token-goat 2.9.18 → 2.9.20

package/dist/{token-goat-chunk-W3E65QOW.mjs → token-goat-chunk-ERGHDBC6.mjs}

@@ -4,7 +4,7 @@ import {
   pathEqClause,
   projectScopeClause,
   querySymbols
-} from "./token-goat-chunk-I67CPTDD.mjs";
+} from "./token-goat-chunk-K3ZNBN4U.mjs";
 import {
   TREE_SITTER_LANGUAGES,
   _detectOpenQuote,
@@ -21,11 +21,11 @@ import {
   stripJsComments,
   stripStringLiterals,
   stripXmlComments
-} from "./token-goat-chunk-L7VCX33I.mjs";
+} from "./token-goat-chunk-T63VHOL3.mjs";
 import {
   fingerprintContent,
   loadConfig
-} from "./token-goat-chunk-CA6V36KW.mjs";
+} from "./token-goat-chunk-74ADIZZX.mjs";
 import {
   registerReset
 } from "./token-goat-chunk-EEIDFMEM.mjs";
@@ -39,11 +39,11 @@ import {
   foldPath,
   globalDbPath,
   isCaseInsensitiveFs
-} from "./token-goat-chunk-QIDLJGXZ.mjs";
+} from "./token-goat-chunk-6LJNM4ZZ.mjs";
 import {
   displaySafeText,
   normalizePath
-} from "./token-goat-chunk-ERTXEKB6.mjs";
+} from "./token-goat-chunk-NMTKNYGF.mjs";
 import {
   DocumentRefusedError,
   MAX_DOCUMENT_WORK_MILLIS
@@ -74,11 +74,11 @@ async function extractEmbeddableDocumentText(filePath) {
       return text;
     }
     case ".docx":
-      return await (await import("./token-goat-chunk-HIZXQVNU.mjs")).docxText(filePath);
+      return await (await import("./token-goat-chunk-VYXGHGUY.mjs")).docxText(filePath);
     case ".pptx":
-      return await (await import("./token-goat-chunk-52IMXJ2L.mjs")).pptxAllSlidesText(filePath, true);
+      return await (await import("./token-goat-chunk-ASTBLE2R.mjs")).pptxAllSlidesText(filePath, true);
     case ".xlsx":
-      return await (await import("./token-goat-chunk-OWSJJY5R.mjs")).allSheetsHeadText(filePath, XLSX_SHEET_ROW_CAP);
+      return await (await import("./token-goat-chunk-DW4VKU7D.mjs")).allSheetsHeadText(filePath, XLSX_SHEET_ROW_CAP);
     default:
       return null;
   }
@@ -104,20 +104,20 @@ var UnsupportedTokenizerError = class extends Error {
 function fail(message) {
   throw new UnsupportedTokenizerError(message);
 }
-function asRecord(value, path5) {
-  if (typeof value !== "object" || value === null || Array.isArray(value)) fail(`${path5} is not an object`);
+function asRecord(value, path7) {
+  if (typeof value !== "object" || value === null || Array.isArray(value)) fail(`${path7} is not an object`);
   return value;
 }
-function asString(value, path5) {
-  if (typeof value !== "string") fail(`${path5} is not a string`);
+function asString(value, path7) {
+  if (typeof value !== "string") fail(`${path7} is not a string`);
   return value;
 }
-function asArray(value, path5) {
-  if (!Array.isArray(value)) fail(`${path5} is not an array`);
+function asArray(value, path7) {
+  if (!Array.isArray(value)) fail(`${path7} is not an array`);
   return value;
 }
-function expectValue(actual, wanted, path5) {
-  if (actual !== wanted) fail(`${path5} is ${JSON.stringify(actual)}, expected ${JSON.stringify(wanted)}`);
+function expectValue(actual, wanted, path7) {
+  if (actual !== wanted) fail(`${path7} is ${JSON.stringify(actual)}, expected ${JSON.stringify(wanted)}`);
 }
 function isChinese(cp) {
   return cp >= 19968 && cp <= 40959 || cp >= 13312 && cp <= 19903 || cp >= 131072 && cp <= 173791 || cp >= 173824 && cp <= 177983 || cp >= 177984 && cp <= 178207 || cp >= 178208 && cp <= 183983 || cp >= 63744 && cp <= 64255 || cp >= 194560 && cp <= 195103;
@@ -309,7 +309,7 @@ var BertWordPiece = class _BertWordPiece {
 var _require = createRequire(import.meta.url);
 var DEFAULT_MODEL = "Xenova/bge-small-en-v1.5";
 var DEFAULT_DIM = 384;
-var DEFAULT_EMBED_THREADS = 2;
+var DEFAULT_EMBED_THREADS = 4;
 var PINNED_MODEL_REVISION = "ea104dacec62c0de699686887e3f920caeb4f3e3";
 var MODEL_FILES = [
   {
@@ -697,7 +697,25 @@ init_define_import_meta_env();
 
 // src/embed_fingerprint.ts
 init_define_import_meta_env();
-var EMBED_FINGERPRINT = "b7b2ff71de288d13";
+var EMBED_FINGERPRINT = "702c126bffd8439e";
+var EMBED_KIND_FINGERPRINTS = /* @__PURE__ */ new Map([
+  ["docx", "a39e82e795d46ed1"],
+  ["markdown", "c65dd62705244608"],
+  ["pdf", "9a0ea76d66dfd358"],
+  ["pptx", "b60b30cb38a801e2"],
+  ["xlsx", "e5b358c460b6507f"]
+]);
+var PRE_KIND_EMBED_FINGERPRINT = "b7b2ff71de288d13";
+var SPLIT_EMBED_FINGERPRINT = "2310db32d9bc8a2b";
+
+// src/embed_stamp.ts
+init_define_import_meta_env();
+import * as path3 from "node:path";
+function embedKindForPath(filePath) {
+  if (isEmbeddableDocument(filePath)) return path3.extname(filePath).toLowerCase().slice(1);
+  if (detectLanguage(filePath) === "markdown") return "markdown";
+  return null;
+}
 
 // src/embeddings.ts
 var QUERY_INSTRUCTION_PREFIX = "Represent this sentence for searching relevant passages: ";
@@ -1345,7 +1363,42 @@ function resetAllEmbeddings(db, keepVectors = false) {
 }
 function embeddingProvenance(modelName = DEFAULT_MODEL) {
   const revision = modelName === DEFAULT_MODEL ? PINNED_MODEL_REVISION.slice(0, 12) : "unpinned";
-  return `${modelName}@${revision}/${backendId()}/embed-${EMBED_FINGERPRINT}`;
+  const kinds = [...EMBED_KIND_FINGERPRINTS].map(([kind, digest]) => `+${kind}-${digest}`).join("");
+  return `${modelName}@${revision}/${backendId()}/embed-${EMBED_FINGERPRINT}${kinds}`;
+}
+function chunkStampsOf(provenance) {
+  const at = provenance.lastIndexOf("/embed-");
+  const kinds = /* @__PURE__ */ new Map();
+  if (at === -1) return { globalDigest: "", kinds };
+  const [globalDigest = "", ...kindParts] = provenance.slice(at + "/embed-".length).split("+");
+  for (const part of kindParts) {
+    const dash = part.lastIndexOf("-");
+    if (dash !== -1) kinds.set(part.slice(0, dash), part.slice(dash + 1));
+  }
+  return { globalDigest, kinds };
+}
+function resetStaleChunking(db, stored, current) {
+  const before = chunkStampsOf(stored);
+  const now = chunkStampsOf(current);
+  if (before.kinds.size === 0 && before.globalDigest === PRE_KIND_EMBED_FINGERPRINT && now.globalDigest === SPLIT_EMBED_FINGERPRINT) return 0;
+  if (before.globalDigest !== now.globalDigest) return resetAllEmbeddings(db, true);
+  const changed = /* @__PURE__ */ new Set();
+  for (const kind of /* @__PURE__ */ new Set([...before.kinds.keys(), ...now.kinds.keys()])) {
+    if (before.kinds.get(kind) !== now.kinds.get(kind)) changed.add(kind);
+  }
+  return changed.size === 0 ? 0 : resetEmbeddingsForKinds(db, changed);
+}
+function resetEmbeddingsForKinds(db, kinds) {
+  const paths = db.prepare("SELECT DISTINCT file_path FROM chunks").pluck().all().filter((p) => {
+    const kind = embedKindForPath(p);
+    return kind !== null && kinds.has(kind);
+  });
+  const clearEmbedSha = db.prepare(`UPDATE files SET embed_sha = NULL WHERE ${pathEqClause("path")}`);
+  const tx = db.transaction(() => {
+    for (const p of paths) clearEmbedSha.run(foldPath(p));
+  });
+  tx.immediate();
+  return paths.length;
 }
 function backendId() {
   return `onnxruntime-node@${majorMinor(runtimeVersion())}`;
@@ -1366,7 +1419,7 @@ function ensureEmbeddingProvenance(db, modelName = DEFAULT_MODEL) {
   const stored = db.prepare("SELECT provenance FROM embedding_provenance WHERE id = 1").pluck().get();
   if (stored === current) return;
   const keepVectors = stored !== void 0 && vectorSpaceOf(stored) === vectorSpaceOf(current);
-  const cleared = resetAllEmbeddings(db, keepVectors);
+  const cleared = stored !== void 0 && keepVectors ? resetStaleChunking(db, stored, current) : resetAllEmbeddings(db, false);
   db.prepare(
     "INSERT INTO embedding_provenance (id, provenance) VALUES (1, ?) ON CONFLICT(id) DO UPDATE SET provenance = excluded.provenance"
   ).run(current);
@@ -1411,15 +1464,154 @@ function _pathPriorityPenalty(filePath) {
   return 0;
 }
 
+// src/parser_stamp.ts
+init_define_import_meta_env();
+
 // src/parser_fingerprint.ts
 init_define_import_meta_env();
-var PARSER_FINGERPRINT = "0d80572d3abdfb19";
+var PARSER_FINGERPRINT = "486e443a693e6cdc";
+var LANGUAGE_PARSER_FINGERPRINTS = /* @__PURE__ */ new Map([
+  ["abap", "3477b349e26ff486"],
+  ["abl", "e79dac17ec1be23e"],
+  ["apache", "d6aa0ead3f7d8936"],
+  ["apex", "140d8f5601f95126"],
+  ["asm", "3e9aa7923f56286d"],
+  ["bash", "5906a50cadf9dc24"],
+  ["batch", "01e9923196d3d66b"],
+  ["caddy", "6aa41b340b3acb36"],
+  ["clojure", "bc2547b735ff453b"],
+  ["cmake", "c30e973a5b0b9628"],
+  ["common_lisp", "74c3d80a715b7507"],
+  ["csharp", "f5f7307d4575fc32"],
+  ["dart", "6657a5eeb20b7b2c"],
+  ["elixir", "88c5ce992a32ad21"],
+  ["emacs_lisp", "b3d8e903be32f3bd"],
+  ["erlang", "2bc6ed62806ffc74"],
+  ["fortran", "56658a0f6d3d9a27"],
+  ["fsharp", "df4c61b50c49aabe"],
+  ["graphql", "f4a0c9155c9649ee"],
+  ["groovy", "b3facf1b9460df44"],
+  ["haskell", "75f0794252047696"],
+  ["jcl", "06199551a218e69d"],
+  ["kotlin", "a99a84e9ae3a1362"],
+  ["liquid", "387d4f7973ef74dd"],
+  ["lua", "3a36b15e74558aa0"],
+  ["makefile", "dd470fdea2050de2"],
+  ["matlab", "65bccffd6a80ddd0"],
+  ["nginx", "0c24489a62c74032"],
+  ["nix", "7a261153f741c56b"],
+  ["objc", "e85cfa09f698e42a"],
+  ["ocaml", "643e216c28a48324"],
+  ["pascal", "4e4ae4aa127e7c4e"],
+  ["perl", "12dff36e9dab0700"],
+  ["php", "6b56ace5008da841"],
+  ["pli", "46626c331cc8c5ed"],
+  ["powershell", "e460d7974d112d63"],
+  ["proto", "a721e303a321bc4c"],
+  ["r", "7d52c82fe21bc414"],
+  ["racket", "d039be035fd8dcb1"],
+  ["rpg", "003b932e718fd3ec"],
+  ["salesforce_metadata", "0d2d649f0825c2b4"],
+  ["sas", "5a5518274e5bf88d"],
+  ["scala", "580d772168d7ec04"],
+  ["scheme", "9008af85ddce7593"],
+  ["solidity", "13e09d7acfc45176"],
+  ["sql", "e7a658d83f5ed705"],
+  ["swift", "bf3fbf0617f7dafd"],
+  ["terraform", "bb7ba075971caf67"],
+  ["thrift", "d39f3c5a82bb5b0b"],
+  ["vb", "4ddc4f794b061bfe"],
+  ["vhdl", "6d9e0f47264942fb"],
+  ["zig", "94d64036a1ad42b0"]
+]);
+
+// src/parser_stamp.ts
+function parserFingerprintForLanguage(language) {
+  return LANGUAGE_PARSER_FINGERPRINTS.get(language) ?? PARSER_FINGERPRINT;
+}
 
 // src/parser.ts
 init_define_import_meta_env();
 import * as fs2 from "node:fs";
 import { createRequire as createRequire2 } from "node:module";
+import * as path6 from "node:path";
+
+// src/asset_extensions.ts
+init_define_import_meta_env();
 import * as path4 from "node:path";
+var IMAGE_EXTENSIONS = /* @__PURE__ */ new Set([
+  ".png",
+  ".jpg",
+  ".jpeg",
+  ".gif",
+  ".webp",
+  ".bmp",
+  ".tif",
+  ".tiff",
+  ".avif",
+  ".heic",
+  ".heif"
+]);
+function isImagePath(p) {
+  return IMAGE_EXTENSIONS.has(path4.extname(p).toLowerCase());
+}
+var NON_TEXT_ASSET_EXTENSIONS = /* @__PURE__ */ new Set([
+  ...IMAGE_EXTENSIONS,
+  ".ico",
+  ".icns",
+  ".psd",
+  ".woff",
+  ".woff2",
+  ".ttf",
+  ".otf",
+  ".eot",
+  ".zip",
+  ".gz",
+  ".tgz",
+  ".bz2",
+  ".xz",
+  ".7z",
+  ".rar",
+  ".jar",
+  ".mp3",
+  ".wav",
+  ".flac",
+  ".ogg",
+  ".m4a",
+  ".aac",
+  ".mp4",
+  ".m4v",
+  ".mov",
+  ".avi",
+  ".mkv",
+  ".webm",
+  ".wasm",
+  ".exe",
+  ".dll",
+  ".so",
+  ".dylib",
+  ".node",
+  ".class",
+  ".pyc",
+  ".pdb",
+  ".bin",
+  ".dat",
+  ".db",
+  ".sqlite",
+  ".sqlite3"
+]);
+function isNonTextAsset(p) {
+  return NON_TEXT_ASSET_EXTENSIONS.has(path4.extname(p).toLowerCase());
+}
+var NON_TEXT_ASSET_SET_ID = (() => {
+  const joined = [...NON_TEXT_ASSET_EXTENSIONS].sort().join(",");
+  let h = 2166136261;
+  for (let i = 0; i < joined.length; i++) {
+    h ^= joined.charCodeAt(i);
+    h = Math.imul(h, 16777619);
+  }
+  return (h >>> 0).toString(16).padStart(8, "0");
+})();
 
 // src/embedding_boundaries.ts
 init_define_import_meta_env();
@@ -1586,14 +1778,14 @@ function buildEmbeddingBoundaries(filePath, content, dbPath) {
 
 // src/languages/salesforce_frontend.ts
 init_define_import_meta_env();
-import * as path3 from "node:path";
+import * as path5 from "node:path";
 function lines(content) {
   return content.split("\n");
 }
 function bundleName(filePath) {
   const normalized = filePath.replaceAll("\\", "/");
-  const parent = path3.posix.basename(path3.posix.dirname(normalized));
-  const base = path3.posix.basename(normalized).replace(/\.[^.]+$/, "");
+  const parent = path5.posix.basename(path5.posix.dirname(normalized));
+  const base = path5.posix.basename(normalized).replace(/\.[^.]+$/, "");
   return parent === "lwc" || parent === "aura" ? base : parent;
 }
 function lwcTagAlias(name) {
@@ -1697,7 +1889,7 @@ var MARKUP_KIND = {
 };
 function markupArtifactName(filePath, extension) {
   if (extension === ".page" || extension === ".component" || extension === ".email") {
-    return path3.posix.basename(filePath.replaceAll("\\", "/")).replace(new RegExp(`${extension.replace(".", "\\.")}$`, "i"), "");
+    return path5.posix.basename(filePath.replaceAll("\\", "/")).replace(new RegExp(`${extension.replace(".", "\\.")}$`, "i"), "");
   }
   return bundleName(filePath);
 }
@@ -1717,7 +1909,7 @@ function attributeRefs(refs, content, filePath, attribute, split = false) {
 }
 function extractSalesforceMarkup(content, filePath) {
   const normalized = filePath.replaceAll("\\", "/");
-  const extension = path3.posix.extname(normalized).toLowerCase();
+  const extension = path5.posix.extname(normalized).toLowerCase();
   const kind = MARKUP_KIND[extension] ?? "salesforce_markup";
   const symbols = [
     symbol(filePath, markupArtifactName(normalized, extension), kind, 1, countContentLines(content))
@@ -1933,6 +2125,41 @@ var TSJS_FN_SCOPE_TYPES = /* @__PURE__ */ new Set([
   "generator_function",
   "generator_function_declaration"
 ]);
+function sameNode(a, b) {
+  return a !== null && b !== null && a.type === b.type && a.startPosition.row === b.startPosition.row && a.startPosition.column === b.startPosition.column && a.endPosition.row === b.endPosition.row && a.endPosition.column === b.endPosition.column;
+}
+function isIifeCallee(node) {
+  let cur = node;
+  let parent = cur.parent;
+  while (parent !== null && parent.type === "parenthesized_expression") {
+    cur = parent;
+    parent = cur.parent;
+  }
+  if (parent === null || parent.type !== "call_expression" || !sameNode(parent.childForFieldName("function"), cur)) return false;
+  let stmt = parent.parent;
+  while (stmt !== null && stmt.type === "parenthesized_expression") stmt = stmt.parent;
+  if (stmt !== null && stmt.type === "unary_expression" && ["!", "void", "+"].includes(stmt.childForFieldName("operator")?.text ?? "")) {
+    stmt = stmt.parent;
+    while (stmt !== null && stmt.type === "parenthesized_expression") stmt = stmt.parent;
+  }
+  return stmt !== null && stmt.type === "expression_statement";
+}
+function isModuleWrapperCallback(node) {
+  const argsNode = node.parent;
+  if (argsNode === null || argsNode.type !== "arguments") return false;
+  const call = argsNode.parent;
+  if (call === null || call.type !== "call_expression" || call.parent === null || call.parent.type !== "expression_statement") return false;
+  const callee = call.childForFieldName("function");
+  if (callee === null) return false;
+  if (callee.type === "identifier" && (callee.text === "$" || callee.text === "define" || callee.text === "require")) return true;
+  if (callee.type === "member_expression") {
+    const object = callee.childForFieldName("object");
+    const property = callee.childForFieldName("property");
+    const first = argsNode.namedChildren[0];
+    return object?.type === "identifier" && object.text === "document" && property?.text === "addEventListener" && first?.type === "string" && first.text.includes("DOMContentLoaded");
+  }
+  return false;
+}
 function extractTsJsSymbols(root, filePath, lines2) {
   const out = [];
   const visit = (node, insideFunction) => {
@@ -1996,12 +2223,6 @@ function extractTsJsSymbols(root, filePath, lines2) {
         out.push(makeSymbol(filePath, fieldName.text, "method", node, lines2, "c"));
       }
     }
-    if (node.type === "method_definition" && node.parent?.type === "object") {
-      const name = nodeName(node);
-      if (name !== null && name !== "") {
-        out.push(makeSymbol(filePath, name, "method", node, lines2, "c"));
-      }
-    }
     if (node.type === "pair" && node.parent?.type === "object") {
       const key = node.childForFieldName("key");
       const value = node.childForFieldName("value");
@@ -2018,7 +2239,9 @@ function extractTsJsSymbols(root, filePath, lines2) {
         }
       }
     }
-    const childInside = insideFunction || TSJS_FN_SCOPE_TYPES.has(node.type);
+    const isAnonymousFn = node.type === "function_expression" || node.type === "arrow_function";
+    const isTransparentWrapper = !insideFunction && isAnonymousFn && (isIifeCallee(node) || isModuleWrapperCallback(node));
+    const childInside = insideFunction || TSJS_FN_SCOPE_TYPES.has(node.type) && !isTransparentWrapper;
     for (const child of node.namedChildren) {
       visit(child, childInside);
     }
@@ -3557,8 +3780,8 @@ function boundSymbolDocstring(docstring) {
   return docstring.slice(0, Math.max(0, end)) + DOCSTRING_TRUNCATION_MARKER;
 }
 function loadGrammar(lang, filePath, content) {
-  const useTsx = lang === "typescript" && filePath !== void 0 && path4.extname(filePath).toLowerCase() === ".tsx";
-  const useCppHeader = lang === "c" && filePath !== void 0 && path4.extname(filePath).toLowerCase() === ".h" && content !== void 0 && CPP_HEADER_SNIFF_RE.test(content);
+  const useTsx = lang === "typescript" && filePath !== void 0 && path6.extname(filePath).toLowerCase() === ".tsx";
+  const useCppHeader = lang === "c" && filePath !== void 0 && path6.extname(filePath).toLowerCase() === ".h" && content !== void 0 && CPP_HEADER_SNIFF_RE.test(content);
   const cacheKey = useTsx ? "typescript:tsx" : useCppHeader ? "c:cpp-header" : lang;
   const cached = _grammarCache.get(cacheKey);
   if (cached !== void 0) return cached;
@@ -3694,7 +3917,7 @@ function parseSourceSymbolsTreeSitterOnly(content, filePath, language) {
 var REGEX_ADAPTERS_SLOT = /* @__PURE__ */ Symbol.for("token-goat.regex-adapters");
 var adapterSlot = globalThis;
 async function loadRegexExtractors() {
-  adapterSlot[REGEX_ADAPTERS_SLOT] ??= await import("./token-goat-chunk-5MMJAJF3.mjs");
+  adapterSlot[REGEX_ADAPTERS_SLOT] ??= await import("./token-goat-chunk-FWYZVS3E.mjs");
 }
 function regexAdapters() {
   const loaded = adapterSlot[REGEX_ADAPTERS_SLOT];
@@ -3756,7 +3979,7 @@ function isUnderSkipDir(filePath, skipDirs) {
 }
 function isParseSkipEligible(filePath, cfg) {
   if (isUnderSkipDir(filePath, cfg.skip_dirs)) return true;
-  if (cfg.skip_files.includes(path4.basename(filePath))) return true;
+  if (cfg.skip_files.includes(path6.basename(filePath))) return true;
   try {
     const stat = fs2.statSync(filePath);
     if (stat.size > cfg.large_file_skip_kb * 1024) return true;
@@ -3764,18 +3987,30 @@ function isParseSkipEligible(filePath, cfg) {
   }
   return false;
 }
+function boundarySpanKey(spans) {
+  return spans.map((s) => `${s.start}-${s.end}`).sort().join(",");
+}
+function embeddingBoundariesMoved(db, filePath, result) {
+  const priorSpans = db.prepare(`SELECT line_start, line_end FROM symbols WHERE ${pathEqClause("file_path")}`).all(foldPath(filePath));
+  const prior = boundarySpanKey(priorSpans.map((r) => ({ start: r.line_start, end: r.line_end })));
+  const next = boundarySpanKey(
+    result.symbols.filter((s) => s.name !== "" && s.kind !== "").map((s) => ({ start: s.lineStart, end: s.lineEnd }))
+  );
+  return prior !== next;
+}
 function writeParseResult(filePath, content, result, dbPath) {
   const db = getDb(dbPath);
   const sha = fingerprintContent(content);
   const mtime = safeMtime(filePath);
   const now = Date.now() / 1e3;
   const priorRow = db.prepare(`SELECT sha, embed_sha FROM files WHERE ${pathEqClause("path")}`).get(foldPath(filePath));
-  const embedShaToCarry = priorRow !== void 0 && priorRow.sha === sha ? priorRow.embed_sha : null;
+  const carriedEmbedSha = priorRow !== void 0 && priorRow.sha === sha ? priorRow.embed_sha : null;
+  const embedShaToCarry = carriedEmbedSha !== null && carriedEmbedSha === sha && detectLanguage(filePath) !== "markdown" && embeddingBoundariesMoved(db, filePath, result) ? null : carriedEmbedSha;
   const writeAll = db.transaction(() => {
     deleteFileRows(db, filePath);
     db.prepare(
       "INSERT INTO files (path, sha, mtime, language, indexed_at, parser_sha, embed_sha) VALUES (?, ?, ?, ?, ?, ?, ?)"
-    ).run(filePath, sha, mtime, result.language, now, PARSER_FINGERPRINT, embedShaToCarry);
+    ).run(filePath, sha, mtime, result.language, now, parserFingerprintForLanguage(result.language), embedShaToCarry);
     const insSym = db.prepare(
       "INSERT INTO symbols (file_path, name, kind, line_start, line_end, body, docstring, parent) VALUES (?, ?, ?, ?, ?, ?, ?, ?)"
     );
@@ -3844,7 +4079,16 @@ var TIMEOUT_EMBED_SHA_PREFIX = "timeout:";
 function timeoutEmbedSha(sha, workMillis) {
   return `${TIMEOUT_EMBED_SHA_PREFIX}${workMillis}:${sha}`;
 }
-function canonicalizeIndexPath(absPath) {
+var ASSET_EMBED_SHA_PREFIX = "asset:";
+function assetEmbedSha(sha) {
+  return `${ASSET_EMBED_SHA_PREFIX}${NON_TEXT_ASSET_SET_ID}:${sha}`;
+}
+var MAX_CHUNKS_EMBED_SHA_PREFIX = "maxchunks:";
+function maxChunksEmbedSha(sha, maxChunks) {
+  return `${MAX_CHUNKS_EMBED_SHA_PREFIX}${maxChunks}:${sha}`;
+}
+function canonicalizeIndexPath(rawPath) {
+  const absPath = normalizePath(rawPath);
   if (!isCaseInsensitiveFs()) return absPath;
   let real;
   try {
@@ -3855,17 +4099,17 @@ function canonicalizeIndexPath(absPath) {
   const cut = Math.max(absPath.lastIndexOf("/"), absPath.lastIndexOf("\\"));
   const base = absPath.slice(cut + 1);
   const realNorm = normalizePath(real);
-  const realBase = path4.basename(realNorm);
+  const realBase = path6.basename(realNorm);
   if (base === realBase) return absPath;
   if (foldPath(base) !== foldPath(realBase)) return absPath;
-  const callerDir = cut < 0 ? normalizePath(path4.resolve(".")) : normalizePath(absPath.slice(0, cut));
-  if (foldPath(callerDir) !== foldPath(path4.dirname(realNorm))) return absPath;
+  const callerDir = cut < 0 ? normalizePath(path6.resolve(".")) : normalizePath(absPath.slice(0, cut));
+  if (foldPath(callerDir) !== foldPath(path6.dirname(realNorm))) return absPath;
   return absPath.slice(0, cut + 1) + realBase;
 }
 function indexedPathSpellingIsStale(storedPath, absPath) {
   if (!isCaseInsensitiveFs()) return false;
   const stored = normalizePath(storedPath);
-  const candidate = normalizePath(path4.resolve(absPath));
+  const candidate = normalizePath(path6.resolve(absPath));
   if (foldPath(stored) !== foldPath(candidate)) return false;
   let real;
   try {
@@ -3891,12 +4135,14 @@ function indexedPathSpellingIsStale(storedPath, absPath) {
   }
   return false;
 }
-function isEmbedFresh(storedEmbedSha, sha, embeddingsEnabled, depsAvailable, symbolOnlyKb) {
+function isEmbedFresh(storedEmbedSha, sha, embeddingsEnabled, depsAvailable, symbolOnlyKb, maxChunksPerFile = 0) {
   if (storedEmbedSha === void 0) return false;
   if (!embeddingsEnabled) return storedEmbedSha === disabledEmbedSha(sha);
   if (storedEmbedSha === sha) return true;
   if (!depsAvailable && storedEmbedSha === unavailableEmbedSha(sha)) return true;
   if (storedEmbedSha === oversizeEmbedSha(sha, symbolOnlyKb)) return true;
+  if (maxChunksPerFile > 0 && storedEmbedSha === maxChunksEmbedSha(sha, maxChunksPerFile)) return true;
+  if (storedEmbedSha === assetEmbedSha(sha)) return true;
   if (storedEmbedSha === timeoutEmbedSha(sha, MAX_DOCUMENT_WORK_MILLIS)) return true;
   return false;
 }
@@ -3913,6 +4159,12 @@ async function indexFileEmbeddings(rawPath, dbPath = globalDbPath(), sha, onErro
     stampEmbedSha(db, filePath, sha, (s) => s);
     return;
   }
+  if (isNonTextAsset(filePath)) {
+    const db = getDb(dbPath);
+    deleteFileEmbeddings(db, filePath);
+    stampEmbedSha(db, filePath, sha, assetEmbedSha);
+    return;
+  }
   if (isEmbeddableDocument(filePath)) {
     let extracted;
     let refusedOnTheClock = false;
@@ -3976,9 +4228,16 @@ async function indexFileEmbeddings(rawPath, dbPath = globalDbPath(), sha, onErro
     stampEmbedSha(db, filePath, sha, (s) => s);
     return;
   }
+  const boundaries = buildEmbeddingBoundaries(filePath, content, dbPath);
+  const chunkCount = chunkFile(filePath, content, void 0, void 0, boundaries).length;
+  if (chunkCount > ixCfg.max_chunks_per_file) {
+    const db = getDb(dbPath);
+    deleteFileEmbeddings(db, filePath);
+    stampEmbedSha(db, filePath, sha, (s) => maxChunksEmbedSha(s, ixCfg.max_chunks_per_file));
+    return;
+  }
   try {
     const db = getDb(dbPath);
-    const boundaries = buildEmbeddingBoundaries(filePath, content, dbPath);
     const outcome = await indexFile(db, filePath, content, boundaries);
     stampEmbedSha(db, filePath, sha, (s) => outcome === "unavailable" ? unavailableEmbedSha(s) : s);
   } catch (err) {
@@ -4530,6 +4789,8 @@ export {
   isEmbeddableDocument,
   ipynbToVirtualSource,
   eachUnfencedLine,
+  isImagePath,
+  isNonTextAsset,
   modelFilesPresent,
   ensureModelFiles,
   checkEmbeddingPreflight,
@@ -4548,7 +4809,7 @@ export {
   formatHeadingTreeParts,
   getWellKnownSections,
   extractChangelogVersionHint,
-  PARSER_FINGERPRINT,
+  parserFingerprintForLanguage,
   treeSitterCoreAvailable,
   treeSitterCoreLoadError,
   isTreeSitterAvailable,
@@ -4558,6 +4819,8 @@ export {
   deleteFileRows,
   isParseSkipEligible,
   indexFileSync,
+  assetEmbedSha,
+  maxChunksEmbedSha,
   indexedPathSpellingIsStale,
   isEmbedFresh,
   indexFileEmbeddings,