/**
* Transcript XML parser
*/
import { TranscriptParseError } from "./errors.js";
/**
* HTML formatting tags to preserve when preserveFormatting is true
*/
const FORMATTING_TAGS = [
"strong",
"em",
"b",
"i",
"mark",
"small",
"del",
"ins",
"sub",
"sup",
];
/**
* Parses YouTube transcript XML data
*/
export class TranscriptParser {
constructor(preserveFormatting = false) {
this.htmlRegex = this.getHtmlRegex(preserveFormatting);
}
/**
* Creates the regex for removing HTML tags
*/
getHtmlRegex(preserveFormatting) {
if (preserveFormatting) {
const formatsRegex = FORMATTING_TAGS.join("|");
const pattern = `<\\/?(?!\\/?(?:${formatsRegex})\\b).*?\\b>`;
return new RegExp(pattern, "gi");
}
return /<[^>]*>/gi;
}
/**
* Decodes HTML entities in text
*/
decodeHtmlEntities(text) {
return text
.replace(/&/g, "&")
.replace(/</g, "<")
.replace(/>/g, ">")
.replace(/"/g, '"')
.replace(/'/g, "'")
.replace(/ /g, " ")
.replace(/&#(\d+);/g, (match, dec) => String.fromCharCode(dec));
}
/**
* Parses XML transcript data into structured snippets
*/
parse(xmlData, videoId) {
try {
// Extract text elements using regex (avoiding XML parser dependencies)
const textRegex = /<text\s+start="([^"]+)"\s+dur="([^"]+)"[^>]*>([^<]*)<\/text>/g;
const snippets = [];
let match;
while ((match = textRegex.exec(xmlData)) !== null) {
const start = parseFloat(match[1]);
const duration = parseFloat(match[2]);
let text = match[3];
// Decode HTML entities
text = this.decodeHtmlEntities(text);
// Remove HTML tags according to preserveFormatting setting
text = text.replace(this.htmlRegex, "");
// Only add non-empty snippets
if (text.trim()) {
snippets.push({
text: text.trim(),
start,
duration,
});
}
}
// Alternative: handle self-closing text tags or different formats
if (snippets.length === 0) {
// Try alternative parsing for edge cases
const altRegex = /<text[^>]+start="([^"]+)"[^>]+dur="([^"]+)"[^>]*>([^<]*)<\/text>/g;
while ((match = altRegex.exec(xmlData)) !== null) {
const start = parseFloat(match[1]);
const duration = parseFloat(match[2]);
let text = this.decodeHtmlEntities(match[3]).replace(this.htmlRegex, "");
if (text.trim()) {
snippets.push({
text: text.trim(),
start,
duration,
});
}
}
}
return snippets;
} catch (error) {
throw new TranscriptParseError(
videoId,
error instanceof Error ? error : new Error(String(error))
);
}
}
}
