parseman 0.61.5 → 0.61.6

package/dist/plugin/entry-supercompiler.d.ts

@@ -36,6 +36,21 @@ export type EntrySupercompileResult = EmitResult & {
  * externally-addressable scan roots passed to the canonical emitter.
  */
 export declare function supercompileEntryAssembly(prog: TableProgram, cfg: EntrySupercompileCfg, extraIps?: readonly number[]): EntrySupercompileResult;
+/**
+ * Functions that read the same are one function: `optional(x)` written in 25 places
+ * is 25 rows and one body, and a trivia skip can be pooled under two labels that
+ * lower alike. This runs after the join, so a site with one caller has
+ * already gone into it, the selection is untouched and every call a parse makes is
+ * still made. Each pass can make the callers of a merged function read alike, so it
+ * repeats until nothing merges.
+ *
+ * Only identifiers are compared or renamed: a grammar's own text reaches this source
+ * as strings and regexes, and they count as written. A temporary is local by the
+ * emitter's contract and is numbered per assembly, so one this function binds is
+ * compared in order of use. A site name that's also bound somewhere, or abbreviated
+ * in `{_pf1}`, can't be renamed by name, so that source stays as it is.
+ */
+export declare function mergeSameBodies(source: string): string;
 /**
  * Plugin-only final splice for an ordinary macro replacement. The table
  * compiler emits its canonical callback-free public artifact first; this pass

package/dist/plugin/entry-supercompiler.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"entry-supercompiler.d.ts","sourceRoot":"","sources":["../../src/plugin/entry-supercompiler.ts"],"names":[],"mappings":"AAYA,OAAO,EAAsB,KAAK,UAAU,EAAE,MAAM,2BAA2B,CAAA;AAE/E,OAAO,EAAkC,KAAK,YAAY,EAAE,MAAM,qBAAqB,CAAA;AAKvF,MAAM,MAAM,oBAAoB,GAAG;IACjC,QAAQ,CAAC,OAAO,EAAE,OAAO,CAAA;IACzB,QAAQ,CAAC,iBAAiB,CAAC,EAAE,OAAO,CAAA;IACpC,QAAQ,CAAC,iBAAiB,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,GAAG,SAAS,CAAA;IACpE,QAAQ,CAAC,UAAU,EAAE,OAAO,CAAA;IAC5B,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAA;IAC1B,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAA;IAC1B,QAAQ,CAAC,KAAK,EAAE,OAAO,CAAA;CACxB,CAAA;AAED,MAAM,MAAM,sBAAsB,GAAG;IACnC,QAAQ,CAAC,SAAS,EAAE,OAAO,CAAA;IAC3B,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAA;IACxB,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAA;IAChC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAA;IAC/B,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAA;IACjC,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAA;IAChC,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAA;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAA;IAC5B,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAA;IACxB,QAAQ,CAAC,mBAAmB,EAAE,MAAM,CAAA;IACpC,QAAQ,CAAC,eAAe,CAAC,EAAE,MAAM,CAAA;IACjC,QAAQ,CAAC,eAAe,EAAE,QAAQ,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,WAAW,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;IACvF,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;CACpD,CAAA;AAED,MAAM,MAAM,uBAAuB,GAAG,UAAU,GAAG;IAAE,QAAQ,CAAC,YAAY,EAAE,sBAAsB,CAAA;CAAE,CAAA;AAyKpG;;;GAGG;AACH,wBAAgB,yBAAyB,CACvC,IAAI,EAAE,YAAY,EAClB,GAAG,EAAE,oBAAoB,EACzB,QAAQ,GAAE,SAAS,MAAM,EAAO,GAC/B,uBAAuB,CAGzB;AAED;;;;;GAKG;AACH,wBAAgB,8BAA8B,CAC5C,IAAI,EAAE,YAAY,EAClB,WAAW,EAAE,MAAM,EACnB,MAAM,CAAC,EAAE,WAAW,CAAC,MAAM,CAAC,GAC3B,MAAM,CA0CR;AAED,8EAA8E;AAC9E,wBAAgB,2BAA2B,CACzC,IAAI,EAAE,YAAY,EAClB,IAAI,EAAE,oBAAoB,EAC1B,QAAQ,EAAE,SAAS,MAAM,EAAE,EAC3B,OAAO,EAAE,UAAU,GAClB,uBAAuB,CAwRzB"}
+{"version":3,"file":"entry-supercompiler.d.ts","sourceRoot":"","sources":["../../src/plugin/entry-supercompiler.ts"],"names":[],"mappings":"AAYA,OAAO,EAAsB,KAAK,UAAU,EAAE,MAAM,2BAA2B,CAAA;AAE/E,OAAO,EAAkC,KAAK,YAAY,EAAE,MAAM,qBAAqB,CAAA;AAKvF,MAAM,MAAM,oBAAoB,GAAG;IACjC,QAAQ,CAAC,OAAO,EAAE,OAAO,CAAA;IACzB,QAAQ,CAAC,iBAAiB,CAAC,EAAE,OAAO,CAAA;IACpC,QAAQ,CAAC,iBAAiB,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,GAAG,SAAS,CAAA;IACpE,QAAQ,CAAC,UAAU,EAAE,OAAO,CAAA;IAC5B,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAA;IAC1B,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAA;IAC1B,QAAQ,CAAC,KAAK,EAAE,OAAO,CAAA;CACxB,CAAA;AAED,MAAM,MAAM,sBAAsB,GAAG;IACnC,QAAQ,CAAC,SAAS,EAAE,OAAO,CAAA;IAC3B,QAAQ,CAAC,MAAM,CAAC,EAAE,MAAM,CAAA;IACxB,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAA;IAChC,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAA;IAC/B,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAA;IACjC,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAA;IAChC,QAAQ,CAAC,YAAY,EAAE,MAAM,CAAA;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAA;IAC5B,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAA;IACxB,QAAQ,CAAC,mBAAmB,EAAE,MAAM,CAAA;IACpC,QAAQ,CAAC,eAAe,CAAC,EAAE,MAAM,CAAA;IACjC,QAAQ,CAAC,eAAe,EAAE,QAAQ,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,WAAW,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;IACvF,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;CACpD,CAAA;AAED,MAAM,MAAM,uBAAuB,GAAG,UAAU,GAAG;IAAE,QAAQ,CAAC,YAAY,EAAE,sBAAsB,CAAA;CAAE,CAAA;AA2KpG;;;GAGG;AACH,wBAAgB,yBAAyB,CACvC,IAAI,EAAE,YAAY,EAClB,GAAG,EAAE,oBAAoB,EACzB,QAAQ,GAAE,SAAS,MAAM,EAAO,GAC/B,uBAAuB,CAGzB;AAgCD;;;;;;;;;;;;;GAaG;AACH,wBAAgB,eAAe,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CA6DtD;AAED;;;;;GAKG;AACH,wBAAgB,8BAA8B,CAC5C,IAAI,EAAE,YAAY,EAClB,WAAW,EAAE,MAAM,EACnB,MAAM,CAAC,EAAE,WAAW,CAAC,MAAM,CAAC,GAC3B,MAAM,CA0CR;AAED,8EAA8E;AAC9E,wBAAgB,2BAA2B,CACzC,IAAI,EAAE,YAAY,EAClB,IAAI,EAAE,oBAAoB,EAC1B,QAAQ,EAAE,SAAS,MAAM,EAAE,EAC3B,OAAO,EAAE,UAAU,GAClB,uBAAuB,CAwRzB"}

package/dist/plugin/index.js

@@ -5630,7 +5630,7 @@ function sourceLeaf(root, fn) {
 }
 
 // src/version.ts
-var PARSEMAN_VERSION = "0.61.5";
+var PARSEMAN_VERSION = "0.61.6";
 
 // src/table/lex-body.ts
 function buildLexBodyRecognizer(spec, constants) {
@@ -19464,21 +19464,70 @@ function runDuplicationDiagnosticRules(ruleMap, opt) {
 var LOOP = 8;
 var ROUNDS = 60;
 var CAP = 1e12;
+var CROSSING_WEIGHT = 4;
+var TAKEN = -1;
 function selectHotSites(code, entries, siteBytes, budget, profile) {
   const selected = /* @__PURE__ */ new Set();
   if (budget <= 0) return selected;
-  const weight = profile === void 0 ? staticWeights(code, entries) : void 0;
-  const w = (ip) => (weight !== void 0 ? weight.get(ip) : profile[ip]) ?? 0;
-  const ranked = [...siteBytes.keys()].sort((a, b) => w(b) / siteBytes.get(b) - w(a) / siteBytes.get(a) || a - b);
   let spent = 0;
-  for (const ip of ranked) {
-    if (w(ip) === 0) break;
+  for (const ip of profile === void 0 ? staticRanking(code, entries, siteBytes) : connectedRanking(code, siteBytes, profile)) {
     spent += siteBytes.get(ip);
     if (spent > budget) break;
     selected.add(ip);
   }
   return selected;
 }
+function edgesAsLowered(code, siteBytes, ip, out) {
+  out.length = 0;
+  if (code[ip] !== OP_CONT) childSlots(code, ip, out);
+  else if (siteBytes.has(ip)) out.push(code[ip + 1], code[ip + 3]);
+  else out.push(code[ip + 2]);
+  for (let i = 0; i < out.length; i++) {
+    let row = out[i];
+    for (; ; ) {
+      const op = code[row];
+      if (op === OP_GAPGATE || op === OP_SEED) row = code[row + 1];
+      else if (op === OP_CONT && !siteBytes.has(row)) row = code[row + 2];
+      else break;
+    }
+    out[i] = row;
+  }
+}
+function staticRanking(code, entries, siteBytes) {
+  const weight = staticWeights(code, entries, siteBytes);
+  const w = (ip) => weight.get(ip) ?? 0;
+  return [...siteBytes.keys()].filter((ip) => w(ip) > 0).sort((a, b) => w(b) / siteBytes.get(b) - w(a) / siteBytes.get(a) || a - b);
+}
+function* connectedRanking(code, siteBytes, calls) {
+  const ran = [...siteBytes.keys()].filter((ip) => calls[ip] > 0).sort((a, b) => a - b);
+  const slot = new Map(ran.map((ip, i) => [ip, i]));
+  const bytes = ran.map((ip) => siteBytes.get(ip));
+  const worth = ran.map((ip) => calls[ip]);
+  const score = worth.map((value, i) => value / bytes[i]);
+  const crossings = ran.map(() => []);
+  const kids = [];
+  ran.forEach((ip, from) => {
+    edgesAsLowered(code, siteBytes, ip, kids);
+    for (const callee of kids) {
+      const to = slot.get(callee);
+      if (to === void 0) continue;
+      const gain = CROSSING_WEIGHT * calls[callee];
+      crossings[from].push([to, gain]);
+      crossings[to].push([from, gain]);
+    }
+  });
+  for (let left = ran.length; left > 0; left--) {
+    let best = 0;
+    for (let i = 1; i < score.length; i++) if (score[i] > score[best]) best = i;
+    score[best] = TAKEN;
+    yield ran[best];
+    for (const [neighbour, gain] of crossings[best]) {
+      if (score[neighbour] === TAKEN) continue;
+      worth[neighbour] = worth[neighbour] + gain;
+      score[neighbour] = worth[neighbour] / bytes[neighbour];
+    }
+  }
+}
 function fitHotSites(code, entries, siteBytes, fixed, ceiling, sourceBytes, profile) {
   const ranked = [...selectHotSites(code, entries, siteBytes, ceiling - fixed, profile)];
   for (; ranked.length > 0; ranked.pop()) {
@@ -19488,25 +19537,17 @@ function fitHotSites(code, entries, siteBytes, fixed, ceiling, sourceBytes, prof
   }
   return void 0;
 }
-function staticWeights(code, entries) {
+function staticWeights(code, entries, siteBytes) {
   let weight = /* @__PURE__ */ new Map();
   const kids = [];
   for (let round = 0; round < ROUNDS; round++) {
     const next = /* @__PURE__ */ new Map();
     for (const ip of entries) next.set(ip, 1);
     for (const [ip, w] of weight) {
-      kids.length = 0;
-      childSlots(code, ip, kids);
+      edgesAsLowered(code, siteBytes, ip, kids);
       const op = code[ip];
-      if (op === OP_CONT && continuationGate(code, code[ip + 3]) !== void 0) {
-        kids.length = 0;
-        kids.push(code[ip + 1], code[ip + 3]);
-      }
       const scale = op === OP_REP || op === OP_REPV ? LOOP : op === OP_CHOICE || op === OP_DISPATCH ? 1 / Math.max(1, kids.length) : 1;
-      for (const k of kids) {
-        const kid = code[k] === OP_GAPGATE ? code[k + 1] : k;
-        next.set(kid, Math.min(CAP, (next.get(kid) ?? 0) + w * scale));
-      }
+      for (const kid of kids) next.set(kid, Math.min(CAP, (next.get(kid) ?? 0) + w * scale));
     }
     weight = next;
   }
@@ -22732,6 +22773,8 @@ function findFreeIdentifiers(code, filename) {
 // src/plugin/entry-supercompiler.ts
 import { parseSync as parseSync5 } from "oxc-parser";
 var PIECE = /^_pf(\d+)$/;
+var MERGEABLE = /^_(?:pf|sk)\d+$/;
+var TEMP = /^_t\d+_/;
 var WRAP = "function __parseman_supercompile__(){\n";
 var SUFFIX = "\n}";
 function ast(value) {
@@ -22865,6 +22908,93 @@ function fallback(emitted, reason, before) {
     }
   };
 }
+function namesUnder(root, uses, bound) {
+  let plain = true;
+  const bind = (id) => {
+    const name = nameOf2(id);
+    if (name !== void 0) bound.add(name);
+  };
+  const walk = (node2, parent) => {
+    if (node2.type === "VariableDeclarator") bind(ast(node2.id));
+    else if (node2.type === "LabeledStatement") bind(ast(node2.label));
+    else if (isFunction(node2)) {
+      if (node2 !== root) bind(ast(node2.id));
+      nodes(node2.params).forEach(bind);
+    }
+    const name = nameOf2(node2);
+    if (name !== void 0 && (MERGEABLE.test(name) || TEMP.test(name))) {
+      const isKey = parent?.type === "Property" && parent.key === node2 && parent.computed !== true;
+      const isProperty = parent?.type === "MemberExpression" && parent.property === node2 && parent.computed !== true;
+      if (parent?.type === "Property" && parent.shorthand === true) plain = false;
+      if (!isKey && !isProperty) uses.push({ ...range(node2), name });
+    }
+    for (const child of children2(node2)) walk(child, node2);
+  };
+  walk(root, void 0);
+  return plain;
+}
+function mergeSameBodies(source) {
+  const parsed = parseSync5("parseman-merge-same-bodies.js", WRAP + source + SUFFIX);
+  if (parsed.errors.length > 0) return source;
+  const statements = nodes(ast(nodes(parsed.program.body)[0]?.body)?.body);
+  const bodies = [];
+  const others = [];
+  const declared = /* @__PURE__ */ new Set();
+  for (const statement of statements) {
+    const name = statement.type === "FunctionDeclaration" ? nameOf2(ast(statement.id)) : void 0;
+    const uses = [];
+    const bound = /* @__PURE__ */ new Set();
+    if (!namesUnder(statement, uses, bound)) return source;
+    for (const local of bound) declared.add(local);
+    if (name === void 0 || !MERGEABLE.test(name)) others.push(...uses);
+    else bodies.push({ ...range(statement), name, uses: uses.sort((a, b) => a.start - b.start), local: bound });
+  }
+  for (const local of declared) if (MERGEABLE.test(local)) return source;
+  const merged = /* @__PURE__ */ new Map();
+  const current = (name) => {
+    let at = name;
+    for (let next = merged.get(at); next !== void 0; next = merged.get(at)) at = next;
+    return at;
+  };
+  let live = bodies;
+  for (let more = true; more; ) {
+    more = false;
+    const first = /* @__PURE__ */ new Map();
+    live = live.filter((body) => {
+      const temps = /* @__PURE__ */ new Map();
+      const edits2 = body.uses.map((use) => {
+        if (MERGEABLE.test(use.name)) {
+          const site = current(use.name);
+          return { ...use, text: site === body.name ? "#self" : site };
+        }
+        if (!body.local.has(use.name)) return { ...use, text: use.name };
+        const seen = temps.get(use.name) ?? temps.size;
+        temps.set(use.name, seen);
+        return { ...use, text: `#t${seen}` };
+      });
+      const key = render2(source, body.start, body.end, edits2);
+      const kept = first.get(key);
+      if (kept === void 0) {
+        first.set(key, body.name);
+        return true;
+      }
+      merged.set(body.name, kept);
+      more = true;
+      return false;
+    });
+  }
+  if (merged.size === 0) return source;
+  const edits = [];
+  const rename = (use) => {
+    if (MERGEABLE.test(use.name) && merged.has(use.name)) edits.push({ ...use, text: current(use.name) });
+  };
+  others.forEach(rename);
+  for (const body of bodies) {
+    if (merged.has(body.name)) edits.push({ start: body.start, end: body.end, text: "" });
+    else body.uses.forEach(rename);
+  }
+  return render2(source, 0, source.length, edits);
+}
 function supercompileRuleMapReplacement(prog, replacement, select) {
   const cfg = defaultAssemblyCfgs(prog)[0];
   if (cfg === void 0) return replacement;
@@ -22899,7 +23029,7 @@ function supercompileRuleMapReplacement(prog, replacement, select) {
   const start = body.start - prefix.length + 1;
   const end = body.end - prefix.length - 1;
   if (replacement.slice(start, end) !== FACTORY_ARITY_GUARD + canonical2.source + "\n") return replacement;
-  const transformed = replacement.slice(0, start) + FACTORY_ARITY_GUARD + candidate.source + "\n" + replacement.slice(end);
+  const transformed = replacement.slice(0, start) + FACTORY_ARITY_GUARD + mergeSameBodies(candidate.source) + "\n" + replacement.slice(end);
   const checked = parseSync5("parseman-entry-supercompile-replacement-result.js", prefix + transformed);
   return checked.errors.length === 0 ? transformed : replacement;
 }