@vibgrate/cli 2026.803.1 → 2026.803.4

package/dist/{chunk-5VH3Y4QC.js → chunk-TVRM57CE.js}

@@ -1,10 +1,5 @@
-import { cacheDir, WEAK_TERMS, expandConcepts } from './chunk-AJ2KE3NT.js';
-import { hashString } from './chunk-JFGYT6BI.js';
-import * as fs2 from 'fs';
-import { createRequire } from 'module';
-import * as os from 'os';
-import * as path2 from 'path';
-import { pathToFileURL } from 'url';
+import { cosine } from './chunk-2A7XCANX.js';
+import { bigramConsumedTokens, WEAK_TERMS, expandConcepts } from './chunk-4LGDQBYK.js';
 
 // src/engine/relations.ts
 var GraphIndex = class {
@@ -72,369 +67,6 @@ function indexFor(graph) {
   }
   return idx;
 }
-var DEFAULT_LOCK_STALE_MS = 15 * 60 * 1e3;
-function isProcessAlive(pid) {
-  try {
-    process.kill(pid, 0);
-    return true;
-  } catch (e) {
-    return e.code === "EPERM";
-  }
-}
-function lockIsStale(file, staleMs) {
-  try {
-    const { pid, at } = JSON.parse(fs2.readFileSync(file, "utf8"));
-    if (typeof at === "number" && Date.now() - at > staleMs) return true;
-    if (typeof pid === "number" && !isProcessAlive(pid)) return true;
-    return false;
-  } catch {
-    return true;
-  }
-}
-function acquireLock(file, staleMs = DEFAULT_LOCK_STALE_MS) {
-  const write = () => {
-    try {
-      fs2.mkdirSync(path2.dirname(file), { recursive: true });
-      const fd = fs2.openSync(file, "wx");
-      fs2.writeFileSync(fd, JSON.stringify({ pid: process.pid, at: Date.now() }));
-      fs2.closeSync(fd);
-      return true;
-    } catch {
-      return false;
-    }
-  };
-  if (write()) return true;
-  if (lockIsStale(file, staleMs)) {
-    try {
-      fs2.rmSync(file, { force: true });
-    } catch {
-    }
-    return write();
-  }
-  return false;
-}
-function releaseLock(file) {
-  try {
-    fs2.rmSync(file, { force: true });
-  } catch {
-  }
-}
-
-// src/engine/embeddings.ts
-function resolveEmbedModel(model) {
-  return model ?? "bge-small-en-v1.5";
-}
-function modelCacheDir() {
-  const base = process.env.XDG_CACHE_HOME || path2.join(os.homedir(), ".cache");
-  return path2.join(base, "vibgrate", "models");
-}
-function modelReadyMarker(modelId) {
-  return path2.join(modelCacheDir(), `.ready-${safe(modelId)}`);
-}
-function isModelReady(modelId = resolveEmbedModel()) {
-  return fs2.existsSync(modelReadyMarker(modelId));
-}
-function unavailableMessage(reason) {
-  switch (reason) {
-    case "no-permission":
-      return `semantic search couldn't write its model cache at ${modelCacheDir()} \u2014 using fast lexical search instead. Relocate it with the standard XDG_CACHE_HOME, or run with --local to stay lexical quietly. (No admin/sudo is ever needed \u2014 the cache lives in your home folder.)`;
-    case "download-failed":
-      return `the semantic model isn't downloaded yet and couldn't be fetched (offline?) \u2014 using fast lexical search. It'll try again next time; run with --local to keep lexical without retrying.`;
-    case "not-installed":
-      return `the optional semantic backend isn't available here \u2014 using fast lexical search.`;
-    case "init-failed":
-    default:
-      return `the semantic model couldn't load \u2014 using fast lexical search.`;
-  }
-}
-function modelCacheInfo(modelId = resolveEmbedModel()) {
-  const dir = modelCacheDir();
-  return { dir, present: isModelReady(modelId) || fs2.existsSync(dir), bytes: dirSize(dir) };
-}
-function clearModelCache() {
-  const dir = modelCacheDir();
-  const bytes = dirSize(dir);
-  try {
-    fs2.rmSync(dir, { recursive: true, force: true });
-  } catch {
-  }
-  return bytes;
-}
-function dirSize(dir) {
-  let total = 0;
-  let entries;
-  try {
-    entries = fs2.readdirSync(dir, { withFileTypes: true });
-  } catch {
-    return 0;
-  }
-  for (const e of entries) {
-    const p = path2.join(dir, e.name);
-    if (e.isDirectory()) total += dirSize(p);
-    else {
-      try {
-        total += fs2.statSync(p).size;
-      } catch {
-      }
-    }
-  }
-  return total;
-}
-function isPermissionError(e) {
-  const code = e?.code;
-  return code === "EACCES" || code === "EPERM" || code === "EROFS";
-}
-function embeddingsCached(root, modelId) {
-  return fs2.existsSync(path2.join(cacheDir(root), `embeddings-${safe(modelId)}.json`));
-}
-async function loadEmbedder(options = {}) {
-  if (options.local) return null;
-  const modelId = resolveEmbedModel(options.model);
-  if (options.noDownload && !isModelReady(modelId)) return null;
-  const fail = (reason) => {
-    options.onUnavailable?.(reason);
-    return null;
-  };
-  const mod = await importEmbedBackend();
-  if (!mod?.FlagEmbedding) return fail("not-installed");
-  const cache = modelCacheDir();
-  try {
-    fs2.mkdirSync(cache, { recursive: true });
-  } catch (e) {
-    return fail(isPermissionError(e) ? "no-permission" : "init-failed");
-  }
-  try {
-    const model = await withTimeout(
-      mod.FlagEmbedding.init({
-        model: mapModel(mod, modelId),
-        cacheDir: cache,
-        showDownloadProgress: options.showDownloadProgress ?? false
-      }),
-      embedInitTimeoutMs(),
-      "embedding model init timed out"
-    );
-    try {
-      fs2.writeFileSync(modelReadyMarker(modelId), (/* @__PURE__ */ new Date(0)).toISOString());
-    } catch {
-    }
-    return {
-      id: modelId,
-      async embed(texts) {
-        const out = [];
-        for await (const batch of model.embed(texts, 64)) {
-          for (const v of batch) out.push(Array.from(v));
-        }
-        return out;
-      },
-      async embedQuery(text) {
-        const v = await model.queryEmbed(text);
-        return Array.from(v);
-      }
-    };
-  } catch (e) {
-    return fail(isPermissionError(e) ? "no-permission" : "download-failed");
-  }
-}
-async function importEmbedBackend() {
-  const hostDir = process.env.VIBGRATE_EMBEDDER_PATH;
-  if (hostDir) {
-    try {
-      const req = createRequire(path2.join(hostDir, "package.json"));
-      const mod = await import(pathToFileURL(req.resolve("fastembed")).href);
-      const resolved = mod?.FlagEmbedding ? mod : mod?.default;
-      if (resolved?.FlagEmbedding) return resolved;
-    } catch {
-    }
-  }
-  return await import('fastembed').catch(() => null);
-}
-function embedInitTimeoutMs() {
-  const n = Number(process.env.VG_EMBED_TIMEOUT_MS);
-  return Number.isFinite(n) && n > 0 ? n : 6e4;
-}
-async function withTimeout(promise, ms, message) {
-  let timer;
-  try {
-    return await Promise.race([
-      promise,
-      new Promise((_, reject) => {
-        timer = setTimeout(() => reject(new Error(message)), ms);
-        timer.unref?.();
-      })
-    ]);
-  } finally {
-    clearTimeout(timer);
-  }
-}
-function mapModel(mod, id) {
-  const m = mod.EmbeddingModel ?? {};
-  if (/bge-small/i.test(id)) return m.BGESmallENV15 ?? m.BGESmallEN ?? id;
-  if (/bge-base/i.test(id)) return m.BGEBaseENV15 ?? id;
-  if (/all-minilm/i.test(id)) return m.AllMiniLML6V2 ?? id;
-  return id;
-}
-var PATH_NOISE = /* @__PURE__ */ new Set([
-  "src",
-  "app",
-  "apps",
-  "lib",
-  "libs",
-  "dist",
-  "build",
-  "out",
-  "index",
-  "main",
-  "packages",
-  "pkg",
-  "test",
-  "tests",
-  "__tests__",
-  "spec",
-  "node_modules",
-  "internal",
-  "com",
-  "example",
-  "demo",
-  "java",
-  "net",
-  "org",
-  "www"
-]);
-function pathContext(file) {
-  const parts = file.split("/");
-  const stem = (parts.pop() ?? "").replace(/\.[^.]+$/, "");
-  const dirs = parts.filter((p) => p && !PATH_NOISE.has(p.toLowerCase()));
-  return [...dirs.slice(-2), stem].filter(Boolean).join(" ");
-}
-function nodeEmbedText(node, areaLabel) {
-  if (node.kind === "document") {
-    const category = (node.signature ?? "document").replace(/^document:/, "");
-    const categoryHints = {
-      manifest: "package dependencies scripts install dependencies versions",
-      docker: "dockerfile container image build runtime",
-      compose: "docker-compose services containers networking volumes",
-      ci: "continuous integration github actions workflow pipeline test build deploy",
-      "build-config": "typescript tsconfig bundler vite webpack eslint prettier test",
-      "api-contract": "openapi swagger graphql protobuf api schema endpoints",
-      "task-runner": "makefile justfile tasks build test run commands",
-      workspace: "monorepo workspace packages pnpm turbo nx",
-      infra: "deploy infrastructure kubernetes kubernetes terraform helm kubernetes",
-      "env-example": "environment variables configuration secrets template",
-      markdown: "documentation readme guide"
-    };
-    const hint = categoryHints[category] ?? category;
-    return [
-      node.qualifiedName,
-      "document",
-      category,
-      hint,
-      pathContext(node.file),
-      (node.doc ?? "").slice(0, 5e3)
-    ].filter(Boolean).join("\n");
-  }
-  return [node.qualifiedName, node.kind, node.signature ?? "", node.doc ?? "", pathContext(node.file), areaLabel ?? ""].filter(Boolean).join(" ");
-}
-function cosine(a, b) {
-  let dot = 0;
-  let na = 0;
-  let nb = 0;
-  const n = Math.min(a.length, b.length);
-  for (let i = 0; i < n; i++) {
-    dot += a[i] * b[i];
-    na += a[i] * a[i];
-    nb += b[i] * b[i];
-  }
-  if (na === 0 || nb === 0) return 0;
-  return dot / (Math.sqrt(na) * Math.sqrt(nb));
-}
-var EMBED_CHUNK = 256;
-var CACHE_WRITE_INTERVAL_MS = 1500;
-function vectorCachePath(root, modelId) {
-  return path2.join(cacheDir(root), `embeddings-${safe(modelId)}.json`);
-}
-function countPending(graph, root, modelId) {
-  const entries = readCacheEntries(vectorCachePath(root, modelId), modelId);
-  const areaLabel = new Map(graph.areas.map((a) => [a.id, a.label]));
-  let pending = 0;
-  for (const n of graph.nodes) {
-    if (n.kind === "file" || n.kind === "external") continue;
-    const h = hashString(nodeEmbedText(n, areaLabel.get(n.area)));
-    const cached = entries[n.id];
-    if (!cached || cached.hash !== h) pending++;
-  }
-  return pending;
-}
-function readCacheEntries(file, modelId) {
-  if (!fs2.existsSync(file)) return {};
-  try {
-    const loaded = JSON.parse(fs2.readFileSync(file, "utf8"));
-    if (loaded.model === modelId && loaded.entries) return loaded.entries;
-  } catch {
-  }
-  return {};
-}
-async function getNodeEmbeddings(graph, embedder, root, onProgress) {
-  const file = vectorCachePath(root, embedder.id);
-  let cache = { model: embedder.id, entries: {} };
-  if (fs2.existsSync(file)) {
-    try {
-      const loaded = JSON.parse(fs2.readFileSync(file, "utf8"));
-      if (loaded.model === embedder.id && loaded.entries) cache = loaded;
-    } catch {
-    }
-  }
-  const areaLabel = new Map(graph.areas.map((a) => [a.id, a.label]));
-  const targets = graph.nodes.filter((n) => n.kind !== "file" && n.kind !== "external");
-  const toEmbed = [];
-  const vectors = /* @__PURE__ */ new Map();
-  for (const n of targets) {
-    const text = nodeEmbedText(n, areaLabel.get(n.area));
-    const h = hashString(text);
-    const cached = cache.entries[n.id];
-    if (cached && cached.hash === h) vectors.set(n.id, cached.vec);
-    else toEmbed.push({ id: n.id, text, hash: h });
-  }
-  const persist = () => {
-    try {
-      fs2.mkdirSync(cacheDir(root), { recursive: true });
-      const tmp = `${file}.tmp.${process.pid}`;
-      fs2.writeFileSync(tmp, JSON.stringify(cache));
-      fs2.renameSync(tmp, file);
-    } catch {
-    }
-  };
-  if (toEmbed.length) {
-    const lock = `${file}.lock`;
-    if (!acquireLock(lock)) return vectors;
-    try {
-      onProgress?.(0, toEmbed.length);
-      let lastWrite = Date.now();
-      for (let i = 0; i < toEmbed.length; i += EMBED_CHUNK) {
-        const slice = toEmbed.slice(i, i + EMBED_CHUNK);
-        const vecs = await embedder.embed(slice.map((t) => t.text));
-        slice.forEach((t, j) => {
-          const vec = vecs[j] ?? [];
-          vectors.set(t.id, vec);
-          cache.entries[t.id] = { hash: t.hash, vec };
-        });
-        onProgress?.(Math.min(i + EMBED_CHUNK, toEmbed.length), toEmbed.length);
-        if (Date.now() - lastWrite >= CACHE_WRITE_INTERVAL_MS) {
-          persist();
-          lastWrite = Date.now();
-        }
-      }
-      const live = new Set(targets.map((n) => n.id));
-      for (const id of Object.keys(cache.entries)) if (!live.has(id)) delete cache.entries[id];
-      persist();
-    } finally {
-      releaseLock(lock);
-    }
-  }
-  return vectors;
-}
-function safe(id) {
-  return id.replace(/[^a-z0-9.-]+/gi, "_");
-}
 
 // src/engine/query.ts
 var STOPWORDS = /* @__PURE__ */ new Set([
@@ -587,7 +219,7 @@ function queryGraph(graph, question, options = {}) {
   const limit = options.limit ?? 12;
   const literals = extractLiteralNeedles(question);
   const forTokens = literals.length > 0 ? stripLiteralNeedles(question) : question;
-  const prep = prepareTerms(forTokens, options.relevance);
+  const prep = prepareTerms(forTokens, options.relevance, priorResidual(options.priorQuestion));
   const weightOf = termWeights(graph, prep.terms.map((t) => t.term));
   const index = indexFor(graph);
   const scored = [];
@@ -620,7 +252,7 @@ async function queryGraphSemantic(graph, question, options) {
   const limit = options.limit ?? 12;
   const literals = extractLiteralNeedles(question);
   const forTokens = literals.length > 0 ? stripLiteralNeedles(question) : question;
-  const prep = prepareTerms(forTokens, options.relevance);
+  const prep = prepareTerms(forTokens, options.relevance, priorResidual(options.priorQuestion));
   const terms = prep.terms.map((t) => t.term);
   const weightOf = termWeights(graph, terms);
   const index = indexFor(graph);
@@ -679,11 +311,13 @@ function tokenizeOrdered(q) {
 var WEAK_TERM_WEIGHT = 0.25;
 var EXPANSION_WEIGHT = 0.7;
 var COVERAGE_BONUS = 0.35;
-function prepareTerms(forTokens, relevance) {
+var CARRY_WEIGHT = 0.5;
+function prepareTerms(forTokens, relevance, prior) {
   const base = tokenize(forTokens);
   const ordered = tokenizeOrdered(forTokens);
   const baseSet = new Set(base);
-  const terms = base.map((t) => ({ term: t, weak: WEAK_TERMS.has(t) }));
+  const consumed = bigramConsumedTokens(ordered);
+  const terms = base.map((t) => ({ term: t, weak: WEAK_TERMS.has(t) || consumed.has(t) }));
   const have = new Set(base);
   for (const e of expandConcepts(ordered)) {
     if (baseSet.has(e.term)) continue;
@@ -695,9 +329,72 @@ function prepareTerms(forTokens, relevance) {
     terms.push({ term: e.term, weak: false, from: e.from, providerWeight: e.weight });
     have.add(e.term);
   }
+  if (prior?.trim()) {
+    const priorOrdered = tokenizeOrdered(prior);
+    const priorConsumed = bigramConsumedTokens(priorOrdered);
+    for (const t of tokenize(prior)) {
+      if (have.has(t) || WEAK_TERMS.has(t) || priorConsumed.has(t)) continue;
+      terms.push({ term: t, weak: false, carried: true });
+      have.add(t);
+    }
+    for (const e of expandConcepts(priorOrdered)) {
+      if (have.has(e.term)) continue;
+      terms.push({ term: e.term, weak: false, from: e.from, carried: true });
+      have.add(e.term);
+    }
+  }
   const hasContent = terms.some((t) => !t.weak);
   return { terms, hasContent };
 }
+function priorResidual(prior) {
+  const raw = prior?.trim();
+  if (!raw) return null;
+  return extractLiteralNeedles(raw).length > 0 ? stripLiteralNeedles(raw) : raw;
+}
+function conceptMapLines(question, relevance, priorQuestion) {
+  const cap = (xs) => [...new Set(xs)].slice(0, 8).join(", ");
+  const literals = extractLiteralNeedles(question);
+  const residual = literals.length > 0 ? stripLiteralNeedles(question) : question;
+  const lines = [];
+  const byFrom = /* @__PURE__ */ new Map();
+  for (const e of expandConcepts(tokenizeOrdered(residual))) {
+    const list = byFrom.get(e.from) ?? [];
+    list.push(e.term);
+    byFrom.set(e.from, list);
+  }
+  for (const [from, terms] of byFrom) {
+    lines.push(`- "${from}" in the ask implies these codebase identifiers: ${cap(terms)}.`);
+  }
+  if (relevance?.topics?.length) {
+    const topicVocab = /* @__PURE__ */ new Map();
+    for (const e of relevance.expansions ?? []) {
+      const list = topicVocab.get(e.from) ?? [];
+      list.push(e.term);
+      topicVocab.set(e.from, list);
+    }
+    for (const t of relevance.topics.slice(0, 3)) {
+      const vocab = topicVocab.get(t.id);
+      lines.push(
+        vocab?.length ? `- The ask is about the "${t.id}" domain; code for it typically uses: ${cap(vocab)}.` : `- The ask is about the "${t.id}" domain.`
+      );
+    }
+  }
+  const priorText = priorResidual(priorQuestion);
+  if (priorText) {
+    const current = new Set(tokenize(residual));
+    const consumed = bigramConsumedTokens(tokenizeOrdered(priorText));
+    const carried = tokenize(priorText).filter((t) => !current.has(t) && !WEAK_TERMS.has(t) && !consumed.has(t));
+    if (carried.length) {
+      lines.push(`- This is a follow-up; topic words carried from the previous ask: ${cap(carried)}.`);
+    }
+  }
+  if (lines.length) {
+    lines.push(
+      '- Seed notation below: `a\u2192b` = ask term "a" implied identifier "b"; `~t` = word-form match; `\u21A9t` = carried from the previous ask.'
+    );
+  }
+  return lines;
+}
 function scoreNode(node, terms, weightOf = () => 1) {
   let score = 0;
   let contentHits = 0;
@@ -707,10 +404,10 @@ function scoreNode(node, terms, weightOf = () => 1) {
   const file = node.file.toLowerCase();
   const dirSegments = new Set(file.split("/").slice(0, -1));
   const nameParts = identifierParts(node.name);
-  for (const { term: t, weak, from, providerWeight } of terms) {
+  for (const { term: t, weak, from, providerWeight, carried } of terms) {
     const expansionW = providerWeight !== void 0 ? Math.min(providerWeight, EXPANSION_WEIGHT) : from ? EXPANSION_WEIGHT : 1;
-    const w = weightOf(t) * (weak ? WEAK_TERM_WEIGHT : 1) * expansionW;
-    const label = from ? `${from}\u2192${t}` : t;
+    const w = weightOf(t) * (weak ? WEAK_TERM_WEIGHT : 1) * expansionW * (carried ? CARRY_WEIGHT : 1);
+    const label = (carried ? "\u21A9" : "") + (from ? `${from}\u2192${t}` : t);
     let contribution = 0;
     if (name === t) {
       contribution = 10 * w;
@@ -726,7 +423,7 @@ function scoreNode(node, terms, weightOf = () => 1) {
       hits.push(label);
     } else if (fuzzyPartMatch(t, nameParts)) {
       contribution = 2 * w;
-      hits.push(from ? `~${from}\u2192${t}` : `~${t}`);
+      hits.push((carried ? "\u21A9" : "") + (from ? `~${from}\u2192${t}` : `~${t}`));
     } else if (dirSegments.has(t)) {
       contribution = 2 * w;
       hits.push(label);
@@ -895,6 +592,6 @@ function round2(x) {
   return Math.round(x * 1e6) / 1e6;
 }
 
-export { GraphIndex, acquireLock, clearModelCache, cosine, countPending, embeddingsCached, extractLiteralNeedles, getNodeEmbeddings, identifierParts, impactOf, indexFor, isLocateOnlyInstruction, isModelReady, loadEmbedder, modelCacheInfo, nodeEmbedText, queryGraph, queryGraphSemantic, releaseLock, residualForSymbolSearch, resolveEmbedModel, unavailableMessage, withTimeout };
-//# sourceMappingURL=chunk-5VH3Y4QC.js.map
-//# sourceMappingURL=chunk-5VH3Y4QC.js.map
+export { GraphIndex, conceptMapLines, extractLiteralNeedles, identifierParts, impactOf, indexFor, isLocateOnlyInstruction, queryGraph, queryGraphSemantic, residualForSymbolSearch };
+//# sourceMappingURL=chunk-TVRM57CE.js.map
+//# sourceMappingURL=chunk-TVRM57CE.js.map