@yurtsever/capsa 0.1.0-alpha.3 → 0.1.0-alpha.5

package/dist/index.js

@@ -7228,6 +7228,9 @@ var DEFAULT_EXCLUDES = [
   ".Trash"
 ];
 var DEFAULT_MAX_DEPTH = 12;
+function isSourceAdapter(adapter) {
+  return typeof adapter.collect === "function";
+}
 function stableId(absPath) {
   return createHash("sha1").update(absPath).digest("hex").slice(0, 12);
 }
@@ -7246,16 +7249,24 @@ async function scan(adapters, options = {}) {
   const items = [];
   const seenItemPaths = /* @__PURE__ */ new Set();
   const scannedRoots = [];
+  const formatAdapters = adapters.filter((a) => !isSourceAdapter(a));
+  const sourceAdapters = adapters.filter(isSourceAdapter);
   const rootSet = /* @__PURE__ */ new Set();
   for (const r of options.roots ?? [env.homeDir])
     rootSet.add(resolve(r));
-  for (const adapter of adapters) {
+  for (const adapter of formatAdapters) {
     for (const r of adapter.globalRoots?.(env) ?? [])
       rootSet.add(resolve(r));
   }
   const roots = [...rootSet].filter((r) => ![...rootSet].some((other) => other !== r && r.startsWith(other + "/")));
+  const addItem = (item) => {
+    if (seenItemPaths.has(item.path))
+      return;
+    seenItemPaths.add(item.path);
+    items.push(item);
+  };
   const handleEntry = async (entry) => {
-    for (const adapter of adapters) {
+    for (const adapter of formatAdapters) {
       let matched = false;
       try {
         matched = adapter.matches(entry);
@@ -7270,12 +7281,8 @@ async function scan(adapters, options = {}) {
       if (!matched)
         continue;
       try {
-        for (const item of await adapter.parse(entry, env)) {
-          if (seenItemPaths.has(item.path))
-            continue;
-          seenItemPaths.add(item.path);
-          items.push(item);
-        }
+        for (const item of await adapter.parse(entry, env))
+          addItem(item);
       } catch (err) {
         issues.push({
           path: entry.absPath,
@@ -7346,6 +7353,18 @@ async function scan(adapters, options = {}) {
     }
     scannedRoots.push(root);
     await walk(root, root, 0);
+    for (const adapter of sourceAdapters) {
+      try {
+        for (const item of await adapter.collect(root, env))
+          addItem(item);
+      } catch (err) {
+        issues.push({
+          path: root,
+          adapter: adapter.id,
+          message: `collect() threw: ${String(err)}`
+        });
+      }
+    }
   }
   return { items, issues, scannedRoots, durationMs: Date.now() - started };
 }
@@ -10227,6 +10246,10 @@ var Store = class {
       }
     }));
   }
+  /** All chunks of one item, in document order. */
+  chunksForItem(itemId) {
+    return this.db.prepare("SELECT id, item_id, idx, heading_path, text FROM chunks WHERE item_id = ? ORDER BY idx").all(itemId);
+  }
   listItems() {
     return this.db.prepare("SELECT * FROM items ORDER BY kind, rel_path").all();
   }
@@ -10246,9 +10269,164 @@ var Store = class {
 import { createHash as createHash2 } from "crypto";
 import { relative as relative5 } from "path";
 
+// ../adapter-git-log/dist/index.js
+import { execFile } from "child_process";
+import { existsSync as existsSync2 } from "fs";
+import { basename as basename4, join as join7 } from "path";
+import { promisify } from "util";
+var execFileAsync = promisify(execFile);
+var DEFAULTS = {
+  limit: 200,
+  maxFiles: 50,
+  maxBodyChars: 4e3,
+  includeMerges: false
+};
+var RECORD = "";
+var UNIT = "";
+var FORMAT = `${RECORD}%H${UNIT}%h${UNIT}%an${UNIT}%cI${UNIT}%s${UNIT}%b${UNIT}`;
+var ITEM_DIR = join7(".git", "commits");
+function gitLogAdapter(options = {}) {
+  const opts = { ...DEFAULTS, ...options };
+  return {
+    id: "git-log",
+    displayName: "Commit",
+    async collect(root, _env) {
+      if (!existsSync2(join7(root, ".git")))
+        return [];
+      const stdout = await runGitLog(root, opts);
+      const items = [];
+      for (const record2 of stdout.split(RECORD)) {
+        const commit = parseRecord(record2);
+        if (commit)
+          items.push(toItem(root, commit, opts));
+      }
+      return items;
+    }
+  };
+}
+async function runGitLog(root, opts) {
+  const args = [
+    "--no-pager",
+    "log",
+    `--max-count=${opts.limit}`,
+    "--name-only",
+    "--no-color",
+    `--format=${FORMAT}`
+  ];
+  if (!opts.includeMerges)
+    args.push("--no-merges");
+  try {
+    const { stdout } = await execFileAsync("git", args, {
+      cwd: root,
+      timeout: 3e4,
+      maxBuffer: 32 * 1024 * 1024,
+      windowsHide: true
+    });
+    return stdout;
+  } catch (err) {
+    const e = err;
+    if (/does not have any commits yet|bad default revision/i.test(String(e.stderr ?? ""))) {
+      return "";
+    }
+    if (e.code === "ENOENT") {
+      throw new Error(`${root} has a .git but git is not on PATH; no commits indexed`);
+    }
+    throw new Error(`git log failed in ${root}: ${firstLine2(e.stderr) || String(err)}`);
+  }
+}
+function parseRecord(record2) {
+  if (!record2.trim())
+    return void 0;
+  const parts = record2.split(UNIT);
+  const [sha, shortSha, author, date5, subject, body] = parts;
+  if (!sha || !shortSha)
+    return void 0;
+  return {
+    sha,
+    shortSha,
+    author: author ?? "",
+    date: date5 ?? "",
+    subject: subject ?? "",
+    body: (body ?? "").trim(),
+    // Everything past the last field is the file list, one path per line.
+    files: parts.slice(6).join(UNIT).split("\n").map((l) => l.trim()).filter(Boolean)
+  };
+}
+function toItem(root, c, opts) {
+  const path = join7(root, ITEM_DIR, `${c.shortSha}.md`);
+  const files = c.files.slice(0, opts.maxFiles);
+  const hidden = c.files.length - files.length;
+  const subject = c.subject || "(no subject)";
+  return {
+    id: stableId(path),
+    format: "git-log",
+    formatLabel: "Commit",
+    name: `${c.shortSha} ${subject}`,
+    // The description becomes the provenance line of every embedded chunk,
+    // so it carries the subject, not the sha.
+    description: subject,
+    path,
+    scope: "project",
+    projectRoot: root,
+    content: renderCommit(c, subject, files, hidden, opts.maxBodyChars),
+    metadata: {
+      sha: c.sha,
+      shortSha: c.shortSha,
+      author: c.author,
+      subject,
+      files: c.files,
+      repo: basename4(root)
+    },
+    knowledge: {
+      kind: "commit",
+      updatedAt: isoDate(c.date),
+      relatedPaths: c.files.length ? c.files : void 0,
+      tags: conventionalTags(subject)
+    }
+  };
+}
+function renderCommit(c, subject, files, hidden, maxBodyChars) {
+  const body = c.body.length > maxBodyChars ? `${c.body.slice(0, maxBodyChars)}
+
+\u2026` : c.body;
+  const lines = [
+    `# ${subject}`,
+    "",
+    `commit \`${c.shortSha}\` \xB7 ${c.author} \xB7 ${c.date.slice(0, 10)}`
+  ];
+  if (body)
+    lines.push("", body);
+  if (files.length) {
+    lines.push("", "## Files", "");
+    for (const f of files)
+      lines.push(`- ${f}`);
+    if (hidden > 0)
+      lines.push(`- \u2026 and ${hidden} more`);
+  }
+  return `${lines.join("\n")}
+`;
+}
+function isoDate(raw2) {
+  if (!raw2)
+    return void 0;
+  const t = Date.parse(raw2);
+  return Number.isNaN(t) ? void 0 : new Date(t).toISOString();
+}
+var CONVENTIONAL = /^([a-z]{3,10})(?:\(([^)]{1,40})\))?!?:\s/;
+function conventionalTags(subject) {
+  const m = CONVENTIONAL.exec(subject.toLowerCase());
+  if (!m)
+    return void 0;
+  const tags = [m[1], ...m[2] ? [m[2]] : []];
+  return tags;
+}
+function firstLine2(text) {
+  return (text ?? "").split("\n").find((l) => l.trim()) ?? "";
+}
+
 // ../adapter-runbooks/dist/index.js
 import { readFile as readFile5, stat as stat3 } from "fs/promises";
-import { basename as basename4, dirname as dirname7, relative as relative3, sep as sep4 } from "path";
+import { basename as basename5, dirname as dirname7, relative as relative3, sep as sep4 } from "path";
 var DOC_DIRS = /* @__PURE__ */ new Set(["docs", "doc", "runbooks", "runbook", "wiki", "adr", "decisions"]);
 var EXCLUDED_DIRS = /* @__PURE__ */ new Set(["tickets", "issues", "todo", "backlog", "plans", "plan", "roadmap", "milestones", "milestone", "node_modules"]);
 function segments(relPath) {
@@ -10279,7 +10457,7 @@ function runbooksAdapter() {
       const relPathInProject = relative3(projectRoot, entry.absPath);
       const fm = parseFrontmatter(raw2);
       const data = fm.data;
-      const title = data["title"] || firstHeading(raw2) || basename4(entry.name, ".md");
+      const title = data["title"] || firstHeading(raw2) || basename5(entry.name, ".md");
       const kindFromPath = segments(relPathInProject).some((p) => ["adr", "decisions"].includes(p.toLowerCase())) ? "decision" : "runbook";
       const tags = parseTags(data["tags"]);
       return [
@@ -10287,7 +10465,7 @@ function runbooksAdapter() {
           id: stableId(entry.absPath),
           format: "runbooks",
           formatLabel: "Runbook",
-          name: `${basename4(projectRoot)}${sep4}${relPathInProject}`,
+          name: `${basename5(projectRoot)}${sep4}${relPathInProject}`,
           description: title,
           path: entry.absPath,
           scope: "project",
@@ -10321,7 +10499,7 @@ function firstHeading(raw2) {
 
 // ../adapter-tickets/dist/index.js
 import { readFile as readFile6, stat as stat4 } from "fs/promises";
-import { basename as basename5, dirname as dirname8, relative as relative4, sep as sep5 } from "path";
+import { basename as basename6, dirname as dirname8, relative as relative4, sep as sep5 } from "path";
 var TICKET_DIRS = /* @__PURE__ */ new Set(["tickets", "issues", "todo", "backlog"]);
 var PLAN_DIRS = /* @__PURE__ */ new Set(["plans", "plan", "roadmap", "milestones", "milestone"]);
 var CHECKBOX = /^\s*[-*+]\s+\[( |x|X)\]\s+/gm;
@@ -10382,14 +10560,14 @@ function ticketsAdapter() {
       const data = fm.data;
       const derived2 = deriveStatus(raw2);
       const status = normaliseStatus(data["status"]) ?? derived2.status;
-      const title = data["title"] || firstHeading2(raw2) || basename5(entry.name, ".md");
+      const title = data["title"] || firstHeading2(raw2) || basename6(entry.name, ".md");
       const tags = parseTags2(data["tags"]);
       return [
         {
           id: stableId(entry.absPath),
           format: "tickets",
           formatLabel: isPlan ? "Plan" : "Ticket",
-          name: `${basename5(projectRoot)}${sep5}${relPathInProject}`,
+          name: `${basename6(projectRoot)}${sep5}${relPathInProject}`,
           description: title,
           path: entry.absPath,
           scope: "project",
@@ -10433,7 +10611,8 @@ function knowledgeAdapters() {
     claudeSkillsAdapter(),
     cursorRulesAdapter(),
     runbooksAdapter(),
-    ticketsAdapter()
+    ticketsAdapter(),
+    gitLogAdapter()
   ];
 }
 function openStore(projectRoot, embedder) {
@@ -10566,6 +10745,40 @@ function rrf(lists, k = DEFAULT_WEIGHTS.rrfK) {
   return scores;
 }
 
+// ../index/dist/pin.js
+var LEADING_ID = /^([A-Za-z]{1,8}-?\d{1,5}(?:-\d{1,5})?|\d{2,6})(?=[-_.\s]|$)/;
+function leadingId(relPath) {
+  const base = relPath.split(/[\\/]/).pop() ?? "";
+  return LEADING_ID.exec(base)?.[1];
+}
+function pinnedItems(query, items) {
+  const q = query.toLowerCase();
+  const idCount = /* @__PURE__ */ new Map();
+  for (const item of items) {
+    const id = leadingId(item.rel_path)?.toLowerCase();
+    if (id)
+      idCount.set(id, (idCount.get(id) ?? 0) + 1);
+  }
+  const out = [];
+  for (const item of items) {
+    const rel = item.rel_path.toLowerCase();
+    if (q.includes(rel)) {
+      out.push(item);
+      continue;
+    }
+    const id = leadingId(item.rel_path);
+    if (!id || (idCount.get(id.toLowerCase()) ?? 0) > 1)
+      continue;
+    const re = new RegExp(`(^|[^\\w-])${escape2(id)}(?=$|[^\\w-])`, "i");
+    if (re.test(query))
+      out.push(item);
+  }
+  return out;
+}
+function escape2(s) {
+  return s.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}
+
 // ../index/dist/retrieve.js
 async function retrieve(store, embedder, query, options = {}) {
   const started = Date.now();
@@ -10573,31 +10786,73 @@ async function retrieve(store, embedder, query, options = {}) {
   const candidates = options.candidates ?? 30;
   const weights = options.weights ?? DEFAULT_WEIGHTS;
   const minRelative = options.minRelativeScore ?? 0.5;
+  const perItem = /* @__PURE__ */ new Map();
+  const chunks = [];
+  const debug = [];
+  let tokens = 0;
+  const pinned = pinnedItems(query, store.listItems());
+  for (const item of pinned) {
+    for (const c of store.chunksForItem(item.id)) {
+      const t = estimateTokens(c.text);
+      const fits = tokens + t <= maxTokens || chunks.length === 0;
+      if (options.debug) {
+        debug.push({
+          chunkId: c.id,
+          relPath: item.rel_path,
+          headingPath: JSON.parse(c.heading_path),
+          kind: item.kind,
+          status: item.status,
+          vecRank: null,
+          ftsRank: null,
+          fused: 0,
+          factor: 1,
+          score: Number.POSITIVE_INFINITY,
+          kept: fits,
+          why: fits ? "pinned" : "budget"
+        });
+      }
+      if (!fits)
+        continue;
+      perItem.set(item.id, (perItem.get(item.id) ?? 0) + 1);
+      tokens += t;
+      chunks.push({
+        chunkId: c.id,
+        score: Number.POSITIVE_INFINITY,
+        kind: item.kind,
+        status: item.status,
+        updatedAt: item.updated_at,
+        relPath: item.rel_path,
+        headingPath: JSON.parse(c.heading_path),
+        text: c.text,
+        tokens: t
+      });
+    }
+  }
+  const pinnedIds = new Set(pinned.map((i) => i.id));
+  const neighbourCutoff = pinned.length ? Math.max(minRelative, 0.75) : minRelative;
+  const neighbourBudget = pinned.length ? Math.min(maxTokens, tokens + Math.floor(maxTokens / 3)) : maxTokens;
   const [qv] = await embedder.embed([query]);
   const vec = qv ? store.vectorSearch(qv, candidates).map((r) => r.chunkId) : [];
   const fts = store.textSearch(query, candidates).map((r) => r.chunkId);
   const fused = rrf([vec, fts], weights.rrfK);
-  const rows = store.chunksWithItems([...fused.keys()]);
+  const rows = store.chunksWithItems([...fused.keys()]).filter((r) => !pinnedIds.has(r.item_id));
   const now = Date.now();
   const scored = rows.map((r) => {
     const fusedScore = fused.get(r.id) ?? 0;
     const factor = signalFactor({ kind: r.item.kind, status: r.item.status, updatedAt: r.item.updated_at }, weights, now);
     return { row: r, fusedScore, factor, score: fusedScore * factor };
   }).sort((a, b) => b.score - a.score);
-  const perItem = /* @__PURE__ */ new Map();
-  const chunks = [];
-  const debug = [];
-  let tokens = 0;
   const best = scored[0]?.score ?? 0;
-  let cut = false;
+  let cut = tokens >= maxTokens;
+  let kept = 0;
   for (const { row, fusedScore, factor, score } of scored) {
     let why = "kept";
-    if (cut || chunks.length > 0 && score < best * minRelative) {
+    if (cut || kept > 0 && score < best * neighbourCutoff) {
       cut = true;
       why = "cutoff";
     } else if ((perItem.get(row.item_id) ?? 0) >= 2) {
       why = "per-item cap";
-    } else if (tokens + estimateTokens(row.text) > maxTokens && chunks.length > 0) {
+    } else if (tokens + estimateTokens(row.text) > neighbourBudget && chunks.length > 0) {
       why = "budget";
     }
     if (options.debug) {
@@ -10621,6 +10876,7 @@ async function retrieve(store, embedder, query, options = {}) {
     const t = estimateTokens(row.text);
     perItem.set(row.item_id, (perItem.get(row.item_id) ?? 0) + 1);
     tokens += t;
+    kept++;
     chunks.push({
       chunkId: row.id,
       score,
@@ -10632,7 +10888,7 @@ async function retrieve(store, embedder, query, options = {}) {
       text: row.text,
       tokens: t
     });
-    if (tokens >= maxTokens)
+    if (tokens >= neighbourBudget)
       cut = true;
   }
   const durationMs = Date.now() - started;
@@ -10652,15 +10908,15 @@ function rankIn(list, id) {
 function formatDebug(result) {
   if (!result.debug)
     return "";
-  const lines = ["  vec  fts  factor  score    keep  where"];
+  const lines = ["  vec  fts  factor  score    keep         where"];
   for (const d of result.debug) {
     const where = [d.relPath, ...d.headingPath].join(" \u203A ").slice(0, 70);
     lines.push([
       String(d.vecRank ?? "-").padStart(5),
       String(d.ftsRank ?? "-").padStart(4),
       d.factor.toFixed(2).padStart(7),
-      d.score.toFixed(4).padStart(8),
-      (d.kept ? "yes" : d.why).padEnd(12),
+      (Number.isFinite(d.score) ? d.score.toFixed(4) : "pin").padStart(8),
+      (d.kept ? d.why === "pinned" ? "pinned" : "yes" : d.why).padEnd(12),
       where
     ].join(" "));
   }
@@ -10688,7 +10944,7 @@ function projectState(store) {
   const live = (i) => (i.kind === "ticket" || i.kind === "plan") && (i.status === "open" || i.status === "in-progress");
   const rank = (i) => i.status === "in-progress" ? 0 : 1;
   const open2 = items.filter(live).sort((a, b) => rank(a) - rank(b) || a.rel_path.localeCompare(b.rel_path)).map((i) => ({ relPath: i.rel_path, name: i.name, status: i.status ?? "unknown", updatedAt: i.updated_at }));
-  const recentlyChanged = [...items].sort((a, b) => (b.updated_at ?? "").localeCompare(a.updated_at ?? "")).slice(0, 10).map((i) => ({ relPath: i.rel_path, kind: i.kind, updatedAt: i.updated_at }));
+  const recentlyChanged = items.filter((i) => i.kind !== "commit").sort((a, b) => (b.updated_at ?? "").localeCompare(a.updated_at ?? "")).slice(0, 10).map((i) => ({ relPath: i.rel_path, kind: i.kind, updatedAt: i.updated_at }));
   const decisions = store.db.prepare("SELECT id, created_at, title FROM decisions ORDER BY id DESC LIMIT 10").all().map((d) => ({ id: d.id, createdAt: d.created_at, title: d.title }));
   return { counts, open: open2, recentlyChanged, decisions };
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yurtsever/capsa",
-  "version": "0.1.0-alpha.3",
+  "version": "0.1.0-alpha.5",
   "description": "Your project's knowledge in one capsule. Local-first project memory and cockpit for AI coding agents: indexes runbooks, tickets, plans and instruction files, serves the smallest relevant context over MCP.",
   "type": "module",
   "license": "MIT",