch !== void 0) {
         pushCstChild(ctx, built, rawEntry);
@@ -3354,15 +3454,16 @@ function normalizeCstCollapse(collapse) {
   }
   return typeof collapse === "function" ? collapse : void 0;
 }
-function buildCstNode(type, children, _fields, span, _rawChildren, _triviaLog, state) {
-  return { _tag: "node", type, span: { ...span }, state: state ?? null, children: [...children] };
+function buildCstNode(type, children, _fields, span, _rawChildren, _triviaLog, state, tags) {
+  return tags !== void 0 && tags.length > 0 ? { _tag: "node", type, tags, span: { ...span }, state: state ?? null, children: [...children] } : { _tag: "node", type, span: { ...span }, state: state ?? null, children: [...children] };
 }
-function cstBuildHost(typeOrOptions, children, _fields, span, rawChildren, triviaLog, state) {
+function cstBuildHost(typeOrOptions, children, _fields, span, rawChildren, triviaLog, state, _tags) {
   if (typeof typeOrOptions === "string") {
     return buildCstNode(typeOrOptions, children ?? [], _fields, span ?? { start: 0, end: 0 }, rawChildren, triviaLog, state);
   }
   const collapse = normalizeCstCollapse(typeOrOptions?.collapse);
-  const host = (type, children2, fields, span2, rawChildren2, triviaLog2, state2) => buildCstNode(type, children2 ?? [], fields, span2, rawChildren2, triviaLog2, state2);
+  const materializeTags = typeOrOptions?.tags === true;
+  const host = (type, children2, fields, span2, rawChildren2, triviaLog2, state2, tags) => buildCstNode(type, children2 ?? [], fields, span2, rawChildren2, triviaLog2, state2, materializeTags ? tags : void 0);
   host._parsemanCstOutput = true;
   if (collapse) host._parsemanCstCollapse = collapse;
   return host;