` text. */\nexport const extractTitle = (html: string): string => {\n const match = html.match(/<title[^>]*>([\\s\\S]*?)<\\/title>/i);\n return match ? match[1].trim() : '';\n};\n\n/** Extract the `<meta name=\"description\">` content. */\nexport const extractMetaDescription = (html: string): string => {\n const metas = html.match(/<meta\\b[^>]*>/gi) ?? [];\n for (const meta of metas) {\n if (/name\\s*=\\s*(\"|')?description\\1?/i.test(meta)) {\n return readAttribute(meta, 'content') ?? '';\n }\n }\n return '';\n};\n\n/** Extract the `<meta property=\"og:image\">` content. */\nexport const extractOgImage = (html: string): string | undefined => {\n const metas = html.match(/<meta\\b[^>]*>/gi) ?? [];\n for (const meta of metas) {\n if (/property\\s*=\\s*(\"|')?og:image\\1?/i.test(meta)) {\n return readAttribute(meta, 'content');\n }\n }\n return undefined;\n};\n\n/** Whether a `<link rel=\"canonical\">` element is present. */\nexport const hasCanonical = (html: string): boolean => {\n const links = html.match(/<link\\b[^>]*>/gi) ?? [];\n return links.some((link) => /rel\\s*=\\s*(\"|')?canonical\\1?/i.test(link));\n};\n\n/** A parsed `<link rel=\"alternate\" hreflang>` element. */\nexport type HreflangLink = { hreflang: string; href: string };\n\n/** Extract every `<link rel=\"alternate\" hreflang=\"…\" href=\"…\">` element. */\nexport const extractHreflangs = (html: string): HreflangLink[] => {\n const links = html.match(/<link\\b[^>]*>/gi) ?? [];\n const result: HreflangLink[] = [];\n for (const link of links) {\n if (!/rel\\s*=\\s*(\"|')?alternate\\1?/i.test(link)) continue;\n const hreflang = readAttribute(link, 'hreflang');\n const href = readAttribute(link, 'href');\n if (hreflang && href) result.push({ hreflang, href });\n }\n return result;\n};\n\n/**\n * Extract every eagerly-loaded script URL: `<script src>`,\n * `<link rel=\"modulepreload\">` and `<link rel=\"preload\" as=\"script\">`.\n *\n * @param html - The raw HTML document.\n * @param baseUrl - Base URL used to resolve relative script URLs.\n * @returns Absolute, de-duplicated script URLs.\n */\nexport const extractScriptUrls = (html: string, baseUrl: string): string[] => {\n const urls = new Set<string>();\n\n const add = (raw: string | undefined) => {\n if (!raw) return;\n try {\n urls.add(new URL(raw, baseUrl).href);\n } catch {\n /* ignore malformed URLs */\n }\n };\n\n for (const script of html.match(/<script\\b[^>]*>/gi) ?? []) {\n add(readAttribute(script, 'src'));\n }\n\n for (const link of html.match(/<link\\b[^>]*>/gi) ?? []) {\n const rel = readAttribute(link, 'rel')?.toLowerCase();\n const as = readAttribute(link, 'as')?.toLowerCase();\n if (rel === 'modulepreload' || (rel === 'preload' && as === 'script')) {\n add(readAttribute(link, 'href'));\n }\n }\n\n return Array.from(urls);\n};\n\n/** A parsed `<a href>` anchor. */\nexport type Anchor = { href: string; text: string };\n\n/** Extract every `<a href=\"…\">text</a>` anchor from the document. */\nexport const extractAnchors = (html: string): Anchor[] => {\n const anchors: Anchor[] = [];\n const anchorPattern = /<a\\b([^>]*)>([\\s\\S]*?)<\\/a>/gi;\n let match = anchorPattern.exec(html);\n while (match !== null) {\n const href = readAttribute(match[1], 'href');\n if (href) {\n const text = match[2]\n .replace(/<[^>]+>/g, ' ')\n .replace(/\\s+/g, ' ')\n .trim();\n anchors.push({ href, text });\n }\n match = anchorPattern.exec(html);\n }\n return anchors;\n};\n\n/**\n * Extract visible text snippets from an HTML document (scripts, styles and\n * tags stripped). Used to approximate the rendered content size without a DOM.\n */\nexport const extractVisibleTextStrings = (html: string): string[] => {\n const withoutNonVisible = html\n .replace(/<script[\\s\\S]*?<\\/script>/gi, ' ')\n .replace(/<style[\\s\\S]*?<\\/style>/gi, ' ')\n .replace(/<noscript[\\s\\S]*?<\\/noscript>/gi, ' ')\n .replace(//g, ' ');\n\n return withoutNonVisible\n .replace(/<[^>]+>/g, '\\n')\n .split('\\n')\n .map((line) => line.replace(/\\s+/g, ' ').trim())\n .filter((line) => line.length > 1);\n};\n"],"mappings":";;;;;;;;;;;AAUA,MAAa,cAAc,SACzB,OAAO,WAAW,cACd,OAAO,WAAW,MAAM,OAAO,IAC/B,IAAI,YAAY,CAAC,CAAC,OAAO,IAAI,CAAC,CAAC;;AAGrC,MAAM,iBACJ,YACA,kBACuB;CACvB,MAAM,QAAQ,WAAW,MACvB,IAAI,OAAO,GAAG,cAAc,4CAA4C,GAAG,CAC7E;CACA,IAAI,CAAC,OAAO,OAAO;CACnB,OAAO,MAAM,MAAM,MAAM,MAAM,MAAM;AACvC;;AAGA,MAAa,mBAAmB,SAAqC;CACnE,MAAM,UAAU,KAAK,MAAM,kBAAkB;CAC7C,OAAO,UAAU,cAAc,QAAQ,IAAI,MAAM,IAAI;AACvD;;AAGA,MAAa,kBAAkB,SAAqC;CAClE,MAAM,UAAU,KAAK,MAAM,kBAAkB;CAC7C,OAAO,UAAU,cAAc,QAAQ,IAAI,KAAK,IAAI;AACtD;;AAGA,MAAa,gBAAgB,SAAyB;CACpD,MAAM,QAAQ,KAAK,MAAM,kCAAkC;CAC3D,OAAO,QAAQ,MAAM,EAAE,CAAC,KAAK,IAAI;AACnC;;AAGA,MAAa,0BAA0B,SAAyB;CAC9D,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,KAAK,MAAM,QAAQ,OACjB,IAAI,mCAAmC,KAAK,IAAI,GAC9C,OAAO,cAAc,MAAM,SAAS,KAAK;CAG7C,OAAO;AACT;;AAGA,MAAa,kBAAkB,SAAqC;CAClE,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,KAAK,MAAM,QAAQ,OACjB,IAAI,oCAAoC,KAAK,IAAI,GAC/C,OAAO,cAAc,MAAM,SAAS;AAI1C;;AAGA,MAAa,gBAAgB,SAA0B;CAErD,QADc,KAAK,MAAM,iBAAiB,KAAK,CAAC,EACpC,CAAC,MAAM,SAAS,gCAAgC,KAAK,IAAI,CAAC;AACxE;;AAMA,MAAa,oBAAoB,SAAiC;CAChE,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,MAAM,SAAyB,CAAC;CAChC,KAAK,MAAM,QAAQ,OAAO;EACxB,IAAI,CAAC,gCAAgC,KAAK,IAAI,GAAG;EACjD,MAAM,WAAW,cAAc,MAAM,UAAU;EAC/C,MAAM,OAAO,cAAc,MAAM,MAAM;EACvC,IAAI,YAAY,MAAM,OAAO,KAAK;GAAE;GAAU;EAAK,CAAC;CACtD;CACA,OAAO;AACT;;;;;;;;;AAUA,MAAa,qBAAqB,MAAc,YAA8B;CAC5E,MAAM,uBAAO,IAAI,IAAY;CAE7B,MAAM,OAAO,QAA4B;EACvC,IAAI,CAAC,KAAK;EACV,IAAI;GACF,KAAK,IAAI,IAAI,IAAI,KAAK,OAAO,CAAC,CAAC,IAAI;EACrC,QAAQ,CAER;CACF;CAEA,KAAK,MAAM,UAAU,KAAK,MAAM,mBAAmB,KAAK,CAAC,GACvD,IAAI,cAAc,QAAQ,KAAK,CAAC;CAGlC,KAAK,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC,GAAG;EACtD,MAAM,MAAM,cAAc,MAAM,KAAK,CAAC,EAAE,YAAY;EACpD,MAAM,KAAK,cAAc,MAAM,IAAI,CAAC,EAAE,YAAY;EAClD,IAAI,QAAQ,mBAAoB,QAAQ,aAAa,OAAO,UAC1D,IAAI,cAAc,MAAM,MAAM,CAAC;CAEnC;CAEA,OAAO,MAAM,KAAK,IAAI;AACxB;;AAMA,MAAa,kBAAkB,SAA2B;CACxD,MAAM,UAAoB,CAAC;CAC3B,MAAM,gBAAgB;CACtB,IAAI,QAAQ,cAAc,KAAK,IAAI;CACnC,OAAO,UAAU,MAAM;EACrB,MAAM,OAAO,cAAc,MAAM,IAAI,MAAM;EAC3C,IAAI,MAAM;GACR,MAAM,OAAO,MAAM,EAAE,CAClB,QAAQ,YAAY,GAAG,CAAC,CACxB,QAAQ,QAAQ,GAAG,CAAC,CACpB,KAAK;GACR,QAAQ,KAAK;IAAE;IAAM;GAAK,CAAC;EAC7B;EACA,QAAQ,cAAc,KAAK,IAAI;CACjC;CACA,OAAO;AACT;;;;;AAMA,MAAa,6BAA6B,SAA2B;CAOnE,OAN0B,KACvB,QAAQ,+BAA+B,GAAG,CAAC,CAC3C,QAAQ,6BAA6B,GAAG,CAAC,CACzC,QAAQ,mCAAmC,GAAG,CAAC,CAC/C,QAAQ,oBAAoB,GAER,CAAC,CACrB,QAAQ,YAAY,IAAI,CAAC,CACzB,MAAM,IAAI,CAAC,CACX,KAAK,SAAS,KAAK,QAAQ,QAAQ,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAC/C,QAAQ,SAAS,KAAK,SAAS,CAAC;AACrC"}

{"version":3,"file":"parseHtml.cjs","names":[],"sources":["../../../src/scan/parseHtml.ts"],"sourcesContent":["/**\n * Tiny dependency-free HTML extraction helpers.\n *\n * The hosted backend audit relies on Cheerio + a real browser, but the CLI scan\n * must stay dependency-light. These regex-based helpers cover the handful of\n * head/anchor signals the score needs. They are intentionally forgiving: when a\n * tag can't be parsed it is simply skipped rather than throwing.\n */\n\n/** Compute the UTF-8 byte length of a string in both Node and browser builds. */\nexport const byteLength = (text: string): number =>\n typeof Buffer !== 'undefined'\n ? Buffer.byteLength(text, 'utf-8')\n : new TextEncoder().encode(text).length;\n\n/** Read an attribute value off a single tag's attribute string. */\nconst readAttribute = (\n attributes: string,\n attributeName: string\n): string | undefined => {\n const match = attributes.match(\n new RegExp(`${attributeName}\\\\s*=\\\\s*(\"([^\"]*)\"|'([^']*)'|([^\\\\s>]+))`, 'i')\n );\n if (!match) return undefined;\n return match[2] ?? match[3] ?? match[4];\n};\n\n/** Extract the `lang` attribute of the `` element, if present. */\nexport const extractHtmlLang = (html: string): string | undefined => {\n const htmlTag = html.match(/]*)>/i);\n return htmlTag ? readAttribute(htmlTag[1], 'lang') : undefined;\n};\n\n/** Extract the `dir` attribute of the `` element, if present. */\nexport const extractHtmlDir = (html: string): string | undefined => {\n const htmlTag = html.match(/]*)>/i);\n return htmlTag ? readAttribute(htmlTag[1], 'dir') : undefined;\n};\n\n/** Extract the document `` text. */\nexport const extractTitle = (html: string): string => {\n const match = html.match(/<title[^>]*>([\\s\\S]*?)<\\/title>/i);\n return match ? match[1].trim() : '';\n};\n\n/** Extract the `<meta name=\"description\">` content. */\nexport const extractMetaDescription = (html: string): string => {\n const metas = html.match(/<meta\\b[^>]*>/gi) ?? [];\n for (const meta of metas) {\n if (/name\\s*=\\s*(\"|')?description\\1?/i.test(meta)) {\n return readAttribute(meta, 'content') ?? '';\n }\n }\n return '';\n};\n\n/** Extract the `<meta property=\"og:image\">` content. */\nexport const extractOgImage = (html: string): string | undefined => {\n const metas = html.match(/<meta\\b[^>]*>/gi) ?? [];\n for (const meta of metas) {\n if (/property\\s*=\\s*(\"|')?og:image\\1?/i.test(meta)) {\n return readAttribute(meta, 'content');\n }\n }\n return undefined;\n};\n\n/** Whether a `<link rel=\"canonical\">` element is present. */\nexport const hasCanonical = (html: string): boolean => {\n const links = html.match(/<link\\b[^>]*>/gi) ?? [];\n return links.some((link) => /rel\\s*=\\s*(\"|')?canonical\\1?/i.test(link));\n};\n\n/** A parsed `<link rel=\"alternate\" hreflang>` element. */\nexport type HreflangLink = { hreflang: string; href: string };\n\n/** Extract every `<link rel=\"alternate\" hreflang=\"…\" href=\"…\">` element. */\nexport const extractHreflangs = (html: string): HreflangLink[] => {\n const links = html.match(/<link\\b[^>]*>/gi) ?? [];\n const result: HreflangLink[] = [];\n for (const link of links) {\n if (!/rel\\s*=\\s*(\"|')?alternate\\1?/i.test(link)) continue;\n const hreflang = readAttribute(link, 'hreflang');\n const href = readAttribute(link, 'href');\n if (hreflang && href) result.push({ hreflang, href });\n }\n return result;\n};\n\n/**\n * Extract every eagerly-loaded script URL: `<script src>`,\n * `<link rel=\"modulepreload\">` and `<link rel=\"preload\" as=\"script\">`.\n *\n * @param html - The raw HTML document.\n * @param baseUrl - Base URL used to resolve relative script URLs.\n * @returns Absolute, de-duplicated script URLs.\n */\nexport const extractScriptUrls = (html: string, baseUrl: string): string[] => {\n const urls = new Set<string>();\n\n const add = (raw: string | undefined) => {\n if (!raw) return;\n try {\n urls.add(new URL(raw, baseUrl).href);\n } catch {\n /* ignore malformed URLs */\n }\n };\n\n for (const script of html.match(/<script\\b[^>]*>/gi) ?? []) {\n add(readAttribute(script, 'src'));\n }\n\n for (const link of html.match(/<link\\b[^>]*>/gi) ?? []) {\n const rel = readAttribute(link, 'rel')?.toLowerCase();\n const as = readAttribute(link, 'as')?.toLowerCase();\n if (rel === 'modulepreload' || (rel === 'preload' && as === 'script')) {\n add(readAttribute(link, 'href'));\n }\n }\n\n return Array.from(urls);\n};\n\n/** A parsed `<a href>` anchor. */\nexport type Anchor = { href: string; text: string };\n\n/** Extract every `<a href=\"…\">text</a>` anchor from the document. */\nexport const extractAnchors = (html: string): Anchor[] => {\n const anchors: Anchor[] = [];\n const anchorPattern = /<a\\b([^>]*)>([\\s\\S]*?)<\\/a>/gi;\n let match = anchorPattern.exec(html);\n while (match !== null) {\n const href = readAttribute(match[1], 'href');\n if (href) {\n const text = match[2]\n .replace(/<[^>]+>/g, ' ')\n .replace(/\\s+/g, ' ')\n .trim();\n anchors.push({ href, text });\n }\n match = anchorPattern.exec(html);\n }\n return anchors;\n};\n\n/**\n * Extract visible text snippets from an HTML document (scripts, styles and\n * tags stripped). Used to approximate the rendered content size without a DOM.\n */\nexport const extractVisibleTextStrings = (html: string): string[] => {\n const withoutNonVisible = html\n .replace(/<script[\\s\\S]*?<\\/script>/gi, ' ')\n .replace(/<style[\\s\\S]*?<\\/style>/gi, ' ')\n .replace(/<noscript[\\s\\S]*?<\\/noscript>/gi, ' ')\n .replace(//g, ' ');\n\n return withoutNonVisible\n .replace(/<[^>]+>/g, '\\n')\n .split('\\n')\n .map((line) => line.replace(/\\s+/g, ' ').trim())\n .filter((line) => line.length > 1);\n};\n"],"mappings":";;;;;;;;;;;AAUA,MAAa,cAAc,SACzB,OAAO,WAAW,cACd,OAAO,WAAW,MAAM,OAAO,IAC/B,IAAI,YAAY,CAAC,CAAC,OAAO,IAAI,CAAC,CAAC;;AAGrC,MAAM,iBACJ,YACA,kBACuB;CACvB,MAAM,QAAQ,WAAW,MACvB,IAAI,OAAO,GAAG,cAAc,4CAA4C,GAAG,CAC7E;CACA,IAAI,CAAC,OAAO,OAAO;CACnB,OAAO,MAAM,MAAM,MAAM,MAAM,MAAM;AACvC;;AAGA,MAAa,mBAAmB,SAAqC;CACnE,MAAM,UAAU,KAAK,MAAM,kBAAkB;CAC7C,OAAO,UAAU,cAAc,QAAQ,IAAI,MAAM,IAAI;AACvD;;AAGA,MAAa,kBAAkB,SAAqC;CAClE,MAAM,UAAU,KAAK,MAAM,kBAAkB;CAC7C,OAAO,UAAU,cAAc,QAAQ,IAAI,KAAK,IAAI;AACtD;;AAGA,MAAa,gBAAgB,SAAyB;CACpD,MAAM,QAAQ,KAAK,MAAM,kCAAkC;CAC3D,OAAO,QAAQ,MAAM,EAAE,CAAC,KAAK,IAAI;AACnC;;AAGA,MAAa,0BAA0B,SAAyB;CAC9D,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,KAAK,MAAM,QAAQ,OACjB,IAAI,mCAAmC,KAAK,IAAI,GAC9C,OAAO,cAAc,MAAM,SAAS,KAAK;CAG7C,OAAO;AACT;;AAGA,MAAa,kBAAkB,SAAqC;CAClE,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,KAAK,MAAM,QAAQ,OACjB,IAAI,oCAAoC,KAAK,IAAI,GAC/C,OAAO,cAAc,MAAM,SAAS;AAI1C;;AAGA,MAAa,gBAAgB,SAA0B;CAErD,QADc,KAAK,MAAM,iBAAiB,KAAK,CAAC,EACpC,CAAC,MAAM,SAAS,gCAAgC,KAAK,IAAI,CAAC;AACxE;;AAMA,MAAa,oBAAoB,SAAiC;CAChE,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC;CAChD,MAAM,SAAyB,CAAC;CAChC,KAAK,MAAM,QAAQ,OAAO;EACxB,IAAI,CAAC,gCAAgC,KAAK,IAAI,GAAG;EACjD,MAAM,WAAW,cAAc,MAAM,UAAU;EAC/C,MAAM,OAAO,cAAc,MAAM,MAAM;EACvC,IAAI,YAAY,MAAM,OAAO,KAAK;GAAE;GAAU;EAAK,CAAC;CACtD;CACA,OAAO;AACT;;;;;;;;;AAUA,MAAa,qBAAqB,MAAc,YAA8B;CAC5E,MAAM,uBAAO,IAAI,IAAY;CAE7B,MAAM,OAAO,QAA4B;EACvC,IAAI,CAAC,KAAK;EACV,IAAI;GACF,KAAK,IAAI,IAAI,IAAI,KAAK,OAAO,CAAC,CAAC,IAAI;EACrC,QAAQ,CAER;CACF;CAEA,KAAK,MAAM,UAAU,KAAK,MAAM,mBAAmB,KAAK,CAAC,GACvD,IAAI,cAAc,QAAQ,KAAK,CAAC;CAGlC,KAAK,MAAM,QAAQ,KAAK,MAAM,iBAAiB,KAAK,CAAC,GAAG;EACtD,MAAM,MAAM,cAAc,MAAM,KAAK,CAAC,EAAE,YAAY;EACpD,MAAM,KAAK,cAAc,MAAM,IAAI,CAAC,EAAE,YAAY;EAClD,IAAI,QAAQ,mBAAoB,QAAQ,aAAa,OAAO,UAC1D,IAAI,cAAc,MAAM,MAAM,CAAC;CAEnC;CAEA,OAAO,MAAM,KAAK,IAAI;AACxB;;AAMA,MAAa,kBAAkB,SAA2B;CACxD,MAAM,UAAoB,CAAC;CAC3B,MAAM,gBAAgB;CACtB,IAAI,QAAQ,cAAc,KAAK,IAAI;CACnC,OAAO,UAAU,MAAM;EACrB,MAAM,OAAO,cAAc,MAAM,IAAI,MAAM;EAC3C,IAAI,MAAM;GACR,MAAM,OAAO,MAAM,EAAE,CAClB,QAAQ,YAAY,GAAG,CAAC,CACxB,QAAQ,QAAQ,GAAG,CAAC,CACpB,KAAK;GACR,QAAQ,KAAK;IAAE;IAAM;GAAK,CAAC;EAC7B;EACA,QAAQ,cAAc,KAAK,IAAI;CACjC;CACA,OAAO;AACT;;;;;AAMA,MAAa,6BAA6B,SAA2B;CAOnE,OAN0B,KACvB,QAAQ,+BAA+B,GAAG,CAAC,CAC3C,QAAQ,6BAA6B,GAAG,CAAC,CACzC,QAAQ,mCAAmC,GAAG,CAAC,CAC/C,QAAQ,oBAAoB,GAER,CAAC,CACrB,QAAQ,YAAY,IAAI,CAAC,CACzB,MAAM,IAAI,CAAC,CACX,KAAK,SAAS,KAAK,QAAQ,QAAQ,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAC/C,QAAQ,SAAS,KAAK,SAAS,CAAC;AACrC"}