magpie-html 0.2.2 → 0.2.3

package/dist/index.cjs

@@ -3101,6 +3101,12 @@ function parseHTML(html, baseUrl) {
   });
   return document;
 }
+function ensureDocument(input, baseUrl) {
+  if (typeof input === "string") {
+    return parseHTML(input, baseUrl);
+  }
+  return input;
+}
 
 // src/utils/meta-helpers.ts
 function getMetaContent(doc, name) {
@@ -3128,7 +3134,8 @@ function getMetaHttpEquiv(doc, httpEquiv) {
 }
 
 // src/metadata/opengraph/extract.ts
-function extractOpenGraph(doc) {
+function extractOpenGraph(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.title = getMetaProperty(doc, "og:title");
   metadata.type = getMetaProperty(doc, "og:type");
@@ -3341,7 +3348,8 @@ function matchesAnyType(obj, targetTypes) {
 }
 
 // src/metadata/schema-org/extract.ts
-function extractSchemaOrg(doc) {
+function extractSchemaOrg(input) {
+  const doc = ensureDocument(input);
   const metadata = {
     jsonLd: []
   };
@@ -3418,7 +3426,8 @@ function organizeByType(metadata) {
 }
 
 // src/metadata/seo/extract.ts
-function extractSEO(doc) {
+function extractSEO(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const titleElement = doc.querySelector("title");
   if (titleElement?.textContent) {
@@ -3450,7 +3459,8 @@ function extractSEO(doc) {
 }
 
 // src/metadata/twitter-card/extract.ts
-function extractTwitterCard(doc) {
+function extractTwitterCard(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.card = getMetaContent(doc, "twitter:card");
   metadata.site = getMetaContent(doc, "twitter:site");
@@ -3607,7 +3617,8 @@ function getAllLinksByPrefix(doc, relPrefix) {
 }
 
 // src/metadata/icons/extract.ts
-function extractIcons(doc) {
+function extractIcons(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const iconLinks = getAllLinksByRels(doc, ["icon", "shortcut icon"]);
   for (const link of iconLinks) {
@@ -3788,7 +3799,8 @@ function parseSizeString(sizeStr) {
 }
 
 // src/metadata/language/extract.ts
-function extractLanguage(doc) {
+function extractLanguage(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const htmlElement = doc.querySelector("html");
   if (htmlElement) {
@@ -3840,7 +3852,8 @@ function extractBestLanguage(doc) {
 }
 
 // src/metadata/links/extract.ts
-function extractLinks3(doc, baseUrl, options = {}) {
+function extractLinks3(input, baseUrl, options = {}) {
+  const doc = ensureDocument(input);
   const opts = normalizeOptions3(options);
   const effectiveBaseUrl = getEffectiveBaseUrl(doc, baseUrl);
   const baseOrigin = effectiveBaseUrl ? getOrigin(effectiveBaseUrl) : null;
@@ -4171,7 +4184,8 @@ function getStringProperty3(obj, prop) {
 }
 
 // src/metadata/canonical/extract.ts
-function extractCanonical(doc) {
+function extractCanonical(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.canonical = getLinkHref(doc, "canonical");
   const alternateLinks = getAllLinks(doc, "alternate");
@@ -4407,7 +4421,8 @@ function generateFeedSuggestions(documentUrl) {
 }
 
 // src/metadata/feed-discovery/extract.ts
-function extractFeedDiscovery(doc, documentUrl) {
+function extractFeedDiscovery(input, documentUrl) {
+  const doc = ensureDocument(input);
   const metadata = {
     feeds: []
   };
@@ -4584,7 +4599,8 @@ async function gatherWebsite(url) {
 }
 
 // src/metadata/analytics/extract.ts
-function extractAnalytics(doc) {
+function extractAnalytics(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const scripts = doc.querySelectorAll("script");
   const googleAnalytics = /* @__PURE__ */ new Set();
@@ -4676,7 +4692,8 @@ function extractAnalytics(doc) {
 }
 
 // src/metadata/assets/extract.ts
-function extractAssets(doc, baseUrl) {
+function extractAssets(input, baseUrl) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const effectiveBaseUrl = getEffectiveBaseUrl2(doc, baseUrl);
   const images = extractImages3(doc, effectiveBaseUrl);
@@ -5003,7 +5020,8 @@ function extractConnectionHints(doc, baseUrl) {
 }
 
 // src/metadata/copyright/extract.ts
-function extractCopyright(doc) {
+function extractCopyright(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.copyright = getMetaContent(doc, "copyright");
   metadata.license = getLinkHref(doc, "license");
@@ -5039,7 +5057,8 @@ function parseCopyright(copyrightString) {
 }
 
 // src/metadata/dublin-core/extract.ts
-function extractDublinCore(doc) {
+function extractDublinCore(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.title = getMetaContent(doc, "DC.title") || getMetaContent(doc, "dcterms.title");
   metadata.description = getMetaContent(doc, "DC.description") || getMetaContent(doc, "dcterms.description");
@@ -5080,7 +5099,8 @@ function extractMultiValue(doc, field) {
 }
 
 // src/metadata/geo/extract.ts
-function extractGeo(doc) {
+function extractGeo(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const geoPosition = getMetaContent(doc, "geo.position");
   if (geoPosition) {
@@ -5137,7 +5157,8 @@ function parseICBM(icbm) {
 }
 
 // src/metadata/monetization/extract.ts
-function extractMonetization(doc) {
+function extractMonetization(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.webMonetization = getMetaContent(doc, "monetization");
   metadata.paypalVerification = getMetaContent(doc, "paypal-site-verification");
@@ -5151,7 +5172,8 @@ function extractMonetization(doc) {
 }
 
 // src/metadata/news/extract.ts
-function extractNews2(doc) {
+function extractNews2(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const newsKeywords = getMetaContent(doc, "news_keywords");
   if (newsKeywords) {
@@ -5169,7 +5191,8 @@ function extractNews2(doc) {
 }
 
 // src/metadata/pagination/extract.ts
-function extractPagination(doc) {
+function extractPagination(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.prev = getLinkHref(doc, "prev") || getLinkHref(doc, "previous");
   metadata.next = getLinkHref(doc, "next");
@@ -5268,7 +5291,8 @@ function parseKeyValueDirective(key, value, result) {
 }
 
 // src/metadata/robots/extract.ts
-function extractRobots(doc) {
+function extractRobots(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   const robotsContent = getMetaContent(doc, "robots");
   if (robotsContent) {
@@ -5302,7 +5326,8 @@ function extractRobots(doc) {
 }
 
 // src/metadata/security/extract.ts
-function extractSecurity(doc) {
+function extractSecurity(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.referrerPolicy = getMetaContent(doc, "referrer");
   metadata.contentSecurityPolicy = getMetaHttpEquiv(doc, "Content-Security-Policy");
@@ -5353,7 +5378,8 @@ function generateSitemapSuggestions(documentUrl) {
 }
 
 // src/metadata/sitemap-discovery/extract.ts
-function extractSitemapDiscovery(doc, documentUrl) {
+function extractSitemapDiscovery(input, documentUrl) {
+  const doc = ensureDocument(input);
   const metadata = {
     sitemaps: []
   };
@@ -5366,7 +5392,8 @@ function extractSitemapDiscovery(doc, documentUrl) {
 }
 
 // src/metadata/social-profiles/extract.ts
-function extractSocialProfiles(doc) {
+function extractSocialProfiles(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.twitter = getMetaContent(doc, "twitter:site") || getMetaContent(doc, "twitter:creator") || extractFromProperty(doc, "twitter:site") || extractFromProperty(doc, "twitter:creator");
   if (metadata.twitter) {
@@ -5519,7 +5546,8 @@ function categorizeSchemaProfile(url, metadata) {
 }
 
 // src/metadata/verification/extract.ts
-function extractVerification(doc) {
+function extractVerification(input) {
+  const doc = ensureDocument(input);
   const metadata = {};
   metadata.googleSiteVerification = getMetaContent(doc, "google-site-verification");
   metadata.msvalidate = getMetaContent(doc, "msvalidate.01");