parseman 0.61.0 → 0.61.2

package/dist/plugin/index.d.ts

@@ -84,7 +84,9 @@ export type AssemblyProfile = {
     /** The rule each corpus file is parsed from, e.g. `'Stylesheet'`. A grammar
      * without this rule keeps the static ranking. */
     readonly entry: string;
-    /** Paths of representative inputs. */
+    /** Paths of representative inputs. At least one must have content, and each that
+     * does has to parse in full without recovery, or the build fails. A file with only
+     * whitespace counts as empty. */
     readonly files: readonly string[];
 };
 declare const _default: import("unplugin").UnpluginInstance<ParsecraftPluginOptions, boolean>;

package/dist/plugin/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/plugin/index.ts"],"names":[],"mappings":"AAyBA,OAAO,WAAW,MAAM,cAAc,CAAA;AActC,OAAO,EAAsC,KAAK,OAAO,EAAE,MAAM,uBAAuB,CAAA;AA6BxF;;;;;;;;;;;;;;;GAeG;AACH,MAAM,MAAM,kBAAkB,GAAG,OAAO,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAA;AACzE;;;;;GAKG;AACH,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,EAAE,SAAS,kBAAkB,EAAE,CAAA;CAAE,CAAA;AAChH,kDAAkD;AAClD,MAAM,MAAM,WAAW,GAAG,GAAG,CAAC,MAAM,EAAE,eAAe,EAAE,CAAC,CAAA;AACxD,eAAO,MAAM,sBAAsB,mBAAmB,CAAA;AAQtD;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAGnD;AA+BD,MAAM,MAAM,uBAAuB,GAAG;IACpC,8DAA8D;IAC9D,aAAa,CAAC,EAAE,MAAM,EAAE,CAAA;IACxB;;;;;;OAMG;IACH,kBAAkB,CAAC,EAAE,OAAO,CAAA;IAC5B;;;;OAIG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;8CAC0C;IAC1C,eAAe,CAAC,EAAE,OAAO,CAAA;IACzB;;;;;;;;OAQG;IACH,eAAe,CAAC,EAAE,eAAe,CAAA;IACjC;;;;;;OAMG;IACH,cAAc,CAAC,EAAE,MAAM,CAAA;CACxB,CAAA;AAED,wFAAwF;AACxF,MAAM,MAAM,eAAe,GAAG;IAC5B;qDACiD;IACjD,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;IACtB,sCAAsC;IACtC,QAAQ,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,CAAA;CAClC,CAAA;;;AAgkBD,MAAM,MAAM,oBAAoB,GAAG;IACjC,IAAI,EAAE,MAAM,CAAA;IACZ,GAAG,EAAE,UAAU,CAAC,WAAW,CAAC,aAAa,CAAC,CAAC,CAAA;IAC3C,kFAAkF;IAClF,QAAQ,EAAE,MAAM,EAAE,CAAA;CACnB,CAAA;AA0CD;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,EAAE,EAAE,MAAM,EACV,aAAa,cAA6B,EAC1C,kBAAkB,UAAQ,EAC1B,QAAQ,UAAQ,EAChB,eAAe,UAAQ,EACvB,eAAe,CAAC,EAAE,eAAe,EACjC,cAAc,SAA0B,EACxC,WAAW,CAAC,EAAE,WAAW,GACxB,oBAAoB,GAAG,IAAI,CAW7B"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/plugin/index.ts"],"names":[],"mappings":"AAyBA,OAAO,WAAW,MAAM,cAAc,CAAA;AActC,OAAO,EAAsC,KAAK,OAAO,EAAE,MAAM,uBAAuB,CAAA;AA6BxF;;;;;;;;;;;;;;;GAeG;AACH,MAAM,MAAM,kBAAkB,GAAG,OAAO,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAA;AACzE;;;;;GAKG;AACH,MAAM,MAAM,eAAe,GAAG,OAAO,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,EAAE,SAAS,kBAAkB,EAAE,CAAA;CAAE,CAAA;AAChH,kDAAkD;AAClD,MAAM,MAAM,WAAW,GAAG,GAAG,CAAC,MAAM,EAAE,eAAe,EAAE,CAAC,CAAA;AACxD,eAAO,MAAM,sBAAsB,mBAAmB,CAAA;AAQtD;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAGnD;AA+BD,MAAM,MAAM,uBAAuB,GAAG;IACpC,8DAA8D;IAC9D,aAAa,CAAC,EAAE,MAAM,EAAE,CAAA;IACxB;;;;;;OAMG;IACH,kBAAkB,CAAC,EAAE,OAAO,CAAA;IAC5B;;;;OAIG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;8CAC0C;IAC1C,eAAe,CAAC,EAAE,OAAO,CAAA;IACzB;;;;;;;;OAQG;IACH,eAAe,CAAC,EAAE,eAAe,CAAA;IACjC;;;;;;OAMG;IACH,cAAc,CAAC,EAAE,MAAM,CAAA;CACxB,CAAA;AAED,wFAAwF;AACxF,MAAM,MAAM,eAAe,GAAG;IAC5B;qDACiD;IACjD,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;IACtB;;qCAEiC;IACjC,QAAQ,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,CAAA;CAClC,CAAA;;;AAgkBD,MAAM,MAAM,oBAAoB,GAAG;IACjC,IAAI,EAAE,MAAM,CAAA;IACZ,GAAG,EAAE,UAAU,CAAC,WAAW,CAAC,aAAa,CAAC,CAAC,CAAA;IAC3C,kFAAkF;IAClF,QAAQ,EAAE,MAAM,EAAE,CAAA;CACnB,CAAA;AA0CD;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,cAAc,CAC5B,IAAI,EAAE,MAAM,EACZ,EAAE,EAAE,MAAM,EACV,aAAa,cAA6B,EAC1C,kBAAkB,UAAQ,EAC1B,QAAQ,UAAQ,EAChB,eAAe,UAAQ,EACvB,eAAe,CAAC,EAAE,eAAe,EACjC,cAAc,SAA0B,EACxC,WAAW,CAAC,EAAE,WAAW,GACxB,oBAAoB,GAAG,IAAI,CAW7B"}

package/dist/plugin/index.js

@@ -5619,7 +5619,7 @@ function sourceLeaf(root, fn) {
 }
 
 // src/version.ts
-var PARSEMAN_VERSION = "0.61.0";
+var PARSEMAN_VERSION = "0.61.2";
 
 // src/table/lex-body.ts
 function buildLexBodyRecognizer(spec, constants) {
@@ -10428,6 +10428,12 @@ function emitShapeMatch(shape, start, mint, ind, firstChar) {
 
 // src/table/child-slots.ts
 function childSlots(code, ip, out) {
+  const from = out.length;
+  if (!childSlotOffsets(code, ip, out)) return false;
+  for (let i = from; i < out.length; i++) out[i] = code[out[i]];
+  return true;
+}
+function childSlotOffsets(code, ip, out) {
   switch (code[ip]) {
     // ZERO-WIDTH / TERMINAL ROWS — no successor. `OP_GUARD` and `OP_ADJ` are both
     // childless tests. `OP_LIVE` holds a hand-written combinator in `prog.fns`; its
@@ -10460,7 +10466,7 @@ function childSlots(code, ip, out) {
     case OP_COV:
     case OP_REJECT:
     case OP_GAPGATE:
-      out.push(code[ip + 1]);
+      out.push(ip + 1);
       return true;
     // Single child at `ip + 2`.
     case OP_GATE:
@@ -10477,55 +10483,55 @@ function childSlots(code, ip, out) {
     case OP_ARMGATE:
     case OP_RGATE:
     case OP_RGATE2:
-      out.push(code[ip + 2]);
+      out.push(ip + 2);
       return true;
     // `STRICT strict tolerant` — both encodings are reachable; an assembly keeps one.
     case OP_STRICT:
-      out.push(code[ip + 1], code[ip + 2]);
+      out.push(ip + 1, ip + 2);
       return true;
     // `ROUTED fallback` — the operand is an offset or −1. The routed token itself is
     // data the enclosing `dispatch()` already consumed, so the only edge is the
     // fallback.
     case OP_ROUTED:
-      if (code[ip + 1] >= 0) out.push(code[ip + 1]);
+      if (code[ip + 1] >= 0) out.push(ip + 1);
       return true;
     // `DISPATCH sel d other otherRouted n a1 … an` — the SELECTOR is a child on
     // exactly the same footing as the arms: it runs inside whatever node and scope
     // this site sits in.
     case OP_DISPATCH: {
-      out.push(code[ip + 1]);
-      if (code[ip + 3] >= 0) out.push(code[ip + 3]);
+      out.push(ip + 1);
+      if (code[ip + 3] >= 0) out.push(ip + 3);
       const n = code[ip + 5];
-      for (let i = 0; i < n; i++) out.push(code[ip + 6 + i]);
+      for (let i = 0; i < n; i++) out.push(ip + 6 + i);
       return true;
     }
     case OP_SEQ:
     case OP_SEQV: {
       const n = code[ip + 1];
-      for (let i = 0; i < n; i++) out.push(code[ip + 2 + i]);
+      for (let i = 0; i < n; i++) out.push(ip + 2 + i);
       return true;
     }
     case OP_SEQX: {
       const n = code[ip + 2];
-      for (let i = 0; i < n; i++) out.push(code[ip + 3 + i]);
+      for (let i = 0; i < n; i++) out.push(ip + 3 + i);
       return true;
     }
     case OP_CHOICE: {
       const n = code[ip + 2];
-      for (let i = 0; i < n; i++) out.push(code[ip + 4 + i]);
+      for (let i = 0; i < n; i++) out.push(ip + 4 + i);
       return true;
     }
     // Arms are stride-2 from `ip + 4` — the gate word sits beside each arm offset.
     case OP_GREEDY: {
-      out.push(code[ip + 1]);
+      out.push(ip + 1);
       const n = code[ip + 2];
-      for (let i = 0; i < n; i++) out.push(code[ip + 4 + 2 * i]);
+      for (let i = 0; i < n; i++) out.push(ip + 4 + 2 * i);
       return true;
     }
     case OP_REP:
     case OP_REPV:
-      out.push(code[ip + 1]);
-      if (code[ip + 4] >= 0) out.push(code[ip + 4]);
+      out.push(ip + 1);
+      if (code[ip + 4] >= 0) out.push(ip + 4);
       return true;
     default:
       return false;
@@ -24326,23 +24332,45 @@ ${names.map((n) => `  - ${n}`).join("\n")}` : " (it is not a static combinator/v
   };
   const profileOf = (compiled) => {
     if (assemblyProfile === void 0 || compiled.hostMode !== "ast" || compiled.prog.lines === 1 || !(assemblyProfile.entry in compiled.prog.rules)) return void 0;
-    let parsed = 0;
+    const refuse = (why) => {
+      throw new Error(`${id} \u2014 assemblyProfile: ${why}`);
+    };
+    if (corpusTexts(assemblyProfile).every((text) => text.trim() === "")) refuse("`files` has no non-empty file");
+    let thrower;
     const fns = compiled.prog.fns.map((fn) => {
       const type = fn?.[INERT_BUILDER];
-      return type === void 0 ? fn : (children3, _fields, span) => ({ type, span, children: children3 });
+      if (type !== void 0) return (children3, _fields, span) => ({ type, span, children: children3 });
+      return typeof fn !== "function" ? fn : new Proxy(fn, {
+        apply(target, self, args) {
+          try {
+            return Reflect.apply(target, self, args);
+          } catch (e) {
+            thrower ??= target;
+            throw e;
+          }
+        }
+      });
     });
-    const counts = countSiteVisits(compiled.prog, (rules3) => {
+    return countSiteVisits(compiled.prog, (rules3) => {
       const entry = rules3[assemblyProfile.entry];
-      for (const text of corpusTexts(assemblyProfile)) {
+      corpusTexts(assemblyProfile).forEach((text, i) => {
+        const file = assemblyProfile.files[i];
+        if (text.trim() === "") return;
+        let result2;
         try {
-          if (run(entry, text).ok) parsed++;
-        } catch {
+          result2 = run(entry, text);
+        } catch (e) {
+          const source = String(thrower).replace(/\s+/g, " ").slice(0, 200);
+          throw new Error(`${id} \u2014 assemblyProfile: ${thrower === void 0 ? "the parse" : `gate or reducer \`${source}\``} threw on ${file}: ${String(e)}`, { cause: e });
         }
-      }
+        if (!result2.ok || result2.unconsumedFrom !== null) {
+          refuse(`${file} does not parse from rule "${assemblyProfile.entry}": it stops at offset ${result2.ok ? result2.unconsumedFrom : result2.span.end}`);
+        }
+        if (result2.errors.length > 0) {
+          refuse(`${file} parses only by recovering from ${result2.errors.length} error(s), first at offset ${result2.errors[0].span.start}`);
+        }
+      });
     }, fns);
-    if (parsed > 0) return counts;
-    warnings.push(`${id} \u2014 assemblyProfile: no corpus file parsed from rule "${assemblyProfile.entry}"; using the static ranking`);
-    return void 0;
   };
   const assembledReplacement = (compiled, metadata, select) => select.size === 0 ? compiled.replacementWithMetadata(metadata) : supercompileRuleMapReplacement(compiled.prog, compiled.replacementWithMetadata(metadata, { select }), select);
   const compileComposeLeafCall = (init) => {