@absolutejs/absolute 0.19.0-beta.507 → 0.19.0-beta.509

package/dist/ai/index.js

@@ -2221,6 +2221,16 @@ var extractWeightedLexicalFields = (result) => {
   const source = result.source ?? "";
   const archivePath = typeof metadata.archivePath === "string" ? metadata.archivePath : source.includes("#") ? source.split("#")[1] ?? "" : "";
   const mediaSegments = Array.isArray(metadata.mediaSegments) ? metadata.mediaSegments.map((segment) => segment && typeof segment === "object" ? toFieldText(segment) : "").filter(Boolean).join(" ") : "";
+  const mediaTimestampFocus = metadata.sourceNativeKind === "media_segment" ? [
+    typeof metadata.mediaKind === "string" ? metadata.mediaKind : "",
+    "audio",
+    "video",
+    "media",
+    "timestamp",
+    "segment",
+    typeof metadata.mediaSegmentStartMs === "number" ? `timestamp ${metadata.mediaSegmentStartMs}` : "",
+    typeof metadata.mediaSegmentEndMs === "number" ? `timestamp ${metadata.mediaSegmentEndMs}` : ""
+  ].filter(Boolean).join(" ") : "";
   const spreadsheetFocus = metadata.sourceNativeKind === "spreadsheet_sheet" ? [
     "spreadsheet",
     "workbook",
@@ -2231,6 +2241,7 @@ var extractWeightedLexicalFields = (result) => {
   ].filter(Boolean).join(" ") : "";
   const metadataFocus = [
     metadata.sourceNativeKind,
+    mediaTimestampFocus,
     spreadsheetFocus,
     metadata.sheetName,
     metadata.sheetNames,
@@ -2445,7 +2456,9 @@ var resolveFileKindBoost = (queryTokens, metadata) => {
     "framework",
     "transcript",
     "audio",
-    "video"
+    "video",
+    "timestamp",
+    "segment"
   ])) {
     return 0.75;
   }
@@ -2461,7 +2474,8 @@ var resolveTranscriptBoost = (queryTokens, metadata) => {
     return 0;
   }
   const overlap = queryTokens.filter((token) => segmentText.includes(token)).length;
-  return overlap / Math.max(1, queryTokens.length);
+  const timestampBoost = queryTokens.includes("timestamp") ? 0.35 : 0;
+  return Math.min(1, overlap / Math.max(1, queryTokens.length) + timestampBoost);
 };
 var resolveArchiveBoost = (queryTokens, result) => {
   const archivePath = typeof result.metadata?.archivePath === "string" ? result.metadata.archivePath.toLowerCase() : typeof result.source === "string" && result.source.includes("#") ? result.source.split("#")[1]?.toLowerCase() ?? "" : "";
@@ -2701,6 +2715,17 @@ var detectDomains = (tokens) => {
   return [...domains];
 };
 var uniqueQueryStrings = (values) => Array.from(new Set(values.map((value) => value.trim()).filter((value) => value.length > 0)));
+var hasExplicitTimestamp = (query) => /\b\d{1,2}:\d{2}\b/.test(query) || /\b\d{1,2}:\d{2}\s*(?:to|-)\s*\d{1,2}:\d{2}\b/i.test(query);
+var hasNamedSourcePhrase = (query) => /\bnamed\s+[a-z0-9]/i.test(query) || /\btitled\s+[a-z0-9]/i.test(query);
+var shouldPreservePrimaryQuery = (query, domains) => {
+  if (hasExplicitTimestamp(query) && (domains.includes("audio") || domains.includes("video"))) {
+    return true;
+  }
+  if (hasNamedSourcePhrase(query) && domains.includes("spreadsheet")) {
+    return true;
+  }
+  return false;
+};
 var createHeuristicRAGQueryTransform = (options = {}) => createRAGQueryTransform({
   defaultModel: options.defaultModel ?? "absolute-heuristic-query-transform",
   providerName: options.providerName ?? "absolute_heuristic",
@@ -2744,9 +2769,10 @@ var createHeuristicRAGQueryTransform = (options = {}) => createRAGQueryTransform
     if (mediaTimestampVariant.length > 0) {
       variants.push(mediaTimestampVariant);
     }
+    const preservePrimaryQuery = shouldPreservePrimaryQuery(query, domains);
     return {
-      query: rewrittenQuery,
-      variants: uniqueQueryStrings(variants)
+      query: preservePrimaryQuery ? query : rewrittenQuery,
+      variants: uniqueQueryStrings(preservePrimaryQuery ? [rewrittenQuery, ...variants] : variants)
     };
   }
 });
@@ -3055,6 +3081,16 @@ var normalizeWhitespace = (value) => value.replace(/\r\n?/g, `
 `).replace(/\n{3,}/g, `
 
 `).trim();
+var formatMediaTimestampForIngest = (value) => {
+  if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
+    return;
+  }
+  const totalSeconds = Math.floor(value / 1000);
+  const minutes = Math.floor(totalSeconds / 60);
+  const seconds = totalSeconds % 60;
+  const milliseconds = Math.floor(value % 1000);
+  return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
+};
 var decodeHtmlEntities = (value) => {
   let output = value;
   for (const [pattern, replacement] of HTML_ENTITY_REPLACEMENTS) {
@@ -3687,6 +3723,9 @@ var createRAGMediaFileExtractor = (transcriber) => ({
       }
       const startMs = typeof segment.startMs === "number" ? segment.startMs : undefined;
       const endMs = typeof segment.endMs === "number" ? segment.endMs : undefined;
+      const startLabel = formatMediaTimestampForIngest(startMs);
+      const endLabel = formatMediaTimestampForIngest(endMs);
+      const mediaKind = typeof result.metadata?.mediaKind === "string" ? result.metadata.mediaKind : "media";
       segmentDocuments.push({
         chunking: input.chunking,
         contentType: input.contentType,
@@ -3703,9 +3742,9 @@ var createRAGMediaFileExtractor = (transcriber) => ({
           speaker: typeof segment.speaker === "string" ? segment.speaker : undefined
         },
         source,
-        text: normalizeWhitespace(`Media transcript segment${typeof startMs === "number" ? ` ${startMs}-${endMs ?? startMs}ms` : ""} from ${input.title ?? input.name ?? input.path ?? DEFAULT_BINARY_NAME}.
+        text: normalizeWhitespace(`${mediaKind} transcript segment${startLabel ? ` at timestamp ${startLabel}${endLabel ? ` to ${endLabel}` : ""}` : ""} from ${input.title ?? input.name ?? input.path ?? DEFAULT_BINARY_NAME}. ` + `${mediaKind} timestamp evidence${startLabel ? ` ${startLabel}${endLabel ? ` to ${endLabel}` : ""}` : ""}.` + `
 ${text}`),
-        title: input.title ? `${input.title} \xB7 Segment ${index + 1}` : `Segment ${index + 1}`
+        title: input.title ? `${input.title} \xB7 ${mediaKind[0]?.toUpperCase() + mediaKind.slice(1)} segment ${index + 1}` : `${mediaKind[0]?.toUpperCase() + mediaKind.slice(1)} segment ${index + 1}`
       });
     }
     const summaryDocument = {
@@ -10238,5 +10277,5 @@ export {
   aiChat
 };
 
-//# debugId=80FE4F97C8F27FEB64756E2164756E21
+//# debugId=E4C202F91B1AEC6364756E2164756E21
 //# sourceMappingURL=index.js.map