@gmickel/gno 1.12.0 → 1.12.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@gmickel/gno",
-  "version": "1.12.0",
+  "version": "1.12.2",
   "description": "Local semantic search for your documents. Index Markdown, PDF, and Office files with hybrid BM25 + vector search.",
   "keywords": [
     "embeddings",

package/src/converters/adapters/markitdownTs/adapter.ts

@@ -40,6 +40,7 @@ const CFB_SIGNATURE = new Uint8Array([
 ]);
 const ENCRYPTION_INFO = utf16le("EncryptionInfo");
 const ENCRYPTED_PACKAGE = utf16le("EncryptedPackage");
+const PDF_TRAILER_SCAN_BYTES = 2048;
 const MAX_MESSAGE_LENGTH = 200;
 const PASSWORD_ERROR_REGEX = /password(?:-protected)?|no password given/i;
 
@@ -100,6 +101,19 @@ function isPasswordProtectedPdf(bytes: Uint8Array): boolean {
   return /\/Encrypt\b/.test(tail);
 }
 
+function hasCompletePdfTrailer(bytes: Uint8Array): boolean {
+  if (!hasPrefix(bytes, PDF_SIGNATURE)) {
+    return false;
+  }
+
+  const tailStart = Math.max(0, bytes.length - PDF_TRAILER_SCAN_BYTES);
+  const tail = Buffer.from(bytes.subarray(tailStart)).toString("latin1");
+  const eofIndex = tail.lastIndexOf("%%EOF");
+  const startXrefIndex = tail.lastIndexOf("startxref");
+
+  return startXrefIndex >= 0 && eofIndex > startXrefIndex;
+}
+
 function isPasswordProtectedXlsx(bytes: Uint8Array): boolean {
   return (
     hasPrefix(bytes, CFB_SIGNATURE) &&
@@ -162,7 +176,20 @@ export const markitdownAdapter: Converter = {
       return { ok: false, error: tooLargeError(input, CONVERTER_ID) };
     }
 
-    // 1b. Detect password-protected documents before calling markitdown-ts.
+    // 1b. Reject obviously truncated PDFs before markitdown-ts logs a parser
+    // stack trace. ISO 32000 requires the final %%EOF marker near the file end.
+    if (input.ext === ".pdf" && !hasCompletePdfTrailer(input.bytes)) {
+      return {
+        ok: false,
+        error: corruptError(
+          input,
+          CONVERTER_ID,
+          "Invalid or incomplete PDF structure"
+        ),
+      };
+    }
+
+    // 1c. Detect password-protected documents before calling markitdown-ts.
     // markitdown-ts logs dependency stack traces to stderr for these files.
     if (isPasswordProtected(input)) {
       return {

package/src/ingestion/sync.ts

@@ -75,6 +75,12 @@ const MAX_CONCURRENCY = 16;
 export const INGEST_VERSION = 6;
 const EMPTY_CONTENT_TYPE_RULES_FINGERPRINT = fingerprintContentTypeRules([]);
 const RELATION_EDGE_TYPE_PATTERN = /^[a-z][a-z0-9_]*$/;
+const NON_RETRYABLE_CONVERSION_ERROR_CODES = new Set([
+  "CORRUPT",
+  "PERMISSION",
+  "TOO_LARGE",
+  "UNSUPPORTED",
+]);
 
 type RelationMap = Record<string, string[]>;
 
@@ -219,6 +225,20 @@ function decideAction(
 
   // Source unchanged, but check for repair cases:
 
+  // Preserve non-retryable conversion failures until the source or ingest
+  // version changes. Re-running an unchanged corrupt/protected file on every
+  // sync only repeats expensive work and noisy diagnostics.
+  if (
+    existing.lastErrorCode &&
+    NON_RETRYABLE_CONVERSION_ERROR_CODES.has(existing.lastErrorCode) &&
+    existing.ingestVersion === INGEST_VERSION
+  ) {
+    return {
+      kind: "skip",
+      reason: "unchanged non-retryable conversion failure",
+    };
+  }
+
   // 1. Previous conversion failed (mirrorHash is null)
   if (!existing.mirrorHash) {
     return { kind: "repair", reason: "previous conversion failed" };
@@ -738,6 +758,8 @@ export class SyncService {
           sourceCtime,
           lastErrorCode: convertResult.error.code,
           lastErrorMessage: convertResult.error.message,
+          ingestVersion: INGEST_VERSION,
+          contentTypeRulesFingerprint,
           // mirrorHash intentionally omitted (will be null)
         });