@kimsungwhee/apple-docs-mcp
Version:
MCP server for Apple Developer Documentation - Search iOS/macOS/SwiftUI/UIKit docs, WWDC videos, Swift/Objective-C APIs & code examples in Claude, Cursor & AI assistants
519 lines • 19.2 kB
JavaScript
/**
* WWDC video content extractor
*/
import { JSDOM } from 'jsdom';
import { httpClient } from '../../utils/http-client.js';
import { logger } from '../../utils/logger.js';
import { WWDC_URLS, WWDC_CONFIG } from '../../utils/constants.js';
import { inferTopics } from '../../utils/topic-mapper.js';
/**
* Extract complete content for a single WWDC video
*/
export async function extractVideoContent(videoUrl, videoId, year, knownTopics) {
try {
logger.info(`Extracting content for WWDC${year} video ${videoId}`);
// Get page HTML
const response = await httpClient.get(videoUrl);
const html = await response.text();
const dom = new JSDOM(html);
const document = dom.window.document;
// Extract basic information
const title = extractTitle(document);
const speakers = extractSpeakers(document);
const duration = extractDuration(document);
// Use known topics or infer topics
const topics = knownTopics || inferTopics(title, extractDescription(document));
// Detect if there's a Code tab (WWDC22 and later) - now only used for logging
const hasCodeTab = detectCodeTab(document, year);
logger.info(`Video ${videoId} has code tab: ${hasCodeTab}`);
// Extract transcript
const transcript = await extractTranscript(document, videoUrl);
const hasTranscript = transcript !== undefined;
// Extract code examples
let codeExamples;
// First try to extract from dedicated code page (WWDC22 and later)
codeExamples = await extractCodeExamples(document, videoUrl);
// If no code found and it's an early year, try extracting from transcript
if ((!codeExamples || codeExamples.length === 0) && parseInt(year) < WWDC_CONFIG.CODE_TAB_INTRODUCED_YEAR) {
codeExamples = extractCodeFromTranscript(document);
}
const hasCode = codeExamples !== undefined && codeExamples.length > 0;
// Extract chapter information
const chapters = extractChapters(document);
// Extract resource links
const resources = await extractResources(document, videoUrl);
// Extract related videos
const relatedVideos = await extractRelatedVideos(document, videoUrl);
const video = {
id: videoId,
year,
url: videoUrl,
title,
speakers,
duration,
topics,
hasTranscript,
hasCode,
transcript: hasTranscript ? transcript : undefined,
codeExamples: hasCode ? codeExamples : undefined,
chapters: chapters.length > 0 ? chapters : undefined,
resources,
relatedVideos: relatedVideos.length > 0 ? relatedVideos : undefined,
extractedAt: new Date().toISOString(),
};
return video;
}
catch (error) {
logger.error(`Failed to extract video ${videoId}:`, error);
throw error;
}
}
/**
* Extract title
*/
function extractTitle(document) {
// Try multiple selectors
const selectors = [
'h1.video-title',
'h1[itemprop="name"]',
'.hero-title',
'h1',
];
for (const selector of selectors) {
const element = document.querySelector(selector);
if (element?.textContent) {
return element.textContent.trim();
}
}
return 'Untitled Video';
}
/**
* Extract description
*/
function extractDescription(document) {
const selectors = [
'.video-description',
'.description',
'[itemprop="description"]',
'meta[name="description"]',
];
for (const selector of selectors) {
const element = document.querySelector(selector);
if (element) {
if (selector.includes('meta')) {
return element.getAttribute('content') || '';
}
return element.textContent?.trim() || '';
}
}
return '';
}
/**
* Extract speakers
*/
function extractSpeakers(document) {
const speakers = [];
// Try to extract from description
const description = document.querySelector('.video-description, .description');
if (description) {
const text = description.textContent || '';
// Match "with [Speaker]" or "by [Speaker]" patterns
const speakerMatch = text.match(/(?:with|by)\s+([^.]+?)(?:\.|$)/i);
if (speakerMatch) {
const speakerText = speakerMatch[1];
// Split multiple speakers
speakers.push(...speakerText.split(/,\s*and\s*|,\s*|\s+and\s+/).map((s) => s.trim()));
}
}
return speakers.filter(s => s.length > 0);
}
/**
* Extract duration
*/
function extractDuration(document) {
const selectors = [
'.video-duration',
'[itemprop="duration"]',
'.duration',
];
for (const selector of selectors) {
const element = document.querySelector(selector);
if (element?.textContent) {
return element.textContent.trim();
}
}
return '';
}
/**
* Detect if there's a Code tab
*/
function detectCodeTab(document, year) {
// WWDC22 and later typically have Code tabs
if (parseInt(year) >= WWDC_CONFIG.CODE_TAB_INTRODUCED_YEAR) {
const codeTab = document.querySelector('[data-supplement="code"], .code-tab, #code');
return codeTab !== null;
}
return false;
}
/**
* Extract transcript
*/
async function extractTranscript(document, videoUrl) {
try {
// Find transcript container
const transcriptContainer = document.querySelector('.transcript, [data-supplement="transcript"], #transcript');
if (!transcriptContainer) {
// If page doesn't have transcript, try loading transcript tab
const transcriptUrl = videoUrl.replace(/\/$/, '') + '/transcript';
const transcriptResponse = await httpClient.get(transcriptUrl).catch(() => null);
const transcriptHtml = transcriptResponse ? await transcriptResponse.text() : null;
if (transcriptHtml) {
const transcriptDom = new JSDOM(transcriptHtml);
const transcriptDoc = transcriptDom.window.document;
return extractTranscriptFromDocument(transcriptDoc);
}
return undefined;
}
return extractTranscriptFromDocument(document);
}
catch (error) {
logger.warn('Failed to extract transcript:', error);
return undefined;
}
}
/**
* Extract transcript data from document
*/
function extractTranscriptFromDocument(document) {
const segments = [];
let fullText = '';
// Find all transcript paragraphs
const paragraphs = document.querySelectorAll('.transcript p, .transcript-line, [data-timestamp]');
paragraphs.forEach((p) => {
const timestamp = p.getAttribute('data-timestamp') ||
p.querySelector('.timestamp')?.textContent || '';
const text = p.textContent?.replace(/^\d+:\d+\s*/, '').trim() || '';
if (text) {
if (timestamp) {
segments.push({ timestamp, text });
}
fullText += text + '\n\n';
}
});
return {
fullText: fullText.trim(),
segments,
};
}
/**
* Extract code examples (new page format)
*/
async function extractCodeExamples(_document, videoUrl) {
const examples = [];
try {
// Try to load code tab page
const codeUrl = videoUrl.replace(/\/$/, '') + '/code';
logger.info(`Fetching code page: ${codeUrl}`);
const codeResponse = await httpClient.get(codeUrl).catch((err) => {
logger.warn(`Failed to fetch code page: ${err.message}`);
return null;
});
const codeHtml = codeResponse ? await codeResponse.text() : null;
if (codeHtml) {
logger.info(`Code page loaded, length: ${codeHtml.length}`);
const codeDom = new JSDOM(codeHtml);
const codeDoc = codeDom.window.document;
// Find all code blocks - WWDC2025 uses pre.code-source
const codeBlocks = codeDoc.querySelectorAll('pre.code-source, .code-listing, pre code, .code-sample');
logger.info(`Found ${codeBlocks.length} code blocks`);
codeBlocks.forEach((block) => {
// Get code content
const codeEl = block.querySelector('code') || block;
let code = codeEl.textContent || '';
// Clean up code indentation
code = cleanCodeIndentation(code);
if (!code || code.length < 10) {
return;
} // Skip too short code
// Find title and timestamp - usually in previous sibling element
let title = '';
let timestamp = '';
const prevSibling = block.previousElementSibling;
if (prevSibling) {
const text = prevSibling.textContent || '';
// Extract timestamp format like "11:02 - Speech Transcriber setup"
const match = text.match(/^(\d+:\d+)\s*-\s*(.+)$/);
if (match) {
timestamp = match[1];
title = match[2].trim();
}
else {
title = text.trim();
}
}
// If not found, try other methods
if (!timestamp) {
timestamp = block.closest('[data-timestamp]')?.getAttribute('data-timestamp') || '';
}
const language = detectLanguage(block);
examples.push({
timestamp,
title,
language,
code,
});
});
// Remove duplicates - sometimes same code block appears multiple times
const uniqueExamples = removeDuplicateCodeExamples(examples);
logger.info(`Extracted ${uniqueExamples.length} unique code examples (from ${examples.length} total)`);
return uniqueExamples;
}
else {
logger.warn('No code page HTML received');
}
return examples;
}
catch (error) {
logger.error('Failed to extract code examples:', error);
}
logger.info(`Returning ${examples.length} code examples`);
return examples;
}
/**
* Extract code from transcript (legacy pages)
*/
function extractCodeFromTranscript(document) {
const examples = [];
// Find code blocks in transcript
const codeBlocks = document.querySelectorAll('.transcript pre code, .transcript .code-voice');
codeBlocks.forEach((block) => {
let code = block.textContent || '';
code = cleanCodeIndentation(code);
if (code && code.length > 20) { // Filter out too short code snippets
examples.push({
language: detectLanguage(block),
code,
});
}
});
return examples;
}
/**
* Detect code language
*/
function detectLanguage(element) {
// Detect from class attribute
const className = element.className;
if (className.includes('swift')) {
return 'swift';
}
if (className.includes('objc') || className.includes('objective-c')) {
return 'objc';
}
if (className.includes('javascript') || className.includes('js')) {
return 'javascript';
}
// Detect from data attributes
const dataLang = element.getAttribute('data-language') ||
element.getAttribute('data-lang');
if (dataLang) {
return dataLang.toLowerCase();
}
// Default to Swift (most WWDC code is Swift)
return 'swift';
}
/**
* Extract chapter information
*/
function extractChapters(document) {
const chapters = [];
const chapterElements = document.querySelectorAll('.chapter, .timeline-chapter, [data-chapter]');
chapterElements.forEach((element) => {
const title = element.querySelector('.chapter-title')?.textContent ||
element.textContent?.trim() || '';
const timestamp = element.getAttribute('data-timestamp') ||
element.querySelector('.timestamp')?.textContent || '';
if (title && timestamp) {
chapters.push({ title, timestamp });
}
});
return chapters;
}
/**
* Extract resource links
*/
async function extractResources(document, videoUrl) {
const resources = {
resourceLinks: [],
};
try {
// Try to extract resources section from page
const resourceSection = document.querySelector('.resources-section, [data-resources], #resources');
if (!resourceSection) {
// If page doesn't have resources section, try loading resources tab
const resourcesUrl = videoUrl.replace(/\/$/, '') + '/resources';
logger.info(`Attempting to load resources page: ${resourcesUrl}`);
const resourcesResponse = await httpClient.get(resourcesUrl).catch(() => null);
if (resourcesResponse) {
const resourcesHtml = await resourcesResponse.text();
const resourcesDom = new JSDOM(resourcesHtml);
const resourcesDoc = resourcesDom.window.document;
return extractResourcesFromDocument(resourcesDoc);
}
}
else {
return extractResourcesFromDocument(document);
}
}
catch (error) {
logger.warn('Failed to extract resources:', error);
}
// Fallback: at least try to extract video download links
const downloadLinks = document.querySelectorAll('a[href*=".mp4"], a[download]');
downloadLinks.forEach((link) => {
const href = link.getAttribute('href') || '';
const text = link.textContent?.toLowerCase() || '';
if (href.includes('_hd.mp4') || text.includes('hd')) {
resources.hdVideo = href.startsWith('http') ? href : `${WWDC_URLS.BASE.replace('/videos', '')}${href}`;
}
else if (href.includes('_sd.mp4') || text.includes('sd')) {
resources.sdVideo = href.startsWith('http') ? href : `${WWDC_URLS.BASE.replace('/videos', '')}${href}`;
}
});
return resources;
}
/**
* Extract resources from document
*/
function extractResourcesFromDocument(document) {
const resources = {
resourceLinks: [],
};
// Extract all resource links
const resourceLinks = document.querySelectorAll('a[href]');
const processedUrls = new Set();
resourceLinks.forEach((link) => {
const href = link.getAttribute('href') || '';
const text = link.textContent?.trim() || '';
if (!href || !text) {
return;
}
// Build complete URL
const fullUrl = href.startsWith('http') ? href : `${WWDC_URLS.BASE.replace('/videos', '')}${href}`;
// Avoid duplicates
if (processedUrls.has(fullUrl)) {
return;
}
processedUrls.add(fullUrl);
// Identify different types of resources
if (href.includes('_hd.mp4') || text.toLowerCase().includes('hd video')) {
resources.hdVideo = fullUrl;
}
else if (href.includes('_sd.mp4') || text.toLowerCase().includes('sd video')) {
resources.sdVideo = fullUrl;
}
else if (href.includes('sample-code') || href.endsWith('.zip')) {
resources.sampleProject = fullUrl;
}
else if (href.includes('/documentation/') ||
href.includes('/design/') ||
href.includes('/technotes/') ||
href.includes('/forums/') ||
text.includes('Human Interface Guidelines') ||
text.includes('Documentation') ||
text.includes('Forum')) {
// Add to resource links list
resources.resourceLinks.push({
title: text,
url: fullUrl,
});
}
});
return resources;
}
/**
* Extract related videos
*/
async function extractRelatedVideos(document, videoUrl) {
const relatedVideos = [];
try {
// Find related videos section
const relatedSection = document.querySelector('.related-videos, [data-related], #related');
if (!relatedSection) {
// Try to find related video links from page bottom
const videoLinks = document.querySelectorAll('a[href*="/videos/play/wwdc"]');
const currentVideoId = videoUrl.match(/\/(\d+)\/?$/)?.[1];
videoLinks.forEach((link) => {
const href = link.getAttribute('href') || '';
const match = href.match(/\/videos\/play\/wwdc(\d{4})\/(\d+)\/?/);
if (match && match[2] !== currentVideoId) {
const year = match[1];
const id = match[2];
const title = link.textContent?.trim() || '';
// Filter out current video and already added videos
if (title && !relatedVideos.find(v => v.id === id)) {
relatedVideos.push({
id,
year,
title,
url: href.startsWith('http') ? href : `${WWDC_URLS.BASE.replace('/videos', '')}${href}`,
});
}
}
});
}
}
catch (error) {
logger.warn('Failed to extract related videos:', error);
}
return relatedVideos;
}
/**
* Clean up code indentation
*/
function cleanCodeIndentation(code) {
const lines = code.split('\n');
// Remove empty lines at start and end
while (lines.length > 0 && lines[0].trim() === '') {
lines.shift();
}
while (lines.length > 0 && lines[lines.length - 1].trim() === '') {
lines.pop();
}
if (lines.length === 0)
return '';
// Find minimum indentation (excluding empty lines)
const minIndent = lines
.filter(line => line.trim().length > 0)
.reduce((min, line) => {
const indent = line.match(/^(\s*)/)?.[1].length || 0;
return Math.min(min, indent);
}, Infinity);
// Remove common indentation
if (minIndent > 0 && minIndent !== Infinity) {
return lines
.map(line => line.substring(minIndent))
.join('\n')
.trim();
}
return lines.join('\n').trim();
}
/**
* Remove duplicate code examples
*/
function removeDuplicateCodeExamples(examples) {
const seen = new Map();
for (const example of examples) {
// Use code content as key
const key = example.code.trim();
// If haven't seen this code, or current one has better metadata (title or timestamp)
const existing = seen.get(key);
if (!existing ||
(example.title && !existing.title) ||
(example.timestamp && !existing.timestamp)) {
seen.set(key, example);
}
}
return Array.from(seen.values());
}
//# sourceMappingURL=content-extractor.js.map