@inerrata-corporation/errata 2.0.22-dev.4020 → 2.0.22-dev.4054

package/generalize-worker.mjs

@@ -1235,10 +1235,10 @@ function mergeDefs(...defs) {
 function cloneDef(schema) {
   return mergeDefs(schema._zod.def);
 }
-function getElementAtPath(obj, path) {
-  if (!path)
+function getElementAtPath(obj, path2) {
+  if (!path2)
     return obj;
-  return path.reduce((acc, key) => acc?.[key], obj);
+  return path2.reduce((acc, key) => acc?.[key], obj);
 }
 function promiseAllObject(promisesObj) {
   const keys = Object.keys(promisesObj);
@@ -1621,11 +1621,11 @@ function aborted(x, startIndex = 0) {
   }
   return false;
 }
-function prefixIssues(path, issues) {
+function prefixIssues(path2, issues) {
   return issues.map((iss) => {
     var _a2;
     (_a2 = iss).path ?? (_a2.path = []);
-    iss.path.unshift(path);
+    iss.path.unshift(path2);
     return iss;
   });
 }
@@ -1808,7 +1808,7 @@ function formatError(error48, mapper = (issue2) => issue2.message) {
 }
 function treeifyError(error48, mapper = (issue2) => issue2.message) {
   const result = { errors: [] };
-  const processError = (error49, path = []) => {
+  const processError = (error49, path2 = []) => {
     var _a2, _b;
     for (const issue2 of error49.issues) {
       if (issue2.code === "invalid_union" && issue2.errors.length) {
@@ -1818,7 +1818,7 @@ function treeifyError(error48, mapper = (issue2) => issue2.message) {
       } else if (issue2.code === "invalid_element") {
         processError({ issues: issue2.issues }, issue2.path);
       } else {
-        const fullpath = [...path, ...issue2.path];
+        const fullpath = [...path2, ...issue2.path];
         if (fullpath.length === 0) {
           result.errors.push(mapper(issue2));
           continue;
@@ -1850,8 +1850,8 @@ function treeifyError(error48, mapper = (issue2) => issue2.message) {
 }
 function toDotPath(_path) {
   const segs = [];
-  const path = _path.map((seg) => typeof seg === "object" ? seg.key : seg);
-  for (const seg of path) {
+  const path2 = _path.map((seg) => typeof seg === "object" ? seg.key : seg);
+  for (const seg of path2) {
     if (typeof seg === "number")
       segs.push(`[${seg}]`);
     else if (typeof seg === "symbol")
@@ -13828,13 +13828,13 @@ function resolveRef(ref, ctx) {
   if (!ref.startsWith("#")) {
     throw new Error("External $ref is not supported, only local refs (#/...) are allowed");
   }
-  const path = ref.slice(1).split("/").filter(Boolean);
-  if (path.length === 0) {
+  const path2 = ref.slice(1).split("/").filter(Boolean);
+  if (path2.length === 0) {
     return ctx.rootSchema;
   }
   const defsKey = ctx.version === "draft-2020-12" ? "$defs" : "definitions";
-  if (path[0] === defsKey) {
-    const key = path[1];
+  if (path2[0] === defsKey) {
+    const key = path2[1];
     if (!key || !ctx.defs[key]) {
       throw new Error(`Reference not found: ${ref}`);
     }
@@ -14773,13 +14773,13 @@ function loadNodeSqlite() {
   }
 }
 var DatabaseSync = loadNodeSqlite().DatabaseSync;
-function openDatabase(path, opts) {
-  const db = opts ? new DatabaseSync(path, opts) : new DatabaseSync(path);
+function openDatabase(path2, opts) {
+  const db = opts ? new DatabaseSync(path2, opts) : new DatabaseSync(path2);
   try {
     db.exec("PRAGMA foreign_keys = ON");
   } catch {
   }
-  if (path !== ":memory:" && !opts?.readOnly) {
+  if (path2 !== ":memory:" && !opts?.readOnly) {
     try {
       db.exec("PRAGMA journal_mode = WAL");
       db.exec("PRAGMA synchronous = NORMAL");
@@ -15093,6 +15093,38 @@ function installHistoryAttrSlim(db) {
   db.prepare("INSERT INTO store_meta(key,value) VALUES (?,?)").run(HISTORY_ATTR_SLIM_KEY, "done");
 }
 
+// ../../packages/local-graph/src/text-index.ts
+var TEXT_INDEX_KEY = "text-index-trigram-1";
+var TEXT_INDEX_MIN_TERM = 3;
+function installTextIndex(db) {
+  if (db.prepare("SELECT value FROM store_meta WHERE key = ?").get(TEXT_INDEX_KEY)) return;
+  try {
+    db.exec(`CREATE VIRTUAL TABLE temp.fts5_probe USING fts5(x, tokenize='trigram'); DROP TABLE temp.fts5_probe`);
+  } catch {
+    db.prepare("INSERT INTO store_meta(key,value) VALUES (?,?)").run(TEXT_INDEX_KEY, "unavailable");
+    return;
+  }
+  db.transaction(() => {
+    db.exec(`CREATE VIRTUAL TABLE IF NOT EXISTS nodes_text
+      USING fts5(description, content='nodes', content_rowid='rowid', tokenize='trigram')`);
+    db.exec(`CREATE TRIGGER IF NOT EXISTS nodes_text_insert AFTER INSERT ON nodes WHEN NEW.valid_to IS NULL
+      BEGIN INSERT INTO nodes_text(rowid, description) VALUES (NEW.rowid, NEW.description); END;`);
+    db.exec(`CREATE TRIGGER IF NOT EXISTS nodes_text_update_out AFTER UPDATE OF description, valid_to ON nodes
+      WHEN OLD.valid_to IS NULL
+      BEGIN INSERT INTO nodes_text(nodes_text, rowid, description) VALUES ('delete', OLD.rowid, OLD.description); END;`);
+    db.exec(`CREATE TRIGGER IF NOT EXISTS nodes_text_update_in AFTER UPDATE OF description, valid_to ON nodes
+      WHEN NEW.valid_to IS NULL
+      BEGIN INSERT INTO nodes_text(rowid, description) VALUES (NEW.rowid, NEW.description); END;`);
+    db.exec(`CREATE TRIGGER IF NOT EXISTS nodes_text_delete AFTER DELETE ON nodes WHEN OLD.valid_to IS NULL
+      BEGIN INSERT INTO nodes_text(nodes_text, rowid, description) VALUES ('delete', OLD.rowid, OLD.description); END;`);
+    db.exec(`INSERT INTO nodes_text(rowid, description) SELECT rowid, description FROM nodes WHERE valid_to IS NULL`);
+    db.prepare("INSERT INTO store_meta(key,value) VALUES (?,?)").run(TEXT_INDEX_KEY, "installed");
+  });
+}
+function textIndexReady(db) {
+  return db.prepare("SELECT value FROM store_meta WHERE key = ?").get(TEXT_INDEX_KEY)?.["value"] === "installed";
+}
+
 // ../../packages/math/src/relevance.ts
 function pageRankScale(ranks) {
   const positive = [...ranks].filter((r) => Number.isFinite(r) && r > 0).sort((a, b) => a - b);
@@ -15103,6 +15135,140 @@ function pageRankScale(ranks) {
 var DEMAND_HOT = Math.log1p(10);
 var DEMAND_WARM = Math.log1p(5);
 
+// ../../packages/local-graph/src/node-select.ts
+var EMBEDDING_DIM_SQL = "(CASE WHEN length(embedding_blob) >= 4 THEN length(embedding_blob) / 4 ELSE COALESCE(json_array_length(embedding_json), 0) END)";
+var KEY = /^[A-Za-z_][A-Za-z0-9_]*(\.[A-Za-z_][A-Za-z0-9_]*)*$/;
+var COLUMNS = /* @__PURE__ */ new Set([
+  "created_at",
+  "last_updated_at",
+  "page_rank",
+  "created_at_seq",
+  "last_reinforced_at_seq",
+  "cumulative_hits",
+  "extraction_confidence"
+]);
+function path(key) {
+  if (!KEY.test(key)) throw new Error(`selectNodes: bad attr key ${key}`);
+  return `'$.${key}'`;
+}
+function column(c) {
+  if (!COLUMNS.has(c)) throw new Error(`selectNodes: bad column ${c}`);
+  return c;
+}
+function compileNodeWhere(q) {
+  if (q.labels.length === 0) throw new Error("selectNodes: no labels");
+  const params = [...q.labels];
+  const conds = [`label IN (${q.labels.map(() => "?").join(", ")})`];
+  if (!q.includeClosed) conds.push("valid_to IS NULL");
+  for (const w of q.where ?? []) {
+    if ("embedding" in w) {
+      conds.push(`${EMBEDDING_DIM_SQL} ${w.embedding === "present" ? ">" : "="} 0`);
+    } else if ("idPrefix" in w) {
+      conds.push("substr(id, 1, ?) = ?");
+      params.push(w.idPrefix.length, w.idPrefix);
+    } else if ("descriptionIn" in w) {
+      if (w.descriptionIn.length === 0) conds.push("0");
+      else {
+        conds.push(`description IN (${w.descriptionIn.map(() => "?").join(", ")})`);
+        params.push(...w.descriptionIn);
+      }
+    } else if ("column" in w) {
+      const c = column(w.column);
+      for (const [op, sym] of [["gte", ">="], ["gt", ">"], ["lte", "<="], ["lt", "<"]]) {
+        const v = w[op];
+        if (v !== void 0) {
+          conds.push(`${c} ${sym} ?`);
+          params.push(v);
+        }
+      }
+    } else {
+      const p = path(w.attr);
+      if ("absent" in w) conds.push(`json_type(attrs_json, ${p}) IS NULL`);
+      else if ("present" in w) conds.push(`json_type(attrs_json, ${p}) IS NOT NULL`);
+      else if ("nullish" in w) conds.push(`json_extract(attrs_json, ${p}) IS NULL`);
+      else if ("notNullish" in w) conds.push(`json_extract(attrs_json, ${p}) IS NOT NULL`);
+      else if ("eq" in w || "ne" in w) {
+        const v = "eq" in w ? w.eq : w.ne;
+        const neg = "ne" in w;
+        if (typeof v === "boolean") {
+          conds.push(`json_type(attrs_json, ${p}) ${neg ? "IS NOT" : "="} ?`);
+          params.push(v ? "true" : "false");
+        } else {
+          conds.push(neg ? `json_extract(attrs_json, ${p}) IS NOT ?` : `json_extract(attrs_json, ${p}) = ?`);
+          params.push(v);
+        }
+      }
+    }
+  }
+  return { where: conds.join(" AND "), params };
+}
+function attrValueSql(p) {
+  return `CASE json_type(attrs_json, ${p}) WHEN 'integer' THEN CAST(json_extract(attrs_json, ${p}) AS REAL) ELSE json_extract(attrs_json, ${p}) END`;
+}
+function compileNodeSelect(q) {
+  const { where, params } = compileNodeWhere(q);
+  const attrs = q.attrs ?? [];
+  const attrCols = attrs.map((k, i) => `, ${attrValueSql(path(k))} AS a${i}, json_type(attrs_json, ${path(k)}) AS t${i}`).join("");
+  let order;
+  if (q.orderBy && "column" in q.orderBy) {
+    order = `${column(q.orderBy.column)} ${q.orderBy.desc ? "DESC" : "ASC"}, rowid`;
+  } else if (q.orderBy) {
+    const p = path(q.orderBy.attr);
+    order = `json_extract(attrs_json, ${p}) IS NULL, json_extract(attrs_json, ${p}) ${q.orderBy.desc ? "DESC" : "ASC"}, rowid`;
+  } else if (q.labels.length > 1) {
+    order = `CASE label ${q.labels.map((_, i) => `WHEN ? THEN ${i}`).join(" ")} END, rowid`;
+  } else {
+    order = "rowid";
+  }
+  const orderParams = !q.orderBy && q.labels.length > 1 ? [...q.labels] : [];
+  const limit = q.limit !== void 0 ? ` LIMIT ${Math.max(0, Math.floor(q.limit))}` : "";
+  const sql = `SELECT id, label, description, created_at, last_updated_at, page_rank, cumulative_hits, extraction_confidence, extraction_source, community, created_at_seq, last_reinforced_at_seq, valid_to, ${EMBEDDING_DIM_SQL} AS embedding_dim${q.withEmbedding ? ", embedding_blob, embedding_json" : ""}${attrCols} FROM nodes WHERE ${where} ORDER BY ${order}${limit}`;
+  return { sql, params: [...params, ...orderParams] };
+}
+function decodeAttr(value, type) {
+  switch (type) {
+    case null:
+    case void 0:
+      return void 0;
+    case "true":
+      return true;
+    case "false":
+      return false;
+    case "null":
+      return null;
+    case "object":
+    case "array":
+      return typeof value === "string" ? JSON.parse(value) : value;
+    default:
+      return value;
+  }
+}
+function decodeSelectedRow(r, attrs, decodeVector) {
+  const out = {};
+  attrs.forEach((k, i) => {
+    const v = decodeAttr(r[`a${i}`], r[`t${i}`]);
+    if (v !== void 0) out[k] = v;
+  });
+  return {
+    id: r["id"],
+    label: r["label"],
+    description: r["description"] ?? "",
+    createdAt: r["created_at"],
+    lastUpdatedAt: r["last_updated_at"],
+    pageRank: r["page_rank"],
+    cumulativeHits: r["cumulative_hits"],
+    extractionConfidence: r["extraction_confidence"],
+    extractionSource: r["extraction_source"],
+    community: r["community"] ?? null,
+    ...r["created_at_seq"] != null ? { createdAtSeq: r["created_at_seq"] } : {},
+    ...r["last_reinforced_at_seq"] != null ? { lastReinforcedAtSeq: r["last_reinforced_at_seq"] } : {},
+    validTo: r["valid_to"] ?? null,
+    embeddingDim: Number(r["embedding_dim"] ?? 0),
+    ...decodeVector ? { embedding: decodeVector(r["embedding_blob"], r["embedding_json"] ?? "[]") } : {},
+    attrs: out
+  };
+}
+
 // ../../packages/local-graph/src/store.ts
 var PAGE_RANK_SCALE_KEY = "pageRankScale.p95.v1";
 var TEXT_SEARCH_COLUMNS = `id, label, description, page_rank,
@@ -15291,6 +15457,11 @@ CREATE INDEX IF NOT EXISTS nodes_cloud_node_id ON nodes(json_extract(attrs_json,
 -- Solution's and Problem's attrs per call. Not partial: the Problem pull reads
 -- closed versions too.
 CREATE INDEX IF NOT EXISTS nodes_enrichment_pending ON nodes(json_extract(attrs_json, '$.enrichmentPending'));
+-- Revisit flags, any label, any version: listNeedsRevisit's as-of read runs on
+-- every context render and otherwise json-checks every row (10.3 s on a 324k-row
+-- store, to return zero flags). The predicate must stay textually identical to
+-- nodesAsOfWhereTrue's or the planner will not use it.
+CREATE INDEX IF NOT EXISTS nodes_revisit_true ON nodes(valid_from) WHERE json_type(attrs_json, '$.revisit') = 'true';
 `;
 function encodeEmbedding(emb) {
   if (!emb || emb.length === 0) return null;
@@ -15391,6 +15562,7 @@ var SqliteGraphStore = class {
     installEditPlaceholderRetirement(this.db);
     installCloudAttrCompaction(this.db);
     installHistoryAttrSlim(this.db);
+    installTextIndex(this.db);
     if (!this.db.prepare("SELECT value FROM store_meta WHERE key = ?").get(SESSION_REPAIR_KEY)) {
       this.db.transaction(() => {
         this.repairSessionAliases();
@@ -15768,8 +15940,8 @@ var SqliteGraphStore = class {
   setLandmark(id, isLandmark) {
     this.stmts.setLandmark.run({ id, v: isLandmark ? 1 : 0 });
   }
-  setLocalCommunity(id, community, column) {
-    this.stmts.setLocalCommunity.run({ id, v: community, col: column ? 1 : 0 });
+  setLocalCommunity(id, community, column2) {
+    this.stmts.setLocalCommunity.run({ id, v: community, col: column2 ? 1 : 0 });
   }
   fieldStmts = /* @__PURE__ */ new Map();
   findNodeFieldsByLabel(label, attrKeys) {
@@ -15777,7 +15949,7 @@ var SqliteGraphStore = class {
     const cacheKey = attrKeys.join(",");
     let stmt = this.fieldStmts.get(cacheKey);
     if (!stmt) {
-      const cols = attrKeys.map((k, i) => `, json_extract(attrs_json, '$.${k}') AS a${i}`).join("");
+      const cols = attrKeys.map((k, i) => `, ${attrValueSql(`'$.${k}'`)} AS a${i}, json_type(attrs_json, '$.${k}') AS t${i}`).join("");
       stmt = this.db.prepare(
         `SELECT id, description, community, extraction_confidence${cols} FROM nodes WHERE label = ? AND valid_to IS NULL`
       );
@@ -15786,8 +15958,8 @@ var SqliteGraphStore = class {
     return stmt.all(label).map((r) => {
       const attrs = {};
       attrKeys.forEach((k, i) => {
-        const v = r[`a${i}`];
-        if (v !== null && v !== void 0) attrs[k] = v;
+        const v = decodeAttr(r[`a${i}`], r[`t${i}`]);
+        if (v !== void 0) attrs[k] = v;
       });
       return {
         id: r["id"],
@@ -15798,6 +15970,15 @@ var SqliteGraphStore = class {
       };
     });
   }
+  selectNodes(q) {
+    const { sql, params } = compileNodeSelect(q);
+    const rows = this.db.prepare(sql).all(...params);
+    return rows.map((r) => decodeSelectedRow(r, q.attrs ?? [], q.withEmbedding ? decodeEmbedding : void 0));
+  }
+  countNodes(q) {
+    const { where, params } = compileNodeWhere(q);
+    return Number(this.db.prepare(`SELECT COUNT(*) AS n FROM nodes WHERE ${where}`).get(...params).n);
+  }
   setCommunity(id, community) {
     this.stmts.setCommunity.run({ id, v: community });
   }
@@ -16045,6 +16226,20 @@ var SqliteGraphStore = class {
   }
   searchTextCandidates(terms) {
     if (terms.length === 0) return [];
+    const indexed = terms.filter((t) => t.length >= TEXT_INDEX_MIN_TERM);
+    if (indexed.length > 0 && textIndexReady(this.db)) {
+      const short = terms.filter((t) => t.length < TEXT_INDEX_MIN_TERM).map((t) => t.toLowerCase());
+      const rows2 = this.db.prepare(`WITH hits AS MATERIALIZED (
+        SELECT t.rowid AS rid, COUNT(*) AS matched FROM json_each(:terms) q, nodes_text t
+        WHERE t.description LIKE '%' || q.value || '%' GROUP BY t.rowid
+      ) SELECT ${TEXT_SEARCH_COLUMNS}, hits.matched FROM hits JOIN nodes ON nodes.rowid = hits.rid
+        WHERE nodes.valid_to IS NULL ORDER BY page_rank DESC, last_updated_at DESC`).all({ terms: JSON.stringify(indexed) });
+      return rows2.map((row) => {
+        const text = row.description.toLowerCase();
+        const extra = short.filter((t) => text.includes(t)).length;
+        return { node: textSearchNode(row), matched: row.matched + extra };
+      });
+    }
     const rows = this.db.prepare(`WITH matches AS MATERIALIZED (
       SELECT n.id, COUNT(*) AS matched FROM nodes n INDEXED BY nodes_live_description, json_each(:terms) t
       WHERE n.valid_to IS NULL AND n.description LIKE '%' || t.value || '%'
@@ -16053,6 +16248,18 @@ var SqliteGraphStore = class {
       JOIN matches USING (id) ORDER BY page_rank DESC, last_updated_at DESC`).all({ terms: JSON.stringify(terms) });
     return rows.map((row) => ({ node: textSearchNode(row), matched: row.matched }));
   }
+  commentAnchors(commentIds) {
+    const out = /* @__PURE__ */ new Map();
+    if (commentIds.length === 0) return out;
+    const rows = this.db.prepare(`WITH a AS MATERIALIZED (
+      SELECT q.value AS cid, COALESCE(
+        (SELECT to_id FROM edges WHERE from_id = q.value AND type IN ('EXPLAINS', 'ANNOTATES') AND valid_to IS NULL LIMIT 1),
+        (SELECT from_id FROM edges WHERE to_id = q.value AND type = 'DOCUMENTED_BY' AND valid_to IS NULL LIMIT 1)
+      ) AS aid FROM json_each(:ids) q
+    ) SELECT a.cid AS cid, ${TEXT_SEARCH_COLUMNS} FROM a JOIN nodes ON nodes.id = a.aid AND nodes.valid_to IS NULL`).all({ ids: JSON.stringify(commentIds) });
+    for (const row of rows) out.set(row.cid, textSearchNode(row));
+    return out;
+  }
   getTextSearchNode(id) {
     const row = this.db.prepare(`SELECT ${TEXT_SEARCH_COLUMNS} FROM nodes WHERE id = ? AND valid_to IS NULL`).get(id);
     return row ? textSearchNode(row) : null;
@@ -16448,7 +16655,8 @@ var SqliteGraphStore = class {
     if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(attr)) throw new Error(`nodesAsOfWhereTrue: bad attr ${JSON.stringify(attr)}`);
     const rows = this.db.prepare(
       `SELECT * FROM nodes WHERE valid_from <= :t AND (valid_to IS NULL OR valid_to > :t)
-           AND json_type(attrs_json, '$.${attr}') = 'true'`
+           AND json_type(attrs_json, '$.${attr}') = 'true'
+         ORDER BY rowid`
     ).all({ t });
     return rows.map(rowToNode);
   }
@@ -17409,11 +17617,11 @@ function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
 
 // src/generalize-worker.ts
 var stores = /* @__PURE__ */ new Map();
-function storeAt(path) {
-  let s = stores.get(path);
+function storeAt(path2) {
+  let s = stores.get(path2);
   if (!s) {
-    s = openGraphStore({ path });
-    stores.set(path, s);
+    s = openGraphStore({ path: path2 });
+    stores.set(path2, s);
   }
   return s;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.22-dev.4020",
+  "version": "2.0.22-dev.4054",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {