parseman 0.15.0 → 0.17.0

package/dist/index.cjs

@@ -181,6 +181,54 @@ function sequenceFirstSet(parsers) {
   }
   return fs;
 }
+function firstSetOf(p, seen = /* @__PURE__ */ new Set()) {
+  if (seen.has(p)) return any();
+  seen.add(p);
+  const fs = (c) => firstSetOf(c, seen);
+  const d = p._def;
+  switch (d.tag) {
+    case "literal":
+    case "regex":
+    case "keywords":
+      return p._meta.firstSet;
+    // terminals: no refs, cached set is exact
+    case "lazy":
+      try {
+        return fs(d.thunk());
+      } catch {
+        return any();
+      }
+    case "choice": {
+      let out = empty();
+      for (const arm of d.parsers) out = union(out, fs(arm));
+      return out;
+    }
+    case "sequence": {
+      let out = empty();
+      for (const term of d.parsers) {
+        out = union(out, fs(term));
+        if (!matchesEmpty(term)) return out;
+      }
+      return out;
+    }
+    case "oneOrMore":
+    case "many":
+    case "optional":
+    case "transform":
+    case "label":
+    case "trivia":
+    case "node":
+    case "grammar":
+    case "expect":
+      return fs(d.parser);
+    case "sepBy":
+      return fs(d.parser);
+    case "skip":
+      return fs(d.main);
+    default:
+      return p._meta.firstSet;
+  }
+}
 function mergeRanges(ranges) {
   if (ranges.length === 0) return [];
   const sorted = [...ranges].sort((a, b) => a.lo - b.lo);
@@ -358,12 +406,16 @@ function pushTriviaLogEntry(ctx, start, end, kindIndex) {
 }
 
 // src/combinators/literal.ts
-var _collatorCache = null;
-function collator() {
-  if (_collatorCache === null) {
-    _collatorCache = new Intl.Collator(void 0, { sensitivity: "accent" });
+function asciiFoldEq(a, b) {
+  if (a.length !== b.length) return false;
+  for (let i = 0; i < a.length; i++) {
+    let ca = a.charCodeAt(i);
+    let cb = b.charCodeAt(i);
+    if (ca >= 65 && ca <= 90) ca += 32;
+    if (cb >= 65 && cb <= 90) cb += 32;
+    if (ca !== cb) return false;
   }
-  return _collatorCache;
+  return true;
 }
 function literal(value, opts = {}) {
   const caseInsensitive = opts.caseInsensitive ?? false;
@@ -378,8 +430,9 @@ function literal(value, opts = {}) {
     const lower = value.toLowerCase();
     const firstUpper = upper.codePointAt(0);
     const firstLower = lower.codePointAt(0);
-    meta.firstSet = firstLower !== void 0 && firstUpper !== void 0 ? { kind: "ranges", ranges: [
-      { lo: Math.min(firstLower, firstUpper), hi: Math.max(firstLower, firstUpper) }
+    meta.firstSet = firstLower !== void 0 && firstUpper !== void 0 ? firstLower === firstUpper ? { kind: "ranges", ranges: [{ lo: firstLower, hi: firstLower }] } : { kind: "ranges", ranges: [
+      { lo: firstLower, hi: firstLower },
+      { lo: firstUpper, hi: firstUpper }
     ] } : firstSet;
   }
   return {
@@ -392,7 +445,7 @@ function literal(value, opts = {}) {
         return failAt(ctx, [JSON.stringify(value)], pos);
       }
       const slice = input.slice(pos, end);
-      const matched = caseInsensitive ? collator().compare(slice, value) === 0 : slice === value;
+      const matched = caseInsensitive ? asciiFoldEq(slice, value) : slice === value;
       if (matched) {
         const span = { start: pos, end };
         const leaf = { _tag: "leaf", value: slice, span };
@@ -1726,8 +1779,8 @@ function classToRanges(cls) {
   return parseClassRanges(body);
 }
 var foldEq = (cVar, cp) => {
-  if (cp >= 65 && cp <= 90) return `(${cVar} === ${cp} || ${cVar} === ${cp + 32})`;
-  if (cp >= 97 && cp <= 122) return `(${cVar} === ${cp} || ${cVar} === ${cp - 32})`;
+  if (cp >= 65 && cp <= 90) return `(${cVar} | 32) === ${cp + 32}`;
+  if (cp >= 97 && cp <= 122) return `(${cVar} | 32) === ${cp}`;
   return `${cVar} === ${cp}`;
 };
 function literalCodePoints(frag) {
@@ -1978,6 +2031,12 @@ function firstSetsDisjoint(a, b) {
   return classDisjoint(a, b);
 }
 function fixedClassSeq(shape) {
+  if (shape.kind === "litFold") {
+    return shape.open.map((cp) => {
+      const other = cp >= 65 && cp <= 90 ? cp + 32 : cp >= 97 && cp <= 122 ? cp - 32 : cp;
+      return other === cp ? { ranges: [[cp, cp]], negated: false } : { ranges: [[Math.min(cp, other), Math.min(cp, other)], [Math.max(cp, other), Math.max(cp, other)]], negated: false };
+    });
+  }
   if (shape.kind !== "seq") return null;
   const out = [];
   for (const p of shape.parts) {
@@ -2399,12 +2458,34 @@ function parseScanShapeCore(source) {
 }
 function scanShapeFromRegex(source, flags) {
   if (/[msu]/.test(flags)) return null;
-  if (/i/.test(flags)) {
-    const open = literalCodePoints(source);
-    return open ? { kind: "litFold", open } : null;
-  }
+  if (/i/.test(flags)) return caseFoldShape(source);
   return parseScanShape(source);
 }
+function caseFoldShape(source) {
+  const la = stripTrailingLookahead(source);
+  if (la) {
+    const inner = caseFoldLiteralOrAlt(la.base);
+    if (!inner) return null;
+    const operand = { ranges: la.ranges, negated: la.classNegated };
+    if (!lookaheadUnambiguous(inner, operand, la.negative)) return null;
+    return { kind: "lookahead", inner, ranges: la.ranges, classNegated: la.classNegated, negative: la.negative };
+  }
+  return caseFoldLiteralOrAlt(source);
+}
+function caseFoldLiteralOrAlt(source) {
+  const asciiOnly = (cps) => cps.every((cp) => cp <= 127);
+  const open = literalCodePoints(source);
+  if (open) return asciiOnly(open) ? { kind: "litFold", open } : null;
+  const arms = splitTopLevelAlternation(source);
+  if (!arms || arms.length < 2) return null;
+  const litArms = [];
+  for (const arm of arms) {
+    const cp = literalCodePoints(arm);
+    if (!cp || !asciiOnly(cp)) return null;
+    litArms.push({ kind: "litFold", open: cp });
+  }
+  return { kind: "alt", arms: litArms, disjoint: false, firsts: litArms.map(() => null) };
+}
 var classCond = (cVar, ranges) => ranges.map(([lo, hi]) => lo === hi ? `${cVar} === ${lo}` : `(${cVar} >= ${lo} && ${cVar} <= ${hi})`).join(" || ");
 var litCond = (base, cps, firstVar) => cps.map(
   (cp, k) => k === 0 && firstVar ? `${firstVar} === ${cp}` : `input.charCodeAt(${base}${k ? ` + ${k}` : ""}) === ${cp}`
@@ -2416,6 +2497,25 @@ function stringHardStop(shape) {
   );
 }
 var codeAt = (start, firstChar) => firstChar ?? `input.charCodeAt(${start})`;
+var ALT_SWITCH_RANGE_LIMIT = 4;
+var ALT_SWITCH_MAX_CASES = 48;
+var ALT_SWITCH_MIN_CASES = 3;
+function planAltSwitch(firsts) {
+  const cases = [];
+  let total = 0;
+  for (const fs of firsts) {
+    if (!fs || fs.negated) return { kind: "if" };
+    const pts = [];
+    for (const [lo, hi] of fs.ranges) {
+      if (hi - lo + 1 > ALT_SWITCH_RANGE_LIMIT) return { kind: "if" };
+      for (let cp = lo; cp <= hi; cp++) pts.push(cp);
+    }
+    total += pts.length;
+    if (total > ALT_SWITCH_MAX_CASES) return { kind: "if" };
+    cases.push(pts);
+  }
+  return total >= ALT_SWITCH_MIN_CASES ? { kind: "switch", cases } : { kind: "if" };
+}
 function emitShapeMatch(shape, start, mint, ind2, firstChar) {
   if (shape.kind === "lookahead") {
     const inner = emitShapeMatch(shape.inner, start, mint, ind2, firstChar);
@@ -2440,18 +2540,57 @@ function emitShapeMatch(shape, start, mint, ind2, firstChar) {
     const okV = mint("_ok");
     const endV2 = mint("_end");
     const lines = [`${ind2}let ${okV} = false`, `${ind2}let ${endV2} = ${start}`];
+    if (shape.arms.length >= 2 && shape.arms.every((a) => a.kind === "litFold" && a.open.length >= 1)) {
+      const groups = /* @__PURE__ */ new Map();
+      for (const a of shape.arms) {
+        const open = a.open;
+        const f = open[0];
+        const lower = f >= 65 && f <= 90 ? f + 32 : f;
+        (groups.get(lower) ?? groups.set(lower, []).get(lower)).push(open);
+      }
+      lines.push(`${ind2}if (${start} < input.length) switch (${codeAt(start, firstChar)}) {`);
+      for (const [lower, arms] of groups) {
+        const labels = lower >= 97 && lower <= 122 ? `case ${lower}: case ${lower - 32}:` : `case ${lower}:`;
+        lines.push(`${ind2}  ${labels} {`);
+        for (const open of arms) {
+          const tail = open.slice(1).map((cp, i) => foldEq(`input.charCodeAt(${start} + ${i + 1})`, cp));
+          const cond = tail.length ? ` && ${tail.join(" && ")}` : "";
+          lines.push(`${ind2}    if (${start} + ${open.length} <= input.length${cond}) { ${okV} = true; ${endV2} = ${start} + ${open.length}; break }`);
+        }
+        lines.push(`${ind2}    break`);
+        lines.push(`${ind2}  }`);
+      }
+      lines.push(`${ind2}}`);
+      return { setup: lines, ok: okV, end: endV2 };
+    }
     if (shape.disjoint) {
-      const c0 = codeAt(start, firstChar);
-      shape.arms.forEach((arm, k) => {
-        const fs = shape.firsts[k];
-        const cond = fs.negated ? `!(${classCond(c0, fs.ranges)})` : `(${classCond(c0, fs.ranges)})`;
-        const m = emitShapeMatch(arm, start, mint, `${ind2}  `, firstChar);
-        lines.push(`${ind2}${k === 0 ? "if" : "else if"} (${start} < input.length && ${cond}) {`);
-        lines.push(...m.setup);
-        lines.push(`${ind2}  ${okV} = ${m.ok}`);
-        lines.push(`${ind2}  ${endV2} = ${m.ok} ? ${m.end} : ${start}`);
+      const plan = planAltSwitch(shape.firsts);
+      if (plan.kind === "switch") {
+        lines.push(`${ind2}if (${start} < input.length) switch (${codeAt(start, firstChar)}) {`);
+        shape.arms.forEach((arm, k) => {
+          const labels = plan.cases[k].map((cp) => `case ${cp}:`).join(" ");
+          const m = emitShapeMatch(arm, start, mint, `${ind2}    `, firstChar);
+          lines.push(`${ind2}  ${labels} {`);
+          lines.push(...m.setup);
+          lines.push(`${ind2}    ${okV} = ${m.ok}`);
+          lines.push(`${ind2}    ${endV2} = ${m.ok} ? ${m.end} : ${start}`);
+          lines.push(`${ind2}    break`);
+          lines.push(`${ind2}  }`);
+        });
         lines.push(`${ind2}}`);
-      });
+      } else {
+        const c0 = codeAt(start, firstChar);
+        shape.arms.forEach((arm, k) => {
+          const fs = shape.firsts[k];
+          const cond = fs.negated ? `!(${classCond(c0, fs.ranges)})` : `(${classCond(c0, fs.ranges)})`;
+          const m = emitShapeMatch(arm, start, mint, `${ind2}  `, firstChar);
+          lines.push(`${ind2}${k === 0 ? "if" : "else if"} (${start} < input.length && ${cond}) {`);
+          lines.push(...m.setup);
+          lines.push(`${ind2}  ${okV} = ${m.ok}`);
+          lines.push(`${ind2}  ${endV2} = ${m.ok} ? ${m.end} : ${start}`);
+          lines.push(`${ind2}}`);
+        });
+      }
     } else {
       const lbl = mint("_alt");
       lines.push(`${ind2}${lbl}: {`);
@@ -2939,6 +3078,8 @@ function tryInlineDestructureTransform(src, valueVars) {
 }
 
 // src/compiler/codegen.ts
+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") {
   return `${prefix}${ctx.vars++}`;
 }
@@ -3028,16 +3169,12 @@ function namedFnPrelude() {
 function pushNamedFnDecl(ctx, fnName, bodyStmts, valueVar, endVar) {
   ctx.namedFnDecls.push([
     `function ${fnName}(input, _pos, _ctx) {`,
-    `  let _pfok = false, _pfv, _pfe = _pos`,
     `  _pfail: {`,
     ...reindentStmts(bodyStmts, 2),
-    `    _pfv = ${valueVar}`,
-    `    _pfe = ${endVar}`,
-    `    _pfok = true`,
+    `    ${NAMED_FN_END} = ${endVar}`,
+    `    return ${valueVar}`,
     `  }`,
-    `  if (!_pfok) return ${NAMED_FN_FAIL}`,
-    `  ${NAMED_FN_END} = _pfe`,
-    `  return _pfv`,
+    `  return ${NAMED_FN_FAIL}`,
     `}`
   ].join("\n"));
 }
@@ -3337,13 +3474,14 @@ function emitLit(def, ctx, pos) {
   const vv = v(ctx);
   const expectedStr = JSON.stringify(JSON.stringify(value));
   const stmts = [];
-  if (caseInsensitive) {
-    ctx.needsCollator = true;
+  if (caseInsensitive && len > 0) {
+    const match = Array.from(
+      { length: len },
+      (_, i) => `(${foldEq(`input.charCodeAt(${pos}${i > 0 ? ` + ${i}` : ""})`, value.charCodeAt(i))})`
+    ).join(" && ");
     stmts.push(
-      ...emitIfFail(ctx, `${pos} + ${len} > input.length`, failBody(ctx, expectedStr, pos)),
-      `${ind(ctx)}const ${vv}_s = input.slice(${pos}, ${pos} + ${len})`,
-      ...emitIfFail(ctx, `_collator.compare(${vv}_s, ${JSON.stringify(value)}) !== 0`, failBody(ctx, expectedStr, pos)),
-      `${ind(ctx)}const ${vv} = ${vv}_s`
+      ...emitIfFail(ctx, `${pos} + ${len} > input.length || !(${match})`, failBody(ctx, expectedStr, pos)),
+      `${ind(ctx)}const ${vv} = input.slice(${pos}, ${pos} + ${len})`
     );
   } else if (len === 0) {
     stmts.push(`${ind(ctx)}const ${vv} = ''`);
@@ -3464,6 +3602,7 @@ function emitRegex(def, ctx, pos) {
       return { stmts: stmts2, valueVar: vv2, endVar: scanned.endVar };
     }
   }
+  _loweringSink?.add(`/${def.source}/${def.flags}`);
   const flags = "y" + def.flags.replace(/[gy]/g, "");
   const key = `${def.source}/${flags}`;
   let rName = ctx.regexMap.get(key);
@@ -3736,11 +3875,16 @@ function emitFirstMatch(def, ctx, pos) {
     const atStart = failsAtStart(p);
     const staticFx = armStaticExpected(ctx, p);
     const deferRuleName = ctx.deferFirstSetRefs ? p._ruleName : void 0;
-    const fsGuard = deferRuleName !== void 0 ? `/*@FS:${deferRuleName}:${codeV}@*/true` : needsFirstSetGuard(p) ? firstSetCond(codeV, p._meta.firstSet) : null;
+    const fsGuard = deferRuleName !== void 0 ? `/*@FS:${deferRuleName}:${codeV}@*/true` : needsFirstSetGuard(p) ? firstSetCond(codeV, p._meta.firstSet) : (() => {
+      if (p._meta.firstSet.kind !== "any" || matchesEmpty(p)) return null;
+      const deep = firstSetOf(p);
+      return deep.kind === "ranges" ? firstSetCond(codeV, deep) : null;
+    })();
     let gateCond = null;
     if (gate) {
       const gateIdx = ctx.mapFns.length;
       ctx.mapFns.push(gate);
+      ctx.mapFnSrcs.push(null);
       gateCond = `${mfRef(ctx)}[${gateIdx}](_ctx.state)`;
     }
     const skipCond = gateCond ? `!${resOkV} && ${gateCond}` : `!${resOkV}`;
@@ -3799,13 +3943,21 @@ function emitNonDisjoint(def, strategy, allExpected, ctx, pos) {
     return emitLiteralsLongestFirst(def, strategy.sortedIndices, allExpected, ctx, pos);
   return emitFirstMatch(def, ctx, pos);
 }
+function pushMapFn(ctx, fn, src) {
+  if (src !== null && ctx.mapFns.length === ctx.mapFnSrcs.length) {
+    const hit = ctx.mapFnSrcs.indexOf(src);
+    if (hit !== -1) return hit;
+  }
+  const idx = ctx.mapFns.length;
+  ctx.mapFns.push(fn);
+  ctx.mapFnSrcs.push(src);
+  return idx;
+}
 function emitTransformChain(p, baseValue, endV, startPos, ctx) {
   const def = p._def;
   if (def.tag === "transform") {
     const innerR = emitTransformChain(def.parser, baseValue, endV, startPos, ctx);
-    const fnIdx = ctx.mapFns.length;
-    ctx.mapFns.push(def.fn);
-    ctx.mapFnSrcs.push(def.fnSrc ?? null);
+    const fnIdx = pushMapFn(ctx, def.fn, def.fnSrc ?? null);
     const vv = v(ctx);
     return {
       stmts: [...innerR.stmts, `${ind(ctx)}const ${vv} = ${mfRef(ctx)}[${fnIdx}](${innerR.valueVar}, { start: ${startPos}, end: ${endV} })`],
@@ -3988,7 +4140,7 @@ function emitScanTo(def, ctx, pos) {
     `${ind(ctx)}while (${curV} < input.length) {`
   ];
   ctx.indent++;
-  const probeCtx = { ...ctx, capturing: false };
+  const probeCtx = { ...ctx, capturing: false, noHoist: true };
   const { stmts: sentStmts, okVar: sentOk } = emitFallible(def.sentinel, probeCtx, curV);
   stmts.push(...sentStmts, `${ind(ctx)}if (${sentOk}) { ${foundV} = true; break }`);
   if (def.skip.length > 0) {
@@ -4044,15 +4196,19 @@ function emitNode(def, ctx, pos) {
     ctx.buildSrcs.push(def.buildSrc ?? null);
   }
   const i = ind(ctx);
-  const capturesTrivia = mkType !== null || structural || buildReadsTrivia(def);
-  const clonesState = structural || buildReadsState(def);
+  const capturesTrivia = mkType !== null || !structural && buildReadsTrivia(def);
+  const clonesState = !structural && buildReadsState(def);
   const chV = v(ctx, "_ch");
   const rawV = v(ctx, "_raw");
-  const tlV = capturesTrivia ? v(ctx, "_tl") : "_EMPTY_TL";
+  const capTLv = structural ? v(ctx, "_ctl") : null;
+  const capSTv = structural ? v(ctx, "_cst") : null;
+  const tlV = capturesTrivia || structural ? v(ctx, "_tl") : "_EMPTY_TL";
   if (!capturesTrivia) ctx.needsEmptyTl = true;
   const sc = v(ctx, "_sc"), sl = v(ctx, "_sl"), sr = v(ctx, "_sr"), st = v(ctx, "_st"), stl = v(ctx, "_stl");
-  const allocStmt = capturesTrivia ? `${i}const ${chV} = [], ${rawV} = [], ${tlV} = []` : `${i}const ${chV} = [], ${rawV} = []`;
-  const innerTl = capturesTrivia ? tlV : "undefined";
+  if (structural) ctx.needsHostReads = true;
+  const allocStmt = structural ? `${i}const ${capTLv} = _ctx._pmCapTL ??= _hostReads(_ctx.build, 4), ${capSTv} = _ctx._pmCapST ??= (_ctx.build === undefined || _hostReads(_ctx.build, 5))
+${i}const ${chV} = [], ${rawV} = [], ${tlV} = ${capTLv} ? [] : _EMPTY_TL` : capturesTrivia ? `${i}const ${chV} = [], ${rawV} = [], ${tlV} = []` : `${i}const ${chV} = [], ${rawV} = []`;
+  const innerTl = structural ? `${capTLv} ? ${tlV} : undefined` : capturesTrivia ? tlV : "undefined";
   const stmts = [
     allocStmt,
     `${i}const ${sc} = _ctx._cstChildren, ${sl} = _ctx._cstLeaves, ${sr} = _ctx._cstRawChildren, ${st} = _ctx.captureTrivia, ${stl} = _ctx._cstTriviaLog`,
@@ -4063,7 +4219,10 @@ function emitNode(def, ctx, pos) {
   stmts.push(`${i}_ctx._cstChildren = ${sc}; _ctx._cstLeaves = ${sl}; _ctx._cstRawChildren = ${sr}; _ctx.captureTrivia = ${st}; _ctx._cstTriviaLog = ${stl}`);
   stmts.push(...emitIfFail(ctx, `!${okVar}`, propagateFailBody(ctx)));
   let stV = "undefined";
-  if (clonesState) {
+  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 ${stV} = _ctx.state !== undefined ? Object.assign({}, _ctx.state) : undefined`);
   }
@@ -4172,6 +4331,32 @@ function emitExpect(def, ctx, pos) {
   return { stmts, valueVar: valVar, endVar };
 }
 function emit(p, ctx, pos) {
+  const usage = ctx.lazyUsage;
+  if (usage && !ctx.noHoist && !ctx.capAsTrivia && !ctx.ruleNames?.has(p) && isHoistableTag(p._def.tag) && (usage.counts.get(p) ?? 0) > 1 && // Only hoist subtrees big enough that de-duplicating them beats the call
+  // overhead. A tiny shared wrapper (e.g. `optional(literal(';'))`) inlines
+  // faster than it calls and barely moves the byte count; the size explosion
+  // is all in LARGE choices/sequences referenced from many positions.
+  (usage.sizes.get(p) ?? 0) >= HOIST_MIN_SUBTREE) {
+    const existing = ctx.namedParsers.get(p);
+    if (existing !== void 0) return emitNamedFnCall(ctx, existing, pos);
+    const fnName = `${nsp(ctx)}_pf${ctx.namedParsers.size}`;
+    ctx.namedParsers.set(p, fnName);
+    const savedIndent = ctx.indent;
+    const savedFailLabel = ctx.failLabel;
+    const savedRecord = ctx.recordFail;
+    ctx.indent = 1;
+    ctx.failLabel = "_pfail";
+    ctx.recordFail = true;
+    const r = emitDispatch(p, ctx, "_pos");
+    ctx.indent = savedIndent;
+    ctx.failLabel = savedFailLabel;
+    ctx.recordFail = savedRecord;
+    pushNamedFnDecl(ctx, fnName, r.stmts, r.valueVar, r.endVar);
+    return emitNamedFnCall(ctx, fnName, pos);
+  }
+  return emitDispatch(p, ctx, pos);
+}
+function emitDispatch(p, ctx, pos) {
   const def = p._def;
   switch (def.tag) {
     case "literal":
@@ -4217,9 +4402,7 @@ function emit(p, ctx, pos) {
           };
         }
       }
-      const fnIdx = ctx.mapFns.length;
-      ctx.mapFns.push(def.fn);
-      ctx.mapFnSrcs.push(def.fnSrc ?? null);
+      const fnIdx = pushMapFn(ctx, def.fn, def.fnSrc ?? null);
       const mv = v(ctx, "_mapped");
       return {
         stmts: [
@@ -4284,6 +4467,7 @@ function emit(p, ctx, pos) {
     case "guard": {
       const fnIdx = ctx.mapFns.length;
       ctx.mapFns.push(def.predicate);
+      ctx.mapFnSrcs.push(null);
       const vv = v(ctx);
       return {
         stmts: [
@@ -4298,6 +4482,7 @@ function emit(p, ctx, pos) {
       const evIdx = ctx.mapFns.length;
       const extra = def.extra;
       ctx.mapFns.push((() => extra));
+      ctx.mapFnSrcs.push(null);
       const innerParser = def.parser;
       if (!ctx.namedParsers.has(innerParser)) {
         const fnName = `_wcf${ctx.namedParsers.size}`;
@@ -4417,9 +4602,58 @@ function childrenOf(def) {
       return [];
   }
 }
+var HOISTABLE_TAGS = /* @__PURE__ */ new Set([
+  "sequence",
+  "choice",
+  "many",
+  "oneOrMore",
+  "optional",
+  "sepBy",
+  "transform",
+  "node",
+  "not"
+]);
+function isHoistableTag(tag) {
+  return HOISTABLE_TAGS.has(tag);
+}
+var HOIST_MIN_SUBTREE = 3;
 function analyzeLazyUsage(root) {
   return analyzeLazyUsageMulti([root]);
 }
+function subtreeSizes(roots) {
+  const sizes = /* @__PURE__ */ new Map();
+  const sz = (p) => {
+    const cached = sizes.get(p);
+    if (cached !== void 0) return cached;
+    const def = p._def;
+    let s = 1;
+    if (def.tag !== "lazy") {
+      for (const child of childrenOf(def)) s += child._def.tag === "lazy" ? 1 : sz(child);
+    }
+    sizes.set(p, s);
+    return s;
+  };
+  const visitedLazy = /* @__PURE__ */ new Set();
+  const visit = (p) => {
+    const def = p._def;
+    if (def.tag === "lazy") {
+      if (visitedLazy.has(p)) return;
+      visitedLazy.add(p);
+      let resolved;
+      try {
+        resolved = def.thunk();
+      } catch {
+        return;
+      }
+      visit(resolved);
+      return;
+    }
+    sz(p);
+    for (const child of childrenOf(def)) visit(child);
+  };
+  for (const root of roots) visit(root);
+  return sizes;
+}
 function analyzeLazyUsageMulti(roots) {
   const counts = /* @__PURE__ */ new Map();
   const recursive = /* @__PURE__ */ new Set();
@@ -4446,10 +4680,14 @@ function analyzeLazyUsageMulti(roots) {
       active.delete(p);
       return;
     }
+    if (isHoistableTag(def.tag)) {
+      counts.set(p, (counts.get(p) ?? 0) + 1);
+    }
     for (const child of childrenOf(def)) walk2(child);
   }
-  for (const root of roots) walk2(root);
-  return { counts, recursive };
+  const rootsArr = [...roots];
+  for (const root of rootsArr) walk2(root);
+  return { counts, recursive, sizes: subtreeSizes(rootsArr) };
 }
 function ruleDependencies(ruleMap) {
   const nameOf = /* @__PURE__ */ new Map();
@@ -4499,7 +4737,6 @@ function compile(combinator, mapFnSources) {
     buildFns: [],
     buildSrcs: [],
     runtimeParsers: [],
-    needsCollator: false,
     namedParsers: /* @__PURE__ */ new Map(),
     triviaCaptureNames: /* @__PURE__ */ new Map(),
     triviaFnNames: /* @__PURE__ */ new Map(),
@@ -4508,18 +4745,18 @@ function compile(combinator, mapFnSources) {
     lazyUsage: analyzeLazyUsage(combinator)
   };
   const r = emit(combinator, ctx, "_pos");
-  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 hostReadsDecls = ctx.needsHostReads ? [HOST_READS_DECL] : [];
   const source = [
     ...emptyTlDecls,
+    ...hostReadsDecls,
     ...ctx.regexDecls,
     ...ctx.expectedDecls,
     "",
     ...namedPrelude,
     ctx.namedFnDecls.join("\n\n"),
-    `${collatorDecl}function _parse(input, _pos, _rp, _mf, _build, _ctx) {`,
+    `function _parse(input, _pos, _rp, _mf, _build, _ctx) {`,
     `  let pos = _pos`,
     ...r.stmts,
     `  return { ok: true, value: ${r.valueVar}, span: { start: _pos, end: ${r.endVar} } }`,
@@ -4527,9 +4764,9 @@ function compile(combinator, mapFnSources) {
   ].join("\n");
   const fn = new Function("input", "_pos", "_rp", "_mf", "_build", "_ctx", [
     ...emptyTlDecls,
+    ...hostReadsDecls,
     ...ctx.regexDecls,
     ...ctx.expectedDecls,
-    collatorDecl,
     ...namedPrelude,
     ...ctx.namedFnDecls.flatMap((decl, i) => i > 0 ? ["", decl] : [decl]),
     `let pos = _pos`,
@@ -4546,7 +4783,7 @@ function compile(combinator, mapFnSources) {
   const buildSources = ctx.buildFns.length === 0 ? void 0 : ctx.buildSrcs;
   const mfCovered = ctx.mapFns.length === 0 || effectiveSources !== void 0 && effectiveSources.length === ctx.mapFns.length;
   const canInline = ctx.runtimeParsers.length === 0 && mfCovered && buildCovered;
-  const inlineExpression = canInline ? buildInlineExpression(ctx, r, collatorDecl, effectiveSources, buildSources) : null;
+  const inlineExpression = canInline ? buildInlineExpression(ctx, r, effectiveSources, buildSources) : null;
   return {
     source,
     inlineExpression,
@@ -4602,7 +4839,6 @@ function compileLinkable(ruleMapArg, ns) {
     buildFns: [],
     buildSrcs: [],
     runtimeParsers: [],
-    needsCollator: false,
     namedParsers: /* @__PURE__ */ new Map(),
     triviaCaptureNames: /* @__PURE__ */ new Map(),
     triviaFnNames: /* @__PURE__ */ new Map(),
@@ -4733,12 +4969,12 @@ function compileLinkable(ruleMapArg, ns) {
     firstSets,
     deps: ruleDependencies(ruleMap),
     needsEmptyTl: !!ctx.needsEmptyTl,
-    needsCollator: ctx.needsCollator,
+    needsHostReads: !!ctx.needsHostReads,
     mfFns: mfSrcs ? [] : ctx.mapFns,
     buildFns: buildSrcs ? [] : ctx.buildFns
   };
 }
-function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources) {
+function buildInlineExpression(ctx, r, mapFnSources, buildSources) {
   const bodyLines = [
     `  let pos = _pos`,
     ...r.stmts,
@@ -4752,14 +4988,15 @@ function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources)
   const mfDecl = mapFnSources?.length ? `  const _mf = [${mapFnSources.join(", ")}]` : "";
   const buildDecl = buildSources?.length ? `  const _build = [${buildSources.join(", ")}]` : "";
   const emptyTlDecl = ctx.needsEmptyTl ? `  const _EMPTY_TL = Object.freeze([])` : "";
-  const needsWrapper = ctx.regexDecls.length > 0 || ctx.expectedDecls.length > 0 || !!collatorDecl || ctx.namedFnDecls.length > 0 || !!mfDecl || !!buildDecl || !!emptyTlDecl;
+  const hostReadsDecl = ctx.needsHostReads ? `  ${HOST_READS_DECL}` : "";
+  const needsWrapper = ctx.regexDecls.length > 0 || ctx.expectedDecls.length > 0 || ctx.namedFnDecls.length > 0 || !!mfDecl || !!buildDecl || !!emptyTlDecl || !!hostReadsDecl;
   if (!needsWrapper) return innerFn;
   const namedPrelude = ctx.namedFnDecls.length > 0 ? namedFnPrelude() : [];
   const hoistedDecls = [
     emptyTlDecl,
+    hostReadsDecl,
     ...ctx.regexDecls.map((d) => `  ${d}`),
     ...ctx.expectedDecls.map((d) => `  ${d}`),
-    collatorDecl ? `  ${collatorDecl.trim()}` : "",
     mfDecl,
     buildDecl,
     ...namedPrelude.map((l) => `  ${l}`),
@@ -4776,6 +5013,172 @@ function buildInlineExpression(ctx, r, collatorDecl, mapFnSources, buildSources)
   ].join("\n");
 }
 
+// src/combinators/scanTo.ts
+function scanTo(sentinel, { skip: skip2 = [], orEOF = false } = {}) {
+  const meta = {
+    firstSet: any(),
+    canMatchNewline: true,
+    isTrivia: false
+  };
+  return {
+    _tag: "scanTo",
+    _meta: meta,
+    _def: { tag: "scanTo", sentinel, skip: skip2, orEOF },
+    parse(input, pos, ctx) {
+      let cur = pos;
+      const probeCtx = {
+        trackLines: ctx.trackLines,
+        state: ctx.state,
+        ...ctx._errors !== void 0 ? { _errors: ctx._errors } : {}
+      };
+      const emit2 = (end) => {
+        if (end > pos && cstCaptureActive(ctx)) {
+          const leaf = { _tag: "leaf", value: input.slice(pos, end), span: { start: pos, end } };
+          pushCstLeaf(ctx, leaf);
+        }
+      };
+      while (cur < input.length) {
+        const s = sentinel.parse(input, cur, probeCtx);
+        if (s.ok) {
+          emit2(cur);
+          return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
+        }
+        let advanced = false;
+        for (const skipper of skip2) {
+          const r = skipper.parse(input, cur, probeCtx);
+          if (r.ok && r.span.end > cur) {
+            cur = r.span.end;
+            advanced = true;
+            break;
+          }
+        }
+        if (!advanced) cur++;
+      }
+      if (orEOF) {
+        emit2(cur);
+        return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
+      }
+      const sentDef = sentinel._def;
+      const expected = sentDef.tag === "literal" ? [JSON.stringify(sentDef.value)] : ["sentinel"];
+      return { ok: false, expected, span: { start: pos, end: cur } };
+    }
+  };
+}
+function balanced(open, close, options = {}) {
+  const self = ref();
+  const skips = options.skip ?? [];
+  const stop = /* @__PURE__ */ new Set([open, close]);
+  let bounded = true;
+  for (const sk of skips) {
+    const cs = firstSetClassChars(sk);
+    if (cs === null) {
+      bounded = false;
+      break;
+    }
+    for (const ch of cs) stop.add(ch);
+  }
+  const cls = [...stop].map(escapeClassChar).join("");
+  const content = bounded ? regex(new RegExp(`[^${cls}]+`)) : regex(new RegExp(`[^${escapeClassChar(open)}${escapeClassChar(close)}]`));
+  const inner = many(choice(self, ...skips, content));
+  const combi = transform(
+    sequence(literal(open), inner, expect(literal(close), close)),
+    // parts: strings (content/self) or arrays (a sequence-shaped skip) or a
+    // ParseError (recovered close). `c` is the close string or a ParseError.
+    ([o, parts, c]) => o + parts.map((p) => typeof p === "string" ? p : Array.isArray(p) ? p.join("") : "").join("") + (typeof c === "string" ? c : "")
+  );
+  if (combi._def.tag === "transform") {
+    combi._def.fnSrc = '([o, parts, c]) => o + parts.map(p => typeof p === "string" ? p : Array.isArray(p) ? p.join("") : "").join("") + (typeof c === "string" ? c : "")';
+  }
+  self.define(combi);
+  return combi;
+}
+function firstSetClassChars(c) {
+  const fs = c._meta.firstSet;
+  if (fs.kind === "empty") return [];
+  if (fs.kind !== "ranges") return null;
+  const out = [];
+  for (const { lo, hi } of fs.ranges) {
+    if (hi - lo > 8) return null;
+    for (let cp = lo; cp <= hi; cp++) out.push(String.fromCodePoint(cp));
+  }
+  return out;
+}
+function escapeClassChar(ch) {
+  return ch.replace(/[\\\]^-]/g, "\\$&");
+}
+
+// src/compiler/ir-serialize.ts
+function evalRuleMapIR(ir) {
+  const _tf = (child, src) => {
+    let fn2;
+    try {
+      fn2 = (0, eval)(`(${src})`);
+    } catch {
+      fn2 = () => {
+        throw new Error("IR transform fn not materialized");
+      };
+    }
+    const t = transform(child, fn2);
+    t._def.fnSrc = src;
+    return t;
+  };
+  const _nd = (type, child, src, opts) => {
+    const n = node(type, child, void 0, opts);
+    n._def.buildSrc = src;
+    return n;
+  };
+  const fn = new Function(
+    "rules",
+    "ref",
+    "regex",
+    "literal",
+    "keywords",
+    "sequence",
+    "choice",
+    "many",
+    "oneOrMore",
+    "optional",
+    "sepBy",
+    "not",
+    "node",
+    "parser",
+    "scanTo",
+    "transform",
+    "skip",
+    "trivia",
+    "label",
+    "expect",
+    "_tf",
+    "_nd",
+    `return (${ir})`
+  );
+  const map = fn(
+    rules,
+    ref,
+    regex,
+    literal,
+    keywords,
+    sequence,
+    choice,
+    many,
+    oneOrMore,
+    optional,
+    sepBy,
+    not,
+    node,
+    parser,
+    scanTo,
+    transform,
+    skip,
+    trivia,
+    label,
+    expect,
+    _tf,
+    _nd
+  );
+  return Object.entries(map);
+}
+
 // src/compiler/linker.ts
 var _nsCounter = 0;
 function linkable(rulesMap, ns) {
@@ -4818,13 +5221,13 @@ function fusedBody(pieces) {
   }
   const contributing = new Set(winner.values());
   const needsEmptyTl = [...contributing].some((p) => p.needsEmptyTl);
-  const needsCollator = [...contributing].some((p) => p.needsCollator);
+  const needsHostReads = [...contributing].some((p) => p.needsHostReads);
   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' })"] : [],
+    ...needsHostReads ? [HOST_READS_DECL] : [],
     // 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).
@@ -4860,9 +5263,18 @@ function fuseRules(pieces) {
   return new Function("_env", body)(env);
 }
 var COMPOSED_PIECES = /* @__PURE__ */ Symbol.for("parseman.composedPieces");
+function isIRPiece(p) {
+  return !!p && typeof p === "object" && typeof p.ir === "string" && typeof p.ns === "string" && !("ruleFns" in p);
+}
+function materializePiece(p) {
+  if (!isIRPiece(p)) return p;
+  const pieces = compileLinkable(evalRuleMapIR(p.ir), p.ns);
+  if (!pieces) throw new Error(`compose: carried IR for ns "${p.ns}" could not be re-lowered`);
+  return pieces;
+}
 function itemPieces(item) {
   const carried = item[COMPOSED_PIECES];
-  if (Array.isArray(carried)) return carried;
+  if (Array.isArray(carried)) return carried.map(materializePiece);
   if (item.ruleFns instanceof Map) return [item];
   return [linkable(item)];
 }
@@ -4958,100 +5370,6 @@ function deeperFail(a, b) {
   return a.span.start >= b.span.start ? a : b;
 }
 
-// src/combinators/scanTo.ts
-function scanTo(sentinel, { skip: skip2 = [], orEOF = false } = {}) {
-  const meta = {
-    firstSet: any(),
-    canMatchNewline: true,
-    isTrivia: false
-  };
-  return {
-    _tag: "scanTo",
-    _meta: meta,
-    _def: { tag: "scanTo", sentinel, skip: skip2, orEOF },
-    parse(input, pos, ctx) {
-      let cur = pos;
-      const probeCtx = {
-        trackLines: ctx.trackLines,
-        state: ctx.state,
-        ...ctx._errors !== void 0 ? { _errors: ctx._errors } : {}
-      };
-      const emit2 = (end) => {
-        if (end > pos && cstCaptureActive(ctx)) {
-          const leaf = { _tag: "leaf", value: input.slice(pos, end), span: { start: pos, end } };
-          pushCstLeaf(ctx, leaf);
-        }
-      };
-      while (cur < input.length) {
-        const s = sentinel.parse(input, cur, probeCtx);
-        if (s.ok) {
-          emit2(cur);
-          return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
-        }
-        let advanced = false;
-        for (const skipper of skip2) {
-          const r = skipper.parse(input, cur, probeCtx);
-          if (r.ok && r.span.end > cur) {
-            cur = r.span.end;
-            advanced = true;
-            break;
-          }
-        }
-        if (!advanced) cur++;
-      }
-      if (orEOF) {
-        emit2(cur);
-        return { ok: true, value: input.slice(pos, cur), span: { start: pos, end: cur } };
-      }
-      const sentDef = sentinel._def;
-      const expected = sentDef.tag === "literal" ? [JSON.stringify(sentDef.value)] : ["sentinel"];
-      return { ok: false, expected, span: { start: pos, end: cur } };
-    }
-  };
-}
-function balanced(open, close, options = {}) {
-  const self = ref();
-  const skips = options.skip ?? [];
-  const stop = /* @__PURE__ */ new Set([open, close]);
-  let bounded = true;
-  for (const sk of skips) {
-    const cs = firstSetClassChars(sk);
-    if (cs === null) {
-      bounded = false;
-      break;
-    }
-    for (const ch of cs) stop.add(ch);
-  }
-  const cls = [...stop].map(escapeClassChar).join("");
-  const content = bounded ? regex(new RegExp(`[^${cls}]+`)) : regex(new RegExp(`[^${escapeClassChar(open)}${escapeClassChar(close)}]`));
-  const inner = many(choice(self, ...skips, content));
-  const combi = transform(
-    sequence(literal(open), inner, expect(literal(close), close)),
-    // parts: strings (content/self) or arrays (a sequence-shaped skip) or a
-    // ParseError (recovered close). `c` is the close string or a ParseError.
-    ([o, parts, c]) => o + parts.map((p) => typeof p === "string" ? p : Array.isArray(p) ? p.join("") : "").join("") + (typeof c === "string" ? c : "")
-  );
-  if (combi._def.tag === "transform") {
-    combi._def.fnSrc = '([o, parts, c]) => o + parts.map(p => typeof p === "string" ? p : Array.isArray(p) ? p.join("") : "").join("") + (typeof c === "string" ? c : "")';
-  }
-  self.define(combi);
-  return combi;
-}
-function firstSetClassChars(c) {
-  const fs = c._meta.firstSet;
-  if (fs.kind === "empty") return [];
-  if (fs.kind !== "ranges") return null;
-  const out = [];
-  for (const { lo, hi } of fs.ranges) {
-    if (hi - lo > 8) return null;
-    for (let cp = lo; cp <= hi; cp++) out.push(String.fromCodePoint(cp));
-  }
-  return out;
-}
-function escapeClassChar(ch) {
-  return ch.replace(/[\\\]^-]/g, "\\$&");
-}
-
 // src/combinators/recover-list.ts
 function tolerantElement(item, sentinel, until) {
   const el = transform(