parseman 0.9.0 → 0.10.0

package/dist/index.d.ts

@@ -11,7 +11,7 @@ export { rules } from './combinators/parser.ts';
 export { ref } from './combinators/ref.ts';
 export { not } from './combinators/not.ts';
 export { node } from './combinators/node.ts';
-export type { BuildNode } from './combinators/node.ts';
+export type { BuildNode, NodeOptions } from './combinators/node.ts';
 export { transform, skip, trivia, label } from './combinators/map.ts';
 export { parse, parser, noTrivia } from './combinators/grammar.ts';
 export type { ParseOptions, ParserOptions, ParsemanParser } from './combinators/grammar.ts';

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,YAAY,EAAE,UAAU,EAAE,WAAW,EAAE,OAAO,EAAE,SAAS,EAAE,YAAY,EAAE,UAAU,EAAE,IAAI,EAAE,UAAU,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,EAAE,cAAc,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;AAEjM,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAA;AAClD,YAAY,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAE9D,OAAO,EAAE,KAAK,EAAE,MAAM,wBAAwB,CAAA;AAC9C,OAAO,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAA;AACpE,YAAY,EAAE,eAAe,EAAE,MAAM,2BAA2B,CAAA;AAEhE,OAAO,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAA;AACpD,OAAO,EAAE,MAAM,EAAE,MAAM,yBAAyB,CAAA;AAChD,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,yBAAyB,CAAA;AAC1E,OAAO,EAAE,KAAK,EAAE,MAAM,yBAAyB,CAAA;AAC/C,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAA;AAC1C,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAA;AAC1C,OAAO,EAAE,IAAI,EAAE,MAAM,uBAAuB,CAAA;AAC5C,YAAY,EAAE,SAAS,EAAE,MAAM,uBAAuB,CAAA;AAEtD,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAA;AACrE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAA;AAClE,YAAY,EAAE,YAAY,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAE3F,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAA;AAC/C,YAAY,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAE3D,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AACxF,YAAY,EAAE,SAAS,EAAE,MAAM,0BAA0B,CAAA;AAEzD,OAAO,EAAE,KAAK,EAAE,MAAM,wBAAwB,CAAA;AAC9C,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAA;AAClD,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AAChE,OAAO,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AAChE,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAA;AAC5D,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAA;AAC1D,YAAY,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAA;AAE5D,YAAY,EAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAA;AAE5G,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAA;AACvD,YAAY,EAAE,aAAa,EAAE,oBAAoB,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAA;AAChG,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAA;AACxD,YAAY,EAAE,WAAW,EAAE,WAAW,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAA;AACzF,OAAO,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAA;AACvD,YAAY,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAA"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,YAAY,EAAE,UAAU,EAAE,WAAW,EAAE,OAAO,EAAE,SAAS,EAAE,YAAY,EAAE,UAAU,EAAE,IAAI,EAAE,UAAU,EAAE,QAAQ,EAAE,SAAS,EAAE,SAAS,EAAE,cAAc,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAA;AAEjM,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAA;AAClD,YAAY,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAE9D,OAAO,EAAE,KAAK,EAAE,MAAM,wBAAwB,CAAA;AAC9C,OAAO,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAA;AACpE,YAAY,EAAE,eAAe,EAAE,MAAM,2BAA2B,CAAA;AAEhE,OAAO,EAAE,QAAQ,EAAE,MAAM,2BAA2B,CAAA;AACpD,OAAO,EAAE,MAAM,EAAE,MAAM,yBAAyB,CAAA;AAChD,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,yBAAyB,CAAA;AAC1E,OAAO,EAAE,KAAK,EAAE,MAAM,yBAAyB,CAAA;AAC/C,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAA;AAC1C,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAA;AAC1C,OAAO,EAAE,IAAI,EAAE,MAAM,uBAAuB,CAAA;AAC5C,YAAY,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,uBAAuB,CAAA;AAEnE,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAA;AACrE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,0BAA0B,CAAA;AAClE,YAAY,EAAE,YAAY,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAE3F,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAA;AAC/C,YAAY,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAE3D,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AACxF,YAAY,EAAE,SAAS,EAAE,MAAM,0BAA0B,CAAA;AAEzD,OAAO,EAAE,KAAK,EAAE,MAAM,wBAAwB,CAAA;AAC9C,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAA;AAClD,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AAChE,OAAO,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AAChE,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAA;AAC5D,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,yBAAyB,CAAA;AAC1D,YAAY,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAA;AAE5D,YAAY,EAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,gBAAgB,CAAA;AAE5G,OAAO,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAA;AACvD,YAAY,EAAE,aAAa,EAAE,oBAAoB,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAA;AAChG,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAA;AACxD,YAAY,EAAE,WAAW,EAAE,WAAW,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAA;AACzF,OAAO,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAA;AACvD,YAAY,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAA"}

package/dist/index.js

@@ -1127,24 +1127,70 @@ function not(combinator) {
   };
 }
 
+// src/compiler/build-arity.ts
+var PARAM_LIST_RE = (
+  // (params) => ...   |   function name?(params) ...   |   single-ident arrow `x => ...`
+  /^(?:function\b[^(]*\(([^)]*)\)|\(([^)]*)\)\s*=>|([A-Za-z_$][\w$]*)\s*=>)/
+);
+var CONFIRMABLE_PARAM_RE = /^[A-Za-z_$][\w$]*\s*\??\s*(?::[^,=]+)?$/;
+function confirmedBuildArity(src) {
+  const s = src.trim();
+  const m = PARAM_LIST_RE.exec(s);
+  if (!m) return null;
+  if (m[3] !== void 0) return 1;
+  const inner = (m[1] ?? m[2] ?? "").trim();
+  if (inner === "") return 0;
+  const parts = inner.split(",");
+  for (const part of parts) {
+    if (!CONFIRMABLE_PARAM_RE.test(part.trim())) return null;
+  }
+  if (/\barguments\b/.test(s)) return null;
+  return parts.length;
+}
+function buildReadsTrivia(def) {
+  const src = def.buildSrc ?? def.build.toString();
+  const arity = confirmedBuildArity(src);
+  if (arity === null) return true;
+  return arity >= 4;
+}
+function buildReadsState(def) {
+  const src = def.buildSrc ?? def.build.toString();
+  const arity = confirmedBuildArity(src);
+  if (arity === null) return true;
+  return arity >= 5;
+}
+
 // src/combinators/node.ts
-function node(type, combinator, build) {
+function collapseChild(child) {
+  return child !== null && typeof child === "object" && child._tag === "leaf" ? child.value : child;
+}
+function node(type, combinator, build, opts) {
   const meta = {
     firstSet: combinator._meta.firstSet,
     canMatchNewline: combinator._meta.canMatchNewline,
     isTrivia: false
   };
+  const collapse = opts?.collapse === true;
+  const def = collapse ? { tag: "node", type, parser: combinator, build, collapse: true } : { tag: "node", type, parser: combinator, build };
+  const capturesTrivia = buildReadsTrivia(def);
+  const clonesState = buildReadsState(def);
   return {
     _tag: "node",
     _meta: meta,
-    _def: { tag: "node", type, parser: combinator, build },
+    _def: def,
     parse(input, pos, ctx) {
       const saved = beginCstNodeCapture(ctx);
+      if (!capturesTrivia) ctx.captureTrivia = false;
       const r = combinator.parse(input, pos, ctx);
       const { children, rawChildren, triviaLog } = endCstNodeCapture(ctx, saved);
       if (!r.ok) return r;
-      const state = ctx.state !== void 0 ? Object.assign({}, ctx.state) : void 0;
-      const built = build(children, rawChildren, r.span, triviaLog, state);
+      const built = collapse && children.length === 1 ? collapseChild(children[0]) : build(
+        children,
+        rawChildren,
+        r.span,
+        triviaLog,
+        clonesState && ctx.state !== void 0 ? Object.assign({}, ctx.state) : void 0
+      );
       const isNodeLike = typeof built === "object" && built !== null && built._tag === "node";
       const rawEntry = isNodeLike ? built : { _tag: "leaf", value: typeof built === "string" ? built : "", span: r.span };
       if (saved.buf !== void 0 || saved.ch !== void 0) {
@@ -2579,25 +2625,34 @@ function emitNode(def, ctx, pos) {
     ctx.buildSrcs.push(def.buildSrc ?? null);
   }
   const i = ind(ctx);
+  const capturesTrivia = mkType !== null || buildReadsTrivia(def);
+  const clonesState = buildReadsState(def);
   const chV = v(ctx, "_ch");
   const rawV = v(ctx, "_raw");
-  const tlV = v(ctx, "_tl");
+  const tlV = capturesTrivia ? v(ctx, "_tl") : "_EMPTY_TL";
+  if (!capturesTrivia) ctx.needsEmptyTl = true;
   const sc = v(ctx, "_sc"), sl = v(ctx, "_sl"), sr = v(ctx, "_sr"), st = v(ctx, "_st"), stl = v(ctx, "_stl");
+  const allocStmt = capturesTrivia ? `${i}const ${chV} = [], ${rawV} = [], ${tlV} = []` : `${i}const ${chV} = [], ${rawV} = []`;
+  const innerTl = capturesTrivia ? tlV : "undefined";
   const stmts = [
-    `${i}const ${chV} = [], ${rawV} = [], ${tlV} = []`,
+    allocStmt,
     `${i}const ${sc} = _ctx._cstChildren, ${sl} = _ctx._cstLeaves, ${sr} = _ctx._cstRawChildren, ${st} = _ctx.captureTrivia, ${stl} = _ctx._cstTriviaLog`,
-    `${i}_ctx._cstChildren = ${chV}; _ctx._cstLeaves = ${chV}; _ctx._cstRawChildren = ${rawV}; _ctx.captureTrivia = true; _ctx._cstTriviaLog = ${tlV}`
+    `${i}_ctx._cstChildren = ${chV}; _ctx._cstLeaves = ${chV}; _ctx._cstRawChildren = ${rawV}; _ctx.captureTrivia = true; _ctx._cstTriviaLog = ${innerTl}`
   ];
   const { stmts: innerStmts, okVar, endVar } = emitFallible(def.parser, ctx, pos);
   stmts.push(...innerStmts);
   stmts.push(`${i}_ctx._cstChildren = ${sc}; _ctx._cstLeaves = ${sl}; _ctx._cstRawChildren = ${sr}; _ctx.captureTrivia = ${st}; _ctx._cstTriviaLog = ${stl}`);
   stmts.push(...emitIfFail(ctx, `!${okVar}`, failBody(ctx, '"node"', pos)));
-  const stV = v(ctx, "_nst");
-  stmts.push(`${i}const ${stV} = _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
+  let stV = "undefined";
+  if (clonesState) {
+    stV = v(ctx, "_nst");
+    stmts.push(`${i}const ${stV} = _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
+  }
   const ndV = v(ctx, "_nd");
   const ndExpr = mkType ? emitInlineMkNodeExpr(mkType, chV, rawV, pos, endVar, tlV) : `_build[${buildIdx}](${chV}, ${rawV}, { start: ${pos}, end: ${endVar} }, ${tlV}, ${stV})`;
+  const finalExpr = def.collapse ? `${chV}.length === 1 ? (${chV}[0] !== null && typeof ${chV}[0] === 'object' && ${chV}[0]._tag === 'leaf' ? ${chV}[0].value : ${chV}[0]) : (${ndExpr})` : ndExpr;
   stmts.push(
-    `${i}const ${ndV} = ${ndExpr}`,
+    `${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', value: typeof ${ndV} === 'string' ? ${ndV} : '', span: { start: ${pos}, end: ${endVar} } })`
   );
@@ -2989,7 +3044,9 @@ function compile(combinator, mapFnSources) {
   const collatorDecl = ctx.needsCollator ? `const _collator = new Intl.Collator(undefined, { sensitivity: 'accent' })
 ` : "";
   const namedPrelude = ctx.namedFnDecls.length > 0 ? [...namedFnPrelude(), ""] : [];
+  const emptyTlDecls = ctx.needsEmptyTl ? ["const _EMPTY_TL = Object.freeze([])"] : [];
   const source = [
+    ...emptyTlDecls,
     ...ctx.regexDecls,
     "",
     ...namedPrelude,
@@ -3001,6 +3058,7 @@ function compile(combinator, mapFnSources) {
     `}`
   ].join("\n");
   const fn = new Function("input", "_pos", "_rp", "_mf", "_build", "_ctx", [
+    ...emptyTlDecls,
     ...ctx.regexDecls,
     collatorDecl,
     ...namedPrelude,
@@ -3054,10 +3112,12 @@ function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources)
   ].join("\n");
   const mfDecl = mapFnSources?.length ? `  const _mf = [${mapFnSources.join(", ")}]` : "";
   const buildDecl = buildSources?.length ? `  const _build = [${buildSources.join(", ")}]` : "";
-  const needsWrapper = ctx.regexDecls.length > 0 || !!collatorDecl || ctx.namedFnDecls.length > 0 || !!mfDecl || !!buildDecl;
+  const emptyTlDecl = ctx.needsEmptyTl ? `  const _EMPTY_TL = Object.freeze([])` : "";
+  const needsWrapper = ctx.regexDecls.length > 0 || !!collatorDecl || ctx.namedFnDecls.length > 0 || !!mfDecl || !!buildDecl || !!emptyTlDecl;
   if (!needsWrapper) return innerFn;
   const namedPrelude = ctx.namedFnDecls.length > 0 ? namedFnPrelude() : [];
   const hoistedDecls = [
+    emptyTlDecl,
     ...ctx.regexDecls.map((d) => `  ${d}`),
     collatorDecl ? `  ${collatorDecl.trim()}` : "",
     mfDecl,
@@ -3342,6 +3402,31 @@ function replaceAtPath(rebuild, root, path, newNode) {
 function defaultRebuild(node2, children) {
   return { ...node2, children };
 }
+function shiftSpans(child, delta) {
+  const c = child;
+  const span = { start: c.span.start + delta, end: c.span.end + delta };
+  if (Array.isArray(c.children)) {
+    return { ...c, span, children: c.children.map((g) => shiftSpans(g, delta)) };
+  }
+  return { ...c, span };
+}
+function graftAndShift(root, path, newNode, delta) {
+  if (path.length === 0) return newNode;
+  const [idx, ...rest] = path;
+  const oldChildren = root.children;
+  const newChildren = oldChildren.map((child, i) => {
+    if (i < idx) return child;
+    if (i === idx) {
+      return rest.length === 0 ? newNode : graftAndShift(child, rest, newNode, delta);
+    }
+    return shiftSpans(child, delta);
+  });
+  return {
+    ...root,
+    span: { start: root.span.start, end: root.span.end + delta },
+    children: newChildren
+  };
+}
 var FunctionalDocImpl = class _FunctionalDocImpl {
   _registry;
   _rootRule;
@@ -3379,7 +3464,12 @@ var FunctionalDocImpl = class _FunctionalDocImpl {
       const r = ruleFn(newInput, node2.span.start, ctx);
       if (!r.ok) continue;
       if (r.span.end === node2.span.end + delta) {
-        const newTree = replaceAtPath(rebuild, this.tree, path, r.value);
+        if (delta === 0) {
+          const newTree2 = replaceAtPath(rebuild, this.tree, path, r.value);
+          return new _FunctionalDocImpl(this._registry, this._rootRule, this._opts, newTree2, [], newInput);
+        }
+        if (this._opts.rebuild) return reparse();
+        const newTree = graftAndShift(this.tree, path, r.value, delta);
         return new _FunctionalDocImpl(this._registry, this._rootRule, this._opts, newTree, [], newInput);
       }
     }