codebase overhaul (#697)

apps/web/src/lib/transcription/caption.ts

@@ -0,0 +1,49 @@
+import type { TranscriptionSegment, CaptionChunk } from "@/types/transcription";
+import {
+	DEFAULT_WORDS_PER_CAPTION,
+	MIN_CAPTION_DURATION_SECONDS,
+} from "@/constants/transcription-constants";
+
+export function buildCaptionChunks({
+	segments,
+	wordsPerChunk = DEFAULT_WORDS_PER_CAPTION,
+	minDuration = MIN_CAPTION_DURATION_SECONDS,
+}: {
+	segments: TranscriptionSegment[];
+	wordsPerChunk?: number;
+	minDuration?: number;
+}): CaptionChunk[] {
+	const captions: CaptionChunk[] = [];
+	let globalEndTime = 0;
+
+	for (const segment of segments) {
+		const words = segment.text.trim().split(/\s+/);
+		if (words.length === 0 || (words.length === 1 && words[0] === "")) continue;
+
+		const segmentDuration = segment.end - segment.start;
+		const wordsPerSecond = words.length / segmentDuration;
+
+		const chunks: string[] = [];
+		for (let i = 0; i < words.length; i += wordsPerChunk) {
+			chunks.push(words.slice(i, i + wordsPerChunk).join(" "));
+		}
+
+		let chunkStartTime = segment.start;
+		for (const chunk of chunks) {
+			const chunkWords = chunk.split(/\s+/).length;
+			const chunkDuration = Math.max(minDuration, chunkWords / wordsPerSecond);
+			const adjustedStartTime = Math.max(chunkStartTime, globalEndTime);
+
+			captions.push({
+				text: chunk,
+				startTime: adjustedStartTime,
+				duration: chunkDuration,
+			});
+
+			globalEndTime = adjustedStartTime + chunkDuration;
+			chunkStartTime += chunkDuration;
+		}
+	}
+
+	return captions;
+}