Add post processors

parser.js

@@ -1,6 +1,17 @@
+const urlparse = require('url');
 const {dom, rule, ruleset} = require('fathom-web');
 
-function buildRuleset(name, rules) {
+function makeUrlAbsolute(base, relative) {
+  const relativeParsed = urlparse.parse(relative);
+
+  if (relativeParsed.host === null) {
+    return urlparse.resolve(base, relative);
+  }
+
+  return relative;
+}
+
+function buildRuleset(name, rules, processors) {
   const reversedRules = Array.from(rules).reverse();
   const builtRuleset = ruleset(...reversedRules.map(([query, handler], order) => rule(
     dom(query),
@@ -11,11 +22,19 @@ function buildRuleset(name, rules) {
     }]
   )));
 
-  return doc => {
+  return (doc, context) => {
     const kb = builtRuleset.score(doc);
     const maxNode = kb.max(name);
+
     if (maxNode) {
-      const value = maxNode.flavors.get(name);
+      let value = maxNode.flavors.get(name);
+
+      if (processors) {
+        processors.forEach(processor => {
+          value = processor(value, context);
+        });
+      }
+
       if (value) {
         return value.trim();
       }
@@ -41,6 +60,9 @@ const metadataRules = {
       ['link[rel="Shortcut Icon"]', node => node.element.getAttribute('href')],
       ['link[rel="mask-icon"]', node => node.element.getAttribute('href')],
     ],
+    processors: [
+      (icon_url, context) => makeUrlAbsolute(context.url, icon_url)
+    ]
   },
 
   image_url: {
@@ -51,6 +73,9 @@ const metadataRules = {
       ['meta[property="twitter:image"]', node => node.element.getAttribute('content')],
       ['meta[name="thumbnail"]', node => node.element.getAttribute('content')],
     ],
+    processors: [
+      (image_url, context) => makeUrlAbsolute(context.url, image_url)
+    ],
   },
 
   keywords: {
@@ -82,15 +107,21 @@ const metadataRules = {
   },
 };
 
-function getMetadata(doc, rules) {
+function getMetadata(doc, rules, url) {
   const metadata = {};
   const ruleSet = rules || metadataRules;
 
+  const context = {};
+
+  if (url) {
+    context.url = url;
+  }
+
   Object.keys(ruleSet).map(metadataKey => {
     const metadataRule = ruleSet[metadataKey];
 
     if(Array.isArray(metadataRule.rules)) {
-      metadata[metadataKey] = buildRuleset(metadataKey, metadataRule.rules)(doc);
+      metadata[metadataKey] = buildRuleset(metadataKey, metadataRule.rules, metadataRule.processors)(doc, context);
     } else {
       metadata[metadataKey] = getMetadata(doc, metadataRule);
     }

tests/metadataRules.test.js

@@ -17,8 +17,10 @@ function ruleTest(testName, testRule, expected, testTag) {
   it(`finds ${testName}`, () => {
     const html = buildHTML(testTag);
     const doc = stringToDom(html);
-    const rule = buildRuleset(testName, testRule.rules);
-    const found = rule(doc);
+    const rule = buildRuleset(testName, testRule.rules, testRule.processors);
+    const found = rule(doc, {
+      url: 'http://www.example.com/'
+    });
     assert.equal(found, expected, `Unable to find ${testName} in ${html}`);
   });
 }
@@ -52,6 +54,7 @@ describe('Canonical URL Rule Tests', function() {
 
 describe('Icon Rule Tests', function() {
   const pageIcon = 'http://www.example.com/favicon.ico';
+  const relativeIcon = '/favicon.ico';
 
   const ruleTests = [
     ['apple-touch-icon', `<link rel="apple-touch-icon" href="${pageIcon}" />`],
@@ -61,6 +64,7 @@ describe('Icon Rule Tests', function() {
     ['shortcut icon', `<link rel="shortcut icon" href="${pageIcon}" />`],
     ['Shortcut Icon', `<link rel="Shortcut Icon" href="${pageIcon}" />`],
     ['mask-icon', `<link rel="mask-icon" href="${pageIcon}" />`],
+    ['relative icon', `<link rel="icon" href="${relativeIcon}" />`],
   ];
 
   ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.icon_url, pageIcon, testTag));
@@ -69,13 +73,15 @@ describe('Icon Rule Tests', function() {
 
 describe('Image Rule Tests', function() {
   const pageImage = 'http://www.example.com/image.png';
+  const relativeImage = '/image.png';
 
   const ruleTests = [
     ['og:image', `<meta property="og:image" content="${pageImage}" />`],
     ['og:image:url', `<meta property="og:image:url" content="${pageImage}" /> `],
     ['og:image:secure_url', `<meta property="og:image:secure_url" content="${pageImage}" /> `],
     ['twitter:image', `<meta property="twitter:image" content="${pageImage}" />`],
     ['thumbnail', `<meta name="thumbnail" content="${pageImage}" />`],
+    ['relative image', `<meta name="thumbnail" content="${relativeImage}" />`],
   ];
 
   ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.image_url, pageImage, testTag));