Add caption addition

.gitignore

@@ -19,4 +19,5 @@ uploads/
 uploads/**
 audit_log_creds.json
 build/
-frames/
+frames/
+*.ass

common-utils

@@ -1 +1 @@
-Subproject commit d8e83ba4748273a696d23d0a11df0dc1ae7fae32
+Subproject commit 017243cdd3557fe8e78c195b79bdce52b17f5b92

server-plugins/generate-captions.js

@@ -0,0 +1,261 @@
+import { readFileSync, createWriteStream } from 'fs';
+import path from 'path';
+
+/**
+ * Format seconds to ASS timestamp format (H:MM:SS.cc)
+ * @param {number} seconds 
+ * @returns {string}
+ */
+function formatTimestampASS(seconds) {
+  const hours = Math.floor(seconds / 3600);
+  const minutes = Math.floor((seconds % 3600) / 60);
+  const secs = seconds % 60;
+  return `${hours}:${minutes.toString().padStart(2, '0')}:${secs.toFixed(2).padStart(5, '0')}`;
+}
+
+/**
+ * Split transcript into sentences
+ * @param {string} transcript 
+ * @returns {string[]}
+ */
+function splitIntoSentences(transcript) {
+  const parts = transcript.split(/([.!?]+)\s+/);
+  const result = [];
+
+  for (let i = 0; i < parts.length - 1; i += 2) {
+    if (i + 1 < parts.length) {
+      result.push(parts[i] + parts[i + 1]);
+    } else {
+      result.push(parts[i]);
+    }
+  }
+
+  if (parts.length % 2 === 1) {
+    result.push(parts[parts.length - 1]);
+  }
+
+  return result;
+}
+
+/**
+ * Assign sentence index to each word
+ * @param {Array} words 
+ * @param {string} transcript 
+ * @returns {Array}
+ */
+function assignSentenceToWords(words, transcript) {
+  const sentences = splitIntoSentences(transcript);
+
+  let wordIdx = 0;
+  sentences.forEach((sentence, sentIdx) => {
+    const sentenceWords = sentence.split(/\s+/);
+    sentenceWords.forEach(() => {
+      if (wordIdx < words.length) {
+        words[wordIdx].sentence_idx = sentIdx;
+        wordIdx++;
+      }
+    });
+  });
+
+  return words;
+}
+
+/**
+ * Create ASS file header with styles
+ * @param {number} videoWidth 
+ * @param {number} videoHeight 
+ * @param {string} fontName 
+ * @param {number} fontSize 
+ * @param {number} marginV 
+ * @returns {string}
+ */
+function createASSHeader(videoWidth = 1920, videoHeight = 1080, fontName = 'Impact', fontSize = 72, marginV = 200, marginL = 10, marginR = 10) {
+  return `[Script Info]
+Title: Word-by-Word Captions
+ScriptType: v4.00+
+WrapStyle: 0
+PlayResX: ${videoWidth}
+PlayResY: ${videoHeight}
+ScaledBorderAndShadow: yes
+
+[V4+ Styles]
+Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
+Style: Default,${fontName},${fontSize},&H00FFFFFF,&H000000FF,&H00000000,&H80000000,-1,0,0,0,100,100,0,0,1,3,2,2,${marginL},${marginR},${marginV},1
+Style: Highlight,${fontName},${fontSize},&H0000FF00,&H000000FF,&H00000000,&H80000000,-1,0,0,0,100,100,0,0,1,3,2,2,${marginL},${marginR},${marginV},1
+
+[Events]
+Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
+`;
+}
+
+/**
+ * Generate ASS subtitle file with word highlighting
+ * @param {Object} options
+ * @param {string} options.captionFilePath - Path to input JSON caption file
+ * @param {string} options.outputFilePath - Path to output ASS file
+ * @param {number} [options.tiltDegrees=8] - Tilt angle in degrees (alternates between +/-)
+ * @param {number} [options.translateY=200] - Distance from bottom in pixels
+ * @param {number} [options.widthPercent=80] - Width percentage for text centering (0-100)
+ * @param {string} [options.fontName='Impact'] - Font name
+ * @param {number} [options.fontSize=72] - Font size
+ * @param {number} [options.wordsPerGroup=4] - Number of words per caption group
+ * @param {number} [options.videoWidth=1920] - Video width for positioning
+ * @param {number} [options.videoHeight=1080] - Video height for positioning
+ * @returns {Promise<string>} Path to generated ASS file
+ */
+async function generateCaptions(options) {
+  const {
+    captionFilePath,
+    outputFilePath,
+    tiltDegrees = 8,
+    translateY = 200,
+    widthPercent = 80,
+    fontName = 'Impact',
+    fontSize = 72,
+    wordsPerGroup = 4,
+    videoWidth = 1920,
+    videoHeight = 1080
+  } = options;
+
+  // Read and parse JSON file
+  const jsonData = JSON.parse(readFileSync(captionFilePath, 'utf-8'));
+  const transcript = jsonData.transcript || '';
+  let words = jsonData.words || [];
+
+  if (words.length === 0) {
+    throw new Error('No words found in caption file');
+  }
+
+  // Assign sentence indices to words
+  words = assignSentenceToWords(words, transcript);
+
+  // Calculate margins for centering within width percentage
+  const totalMargin = videoWidth * (1 - widthPercent / 100);
+  const sideMargin = Math.floor(totalMargin / 2);
+
+  // Create output stream
+  const output = createWriteStream(outputFilePath);
+
+  // Write header with calculated margins
+  output.write(createASSHeader(videoWidth, videoHeight, fontName, fontSize, translateY, sideMargin, sideMargin));
+
+  // Process words in groups respecting sentence boundaries
+  let i = 0;
+  let groupIdx = 0;
+
+  while (i < words.length) {
+    const currentSentence = words[i].sentence_idx || 0;
+
+    // Collect words for this group (up to wordsPerGroup, same sentence only)
+    const wordGroup = [];
+    let j = i;
+
+    while (j < words.length && wordGroup.length < wordsPerGroup) {
+      if ((words[j].sentence_idx || 0) === currentSentence) {
+        wordGroup.push(words[j]);
+        j++;
+      } else {
+        break; // Stop at sentence boundary
+      }
+    }
+
+    if (wordGroup.length === 0) {
+      i++;
+      continue;
+    }
+
+    // Alternate tilt
+    const currentTilt = groupIdx % 2 === 0 ? tiltDegrees : -tiltDegrees;
+    const tiltTag = `{\\frz${currentTilt}}`;
+
+    // Calculate positioning for centering
+    const posTag = sideMargin > 0 ? `{\\an2\\pos(${videoWidth / 2},${videoHeight - translateY})}` : '';
+
+    // For each word in the group, create an event with highlighting
+    for (let wordIdx = 0; wordIdx < wordGroup.length; wordIdx++) {
+      const wordObj = wordGroup[wordIdx];
+      const wordStart = wordObj.start;
+      const wordEnd = wordObj.end;
+
+      // Build the caption text with highlighting
+      const captionParts = wordGroup.map((w, idx) => {
+        if (idx === wordIdx) {
+          // Current word - highlighted in green
+          return `{\\c&H00FF00&}${w.word}{\\c&HFFFFFF&}`;
+        } else {
+          // Other words - white
+          return w.word;
+        }
+      });
+
+      const captionText = tiltTag + posTag + captionParts.join(' ');
+
+      // Write dialogue line
+      output.write(`Dialogue: 0,${formatTimestampASS(wordStart)},${formatTimestampASS(wordEnd)},Default,,0,0,0,,${captionText}\n`);
+    }
+
+    i = j;
+    groupIdx++;
+  }
+
+  output.end();
+
+  return new Promise((resolve, reject) => {
+    output.on('finish', () => {
+      console.log(`Created ${outputFilePath} with word-by-word highlighting and tilted groups`);
+      resolve(outputFilePath);
+    });
+    output.on('error', reject);
+  });
+}
+
+// Example usage
+// Check if this is the main module in ESM
+import { fileURLToPath } from 'url';
+import { dirname } from 'path';
+
+const __filename = fileURLToPath(import.meta.url);
+const isMainModule = process.argv[1] === __filename;
+
+if (isMainModule) {
+  const args = process.argv.slice(2);
+
+  if (args.length < 2) {
+    console.log('Usage: node generateCaptions.js <input.json> <output.ass> [options]');
+    console.log('\nOptions:');
+    console.log('  --tilt <degrees>        Tilt angle (default: 8)');
+    console.log('  --translateY <pixels>   Distance from bottom (default: 200)');
+    console.log('  --width <percent>       Width percentage 0-100 (default: 80)');
+    console.log('  --font <name>           Font name (default: Impact)');
+    console.log('  --fontSize <size>       Font size (default: 72)');
+    console.log('  --wordsPerGroup <num>   Words per caption group (default: 4)');
+    console.log('\nExample:');
+    console.log('  node generateCaptions.js input.json output.ass --tilt 10 --width 90');
+    process.exit(1);
+  }
+
+  const captionFilePath = args[0];
+  const outputFilePath = args[1];
+
+  // Parse optional arguments
+  const options = {
+    captionFilePath,
+    outputFilePath
+  };
+
+  for (let i = 2; i < args.length; i += 2) {
+    const key = args[i].replace('--', '');
+    const value = args[i + 1];
+
+    if (key === 'tilt') options.tiltDegrees = parseFloat(value);
+    else if (key === 'translateY') options.translateY = parseInt(value);
+    else if (key === 'width') options.widthPercent = parseFloat(value);
+    else if (key === 'font') options.fontName = value;
+    else if (key === 'fontSize') options.fontSize = parseInt(value);
+    else if (key === 'wordsPerGroup') options.wordsPerGroup = parseInt(value);
+  }
+
+  generateCaptions(options).catch(console.error);
+}
+
+export default { generateCaptions };