pi-goosedump 0.10.8 → 0.10.10

package/README.md

@@ -22,8 +22,8 @@ report that they are unavailable.
 - **`goose_get`** expands entry IDs returned by `goose_search` or `goose_grep`.
 - **`goose_grep`** filters session messages with a glob pattern.
 - **`goose_recall`** searches memory harvested by earlier compactions. It
-  searches the current working directory by default and returns full entries
-  with provenance.
+  searches the current working directory by default and returns remembered
+  excerpts with provenance within a 4,096-byte/token-safe output budget.
 - **`goose_compact`** starts asynchronous compaction of the current Pi session.
   Set `keep` to preserve the last N user turns verbatim.
 
@@ -75,7 +75,8 @@ global setting.
 ## Compaction behavior
 
 The extension replaces Pi's normal summary during manual, scheduled, and
-automatic compaction. It preserves the previous compaction summary and asks
+automatic compaction. It passes `--train` so the compacted range is harvested
+into persistent memory, preserves the previous compaction summary, and asks
 goosedump to summarize only the newly compacted range. A `keep:N` request
 instead keeps the last N active user turns; it is cancelled when no earlier
 history is available to summarize.

package/index.ts

@@ -78,6 +78,13 @@ interface GoosedumpMemoryHit {
   source_path: string;
 }
 
+const RECALL_TOKEN_BUDGET = 4096;
+// Every non-empty token occupies at least one UTF-8 byte, so this byte cap is
+// also a conservative hard cap on the number of tokens returned to Pi.
+const RECALL_MAX_BYTES = RECALL_TOKEN_BUDGET;
+const MEMORY_HIT_SEPARATOR = '\n\n---\n\n';
+const MEMORY_TEXT_TRUNCATED = '\n[Entry text truncated]';
+
 // A `content` part of a pi-native message line.
 interface PiPartJson {
   type: string;
@@ -395,13 +402,44 @@ function parseMemoryHits(output: string): GoosedumpMemoryHit[] {
   );
 }
 
-function formatMemoryHits(hits: GoosedumpMemoryHit[]): string {
-  return hits
-    .map(
-      (hit) =>
-        `[${hit.hash}] ${hit.kind} (${hit.provider}:${hit.context_id}/${hit.entry_id})\nWorking directory: ${hit.path}\nSource: ${hit.source_path}\n${hit.text}`,
-    )
-    .join('\n\n---\n\n');
+function formatMemoryHit(hit: GoosedumpMemoryHit, maxBytes: number): string {
+  const provenance =
+    `[${hit.hash}] ${hit.kind} (${hit.provider}:${hit.context_id}/${hit.entry_id})\n` +
+    `Working directory: ${hit.path}\nSource: ${hit.source_path}\n`;
+  const full = `${provenance}${hit.text}`;
+  if (Buffer.byteLength(full) <= maxBytes) return full;
+
+  const provenanceBytes = Buffer.byteLength(provenance);
+  const markerBytes = Buffer.byteLength(MEMORY_TEXT_TRUNCATED);
+  if (provenanceBytes + markerBytes >= maxBytes) return utf8Prefix(provenance, maxBytes);
+
+  const textBytes = maxBytes - provenanceBytes - markerBytes;
+  return `${provenance}${utf8Prefix(hit.text, textBytes)}${MEMORY_TEXT_TRUNCATED}`;
+}
+
+function formatMemoryHits(hits: GoosedumpMemoryHit[], maxBytes = RECALL_MAX_BYTES): string {
+  const formatted: string[] = [];
+  let remainingBytes = maxBytes;
+  const separatorBytes = Buffer.byteLength(MEMORY_HIT_SEPARATOR);
+
+  for (let index = 0; index < hits.length; index += 1) {
+    if (index > 0) {
+      if (remainingBytes < separatorBytes) break;
+      formatted.push(MEMORY_HIT_SEPARATOR);
+      remainingBytes -= separatorBytes;
+    }
+
+    const remainingHits = hits.length - index;
+    const futureSeparators = separatorBytes * (remainingHits - 1);
+    const hitBytes = Math.floor((remainingBytes - futureSeparators) / remainingHits);
+    if (hitBytes <= 0) break;
+
+    const hit = formatMemoryHit(hits[index], hitBytes);
+    formatted.push(hit);
+    remainingBytes -= Buffer.byteLength(hit);
+  }
+
+  return formatted.join('');
 }
 
 const ANSI_SGR = new RegExp(String.fromCharCode(27) + '\\[[0-9;]*m', 'g');
@@ -612,7 +650,7 @@ async function goosedumpCompact(
   executablePath?: string,
 ): Promise<{ summary: string; warning: string }> {
   const scope = options.scope ?? 'lineage';
-  const args = ['compact', target, '--scope', scope];
+  const args = ['compact', target, '--train', '--scope', scope];
   if (options.from) args.push('--from', options.from);
   args.push('--until', options.until);
 
@@ -1519,7 +1557,7 @@ export function createGoosedumpIntegration(
           name: 'goose_recall',
           label: 'goose_recall',
           description:
-            'Recall durable context harvested from earlier compactions. Searches the current project by default and returns full remembered entries with provenance.',
+            'Recall durable context harvested from earlier compactions. Searches the current project by default and returns bounded remembered excerpts with provenance.',
           promptSnippet: 'Recall durable context from earlier sessions',
           promptGuidelines: [
             'Use goose_recall for decisions, evidence, and context that may have disappeared from the current session.',

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-goosedump",
-  "version": "0.10.8",
+  "version": "0.10.10",
   "description": "Pi extension for goosedump-backed session-history search, entry expansion, and compaction",
   "keywords": [
     "compaction",
@@ -33,7 +33,7 @@
   },
   "dependencies": {
     "@earendil-works/pi-tui": "^0.78.0",
-    "@jarkkojs/goosedump": "^0.10.8",
+    "@jarkkojs/goosedump": "^0.10.10",
     "@sinclair/typebox": "^0.34.49"
   },
   "devDependencies": {