parseman 0.13.0 → 0.14.0

package/dist/index.js

@@ -251,6 +251,36 @@ function literal(value, opts = {}) {
 }
 
 // src/combinators/regex.ts
+var firstSetAnalyzer = null;
+function registerRegexAnalyzer(analyzer) {
+  firstSetAnalyzer = analyzer;
+}
+var permissiveFirstSet = () => ({ firstSet: any(), canMatchNewline: true });
+function regex(pattern, flags = "") {
+  const source = typeof pattern === "string" ? pattern : pattern.source;
+  const resolvedFlags = typeof pattern === "string" ? flags : pattern.flags;
+  const anchored = new RegExp(source, "y" + resolvedFlags.replace(/[gy]/g, ""));
+  const { firstSet, canMatchNewline } = (firstSetAnalyzer ?? permissiveFirstSet)(source);
+  const meta = { firstSet, canMatchNewline, isTrivia: false };
+  return {
+    _tag: "regex",
+    _meta: meta,
+    _def: { tag: "regex", source, flags: resolvedFlags },
+    parse(input, pos, ctx) {
+      anchored.lastIndex = pos;
+      const m = anchored.exec(input);
+      if (m === null) {
+        return failAt(ctx, [`/${source}/`], pos);
+      }
+      const span = { start: pos, end: pos + m[0].length };
+      const leaf = { _tag: "leaf", value: m[0], span };
+      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf);
+      return { ok: true, value: m[0], span };
+    }
+  };
+}
+
+// src/combinators/regex-analyze.ts
 import regexpTree from "regexp-tree";
 function firstSetFromRegex(pattern) {
   try {
@@ -355,40 +385,6 @@ function extractFirstSet(node2) {
       return { firstSet: any(), canMatchNewline: true };
   }
 }
-function optimizeRegex(source, flags) {
-  try {
-    const result = regexpTree.optimize(`/${source}/${flags}`);
-    const str = result.toString();
-    const lastSlash = str.lastIndexOf("/");
-    return str.slice(1, lastSlash);
-  } catch {
-    return source;
-  }
-}
-function regex(pattern, flags = "") {
-  const source = typeof pattern === "string" ? pattern : pattern.source;
-  const resolvedFlags = typeof pattern === "string" ? flags : pattern.flags;
-  const optimizedSource = optimizeRegex(source, resolvedFlags);
-  const anchored = new RegExp(optimizedSource, "y" + resolvedFlags.replace(/[gy]/g, ""));
-  const { firstSet, canMatchNewline } = firstSetFromRegex(source);
-  const meta = { firstSet, canMatchNewline, isTrivia: false };
-  return {
-    _tag: "regex",
-    _meta: meta,
-    _def: { tag: "regex", source, flags: resolvedFlags, optimizedSource },
-    parse(input, pos, ctx) {
-      anchored.lastIndex = pos;
-      const m = anchored.exec(input);
-      if (m === null) {
-        return failAt(ctx, [`/${source}/`], pos);
-      }
-      const span = { start: pos, end: pos + m[0].length };
-      const leaf = { _tag: "leaf", value: m[0], span };
-      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf);
-      return { ok: true, value: m[0], span };
-    }
-  };
-}
 
 // src/combinators/keywords.ts
 function word(str, boundary = "_0-9A-Za-z") {
@@ -1086,7 +1082,9 @@ function rules(factory) {
       if (typeof key !== "string") return void 0;
       const record = target;
       if (!(key in record)) {
-        record[key] = ref();
+        const r = ref();
+        r._ruleName = key;
+        record[key] = r;
       }
       return record[key];
     }
@@ -1151,12 +1149,14 @@ function confirmedBuildArity(src) {
   return parts.length;
 }
 function buildReadsTrivia(def) {
+  if (!def.build) return true;
   const src = def.buildSrc ?? def.build.toString();
   const arity = confirmedBuildArity(src);
   if (arity === null) return true;
   return arity >= 4;
 }
 function buildReadsState(def) {
+  if (!def.build) return true;
   const src = def.buildSrc ?? def.build.toString();
   const arity = confirmedBuildArity(src);
   if (arity === null) return true;
@@ -1175,8 +1175,8 @@ function node(type, combinator, build, opts) {
   };
   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);
+  const capturesTrivia = build ? buildReadsTrivia(def) : true;
+  const clonesState = build ? buildReadsState(def) : true;
   return {
     _tag: "node",
     _meta: meta,
@@ -1187,13 +1187,8 @@ function node(type, combinator, build, opts) {
       const r = combinator.parse(input, pos, ctx);
       const { children, rawChildren, triviaLog } = endCstNodeCapture(ctx, saved);
       if (!r.ok) return r;
-      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 st = clonesState && ctx.state !== void 0 ? Object.assign({}, ctx.state) : void 0;
+      const built = collapse && children.length === 1 ? collapseChild(children[0]) : build ? build(children, rawChildren, r.span, triviaLog, st) : ctx.build ? ctx.build(type, children, rawChildren, r.span, triviaLog, st) : { _tag: "node", type, span: { start: r.span.start, end: r.span.end }, state: st ?? null, children };
       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) {
@@ -1368,8 +1363,14 @@ function staticExpected(c) {
     case "transform":
     case "not":
       return staticExpected(def.parser);
-    case "lazy":
-      return staticExpected(def.thunk());
+    case "lazy": {
+      const name = c._ruleName;
+      try {
+        return staticExpected(def.thunk());
+      } catch {
+        return name ? [name] : [];
+      }
+    }
     default:
       return [];
   }
@@ -1669,6 +1670,15 @@ function seqIsUnambiguous(parts) {
     const p = parts[k];
     const mandatory = p.part === "lit" ? !p.optional : p.min === 1;
     if (mandatory) hasMandatory = true;
+    if (p.part === "group") {
+      const exposure = groupPartExposure(p);
+      if (exposure === "unsupported") {
+        const trailingOnce = k === parts.length - 1 && p.min === 1 && !p.unbounded;
+        if (!trailingOnce) return false;
+      } else if (exposure !== null) {
+        if (!firstSetsDisjoint(exposure, seqFirstAccept(parts.slice(k + 1)))) return false;
+      }
+    }
     const skippableSingle = p.part === "lit" && p.optional || p.part !== "lit" && !p.unbounded && p.min === 0;
     const isUnboundedRepeat = p.part !== "lit" && p.unbounded;
     if (skippableSingle || isUnboundedRepeat) {
@@ -1715,6 +1725,58 @@ function firstSetsDisjoint(a, b) {
   if (a === "any" || b === "any") return false;
   return classDisjoint(a, b);
 }
+function fixedClassSeq(shape) {
+  if (shape.kind !== "seq") return null;
+  const out = [];
+  for (const p of shape.parts) {
+    if (p.part === "lit") {
+      if (p.optional) return null;
+      for (const cp of p.cps) out.push({ ranges: [[cp, cp]], negated: false });
+    } else if (p.part === "run" && !p.unbounded && p.min === 1) {
+      out.push({ ranges: p.ranges, negated: p.negated });
+    } else {
+      return null;
+    }
+  }
+  return out.length ? out : null;
+}
+function altFixedMutuallyExclusive(shape) {
+  if (shape.kind !== "alt" || shape.disjoint) return false;
+  const lists = [];
+  for (const arm of shape.arms) {
+    const f = fixedClassSeq(arm);
+    if (!f) return false;
+    lists.push(f);
+  }
+  for (let i = 0; i < lists.length; i++) {
+    for (let j = 0; j < lists.length; j++) {
+      if (i === j) continue;
+      const a = lists[i];
+      const b = lists[j];
+      if (a.length > b.length) continue;
+      let disjointAt = false;
+      for (let p = 0; p < a.length; p++) {
+        if (classDisjoint(a[p], b[p])) {
+          disjointAt = true;
+          break;
+        }
+      }
+      if (!disjointAt) return false;
+    }
+  }
+  return true;
+}
+function groupPartExposure(part) {
+  const innerWiggle = trailingBacktrackClass(part.inner);
+  if (innerWiggle === "unsupported") return "unsupported";
+  const requiredOnce = !part.unbounded && part.min === 1;
+  if (requiredOnce) return innerWiggle;
+  const dropExposed = shapeFirstAccept(part.inner);
+  if (dropExposed === "any") return "unsupported";
+  if (innerWiggle === null) return dropExposed;
+  const unioned = unionCharSets([innerWiggle, dropExposed]);
+  return unioned === "any" ? "unsupported" : unioned;
+}
 function trailingBacktrackClass(shape) {
   if (shape.kind === "chars") return { ranges: shape.ranges, negated: false };
   if (shape.kind === "ident") return { ranges: shape.tail, negated: false };
@@ -1726,18 +1788,10 @@ function trailingBacktrackClass(shape) {
       if (last.unbounded || last.min === 0) return { ranges: last.ranges, negated: last.negated };
       return null;
     }
-    const innerWiggle = trailingBacktrackClass(last.inner);
-    if (innerWiggle === "unsupported") return "unsupported";
-    const requiredOnce = !last.unbounded && last.min === 1;
-    if (requiredOnce) return innerWiggle;
-    const dropExposed = shapeFirstAccept(last.inner);
-    if (dropExposed === "any") return "unsupported";
-    if (innerWiggle === null) return dropExposed;
-    const unioned = unionCharSets([innerWiggle, dropExposed]);
-    return unioned === "any" ? "unsupported" : unioned;
+    return groupPartExposure(last);
   }
   if (shape.kind === "alt") {
-    if (!shape.disjoint) return "unsupported";
+    if (!shape.disjoint) return altFixedMutuallyExclusive(shape) ? null : "unsupported";
     const classes = shape.arms.map(trailingBacktrackClass);
     if (classes.some((c) => c === "unsupported")) return "unsupported";
     const concrete = classes.filter((c) => c !== null);
@@ -1877,7 +1931,7 @@ function seqFirstAccept(parts) {
 function groupInnerSafe(shape) {
   if (shape.kind === "chars" || shape.kind === "ident" || shape.kind === "seq" || shape.kind === "litFold") return true;
   if (shape.kind === "lookahead") return groupInnerSafe(shape.inner);
-  if (shape.kind === "alt") return shape.disjoint;
+  if (shape.kind === "alt") return true;
   return false;
 }
 function shapeFirstAccept(shape) {
@@ -1905,9 +1959,8 @@ function shapeFirstAccept(shape) {
       return shapeFirstAccept(shape.inner);
     // zero-width — doesn't change first-char acceptance
     case "alt": {
-      if (!shape.disjoint) return "any";
-      const sets = shape.firsts.filter((f) => f !== null);
-      return sets.length === shape.arms.length ? unionCharSets(sets) : "any";
+      const sets = shape.firsts.map((f) => f === null ? "any" : f);
+      return unionCharSets(sets);
     }
   }
 }
@@ -1948,6 +2001,120 @@ function parseScanShape(source) {
   }
   return parseScanShapeCore(source);
 }
+function splitDelimArms(body) {
+  const arms = [];
+  let depth = 0;
+  let last = 0;
+  let i = 0;
+  while (i < body.length) {
+    const ch = body[i];
+    if (ch === "\\") {
+      i += 2;
+      continue;
+    }
+    if (ch === "[") {
+      let j = i + 1;
+      if (body[j] === "^") j++;
+      while (j < body.length && body[j] !== "]") {
+        if (body[j] === "\\") j += 2;
+        else j++;
+      }
+      if (body[j] !== "]") return null;
+      i = j + 1;
+      continue;
+    }
+    if (ch === "(") {
+      depth++;
+      i++;
+      continue;
+    }
+    if (ch === ")") {
+      depth--;
+      if (depth < 0) return null;
+      i++;
+      continue;
+    }
+    if (ch === "|" && depth === 0) {
+      arms.push(body.slice(last, i));
+      last = i + 1;
+    }
+    i++;
+  }
+  if (depth !== 0) return null;
+  arms.push(body.slice(last));
+  return arms;
+}
+function delimitedBodySound(body, close) {
+  const l0 = close[0];
+  if (close.length > 2) return false;
+  const arms = splitDelimArms(body);
+  if (!arms) return false;
+  let bulkSeen = false;
+  let guardSeen = false;
+  for (const arm of arms) {
+    const bulk = /^\[\^((?:\\.|[^\]])+)\]$/.exec(arm);
+    if (bulk) {
+      if (bulkSeen) return false;
+      const ranges = parseClassRanges(bulk[1]);
+      if (!ranges || ranges.length !== 1 || ranges[0][0] !== l0 || ranges[0][1] !== l0) return false;
+      bulkSeen = true;
+      continue;
+    }
+    if (close.length === 2 && arm.endsWith(")")) {
+      const idx = arm.indexOf("(?!");
+      if (idx === -1) return false;
+      const prefix = literalCodePoints(arm.slice(0, idx));
+      const op = parseClassOperand(arm.slice(idx + 3, -1));
+      const prefixIsL0 = prefix?.length === 1 && prefix[0] === l0;
+      const forbidsL1 = op && !op.negated && op.ranges.length === 1 && op.ranges[0][0] === close[1] && op.ranges[0][1] === close[1];
+      if (!prefixIsL0 || !forbidsL1 || guardSeen) return false;
+      guardSeen = true;
+      continue;
+    }
+    return false;
+  }
+  return bulkSeen && guardSeen === (close.length === 2);
+}
+function parseDelimited(source) {
+  if (source.includes("\\\\")) return null;
+  let i = 0;
+  while (i < source.length && source[i] !== "(") {
+    i += source[i] === "\\" ? 2 : 1;
+  }
+  if (source[i] !== "(" || source[i + 1] !== "?" || source[i + 2] !== ":") return null;
+  const openFrag = source.slice(0, i);
+  let k = i + 3;
+  let depth = 1;
+  while (k < source.length && depth > 0) {
+    const ch = source[k];
+    if (ch === "\\") {
+      k += 2;
+      continue;
+    }
+    if (ch === "[") {
+      let j = k + 1;
+      if (source[j] === "^") j++;
+      while (j < source.length && source[j] !== "]") {
+        if (source[j] === "\\") j += 2;
+        else j++;
+      }
+      if (source[j] !== "]") return null;
+      k = j + 1;
+      continue;
+    }
+    if (ch === "(") depth++;
+    else if (ch === ")") depth--;
+    k++;
+  }
+  if (depth !== 0) return null;
+  const body = source.slice(i + 3, k - 1);
+  if (source[k] !== "*") return null;
+  const closeFrag = source.slice(k + 1);
+  const open = literalCodePoints(openFrag);
+  const close = literalCodePoints(closeFrag);
+  if (!open || !close || !delimitedBodySound(body, close)) return null;
+  return { kind: "delimited", open, close };
+}
 function parseScanShapeCore(source) {
   let m = /^\[((?:\\.|[^\]])+)\]([+*])$/.exec(source);
   if (m && !m[1].startsWith("^")) {
@@ -1972,14 +2139,8 @@ function parseScanShapeCore(source) {
     const stop = parseClassRanges(m[2]);
     if (open && stop) return { kind: "until", open, stop };
   }
-  if (!source.includes("\\\\")) {
-    m = /^(.*?)\((?:\?:)?[\s\S]*\)\*(.*?)$/.exec(source);
-    if (m && m[1] && m[2]) {
-      const open = literalCodePoints(m[1]);
-      const close = literalCodePoints(m[2]);
-      if (open && close) return { kind: "delimited", open, close };
-    }
-  }
+  const delim = parseDelimited(source);
+  if (delim) return delim;
   const parts = parseSeqParts(source);
   if (parts && seqIsUnambiguous(parts)) return { kind: "seq", parts };
   return null;
@@ -2434,6 +2595,7 @@ function looseMkBuildRe(type) {
   );
 }
 function analyzeMkInlineBuild(def) {
+  if (!def.build) return null;
   const src = (def.buildSrc ?? def.build.toString()).trim();
   const strict = src.match(MK_BUILD_RE);
   if (strict) {
@@ -2531,6 +2693,15 @@ function v(ctx, prefix = "_v") {
 function ind(ctx) {
   return "  ".repeat(ctx.indent);
 }
+function nsp(ctx) {
+  return ctx.ns ?? "";
+}
+function mfRef(ctx) {
+  return `${nsp(ctx)}_mf`;
+}
+function buildRef(ctx) {
+  return `${nsp(ctx)}_build`;
+}
 function reindentStmts(stmts, targetLevels) {
   const nonEmpty = stmts.filter((s) => s.trim().length > 0);
   if (nonEmpty.length === 0) return stmts;
@@ -2547,7 +2718,7 @@ function failReturnArr(expectedArr, posExpr) {
 function hoistExpected(ctx, constArrSource) {
   let name = ctx.expectedMap.get(constArrSource);
   if (name === void 0) {
-    name = `_fx${ctx.expectedDecls.length}`;
+    name = `${nsp(ctx)}_fx${ctx.expectedDecls.length}`;
     ctx.expectedDecls.push(`const ${name} = ${constArrSource}`);
     ctx.expectedMap.set(constArrSource, name);
   }
@@ -2646,13 +2817,13 @@ function emitLeafCapture(ctx, valExpr, startExpr, endExpr) {
     `${i}}`
   ];
 }
-function ensureRegexDecl(ctx, optimizedSource, flags) {
+function ensureRegexDecl(ctx, source, flags) {
   const f = "y" + flags.replace(/[gy]/g, "");
-  const key = `${optimizedSource}/${f}`;
+  const key = `${source}/${f}`;
   let rName = ctx.regexMap.get(key);
   if (rName === void 0) {
-    rName = `_re${ctx.regexDecls.length}`;
-    ctx.regexDecls.push(`const ${rName} = /${optimizedSource}/${f}`);
+    rName = `${nsp(ctx)}_re${ctx.regexDecls.length}`;
+    ctx.regexDecls.push(`const ${rName} = /${source}/${f}`);
     ctx.regexMap.set(key, rName);
   }
   return rName;
@@ -2661,7 +2832,7 @@ function ensureTriviaFn(ctx) {
   const trivia2 = ctx.activeTrivia;
   const existing = ctx.triviaFnNames.get(trivia2);
   if (existing) return existing;
-  const fnName = `_tf${ctx.triviaFnNames.size}`;
+  const fnName = `${nsp(ctx)}_tf${ctx.triviaFnNames.size}`;
   ctx.triviaFnNames.set(trivia2, fnName);
   ctx.triviaCaptureNames.set(trivia2, fnName);
   const labeledSpec = analyzeLabeledTrivia(trivia2);
@@ -2683,7 +2854,7 @@ function ensureTriviaFn(ctx) {
       for (const arm of regexSpec.arms) {
         const def = arm.parser._def;
         if (def.tag !== "regex") break;
-        reNames.push(ensureRegexDecl(ctx, def.optimizedSource, def.flags));
+        reNames.push(ensureRegexDecl(ctx, def.source, def.flags));
       }
       if (reNames.length === regexSpec.arms.length) {
         ctx.namedFnDecls.push(buildLabeledRegexTriviaFnDecl(fnName, regexSpec, reNames));
@@ -3008,7 +3179,7 @@ function emitKeywords(def, ctx, pos) {
   const key = `${source}/${flags}`;
   let rName = ctx.regexMap.get(key);
   if (rName === void 0) {
-    rName = `_re${ctx.regexDecls.length}`;
+    rName = `${nsp(ctx)}_re${ctx.regexDecls.length}`;
     ctx.regexDecls.push(`const ${rName} = /${source}/${flags}`);
     ctx.regexMap.set(key, rName);
   }
@@ -3042,11 +3213,11 @@ function emitRegex(def, ctx, pos) {
     }
   }
   const flags = "y" + def.flags.replace(/[gy]/g, "");
-  const key = `${def.optimizedSource}/${flags}`;
+  const key = `${def.source}/${flags}`;
   let rName = ctx.regexMap.get(key);
   if (rName === void 0) {
-    rName = `_re${ctx.regexDecls.length}`;
-    ctx.regexDecls.push(`const ${rName} = /${def.optimizedSource}/${flags}`);
+    rName = `${nsp(ctx)}_re${ctx.regexDecls.length}`;
+    ctx.regexDecls.push(`const ${rName} = /${def.source}/${flags}`);
     ctx.regexMap.set(key, rName);
   }
   const mv = v(ctx, "_m");
@@ -3223,7 +3394,7 @@ function emitGreedyClassify(def, superIndex, ctx, pos) {
   const reKey = `${regexDef.source}/${cleanFlags}`;
   let reVar = ctx.regexMap.get(reKey);
   if (reVar === void 0) {
-    reVar = `_re${ctx.regexDecls.length}`;
+    reVar = `${nsp(ctx)}_re${ctx.regexDecls.length}`;
     ctx.regexDecls.push(`const ${reVar} = /${regexDef.source}/${cleanFlags}`);
     ctx.regexMap.set(reKey, reVar);
   }
@@ -3316,7 +3487,7 @@ function emitFirstMatch(def, ctx, pos) {
     if (gate) {
       const gateIdx = ctx.mapFns.length;
       ctx.mapFns.push(gate);
-      gateCond = `_mf[${gateIdx}](_ctx.state)`;
+      gateCond = `${mfRef(ctx)}[${gateIdx}](_ctx.state)`;
     }
     const skipCond = gateCond ? `!${resOkV} && ${gateCond}` : `!${resOkV}`;
     const armHasAutoNot = !!(autoNot && autoNot.length > 0);
@@ -3383,7 +3554,7 @@ function emitTransformChain(p, baseValue, endV, startPos, ctx) {
     ctx.mapFnSrcs.push(def.fnSrc ?? null);
     const vv = v(ctx);
     return {
-      stmts: [...innerR.stmts, `${ind(ctx)}const ${vv} = _mf[${fnIdx}](${innerR.valueVar}, { start: ${startPos}, end: ${endV} })`],
+      stmts: [...innerR.stmts, `${ind(ctx)}const ${vv} = ${mfRef(ctx)}[${fnIdx}](${innerR.valueVar}, { start: ${startPos}, end: ${endV} })`],
       valueVar: vv,
       endVar: endV
     };
@@ -3611,16 +3782,17 @@ function emitNot(def, ctx, pos) {
   };
 }
 function emitNode(def, ctx, pos) {
-  const mkType = analyzeMkInlineBuild(def);
+  const structural = def.build === void 0;
+  const mkType = structural ? null : analyzeMkInlineBuild(def);
   let buildIdx = null;
-  if (!mkType) {
+  if (!mkType && !structural) {
     buildIdx = ctx.buildFns.length;
     ctx.buildFns.push(def.build);
     ctx.buildSrcs.push(def.buildSrc ?? null);
   }
   const i = ind(ctx);
-  const capturesTrivia = mkType !== null || buildReadsTrivia(def);
-  const clonesState = buildReadsState(def);
+  const capturesTrivia = mkType !== null || structural || buildReadsTrivia(def);
+  const clonesState = structural || buildReadsState(def);
   const chV = v(ctx, "_ch");
   const rawV = v(ctx, "_raw");
   const tlV = capturesTrivia ? v(ctx, "_tl") : "_EMPTY_TL";
@@ -3643,7 +3815,8 @@ function emitNode(def, ctx, pos) {
     stmts.push(`${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 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}, ${rawV}, { start: ${pos}, end: ${endVar} }, ${tlV}, ${stV})`;
+  const ndExpr = ctx.ns || structural ? `_ctx.build !== undefined ? _ctx.build(${JSON.stringify(def.type)}, ${chV}, ${rawV}, { start: ${pos}, end: ${endVar} }, ${tlV}, ${stV}) : (${buildExpr})` : buildExpr;
   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} = ${finalExpr}`,
@@ -3668,8 +3841,9 @@ function emitRuntimeFallback(parser2, ctx, pos) {
   return { stmts, valueVar: vv, endVar: ev };
 }
 function emitLazy(p, def, ctx, pos) {
+  const canonical = ctx.ruleNames?.get(p);
   const usage = ctx.lazyUsage;
-  if (usage && (usage.counts.get(p) ?? 0) <= 1 && !usage.recursive.has(p)) {
+  if (!canonical && usage && (usage.counts.get(p) ?? 0) <= 1 && !usage.recursive.has(p)) {
     let resolved;
     try {
       resolved = def.thunk();
@@ -3679,7 +3853,7 @@ function emitLazy(p, def, ctx, pos) {
     return emit(resolved, ctx, pos);
   }
   if (!ctx.namedParsers.has(p)) {
-    const fnName2 = `_pf${ctx.namedParsers.size}`;
+    const fnName2 = canonical ?? `${nsp(ctx)}_pf${ctx.namedParsers.size}`;
     ctx.namedParsers.set(p, fnName2);
     let resolved;
     try {
@@ -3797,7 +3971,7 @@ function emit(p, ctx, pos) {
       return {
         stmts: [
           ...inner.stmts,
-          `${ind(ctx)}const ${mv} = _mf[${fnIdx}](${inner.valueVar}, { start: ${pos}, end: ${inner.endVar} })`
+          `${ind(ctx)}const ${mv} = ${mfRef(ctx)}[${fnIdx}](${inner.valueVar}, { start: ${pos}, end: ${inner.endVar} })`
         ],
         valueVar: mv,
         endVar: inner.endVar
@@ -3860,7 +4034,7 @@ function emit(p, ctx, pos) {
       const vv = v(ctx);
       return {
         stmts: [
-          ...emitIfFail(ctx, `!_mf[${fnIdx}](_ctx.state)`, failBody(ctx, '"gate"', pos)),
+          ...emitIfFail(ctx, `!${mfRef(ctx)}[${fnIdx}](_ctx.state)`, failBody(ctx, '"gate"', pos)),
           `${ind(ctx)}const ${vv} = null`
         ],
         valueVar: vv,
@@ -3893,7 +4067,7 @@ function emit(p, ctx, pos) {
       const ev = v(ctx, "_wce");
       return {
         stmts: [
-          `${ind(ctx)}const ${rv} = { ..._ctx, state: _mf[${evIdx}]() }`,
+          `${ind(ctx)}const ${rv} = { ..._ctx, state: ${mfRef(ctx)}[${evIdx}]() }`,
           `${ind(ctx)}const ${vv} = ${fn}(input, ${pos}, ${rv})`,
           // withCtx runs on a spread child ctx — copy its recorded failure back
           // and propagate the inner failure verbatim (interpreter parity).
@@ -4024,6 +4198,39 @@ function analyzeLazyUsageMulti(roots) {
   for (const root of roots) walk2(root);
   return { counts, recursive };
 }
+function ruleDependencies(ruleMap) {
+  const nameOf = /* @__PURE__ */ new Map();
+  for (const [name, comb] of ruleMap) nameOf.set(comb, name);
+  const deps = /* @__PURE__ */ new Map();
+  for (const [name, comb] of ruleMap) {
+    const found = /* @__PURE__ */ new Set();
+    const seen = /* @__PURE__ */ new Set();
+    const walk2 = (p, isRoot) => {
+      const def = p._def;
+      const boundary = def.tag === "lazy" ? p._ruleName ?? nameOf.get(p) : nameOf.get(p);
+      if (!isRoot && boundary !== void 0) {
+        found.add(boundary);
+        return;
+      }
+      if (seen.has(p)) return;
+      seen.add(p);
+      if (def.tag === "lazy") {
+        let resolved;
+        try {
+          resolved = def.thunk();
+        } catch {
+          return;
+        }
+        walk2(resolved, false);
+        return;
+      }
+      for (const child of childrenOf(def)) walk2(child, false);
+    };
+    walk2(comb, true);
+    deps.set(name, [...found]);
+  }
+  return deps;
+}
 function compile(combinator, mapFnSources) {
   const ctx = {
     vars: 0,
@@ -4107,6 +4314,150 @@ function compile(combinator, mapFnSources) {
     }
   };
 }
+function compileLinkable(ruleMapArg, ns) {
+  if (!ns) throw new Error("compileLinkable: ns must be a non-empty namespace");
+  const ruleMap = ruleMapArg.filter(([, val]) => {
+    const d = val._def;
+    if (d.tag !== "lazy") return true;
+    try {
+      d.thunk();
+      return true;
+    } catch {
+      return false;
+    }
+  });
+  const ctx = {
+    vars: 0,
+    indent: 1,
+    regexDecls: [],
+    regexMap: /* @__PURE__ */ new Map(),
+    expectedDecls: [],
+    expectedMap: /* @__PURE__ */ new Map(),
+    recordFail: true,
+    mapFns: [],
+    mapFnSrcs: [],
+    buildFns: [],
+    buildSrcs: [],
+    runtimeParsers: [],
+    needsCollator: false,
+    namedParsers: /* @__PURE__ */ new Map(),
+    triviaCaptureNames: /* @__PURE__ */ new Map(),
+    triviaFnNames: /* @__PURE__ */ new Map(),
+    namedFnDecls: [],
+    capturing: ruleMap.some(([, rule]) => hasNodeDef(rule)),
+    lazyUsage: analyzeLazyUsageMulti(ruleMap.map(([, rule]) => rule)),
+    ns
+  };
+  const seen = /* @__PURE__ */ new Map();
+  const ruleNames = /* @__PURE__ */ new Map();
+  const fnNameToKey = /* @__PURE__ */ new Map();
+  for (const [key, rule] of ruleMap) {
+    const base = `_r_${key.replace(/[^A-Za-z0-9_$]/g, "_")}`;
+    const n = seen.get(base) ?? 0;
+    seen.set(base, n + 1);
+    const fn = n === 0 ? base : `${base}$${n}`;
+    ruleNames.set(rule, fn);
+    fnNameToKey.set(fn, key);
+  }
+  ctx.ruleNames = ruleNames;
+  {
+    const scanned = /* @__PURE__ */ new Set();
+    const scanExternal = (p) => {
+      if (scanned.has(p)) return;
+      scanned.add(p);
+      const def = p._def;
+      if (def.tag === "lazy") {
+        let resolved;
+        try {
+          resolved = def.thunk();
+        } catch {
+          resolved = void 0;
+        }
+        if (resolved === void 0) {
+          const name = p._ruleName;
+          if (name && !ctx.namedParsers.has(p)) {
+            const fn = `_r_${name.replace(/[^A-Za-z0-9_$]/g, "_")}`;
+            ruleNames.set(p, fn);
+            ctx.namedParsers.set(p, fn);
+          }
+          return;
+        }
+        scanExternal(resolved);
+        return;
+      }
+      for (const child of childrenOf(def)) scanExternal(child);
+    };
+    for (const [, rule] of ruleMap) scanExternal(rule);
+  }
+  const perEntry = [];
+  for (const [key, rule] of ruleMap) {
+    const fn = ruleNames.get(rule);
+    if (!ctx.namedParsers.has(rule)) {
+      ctx.namedParsers.set(rule, fn);
+      let resolved = rule;
+      const def = rule._def;
+      if (def.tag === "lazy") {
+        try {
+          resolved = def.thunk();
+        } catch {
+          return null;
+        }
+      }
+      const savedIndent = ctx.indent, savedFail = ctx.failLabel, savedRec = ctx.recordFail;
+      ctx.indent = 1;
+      ctx.failLabel = "_pfail";
+      ctx.recordFail = true;
+      const body = emit(resolved, ctx, "_pos");
+      ctx.indent = savedIndent;
+      ctx.failLabel = savedFail;
+      ctx.recordFail = savedRec;
+      pushNamedFnDecl(ctx, fn, body.stmts, body.valueVar, body.endVar);
+    }
+    perEntry.push({ key, r: emitNamedFnCall(ctx, fn, "_pos") });
+  }
+  if (ctx.runtimeParsers.length !== 0) return null;
+  const mfSrcs = ctx.mapFns.length > 0 && ctx.mapFnSrcs.every((s) => s !== null) ? ctx.mapFnSrcs : null;
+  const buildSrcs = ctx.buildFns.length > 0 && ctx.buildSrcs.every((s) => s !== null) ? ctx.buildSrcs : null;
+  const ruleFns = /* @__PURE__ */ new Map();
+  const privateFns = [];
+  for (const decl of ctx.namedFnDecls) {
+    const fnName = decl.match(/^function (\S+?)\(/)?.[1];
+    const key = fnName ? fnNameToKey.get(fnName) : void 0;
+    if (key !== void 0) ruleFns.set(key, decl);
+    else privateFns.push(decl);
+  }
+  const mfDecl = ctx.mapFns.length === 0 ? "" : mfSrcs ? `const ${mfRef(ctx)} = [${mfSrcs.join(", ")}]` : `const ${mfRef(ctx)} = _env[${JSON.stringify(nsp(ctx) + "mf")}]`;
+  const buildDecl = ctx.buildFns.length === 0 ? "" : buildSrcs ? `const ${buildRef(ctx)} = [${buildSrcs.join(", ")}]` : `const ${buildRef(ctx)} = _env[${JSON.stringify(nsp(ctx) + "build")}]`;
+  const prelude = [
+    ...ctx.regexDecls,
+    ...ctx.expectedDecls,
+    mfDecl,
+    buildDecl,
+    ...privateFns
+  ].filter(Boolean);
+  const wrappers = /* @__PURE__ */ new Map();
+  for (const { key, r } of perEntry) {
+    wrappers.set(key, [
+      `function(input, _pos, _ctx) {`,
+      `  let pos = _pos`,
+      ...r.stmts,
+      `  return { ok: true, value: ${r.valueVar}, span: { start: _pos, end: ${r.endVar} } }`,
+      `}`
+    ].join("\n"));
+  }
+  return {
+    ns,
+    keys: perEntry.map((e) => e.key),
+    prelude,
+    ruleFns,
+    wrappers,
+    deps: ruleDependencies(ruleMap),
+    needsEmptyTl: !!ctx.needsEmptyTl,
+    needsCollator: ctx.needsCollator,
+    mfFns: mfSrcs ? [] : ctx.mapFns,
+    buildFns: buildSrcs ? [] : ctx.buildFns
+  };
+}
 function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources) {
   const bodyLines = [
     `  let pos = _pos`,
@@ -4145,6 +4496,87 @@ function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources)
   ].join("\n");
 }
 
+// src/compiler/linker.ts
+var _nsCounter = 0;
+function linkable(rulesMap, ns) {
+  const pieces = compileLinkable([...Object.entries(rulesMap)], ns ?? `_lk${_nsCounter++}_`);
+  if (!pieces) throw new Error("linkable(): this grammar cannot be compiled to a linkable artifact (contains a runtime-only parser fallback)");
+  return pieces;
+}
+function cstBuildHost(type, children, _rawChildren, span, _triviaLog, state) {
+  return { _tag: "node", type, span: { start: span.start, end: span.end }, state: state ?? null, children: [...children] };
+}
+function pick(grammar, names) {
+  const p = grammar.ruleFns instanceof Map ? grammar : linkable(grammar);
+  const keep = /* @__PURE__ */ new Set();
+  const has = new Set(p.keys);
+  for (const n of names) {
+    if (!has.has(n)) throw new Error(`pick: rule "${n}" is not in this grammar (available: ${p.keys.join(", ")})`);
+  }
+  const visit = (n) => {
+    if (keep.has(n) || !has.has(n)) return;
+    keep.add(n);
+    for (const d of p.deps.get(n) ?? []) visit(d);
+  };
+  for (const n of names) visit(n);
+  const filt = (m) => new Map([...m].filter(([k]) => keep.has(k)));
+  return {
+    ...p,
+    keys: [...keep],
+    ruleFns: filt(p.ruleFns),
+    wrappers: filt(p.wrappers),
+    deps: filt(p.deps)
+  };
+}
+function fusedBody(pieces) {
+  const winner = /* @__PURE__ */ new Map();
+  for (const p of pieces) for (const k of p.keys) winner.set(k, p);
+  for (const [k, p] of winner) {
+    for (const d of p.deps.get(k) ?? []) {
+      if (!winner.has(d)) throw new Error(`compose: rule "${k}" references missing rule "${d}"`);
+    }
+  }
+  const contributing = new Set(winner.values());
+  const needsEmptyTl = [...contributing].some((p) => p.needsEmptyTl);
+  const needsCollator = [...contributing].some((p) => p.needsCollator);
+  const lines = [
+    // Shared sentinel protocol (must match NAMED_FN_FAIL / NAMED_FN_END in codegen).
+    "const _pfFail = {}",
+    "let _pfEnd",
+    ...needsEmptyTl ? ["const _EMPTY_TL = Object.freeze([])"] : [],
+    ...needsCollator ? ["const _collator = new Intl.Collator(undefined, { sensitivity: 'accent' })"] : [],
+    // Each contributing artifact's namespaced private prelude (regexes, _pf, …).
+    ...[...contributing].flatMap((p) => p.prelude),
+    // The winning `_r_<Name>` function for each rule (one per name → no redeclare).
+    ...[...winner].map(([k, p]) => p.ruleFns.get(k))
+  ];
+  const wrapperEntries = [...winner].map(([k, p]) => `${JSON.stringify(k)}: ${p.wrappers.get(k)}`);
+  const body = [...lines, "return {", wrapperEntries.join(",\n"), "}"].join("\n");
+  const env = {};
+  for (const p of contributing) {
+    if (p.mfFns.length) env[`${p.ns}mf`] = p.mfFns;
+    if (p.buildFns.length) env[`${p.ns}build`] = p.buildFns;
+  }
+  return { body, env };
+}
+function fuseRules(pieces) {
+  const { body, env } = fusedBody(pieces);
+  return new Function("_env", body)(env);
+}
+var COMPOSED_PIECES = /* @__PURE__ */ Symbol.for("parseman.composedPieces");
+function itemPieces(item) {
+  const carried = item[COMPOSED_PIECES];
+  if (Array.isArray(carried)) return carried;
+  if (item.ruleFns instanceof Map) return [item];
+  return [linkable(item)];
+}
+function compose(items) {
+  const pieces = items.flatMap(itemPieces);
+  const map = fuseRules(pieces);
+  Object.defineProperty(map, COMPOSED_PIECES, { value: pieces, enumerable: false });
+  return map;
+}
+
 // src/combinators/guard.ts
 function guard(predicate) {
   const meta = {
@@ -4342,6 +4774,7 @@ function manyRecover(item, until) {
 }
 
 // src/functional/doc.ts
+var REENTRY_MAX_SPAN_FRACTION = 0.5;
 function isNode(x) {
   return typeof x === "object" && x !== null && x._tag === "node";
 }
@@ -4402,6 +4835,47 @@ function graftAndShift(root, path, newNode, delta) {
     children: newChildren
   };
 }
+function structurallyEqual(a, b) {
+  if (a === b) return true;
+  if (typeof a !== "object" || typeof b !== "object" || a === null || b === null) return false;
+  const at = a._tag;
+  const bt = b._tag;
+  if (at !== bt) return false;
+  const as = a.span;
+  const bs = b.span;
+  if (as || bs) {
+    if (!as || !bs || as.start !== bs.start || as.end !== bs.end) return false;
+  }
+  if (at === "leaf" || at === "trivia") {
+    return a.value === b.value;
+  }
+  if (a.type !== b.type) return false;
+  const ac = a.children;
+  const bc = b.children;
+  if (ac || bc) {
+    if (!ac || !bc || ac.length !== bc.length) return false;
+    for (let i = 0; i < ac.length; i++) if (!structurallyEqual(ac[i], bc[i])) return false;
+  }
+  return true;
+}
+function boundaryIsSafe(ruleFn, newInput, start, boundary, state, build, produced) {
+  if (!produced.ok) return false;
+  if (boundary >= newInput.length) return true;
+  for (const sentinel of [" ", "\uFFFF"]) {
+    if (newInput[boundary] === sentinel) continue;
+    const probed = newInput.slice(0, boundary) + sentinel.repeat(newInput.length - boundary);
+    const ctx = { trackLines: false, state, build };
+    let r;
+    try {
+      r = ruleFn(probed, start, ctx);
+    } catch {
+      return false;
+    }
+    if (!r.ok || r.span.end !== produced.span.end) return false;
+    if (!structurallyEqual(r.value, produced.value)) return false;
+  }
+  return true;
+}
 var ParseDocImpl = class _ParseDocImpl {
   _registry;
   _rootRule;
@@ -4433,20 +4907,24 @@ var ParseDocImpl = class _ParseDocImpl {
     }
     const rebuild = this._opts.rebuild ?? defaultRebuild;
     for (const { node: node2, path } of candidates) {
+      if (node2.span.end - node2.span.start > this.input.length * REENTRY_MAX_SPAN_FRACTION) break;
       const ruleFn = this._registry[node2.type];
       if (!ruleFn) continue;
-      const ctx = { trackLines: false, state: node2.state };
+      if (!(node2.span.start <= from && to <= node2.span.end)) continue;
+      const ctx = { trackLines: false, state: node2.state, build: this._opts.build };
       const r = ruleFn(newInput, node2.span.start, ctx);
       if (!r.ok) continue;
-      if (r.span.end === node2.span.end + delta) {
-        if (delta === 0) {
-          const newTree2 = replaceAtPath(rebuild, this.tree, path, r.value);
-          return new _ParseDocImpl(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 _ParseDocImpl(this._registry, this._rootRule, this._opts, newTree, [], newInput);
+      if (r.span.end !== node2.span.end + delta) continue;
+      if (!boundaryIsSafe(ruleFn, newInput, node2.span.start, node2.span.end + delta, node2.state, this._opts.build, r)) {
+        continue;
+      }
+      if (delta === 0) {
+        const newTree2 = replaceAtPath(rebuild, this.tree, path, r.value);
+        return new _ParseDocImpl(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 _ParseDocImpl(this._registry, this._rootRule, this._opts, newTree, [], newInput);
     }
     return reparse();
   }
@@ -4454,7 +4932,7 @@ var ParseDocImpl = class _ParseDocImpl {
 function parseDoc(registry, rootRule, input, opts = {}) {
   const ruleFn = registry[rootRule];
   if (!ruleFn) throw new Error(`No rule '${rootRule}' in registry`);
-  const ctx = { trackLines: false, state: opts.state };
+  const ctx = { trackLines: false, state: opts.state, build: opts.build };
   const r = ruleFn(input, 0, ctx);
   if (r.ok) {
     return new ParseDocImpl(registry, rootRule, opts, r.value, [], input);
@@ -4462,12 +4940,45 @@ function parseDoc(registry, rootRule, input, opts = {}) {
   return new ParseDocImpl(registry, rootRule, opts, null, [{ ok: false, expected: r.expected, span: r.span }], input);
 }
 
+// src/functional/run.ts
+var invoke = (r, input, pos, ctx) => typeof r === "function" ? r(input, pos, ctx) : r.parse(input, pos, ctx);
+function run(entry, input, options = {}) {
+  const triviaLog = [];
+  const errors = [];
+  const ctx = {
+    trackLines: false,
+    _triviaLog: triviaLog,
+    _errors: errors,
+    build: options.build,
+    state: options.state
+  };
+  const r = invoke(entry, input, 0, ctx);
+  let unconsumedFrom = null;
+  if (r.ok) {
+    let pos = r.span?.end ?? 0;
+    if (options.trivia && pos < input.length) {
+      const t = invoke(options.trivia, input, pos, { trackLines: false });
+      if (t.ok && t.span.end > pos) pos = t.span.end;
+    }
+    unconsumedFrom = pos < input.length ? pos : null;
+  }
+  return {
+    ok: r.ok,
+    value: r.ok ? r.value : void 0,
+    span: r.span ?? { start: 0, end: 0 },
+    expected: r.ok ? [] : r.expected ?? [],
+    errors,
+    triviaLog,
+    unconsumedFrom
+  };
+}
+
 // src/cst/trivia-index.ts
-function merge(map, key, run) {
-  if (run.length === 0) return;
+function merge(map, key, run2) {
+  if (run2.length === 0) return;
   const existing = map.get(key);
-  if (existing) existing.push(...run);
-  else map.set(key, [...run]);
+  if (existing) existing.push(...run2);
+  else map.set(key, [...run2]);
 }
 function buildTriviaIndex(root, input = "", opts) {
   const before = /* @__PURE__ */ new Map();
@@ -4479,11 +4990,11 @@ function buildTriviaIndex(root, input = "", opts) {
       for (let i = 0; i < log.length; i += 3) {
         const tStart = log[i], tEnd = log[i + 1], insertIdx = log[i + 2];
         if (tEnd <= tStart) continue;
-        const run = [{ value: input.slice(tStart, tEnd), span: { start: tStart, end: tEnd } }];
+        const run2 = [{ value: input.slice(tStart, tEnd), span: { start: tStart, end: tEnd } }];
         const next = raw[insertIdx];
-        if (next) merge(before, next.span.start, run);
+        if (next) merge(before, next.span.start, run2);
         const prev = insertIdx > 0 ? raw[insertIdx - 1] : void 0;
-        if (prev) merge(after, prev.span.end, run);
+        if (prev) merge(after, prev.span.end, run2);
       }
     }
     const ch = node2.children;
@@ -4500,17 +5011,17 @@ function buildTriviaIndex(root, input = "", opts) {
     const rootSpan = root?.span;
     const re = new RegExp(opts.trivia.source, opts.trivia.flags.replace(/[gy]/g, "") + "y");
     const scanFrom = (from) => {
-      const run = [];
+      const run2 = [];
       let pos = from;
       re.lastIndex = pos;
       let m = re.exec(input);
       while (m && m.index === pos && m[0].length > 0) {
-        run.push({ value: m[0], span: { start: pos, end: pos + m[0].length } });
+        run2.push({ value: m[0], span: { start: pos, end: pos + m[0].length } });
         pos += m[0].length;
         re.lastIndex = pos;
         m = re.exec(input);
       }
-      return run;
+      return run2;
     };
     if (rootSpan && rootSpan.start > 0) {
       const lead = scanFrom(0);
@@ -4596,20 +5107,295 @@ function triviaEntries(log, labels, opts) {
     }
   };
 }
+
+// src/cst/offset-model.ts
+var { isFinite } = Number;
+function floorIndex(arr, len, x) {
+  let lo = 0;
+  let hi = len - 1;
+  let ans = -1;
+  while (lo <= hi) {
+    const mid = lo + hi >>> 1;
+    if (arr[mid] <= x) {
+      ans = mid;
+      lo = mid + 1;
+    } else {
+      hi = mid - 1;
+    }
+  }
+  return ans;
+}
+var OffsetIndex = class {
+  /** Slot starts, sorted ascending. Parallel to `_ends`. */
+  _starts;
+  _ends;
+  count;
+  docStart;
+  docEnd;
+  constructor(slots, docStart = 0, docEnd) {
+    const sorted = [...slots].sort((a, b) => a.start - b.start);
+    const starts = [];
+    const ends = [];
+    let prevEnd = docStart;
+    for (const s of sorted) {
+      if (!isFinite(s.start) || !isFinite(s.end) || s.end < s.start) {
+        throw new RangeError(`OffsetIndex: invalid slot [${s.start}, ${s.end}]`);
+      }
+      if (s.start < prevEnd) {
+        throw new RangeError(`OffsetIndex: overlapping/out-of-order slot [${s.start}, ${s.end}] after end ${prevEnd}`);
+      }
+      starts.push(s.start);
+      ends.push(s.end);
+      prevEnd = s.end;
+    }
+    this._starts = starts;
+    this._ends = ends;
+    this.count = starts.length;
+    this.docStart = docStart;
+    this.docEnd = docEnd ?? (this.count ? Math.max(prevEnd, ends[ends.length - 1]) : docStart);
+  }
+  /** The slot at index `i` (0-based, source order), or undefined. */
+  slot(i) {
+    if (i < 0 || i >= this.count) return void 0;
+    return { start: this._starts[i], end: this._ends[i] };
+  }
+  /**
+   * Index of the gap that CONTAINS `offset`, in `[0, count]`. Gap `i` spans
+   * `[endOf(i-1), startOf(i))` where `endOf(-1)=docStart` and `startOf(count)=docEnd`.
+   * An offset that lands inside a slot returns that slot's TRAILING gap index
+   * (i.e. the gap after it) — callers querying at slot boundaries get the
+   * intuitive answer via `gapBefore`/`gapAfter`.
+   */
+  gapIndexAt(offset) {
+    const i = floorIndex(this._starts, this.count, offset);
+    if (i < 0) return 0;
+    if (offset < this._ends[i]) return i + 1;
+    return i + 1;
+  }
+  /** The gap at index `i` in `[0, count]`. */
+  gap(i) {
+    if (i < 0 || i > this.count) return void 0;
+    const start = i === 0 ? this.docStart : this._ends[i - 1];
+    const end = i === this.count ? this.docEnd : this._starts[i];
+    return { start, end };
+  }
+  /**
+   * The trivia gap immediately BEFORE `offset` (`[prevSlotEnd, offset)`), where
+   * `prevSlotEnd` is the end of the last slot ending at or before `offset` (or
+   * `docStart`). For a real slot start this is the gap preceding that slot; for an
+   * adjacent slot the gap is empty and agrees with `gapAfter` on the same seam.
+   */
+  gapBefore(offset) {
+    const e = floorIndex(this._ends, this.count, offset);
+    const start = e >= 0 ? this._ends[e] : this.docStart;
+    return { start, end: offset };
+  }
+  /**
+   * The trivia gap immediately AFTER `offset` (`[offset, nextSlotStart)`), where
+   * `nextSlotStart` is the least slot start at or after `offset` (or `docEnd`).
+   */
+  gapAfter(offset) {
+    return { start: offset, end: this._ceilStart(offset) };
+  }
+  /** Least slot start `>= offset`, else `docEnd`. */
+  _ceilStart(offset) {
+    let lo = 0;
+    let hi = this.count - 1;
+    let ans = this.docEnd;
+    while (lo <= hi) {
+      const mid = lo + hi >>> 1;
+      if (this._starts[mid] >= offset) {
+        ans = this._starts[mid];
+        hi = mid - 1;
+      } else {
+        lo = mid + 1;
+      }
+    }
+    return ans;
+  }
+  /** Is this gap empty (adjacent slots, no trivia)? */
+  isEmpty(gap) {
+    return gap.end <= gap.start;
+  }
+};
+var NL = 10;
+var CR = 13;
+var SP = 32;
+var TAB = 9;
+function gapText(source, gap) {
+  return gap.end > gap.start ? source.slice(gap.start, gap.end) : "";
+}
+function lineBreaksIn(source, gap) {
+  let n = 0;
+  for (let i = gap.start; i < gap.end; i++) {
+    const c = source.charCodeAt(i);
+    if (c === NL) n++;
+    else if (c === CR && source.charCodeAt(i + 1) !== NL) n++;
+  }
+  return n;
+}
+function blankLinesIn(source, gap) {
+  return Math.max(0, lineBreaksIn(source, gap) - 1);
+}
+function lineStartWithin(source, gap) {
+  for (let i = gap.end - 1; i >= gap.start; i--) {
+    const c = source.charCodeAt(i);
+    if (c === NL || c === CR) return i + 1;
+  }
+  return gap.start;
+}
+function indentWidth(source, gap) {
+  const ls = lineStartWithin(source, gap);
+  if (ls === gap.start) return 0;
+  return gap.end - ls;
+}
+function indentMixed(source, gap) {
+  const ls = lineStartWithin(source, gap);
+  let sawSpace = false;
+  let sawTab = false;
+  for (let i = ls; i < gap.end; i++) {
+    const c = source.charCodeAt(i);
+    if (c === SP) sawSpace = true;
+    else if (c === TAB) sawTab = true;
+  }
+  return sawSpace && sawTab;
+}
+function commentsIn(source, gap, comment) {
+  const re = new RegExp(comment.source, comment.flags.replace(/[gy]/g, "") + "g");
+  re.lastIndex = gap.start;
+  const out = [];
+  let m;
+  while ((m = re.exec(source)) && m.index < gap.end) {
+    if (m[0].length === 0) {
+      re.lastIndex++;
+      continue;
+    }
+    if (m.index + m[0].length > gap.end) break;
+    out.push({ start: m.index, end: m.index + m[0].length });
+  }
+  return out;
+}
+function gapIsSignificant(source, gap, comment) {
+  if (gap.end <= gap.start) return false;
+  if (lineBreaksIn(source, gap) > 0) return true;
+  if (comment && commentsIn(source, gap, comment).length > 0) return true;
+  return false;
+}
+function collectLeafSlots(root) {
+  const slots = [];
+  const visit = (n) => {
+    const kids = n.children;
+    if (kids && kids.length) {
+      for (const c of kids) visit(c);
+    } else if (n.span) {
+      slots.push({ start: n.span.start, end: n.span.end });
+    }
+  };
+  if (root) visit(root);
+  return slots;
+}
+function buildOffsetIndex(root, input = "") {
+  return new OffsetIndex(collectLeafSlots(root), 0, input.length);
+}
+
+// src/cst/relative-spans.ts
+function relativize(node2, parentBase = 0) {
+  return {
+    start: node2.start - parentBase,
+    end: node2.end - parentBase,
+    children: node2.children.map((c) => relativize(c, node2.start))
+  };
+}
+function absolutize(node2, parentBase = 0) {
+  const start = parentBase + node2.start;
+  return {
+    start,
+    end: parentBase + node2.end,
+    children: node2.children.map((c) => absolutize(c, start))
+  };
+}
+function absoluteSpanAt(root, path) {
+  let base = 0;
+  let node2 = root;
+  for (const idx of path) {
+    base += node2.start;
+    const child = node2.children[idx];
+    if (!child) throw new RangeError(`absoluteSpanAt: no child ${idx} on path ${path.join(".")}`);
+    node2 = child;
+  }
+  return { start: base + node2.start, end: base + node2.end };
+}
+function shiftAbsolute(node2, at, delta) {
+  const s = node2.start >= at ? node2.start + delta : node2.start;
+  const e = node2.end >= at ? node2.end + delta : node2.end;
+  return {
+    start: s,
+    end: e,
+    children: node2.children.map((c) => shiftAbsolute(c, at, delta))
+  };
+}
+function applyEdit(root, at, delta, stats) {
+  return editNode(root, at, delta, 0, 0, stats);
+}
+function editNode(node2, at, delta, parentBase, parentShift, stats) {
+  const absStart = parentBase + node2.start;
+  const absEnd = parentBase + node2.end;
+  if (absEnd < at) return node2;
+  const startShifts = absStart >= at;
+  if (startShifts && parentShift === delta) return node2;
+  if (stats) stats.allocated++;
+  const newAbsStart = startShifts ? absStart + delta : absStart;
+  const newAbsEnd = absEnd + delta;
+  const newParentBase = parentBase + parentShift;
+  const newStart = newAbsStart - newParentBase;
+  const newEnd = newAbsEnd - newParentBase;
+  const ownShift = newAbsStart - absStart;
+  let changed = false;
+  const newChildren = [];
+  for (const c of node2.children) {
+    const nc = editNode(c, at, delta, absStart, ownShift, stats);
+    if (nc !== c) changed = true;
+    newChildren.push(nc);
+  }
+  return {
+    start: newStart,
+    end: newEnd,
+    children: changed ? newChildren : node2.children
+  };
+}
+
+// src/index.ts
+registerRegexAnalyzer(firstSetFromRegex);
 export {
+  OffsetIndex,
+  absoluteSpanAt,
+  absolutize,
   annotateSpan,
+  applyEdit,
   balanced,
+  blankLinesIn,
   buildLineIndex,
+  buildOffsetIndex,
   buildTriviaIndex,
   choice,
+  collectLeafSlots,
+  commentsIn,
   compile,
   completionsAt,
+  compose,
   createVisitor,
+  cstBuildHost,
   expect,
+  gapIsSignificant,
+  gapText,
   guard,
+  indentMixed,
+  indentWidth,
   isParseError,
   keywords,
   label,
+  lineBreaksIn,
+  lineStartWithin,
   literal,
   makeWord,
   many,
@@ -4623,14 +5409,18 @@ export {
   parse,
   parseDoc,
   parser,
+  pick,
   recover,
   ref,
   regex,
+  relativize,
   rules,
+  run,
   scanTo,
   sepBy,
   sepByRecover,
   sequence,
+  shiftAbsolute,
   skip,
   staticExpected,
   transform,