@absolutejs/absolute 0.19.0-beta.621 → 0.19.0-beta.622

package/dist/ai/index.js

@@ -4471,6 +4471,8 @@ var buildProvenanceLabel2 = (metadata) => {
   const mediaSegmentGroupIndex = getContextNumber2(metadata.mediaSegmentGroupIndex);
   const mediaChannel = getContextString2(metadata.mediaChannel);
   const mediaDurationLabel = formatMediaDurationLabel2(metadata.mediaDurationMs);
+  const mediaSegmentWindowDurationLabel = formatMediaDurationLabel2(metadata.mediaSegmentGroupDurationMs);
+  const mediaSegmentGapLabel = formatMediaDurationLabel2(metadata.mediaSegmentGapFromPreviousMs);
   const spreadsheetHeaders = getSpreadsheetHeaders(metadata);
   const slideNotesText = getContextString2(metadata.slideNotesText);
   const pdfTextMode = getContextString2(metadata.pdfTextMode);
@@ -4506,6 +4508,8 @@ var buildProvenanceLabel2 = (metadata) => {
     mediaChannel ? `Channel ${mediaChannel}` : "",
     mediaSpeakerCount ? `${mediaSpeakerCount} speakers` : "",
     mediaDurationLabel ? `Duration ${mediaDurationLabel}` : "",
+    mediaSegmentWindowDurationLabel ? `Segment window ${mediaSegmentWindowDurationLabel}` : "",
+    mediaSegmentGapLabel ? `Gap ${mediaSegmentGapLabel} from previous window` : "",
     transcriptSource ? `Transcript ${transcriptSource}` : "",
     threadTopic ? `Thread ${threadTopic}` : "",
     threadRootMessageId ? `Thread root ${threadRootMessageId}` : "",
@@ -8715,6 +8719,10 @@ var scoreStructuredEvidenceMatch = (queryTokens, result) => {
   const mediaSpeakerCount = typeof metadata.mediaSpeakerCount === "number" ? metadata.mediaSpeakerCount : undefined;
   const mediaSegmentCount = typeof metadata.mediaSegmentCount === "number" ? metadata.mediaSegmentCount : undefined;
   const mediaSegmentGroupSize = typeof metadata.mediaSegmentGroupSize === "number" ? metadata.mediaSegmentGroupSize : undefined;
+  const mediaSegmentGroupStartMs = typeof metadata.mediaSegmentGroupStartMs === "number" && Number.isFinite(metadata.mediaSegmentGroupStartMs) ? metadata.mediaSegmentGroupStartMs : typeof metadata.mediaSegmentStartMs === "number" && Number.isFinite(metadata.mediaSegmentStartMs) ? metadata.mediaSegmentStartMs : undefined;
+  const mediaSegmentGroupEndMs = typeof metadata.mediaSegmentGroupEndMs === "number" && Number.isFinite(metadata.mediaSegmentGroupEndMs) ? metadata.mediaSegmentGroupEndMs : typeof metadata.mediaSegmentEndMs === "number" && Number.isFinite(metadata.mediaSegmentEndMs) ? metadata.mediaSegmentEndMs : undefined;
+  const mediaSegmentGroupDurationMs = typeof metadata.mediaSegmentGroupDurationMs === "number" && Number.isFinite(metadata.mediaSegmentGroupDurationMs) ? metadata.mediaSegmentGroupDurationMs : typeof mediaSegmentGroupStartMs === "number" && typeof mediaSegmentGroupEndMs === "number" && mediaSegmentGroupEndMs >= mediaSegmentGroupStartMs ? mediaSegmentGroupEndMs - mediaSegmentGroupStartMs : undefined;
+  const mediaSegmentGapFromPreviousMs = typeof metadata.mediaSegmentGapFromPreviousMs === "number" && Number.isFinite(metadata.mediaSegmentGapFromPreviousMs) ? metadata.mediaSegmentGapFromPreviousMs : undefined;
   const mediaChannel = metadataString(metadata.mediaChannel);
   const speaker = metadataString(metadata.speaker);
   const ocrConfidence = typeof metadata.ocrRegionConfidence === "number" ? metadata.ocrRegionConfidence : typeof metadata.ocrPageAverageConfidence === "number" ? metadata.ocrPageAverageConfidence : typeof metadata.ocrAverageConfidence === "number" ? metadata.ocrAverageConfidence : typeof metadata.ocrConfidence === "number" ? metadata.ocrConfidence : undefined;
@@ -8836,6 +8844,36 @@ var scoreStructuredEvidenceMatch = (queryTokens, result) => {
   ])) {
     score += 0.06;
   }
+  if (typeof mediaSegmentGroupStartMs === "number" && typeof mediaSegmentGroupEndMs === "number" && queryHasAnyToken(queryTokens, [
+    "timestamp",
+    "segment",
+    "segments",
+    "audio",
+    "video",
+    "window",
+    "start",
+    "end"
+  ])) {
+    score += 0.05;
+  }
+  if (typeof mediaSegmentGroupDurationMs === "number" && queryHasAnyToken(queryTokens, ["duration", "timestamp", "segment"])) {
+    score += 0.06;
+    if (queryHasAnyToken(queryTokens, ["long", "longer", "longest"]) && mediaSegmentGroupDurationMs >= 1e4) {
+      score += 0.12;
+    }
+    if (queryHasAnyToken(queryTokens, ["short", "shorter", "shortest"]) && mediaSegmentGroupDurationMs <= 5000) {
+      score += 0.12;
+    }
+  }
+  if (queryHasAnyToken(queryTokens, ["continuous", "gap", "gaps"]) && typeof mediaSegmentGapFromPreviousMs === "number") {
+    if (mediaSegmentGapFromPreviousMs === 0) {
+      score += 0.2;
+    } else if (mediaSegmentGapFromPreviousMs <= 1000) {
+      score += 0.14;
+    } else if (mediaSegmentGapFromPreviousMs <= 3000) {
+      score += 0.06;
+    }
+  }
   if (mediaChannel && queryHasAnyToken(queryTokens, [
     "channel",
     "channels",
@@ -9133,28 +9171,71 @@ var formatMediaTimestampForIngest = (value) => {
   return `${String(minutes).padStart(2, "0")}:${String(seconds).padStart(2, "0")}.${String(milliseconds).padStart(3, "0")}`;
 };
 var normalizeMediaSpeaker = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : undefined;
-var normalizeMediaChannel = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : undefined;
+var normalizeMediaTimestamp = (value) => typeof value === "number" && Number.isFinite(value) && value >= 0 ? value : undefined;
+var normalizeMediaSegmentWindow = (startMs, endMs) => {
+  if (typeof startMs !== "number" && typeof endMs !== "number") {
+    return { endMs, startMs };
+  }
+  if (typeof startMs === "number" && typeof endMs === "number" && endMs < startMs) {
+    return {
+      startMs: undefined,
+      endMs: undefined
+    };
+  }
+  return {
+    startMs,
+    endMs
+  };
+};
+var canonicalMediaChannel = (value) => {
+  if (typeof value !== "string") {
+    return;
+  }
+  switch (value.trim().toLowerCase().replace(/[\s_-]+/g, "")) {
+    case "l":
+    case "left":
+    case "ch1":
+    case "channel1":
+    case "one":
+      return "left";
+    case "r":
+    case "right":
+    case "ch2":
+    case "channel2":
+    case "two":
+      return "right";
+    case "m":
+    case "mono":
+      return "mono";
+    default:
+      return value.trim().toLowerCase();
+  }
+};
+var normalizeMediaChannel = (value) => canonicalMediaChannel(value);
 var buildMediaTimestampBoundary = (segments) => {
   let startMs;
   let endMs;
   for (const segment of segments) {
     if (typeof segment.startMs === "number" && Number.isFinite(segment.startMs)) {
-      startMs = segment.startMs;
-      break;
+      startMs = typeof startMs === "number" ? Math.min(startMs, segment.startMs) : segment.startMs;
     }
   }
-  for (let index = segments.length - 1;index >= 0; index--) {
-    const segment = segments[index];
-    if (!segment) {
-      continue;
-    }
+  for (const segment of segments) {
     if (typeof segment.endMs === "number" && Number.isFinite(segment.endMs)) {
-      endMs = segment.endMs;
-      break;
+      endMs = typeof endMs === "number" ? Math.max(endMs, segment.endMs) : segment.endMs;
     }
   }
   return { endMs, startMs };
 };
+var normalizeMediaSegment = (segment) => {
+  const normalizedTimestampWindow = normalizeMediaSegmentWindow(normalizeMediaTimestamp(segment.startMs), normalizeMediaTimestamp(segment.endMs));
+  return {
+    ...segment,
+    speaker: normalizeMediaSpeaker(segment.speaker),
+    channel: canonicalMediaChannel(segment.channel),
+    ...normalizedTimestampWindow
+  };
+};
 var groupTranscriptSegments = (segments) => {
   const groups = [];
   for (const segment of segments) {
@@ -10743,7 +10824,7 @@ var createRAGMediaFileExtractor = (transcriber) => ({
         return false;
       }
       return normalizeWhitespace(segment.text ?? "").length > 0;
-    });
+    }).map(normalizeMediaSegment);
     const segmentGroups = groupTranscriptSegments(rawSegments);
     const source = input.source ?? input.path ?? input.name ?? `${slugify(input.title ?? DEFAULT_BINARY_NAME)}.media.txt`;
     const segmentCount = rawSegments.length;
@@ -10758,6 +10839,7 @@ var createRAGMediaFileExtractor = (transcriber) => ({
       ...new Set(rawSegments.map((segment) => normalizeMediaSpeaker(segment.speaker)).filter((value) => typeof value === "string"))
     ];
     const segmentDocuments = [];
+    let previousGroupEndMs;
     for (const [index, segmentGroup] of segmentGroups.entries()) {
       const { endMs, startMs } = buildMediaTimestampBoundary(segmentGroup.segments);
       const groupText = normalizeWhitespace(segmentGroup.segments.map((segment) => normalizeWhitespace(segment.text ?? "")).filter((value) => value.length > 0).join(" "));
@@ -10769,6 +10851,9 @@ var createRAGMediaFileExtractor = (transcriber) => ({
       const startLabel = formatMediaTimestampForIngest(startMs);
       const endLabel = formatMediaTimestampForIngest(endMs);
       const mediaKind = typeof result.metadata?.mediaKind === "string" ? result.metadata.mediaKind : "media";
+      const mediaSegmentGapFromPreviousMs = typeof startMs === "number" && typeof previousGroupEndMs === "number" && Number.isFinite(startMs) && Number.isFinite(previousGroupEndMs) ? Math.max(0, startMs - previousGroupEndMs) : undefined;
+      previousGroupEndMs = endMs;
+      const mediaSegmentGroupDurationMs = typeof endMs === "number" && typeof startMs === "number" && Number.isFinite(endMs) && Number.isFinite(startMs) && endMs >= startMs ? endMs - startMs : undefined;
       segmentDocuments.push({
         chunking: input.chunking,
         contentType: input.contentType,
@@ -10787,6 +10872,10 @@ var createRAGMediaFileExtractor = (transcriber) => ({
           mediaSegmentGroupSize: segmentGroup.segments.length,
           mediaSegmentGroupSpeaker: segmentGroup.speaker,
           mediaChannel: segmentGroup.channel,
+          mediaSegmentGroupDurationMs,
+          mediaSegmentGapFromPreviousMs,
+          mediaSegmentGroupStartMs: startMs,
+          mediaSegmentGroupEndMs: endMs,
           mediaSegments: segmentGroup.segments,
           startMs: mediaSegmentStartMs,
           endMs: mediaSegmentEndMs,
@@ -12650,6 +12739,9 @@ var buildRAGContextProvenanceLabel = (metadata) => {
   const threadTopic = getContextString3(metadata.threadTopic);
   const from = getContextString3(metadata.from);
   const speaker = getContextString3(metadata.speaker);
+  const mediaChannel = getContextString3(metadata.mediaChannel);
+  const mediaSegmentGroupDurationMs = getContextNumber3(metadata.mediaSegmentGroupDurationMs);
+  const mediaSegmentGapFromPreviousMs = getContextNumber3(metadata.mediaSegmentGapFromPreviousMs);
   const ocrConfidence = getContextNumber3(metadata.ocrRegionConfidence) ?? getContextNumber3(metadata.ocrConfidence);
   const labels = [
     pdfTextMode ? `PDF ${pdfTextMode}` : "",
@@ -12659,6 +12751,9 @@ var buildRAGContextProvenanceLabel = (metadata) => {
     transcriptSource ? `Transcript ${transcriptSource}` : "",
     threadTopic ? `Thread ${threadTopic}` : "",
     speaker ? `Speaker ${speaker}` : "",
+    mediaChannel ? `Channel ${mediaChannel}` : "",
+    typeof mediaSegmentGroupDurationMs === "number" ? `Segment window ${String(Math.floor(mediaSegmentGroupDurationMs / 1000))}s` : "",
+    typeof mediaSegmentGapFromPreviousMs === "number" ? `Segment gap ${String(Math.floor(mediaSegmentGapFromPreviousMs / 1000))}s` : "",
     from ? `Sender ${from}` : ""
   ].filter((value) => value.length > 0);
   return labels.length > 0 ? labels.join(" \xB7 ") : undefined;
@@ -24836,5 +24931,5 @@ export {
   aiChat
 };
 
-//# debugId=AFAF0A5BC1AB4BC864756E2164756E21
+//# debugId=E93ED81BD7E5AC3C64756E2164756E21
 //# sourceMappingURL=index.js.map