granola-toolkit 0.6.0 → 0.7.0

package/dist/cli.js

@@ -906,7 +906,40 @@ function parseCacheContents(contents) {
 }
 //#endregion
 //#region src/transcripts.ts
-function buildTranscriptExport(document, segments) {
+function transcriptSegmentKey(segment) {
+	if (segment.id) return `id:${segment.id}`;
+	return [
+		segment.documentId,
+		segment.source,
+		segment.startTimestamp,
+		segment.endTimestamp
+	].join("|");
+}
+function compareSegmentTimestamps(left, right) {
+	if (left === right) return 0;
+	const leftTime = Date.parse(left);
+	const rightTime = Date.parse(right);
+	if (!Number.isNaN(leftTime) && !Number.isNaN(rightTime)) return leftTime - rightTime;
+	return compareStrings(left, right);
+}
+function compareTranscriptSegments(left, right) {
+	return compareSegmentTimestamps(left.startTimestamp, right.startTimestamp) || compareSegmentTimestamps(left.endTimestamp, right.endTimestamp) || compareStrings(left.source, right.source) || compareStrings(left.id, right.id) || compareStrings(left.text, right.text);
+}
+function preferredTranscriptSegment(current, candidate) {
+	if (!current) return candidate;
+	if (candidate.isFinal !== current.isFinal) return candidate.isFinal ? candidate : current;
+	return compareSegmentTimestamps(candidate.endTimestamp, current.endTimestamp) > 0 || candidate.text.length > current.text.length ? candidate : current;
+}
+function normaliseTranscriptSegments(segments) {
+	const selected = /* @__PURE__ */ new Map();
+	for (const segment of [...segments].sort(compareTranscriptSegments)) {
+		const key = transcriptSegmentKey(segment);
+		const current = selected.get(key);
+		selected.set(key, preferredTranscriptSegment(current, segment));
+	}
+	return [...selected.values()].sort(compareTranscriptSegments);
+}
+function buildTranscriptExport(document, segments, rawSegments = segments) {
 	const renderedSegments = segments.map((segment) => ({
 		endTimestamp: segment.endTimestamp,
 		id: segment.id,
@@ -921,7 +954,7 @@ function buildTranscriptExport(document, segments) {
 		id: document.id,
 		raw: {
 			document,
-			segments
+			segments: rawSegments
 		},
 		segments: renderedSegments,
 		title: document.title,
@@ -990,7 +1023,7 @@ async function writeTranscripts(cacheData, outputDir, format = "text") {
 				title: documentId,
 				updatedAt: ""
 			};
-			const content = renderTranscriptExport(buildTranscriptExport(document, segments), format);
+			const content = renderTranscriptExport(buildTranscriptExport(document, normaliseTranscriptSegments(segments), segments), format);
 			if (!content) return [];
 			return [{
 				content,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "granola-toolkit",
-  "version": "0.6.0",
+  "version": "0.7.0",
   "description": "CLI toolkit for exporting and working with Granola notes and transcripts",
   "keywords": [
     "cli",