parseman 0.54.0 → 0.55.1

package/dist/combinators/withCtx.d.ts

@@ -1,13 +1,28 @@
 import type { Combinator } from '../types.ts';
 /**
- * Runs `combinator` with `ctx.state` set to `extra` for the duration of the parse.
- * The outer user context is restored on exit (lexical scoping).
+ * Runs `combinator` under a scoped `ctx.state`, restored on exit (lexical scoping).
  *
- *   const functionBody = withCtx({ inFn: true },
- *     sequence(literal('{'), many(statement), literal('}'))
- *   )
+ * Two forms, told apart by whether the first argument is a function:
  *
- * Read back with guard() or from within a transform's span argument.
+ * - **Replace** — `withCtx(extra, c)`: the subtree's state IS `extra`.
+ *
+ *     const functionBody = withCtx({ inFn: true },
+ *       sequence(literal('{'), many(statement), literal('}'))
+ *     )
+ *
+ * - **Derive** — `withCtx(state => next, c)`: the subtree's state is computed
+ *   FROM the state in force at entry, once per entry. Use it to add or override
+ *   a field while every outer field (options, document-wide sinks) stays visible
+ *   to the gates and reducers inside:
+ *
+ *     const argument = withCtx(s => ({ ...s, stop: 'comparison' }), value)
+ *
+ * Because a function argument always selects the derive form, a state that is
+ * itself a function cannot be installed with the replace form; wrap it
+ * (`withCtx(() => fn, c)`).
+ *
+ * Read back with gate() or from within a transform's span argument.
  */
-export declare function withCtx<U, T>(extra: U, combinator: Combinator<T>): Combinator<T>;
+export declare function withCtx<S = Readonly<Record<string, unknown>> | undefined, U = unknown, T = unknown>(derive: (state: S) => U, combinator: Combinator<T>): Combinator<T>;
+export declare function withCtx<U, T>(extra: U extends (...args: never[]) => unknown ? never : U, combinator: Combinator<T>): Combinator<T>;
 //# sourceMappingURL=withCtx.d.ts.map

package/dist/combinators/withCtx.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"withCtx.d.ts","sourceRoot":"","sources":["../../src/combinators/withCtx.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAyC,MAAM,aAAa,CAAA;AAEpF;;;;;;;;;GASG;AACH,wBAAgB,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAgChF"}
+{"version":3,"file":"withCtx.d.ts","sourceRoot":"","sources":["../../src/combinators/withCtx.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAyC,MAAM,aAAa,CAAA;AAEpF;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAIH,wBAAgB,OAAO,CAAC,CAAC,GAAG,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,GAAG,SAAS,EAAE,CAAC,GAAG,OAAO,EAAE,CAAC,GAAG,OAAO,EAAE,MAAM,EAAE,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;AAKvK,wBAAgB,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,SAAS,CAAC,GAAG,IAAI,EAAE,KAAK,EAAE,KAAK,OAAO,GAAG,KAAK,GAAG,CAAC,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA"}

package/dist/index.js

@@ -7259,19 +7259,11 @@ var Encoder = class {
   nonCommittingChoiceSites = /* @__PURE__ */ new Set();
   failureRollbackCleanSites = /* @__PURE__ */ new Set();
   /**
-   * `hasSideWriter` memo for `failureNeedsRollback`, shared across every call
-   * for the lifetime of this `Encoder`. A cached entry is written only when
-   * `complete` is true, which means the answer is a path-independent fact
-   * about that combinator node — it does not depend on which root's traversal
-   * asked, only on the node's own reachable subgraph. Scoping the memo per call
-   * (as a local inside `failureNeedsRollback`) was correct but discarded that
-   * fact after every arm, so a grammar with many ordered-choice arms / optional
-   * / attempt sites re-walked the SAME shared rule graph from scratch for each
-   * site — O(sites × graph size) instead of O(graph size). Measured on the
-   * scss-parser grammar (composed CSS-base superset, the deepest/largest of the
-   * four grammars): 47-50s for one `compileRuleMap` with the per-call memo,
-   * versus the whole macro transform completing in under a second once this
-   * memo is hoisted here.
+   * `reachesSideWriter`'s answers, shared by every `failureNeedsRollback` call
+   * for the lifetime of this `Encoder`. Every entry is exact — it depends only on
+   * the node's own reachable subgraph, never on which root's walk asked — so each
+   * combinator is walked once per encode however many choice arms, optionals and
+   * attempt sites reach it.
    */
   sideWriterMemo = /* @__PURE__ */ new Map();
   labels = void 0;
@@ -7704,63 +7696,96 @@ var Encoder = class {
    * writer census covers the side sinks that proof deliberately does not model:
    * fields, recovered errors, and interpreted scanners. Strict assemblies do
    * not run repeat recovery, but these rows can still publish through a caller's
-   * context. Cycles are safe to stop: revisiting the same combinator cannot
-   * discover a writer the first traversal did not already reach.
+   * context.
    */
   failureNeedsRollback(p, autoNot) {
     const refName = p._ruleName;
     const winner = refName === void 0 ? void 0 : this.winners?.[refName];
     const root = winner !== void 0 && !winnerWrapsReference2(winner, p) ? winner : p;
-    const sideWriterMemo = this.sideWriterMemo;
-    const seen = /* @__PURE__ */ new Set();
-    const hasSideWriter = (current) => {
-      const cached = sideWriterMemo.get(current);
-      if (cached !== void 0) return { found: cached, complete: true };
-      if (seen.has(current)) return { found: false, complete: false };
-      seen.add(current);
-      const def = current._def;
-      let result;
-      switch (def.tag) {
-        case "field":
-        case "expect":
-        case "recover":
-        case "scanTo":
-        case "unknown":
-          result = { found: true, complete: true };
-          break;
-        case "lazy": {
-          const name = current._ruleName;
-          const resolvedWinner = name === void 0 ? void 0 : this.winners?.[name];
-          if (resolvedWinner !== void 0 && !winnerWrapsReference2(resolvedWinner, current)) {
-            result = hasSideWriter(resolvedWinner);
-            break;
-          }
-          try {
-            result = hasSideWriter(def.thunk());
-          } catch {
-            result = { found: true, complete: true };
+    return mayLeavePartialCapture(root, /* @__PURE__ */ new Set(), true) || this.reachesSideWriter(root) || (autoNot?.length ?? 0) > 0 && capturesLeaf(root);
+  }
+  /**
+   * Whether a field, recovered error, interpreted scanner or unresolvable shape
+   * is reachable from `root` — the writer census of `failureNeedsRollback`.
+   *
+   * Reachability over a cyclic graph, decided with Tarjan's strongly connected
+   * components so that every node the walk enters leaves with its exact answer
+   * memoized: the members of one component reach each other, so they share one
+   * answer, settled when the walk returns to the component's first node. A DFS
+   * that instead cuts a cycle at its own ancestor path learns nothing it can
+   * keep about a node whose walk hit the cut, and re-walks that node along
+   * every path that reaches it — exponential over recursive value rules (linking
+   * jess's SCSS grammar cost ~290G instructions). This walk enters each node
+   * once per `Encoder`.
+   */
+  reachesSideWriter(root) {
+    const memo = this.sideWriterMemo;
+    const known = memo.get(root);
+    if (known !== void 0) return known;
+    const order = /* @__PURE__ */ new Map();
+    const low = [];
+    const found = [];
+    const stack = [];
+    const visit = (node2) => {
+      const at = low.length;
+      order.set(node2, at);
+      low.push(at);
+      found.push(false);
+      stack.push(node2);
+      const next = this.sideWriterSuccessors(node2);
+      if (next === true) found[at] = true;
+      else {
+        for (const child of next) {
+          const hit = memo.get(child);
+          if (hit !== void 0) {
+            if (hit) found[at] = true;
+            continue;
           }
-          break;
-        }
-        default: {
-          let complete = true;
-          result = { found: false, complete: true };
-          for (const child of childrenOf2(def)) {
-            const childResult = hasSideWriter(child);
-            if (childResult.found) {
-              result = { found: true, complete: true };
-              break;
-            }
-            complete &&= childResult.complete;
+          const seen = order.get(child);
+          if (seen !== void 0) {
+            if (seen < low[at]) low[at] = seen;
+            continue;
           }
-          if (!result.found) result = { found: false, complete };
+          const k = visit(child);
+          if (low[k] < low[at]) low[at] = low[k];
+          if (found[k]) found[at] = true;
         }
       }
-      seen.delete(current);
-      if (result.found || result.complete) sideWriterMemo.set(current, result.found);
-      return result;
+      if (low[at] === at) {
+        const answer = found[at];
+        let member;
+        do {
+          member = stack.pop();
+          memo.set(member, answer);
+        } while (member !== node2);
+      }
+      return at;
     };
-    return mayLeavePartialCapture(root, /* @__PURE__ */ new Set(), true) || hasSideWriter(root).found || (autoNot?.length ?? 0) > 0 && capturesLeaf(root);
+    return found[visit(root)];
+  }
+  /** The census's edges out of `node`, or `true` when `node` is itself a writer. */
+  sideWriterSuccessors(node2) {
+    const def = node2._def;
+    switch (def.tag) {
+      case "field":
+      case "expect":
+      case "recover":
+      case "scanTo":
+      case "unknown":
+        return true;
+      case "lazy": {
+        const name = node2._ruleName;
+        const winner = name === void 0 ? void 0 : this.winners?.[name];
+        if (winner !== void 0 && !winnerWrapsReference2(winner, node2)) return [winner];
+        try {
+          return [def.thunk()];
+        } catch {
+          return true;
+        }
+      }
+      default:
+        return childrenOf2(def);
+    }
   }
   /** A first set becomes a char class string, or −1 for `any`. */
   charClass(fs) {
@@ -8277,6 +8302,9 @@ var Encoder = class {
           this.expected([adjacencyExpected(d)])
         );
       case "withCtx": {
+        if (typeof d.extra === "function") {
+          return this.emit(OP_WITHCTX, ~this.fn(d.extra, d.extraSrc ?? null), this.node(d.parser).ip);
+        }
         if (!emittableConst(d.extra)) {
           this.runtimeOnly.add("withCtx(extra) \u2014 the state object is not serialisable");
         }
@@ -13233,8 +13261,21 @@ function assemble(t, prog, cfg) {
         };
       }
       case OP_WITHCTX: {
-        const extra = k[code[ip + 1]];
+        const ki = code[ip + 1];
         const child = link(code[ip + 2]);
+        if (ki < 0) {
+          const derive2 = fns[~ki];
+          return (input, pos, ctx) => {
+            const saved = ctx.state;
+            ctx.state = derive2(saved);
+            try {
+              return child(input, pos, ctx);
+            } finally {
+              ctx.state = saved;
+            }
+          };
+        }
+        const extra = k[ki];
         return (input, pos, ctx) => {
           const saved = ctx.state;
           ctx.state = extra;
@@ -17725,7 +17766,7 @@ function fieldOptionsSrc(def) {
 }
 
 // src/version.ts
-var PARSEMAN_VERSION = "0.54.0";
+var PARSEMAN_VERSION = "0.55.1";
 
 // src/compiler/compile-linkable-table.ts
 function isLocal(rule) {
@@ -19313,19 +19354,28 @@ function withCtx(extra, combinator) {
     canMatchNewline: combinator._meta.canMatchNewline,
     isTrivia: false
   };
+  const parse2 = typeof extra === "function" ? (input, pos, ctx) => {
+    const saved = ctx.state;
+    ctx.state = extra(saved);
+    try {
+      return combinator.parse(input, pos, ctx);
+    } finally {
+      ctx.state = saved;
+    }
+  } : (input, pos, ctx) => {
+    const saved = ctx.state;
+    ctx.state = extra;
+    try {
+      return combinator.parse(input, pos, ctx);
+    } finally {
+      ctx.state = saved;
+    }
+  };
   return {
     _tag: "withCtx",
     _meta: meta,
     _def: { tag: "withCtx", extra, parser: combinator },
-    parse(input, pos, ctx) {
-      const saved = ctx.state;
-      ctx.state = extra;
-      try {
-        return combinator.parse(input, pos, ctx);
-      } finally {
-        ctx.state = saved;
-      }
-    }
+    parse: parse2
   };
 }