import { readFile } from 'fs/promises'
const isWebVTTHeader = (line: string): boolean => line === 'WEBVTT'
const isEmpty = (line: string): boolean => line === ''
const isTimestamp = (line: string): boolean => line.includes('-->')
const isLineNumber = (line: string): boolean => /^\d+$/.test(line)
const isCueSetting = (line: string): boolean => line.includes('align:') || line.includes('position:')
const removeHtmlTags = (text: string): string => text.replace(/<[^>]*>/g, '')
const decodeHtmlEntities = (text: string): string =>
text
.replace(/</g, '<')
.replace(/>/g, '>')
.replace(/&/g, '&')
.replace(/"/g, '"')
const cleanSubtitleText = (text: string): string => {
const withoutTags = removeHtmlTags(text)
const decoded = decodeHtmlEntities(withoutTags)
return decoded.trim()
}
const extractTextLines = (lines: string[]): string[] => {
const textLines: string[] = []
let isTextLine = false
for (const line of lines) {
const trimmedLine = line.trim()
// Skip WEBVTT header and empty lines
if (isWebVTTHeader(trimmedLine) || isEmpty(trimmedLine)) {
continue
}
// Skip timestamp lines (format: 00:00:00.000 --> 00:00:02.000)
if (isTimestamp(trimmedLine)) {
isTextLine = true
continue
}
// Skip cue settings and numbers
if (isLineNumber(trimmedLine) || isCueSetting(trimmedLine)) {
continue
}
// Extract actual subtitle text
if (isTextLine && trimmedLine) {
const cleanText = cleanSubtitleText(trimmedLine)
if (cleanText) {
textLines.push(cleanText)
}
}
}
return textLines
}
const normalizeSpaces = (text: string): string => text.replace(/\s+/g, ' ').trim()
export const parseVTTFile = async (filePath: string): Promise<string> => {
try {
const content = await readFile(filePath, 'utf-8')
const lines = content.split('\n')
const textLines = extractTextLines(lines)
const joinedText = textLines.join(' ')
return normalizeSpaces(joinedText)
} catch (error) {
const errorMessage = error instanceof Error ? error.message : 'Unknown error'
throw new Error(`Failed to parse VTT file: ${errorMessage}`)
}
}
