| 1 | import * as cheerio from 'cheerio' |
| 2 | |
| 3 | const NON_TEMPLATE_SKELETON_RESOURCE_RE = |
| 4 | /(?:^|\/)(?:(?:tailwindcss\.v3|anime\.v4|ppt-runtime|chart\.v4|katex(?:\.min)?|katex-auto-render\.min)\.(?:js|css)|assets\/fonts\/.+)(?:[?#].*)?$/i |
| 5 | const SKELETON_HINT_RE = |
| 6 | /\b(?:bg-|background|decor|decoration|texture|mask|overlay|ornament|pattern|backdrop)\b/i |
| 7 | |
| 8 | const normalizeTemplateResourceRef = (value: string): string | null => { |
| 9 | const raw = value.trim().replace(/^['"]|['"]$/g, '').trim() |
| 10 | if (!raw || raw.startsWith('#')) return null |
| 11 | const withoutQuery = raw.split('#')[0].split('?')[0].trim() |
| 12 | if ( |
| 13 | !withoutQuery || |
| 14 | /^data:/i.test(withoutQuery) || |
| 15 | /^blob:/i.test(withoutQuery) || |
| 16 | /^javascript:/i.test(withoutQuery) |
| 17 | ) { |
| 18 | return null |
| 19 | } |
| 20 | if (NON_TEMPLATE_SKELETON_RESOURCE_RE.test(withoutQuery)) return null |
| 21 | return withoutQuery.replace(/^\.\//, '') |
| 22 | } |
| 23 | |
| 24 | const collectTemplateSkeletonResourceRefs = (html: string): string[] => { |
| 25 | const refs = new Set<string>() |
| 26 | const push = (value: string | undefined | null): void => { |
| 27 | if (!value) return |
| 28 | const normalized = normalizeTemplateResourceRef(value) |
| 29 | if (normalized) refs.add(normalized) |
| 30 | } |
| 31 | |
| 32 | const urlRe = /url\(\s*(['"]?)([^'")]+)\1\s*\)/gi |
| 33 | let match: RegExpExecArray | null |
| 34 | while ((match = urlRe.exec(html)) !== null) { |
| 35 | push(match[2]) |
| 36 | } |
| 37 | |
| 38 | try { |
| 39 | const $ = cheerio.load(html, { scriptingEnabled: false }) |
| 40 | $('image').each((_, node) => { |
| 41 | const el = $(node) |
| 42 | push(el.attr('href') || el.attr('xlink:href')) |
| 43 | }) |
| 44 | $('img, video, source').each((_, node) => { |
| 45 | const el = $(node) |
| 46 | const identity = [ |
| 47 | el.attr('class') || '', |
| 48 | el.attr('style') || '', |
| 49 | el.parent().attr('class') || '', |
| 50 | el.parent().attr('style') || '' |
| 51 | ].join(' ') |
| 52 | if (!SKELETON_HINT_RE.test(identity)) return |
| 53 | push(el.attr('src') || el.attr('poster')) |
| 54 | }) |
| 55 | } catch { |
| 56 | // CSS url(...) extraction above still covers the most important template resources. |
| 57 | } |
| 58 | |
| 59 | return Array.from(refs).sort() |
| 60 | } |
| 61 | |
| 62 | export const validateTemplateSkeletonPreserved = (beforeHtml: string, afterHtml: string): string[] => { |
| 63 | const beforeRefs = collectTemplateSkeletonResourceRefs(beforeHtml) |
| 64 | if (beforeRefs.length === 0) return [] |
| 65 | const afterRefs = new Set(collectTemplateSkeletonResourceRefs(afterHtml)) |
| 66 | return beforeRefs.filter((ref) => !afterRefs.has(ref)) |
| 67 | } |
| 68 |