@mjasnikovs/pi-task 0.40.2 → 0.40.3

package/dist/workers/docs-chunk.js

@@ -37,16 +37,25 @@ export const README_SPLIT_RE = /^#{1,2} /m;
 export function splitAtMatches(text, re) {
     const parts = [];
     let lastIndex = 0;
+    let acceptedEnd = 0;
     let m;
     while ((m = re.exec(text))) {
+        // Scan resumes ONE character in, not past the match: the regex's trailing
+        // `\s+` can span a newline, so a genuinely separate line-anchored
+        // declaration may start inside what this match consumed.
+        re.lastIndex = m.index + 1;
+        // But a match that starts inside the last ACCEPTED one is that match's own
+        // tail, not a new declaration, and cutting there severs a declaration from
+        // the modifiers and attributes that open it. `export\nfunction a(){}` is
+        // one declaration; so is a cargo `#[cfg(…)]` above its `impl`, which
+        // CARGO_DECL_SPLIT_RE absorbs on purpose and this used to hand back as a
+        // chunk holding nothing but the attribute.
+        if (m.index < acceptedEnd)
+            continue;
         if (m.index > lastIndex)
             parts.push(text.slice(lastIndex, m.index));
         lastIndex = m.index;
-        // Advance by ONE, not by the match length. The regex's trailing `\s+` can
-        // span a newline, so the next line-anchored declaration may start INSIDE
-        // what this match consumed: `export\nfunction a(){}` is two chunks here
-        // and one if the scan resumes past the match.
-        re.lastIndex = m.index + 1;
+        acceptedEnd = m.index + m[0].length;
     }
     if (lastIndex < text.length)
         parts.push(text.slice(lastIndex));

package/dist/workers/docs-index.d.ts

@@ -7,4 +7,32 @@ export interface IndexResult {
     chunksWritten: number;
     contentHash: string;
 }
+/**
+ * The gate that decides whether a package needs re-indexing.
+ *
+ * The entry file goes in SURFACED, not raw. What is cached is the extractor's
+ * OUTPUT, so a build whose extractor changed has stale chunks even though every
+ * byte on disk is identical — a crate indexed before the braced-`use` fix keeps
+ * `pub use crate::runtime::;` forever, because name, version and file bytes all
+ * still match. Surfacing here costs one file and makes the hash answer the
+ * question actually being asked: would re-reading produce the same chunks?
+ *
+ * The CHUNKER counts too, for the same reason: the rows are chunks, not surface,
+ * so a fix to where a declaration is cut leaves stale rows behind on its own. So
+ * does WHICH FILES are read: dropping a package's duplicate `.d.cts` twins
+ * changes the rows without changing a byte on disk.
+ *
+ * It is not total. An extractor change that alters only files BELOW the entry
+ * goes unnoticed; deleting the cache is still the escape hatch for that.
+ */
+/**
+ * The chunker's own source, so a cut-point fix invalidates every cached package.
+ *
+ * `declSplitRe.source` alone does not do it: the attribute-orphaning bug was in
+ * `splitAtMatches`, not in any profile's regex, so the fingerprint sat still while
+ * the rows it describes changed. A package indexed before that fix keeps its
+ * dangling `#[cfg(...)]` chunks forever, and the hash is the only thing that would
+ * have said so.
+ */
+export declare function chunkerFingerprint(): string;
 export declare function ensureIndexed(cache: CacheHandle, pkg: ResolvedPackage, profile?: EcosystemProfile): IndexResult;

package/dist/workers/docs-index.js

@@ -2,7 +2,7 @@ import { createHash } from 'node:crypto';
 import * as fs from 'node:fs';
 import * as path from 'node:path';
 import {} from './docs-resolve.js';
-import { chunkDeclarations, chunkReadme } from './docs-chunk.js';
+import { chunkDeclarations, chunkReadme, splitAtMatches } from './docs-chunk.js';
 import { ECOSYSTEMS } from './docs-ecosystems.js';
 const ZERO_SEP = Buffer.from([0]);
 /**
@@ -23,12 +23,26 @@ const ZERO_SEP = Buffer.from([0]);
  * It is not total. An extractor change that alters only files BELOW the entry
  * goes unnoticed; deleting the cache is still the escape hatch for that.
  */
+/**
+ * The chunker's own source, so a cut-point fix invalidates every cached package.
+ *
+ * `declSplitRe.source` alone does not do it: the attribute-orphaning bug was in
+ * `splitAtMatches`, not in any profile's regex, so the fingerprint sat still while
+ * the rows it describes changed. A package indexed before that fix keeps its
+ * dangling `#[cfg(...)]` chunks forever, and the hash is the only thing that would
+ * have said so.
+ */
+export function chunkerFingerprint() {
+    return `${String(splitAtMatches)}\u0000${String(chunkDeclarations)}\u0000${String(chunkReadme)}`;
+}
 function computeContentHash(pkg, profile) {
     const hash = createHash('sha256');
     hash.update(Buffer.from(`${pkg.name}@${pkg.version}`, 'utf8'));
     hash.update(ZERO_SEP);
     hash.update(Buffer.from(`${profile.declSplitRe.source}\u0000${profile.commentPrefix}`, 'utf8'));
     hash.update(ZERO_SEP);
+    hash.update(Buffer.from(chunkerFingerprint(), 'utf8'));
+    hash.update(ZERO_SEP);
     // Source text, the same trick as `declSplitRe.source`: the fingerprint moves
     // whenever the selection rule does, with nothing to remember to bump.
     hash.update(Buffer.from(`${String(profile.isSurfaceFile)}\u0000${String(dropParallelDeclarations)}`

package/dist/workers/pi-worker-docs.js

@@ -194,7 +194,7 @@ export function registerPiWorkerDocs(pi, internals = {}) {
                 const r = await lookup(projectCorpus(projectName), chunks);
                 if (r.kind === 'failed')
                     return docsFailureResult(r.extraction, baseDetails, '');
-                const { extraction, excerptVerified: verified, body: text } = r;
+                const { extraction, excerptVerified: verified, body: text, content: retrieved } = r;
                 // SAME instrumentation channel as the package path below. Both branches
                 // record, or "the last docs answer before the worker stopped" is
                 // unanswerable whenever the last answer came from the branch that does
@@ -213,7 +213,8 @@ export function registerPiWorkerDocs(pi, internals = {}) {
                     reason: 'project-source lookup — the type-only detector is not applied here',
                     excerptVerified: verified,
                     excerptCheck: extraction.excerptCheck,
-                    toolText: text
+                    toolText: text,
+                    retrievedText: retrieved
                 });
                 return workerAnswer(text, {
                     ...baseDetails,
@@ -346,7 +347,8 @@ export function registerPiWorkerDocs(pi, internals = {}) {
                 reason: typeOnly.reason,
                 excerptVerified: verified,
                 excerptCheck: extraction.excerptCheck,
-                toolText: text
+                toolText: text,
+                retrievedText: concatenated
             });
             return workerAnswer(text, {
                 ...baseDetails,

package/dist/workers/typeonly-log.d.ts

@@ -31,6 +31,19 @@ export interface TypeOnlyLogRecord {
      * Optional — older records still parse.
      */
     excerptCheck?: ExcerptVerification;
+    /**
+     * The retrieved chunk text handed to the extraction child, before it wrote a word.
+     *
+     * `toolText` cannot stand in for it. The tool return embeds the child's own prose,
+     * so scoring that prose against it asks whether the answer contains itself — two
+     * full runs read 13/13, 12/12, 10/10, 13/13, 4/4 and 2/2 clean, 54 answers and not
+     * one miss, while one of them shipped `decodeFile`, which aeson 2 does not have.
+     * Removing the prose leaves only the child's own CITED excerpt, a line or two, and
+     * flags `from_str` and `Context` as invented. Neither corpus is the one the question
+     * needs; this is. Optional — records written before it still parse, and a scorer
+     * must say "not computable" rather than guess when it is absent.
+     */
+    retrievedText?: string;
     /**
      * The tool's ENTIRE return text — version banner, npm header, the answer prose, the cited
      * excerpt, and (when it fires) the type-only banner. Optional so logs written before this

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@mjasnikovs/pi-task",
-    "version": "0.40.2",
+    "version": "0.40.3",
     "description": "Deterministic task planning and spec-orchestration for local models — crash-safe /task pipelines with verify/enforce gates, a real-time remote web view, and web/docs/fetch/worker subagent tools.",
     "type": "module",
     "main": "./dist/index.js",