Make parser extendable

mozilla · Jul 15, 2016 · 3e0a015 · 3e0a015
1 parent 82ac24e
commit 3e0a015
Show file tree

Hide file tree

Showing 3 changed files with 79 additions and 100 deletions.
diff --git a/lib/default-ruleset.js b/lib/default-ruleset.js
@@ -0,0 +1,34 @@
+module.exports = {
+  description: [
+    ['meta[property="og:description"]', node => node.element.content],
+    ['meta[name="description"]', node => node.element.content],
+  ],
+  icon_url: [
+    ['link[rel="apple-touch-icon"]', node => node.element.href],
+    ['link[rel="apple-touch-icon-precomposed"]', node => node.element.href],
+    ['link[rel="icon"]', node => node.element.href],
+    ['link[rel="fluid-icon"]', node => node.element.href],
+    ['link[rel="shortcut icon"]', node => node.element.href],
+    ['link[rel="Shortcut Icon"]', node => node.element.href],
+    ['link[rel="mask-icon"]', node => node.element.href],
+  ],
+  image_url: [
+    ['meta[property="og:image"]', node => node.element.content],
+    ['meta[property="twitter:image"]', node => node.element.content],
+    ['meta[name="thumbnail"]', node => node.element.content],
+    ['img', node => node.element.src],
+  ],
+  title: [
+    ['meta[property="og:title"]', node => node.element.content],
+    ['meta[property="twitter:title"]', node => node.element.content],
+    ['meta[name="hdl"]', node => node.element.content],
+    ['title', node => node.element.text],
+  ],
+  type: [
+    ['meta[property="og:type"]', node => node.element.content],
+  ],
+  url: [
+    ['meta[property="og:url"]', node => node.element.content],
+    ['link[rel="canonical"]', node => node.element.href],
+  ]
+};
diff --git a/parser.js b/parser.js
@@ -1,95 +1,40 @@
 const {dom, rule, ruleset} = require('fathom-web');
-
-
-function buildRuleset(name, rules) {
-  const reversedRules = Array.from(rules).reverse();
-  const builtRuleset = ruleset(...reversedRules.map(([query, handler], order) => rule(
-    dom(query),
-    node => [{
-      score: order,
-      flavor: name,
-      notes: handler(node),
-    }]
-  )));
-
-  return doc => {
-    const kb = builtRuleset.score(doc);
-    const maxNode = kb.max(name);
-    if (maxNode) {
-      const value = maxNode.flavors.get(name);
-      if (value) {
-        return value.trim();
+const DEFAULT_RULESET = require('./lib/default-ruleset');
+
+const MetadataParser = {
+  metadataRules: Object.assign({}, DEFAULT_RULESET),
+  buildRuleset(name, rules) {
+    const reversedRules = Array.from(rules).reverse();
+    const builtRuleset = ruleset(...reversedRules.map(([query, handler], order) => rule(
+      dom(query),
+      node => [{
+        score: order,
+        flavor: name,
+        notes: handler(node),
+      }]
+    )));
+
+    return doc => {
+      const kb = builtRuleset.score(doc);
+      const maxNode = kb.max(name);
+      if (maxNode) {
+        const value = maxNode.flavors.get(name);
+        if (value) {
+          return value.trim();
+        }
       }
-    }
-  };
-}
-
-
-const titleRules = buildRuleset('title', [
-  ['meta[property="og:title"]', node => node.element.content],
-  ['meta[property="twitter:title"]', node => node.element.content],
-  ['meta[name="hdl"]', node => node.element.content],
-  ['title', node => node.element.text],
-]);
-
-const canonicalUrlRules = buildRuleset('url', [
-  ['meta[property="og:url"]', node => node.element.content],
-  ['link[rel="canonical"]', node => node.element.href],
-]);
-
-
-const iconRules = buildRuleset('icon', [
-  ['link[rel="apple-touch-icon"]', node => node.element.href],
-  ['link[rel="apple-touch-icon-precomposed"]', node => node.element.href],
-  ['link[rel="icon"]', node => node.element.href],
-  ['link[rel="fluid-icon"]', node => node.element.href],
-  ['link[rel="shortcut icon"]', node => node.element.href],
-  ['link[rel="Shortcut Icon"]', node => node.element.href],
-  ['link[rel="mask-icon"]', node => node.element.href],
-]);
-
-const imageRules = buildRuleset('image', [
-  ['meta[property="og:image"]', node => node.element.content],
-  ['meta[property="twitter:image"]', node => node.element.content],
-  ['meta[name="thumbnail"]', node => node.element.content],
-  ['img', node => node.element.src],
-]);
-
-const descriptionRules = buildRuleset('description', [
-  ['meta[property="og:description"]', node => node.element.content],
-  ['meta[name="description"]', node => node.element.content],
-]);
-
-const typeRules = buildRuleset('type', [
-  ['meta[property="og:type"]', node => node.element.content],
-]);
-
-
-const metadataRules = {
-  description: descriptionRules,
-  icon_url: iconRules,
-  image_url: imageRules,
-  title: titleRules,
-  type: typeRules,
-  url: canonicalUrlRules
+    };
+  },
+  getMetadata(doc) {
+    const metadata = {};
+
+    Object.keys(MetadataParser.metadataRules).forEach(metadataKey => {
+      const metadataRule = MetadataParser.buildRuleset(metadataKey, MetadataParser.metadataRules[metadataKey]);
+      metadata[metadataKey] = metadataRule(doc);
+    });
+
+    return metadata;
+  }
 };
 
-
-function getMetadata(doc, rules) {
-  const metadata = {};
-  const ruleSet = rules || metadataRules;
-
-  Object.keys(ruleSet).map(metadataKey => {
-    const metadataRule = ruleSet[metadataKey];
-    metadata[metadataKey] = typeof metadataRule === 'function' ?
-      metadataRule(doc) :
-      getMetadata(doc, metadataRule);
-  });
-
-  return metadata;
-}
-
-module.exports = {
-  metadataRules,
-  getMetadata
-};
+module.exports = MetadataParser;
diff --git a/tests/metadataRules.test.js b/tests/metadataRules.test.js
@@ -1,6 +1,6 @@
 // Tests for parse.js
 const {assert} = require('chai');
-const {metadataRules} = require('../parser');
+const {metadataRules, buildRuleset} = require('../parser');
 const {stringToDom} = require('./test-utils');
 
 function buildHTML(tag) {
@@ -13,11 +13,11 @@ function buildHTML(tag) {
   `;
 }
 
-function ruleTest(testName, testRule, expected, testTag) {
+function ruleTest(testName, testRuleName, expected, testTag) {
   it(`finds ${testName}`, () => {
     const html = buildHTML(testTag);
     const doc = stringToDom(html);
-    const found = testRule(doc);
+    const found = buildRuleset(testRuleName, metadataRules[testRuleName])(doc);
     assert.equal(found, expected, `Unable to find ${testName} in ${html}`);
   });
 }
@@ -33,7 +33,7 @@ describe('Title Rule Tests', function() {
     ['title', `<title>${pageTitle}</title>`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.title, pageTitle, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'title', pageTitle, testTag));
 });
 
 
@@ -45,7 +45,7 @@ describe('Canonical URL Rule Tests', function() {
     ['rel=canonical', `<link rel="canonical" href="${pageUrl}" />`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.url, pageUrl, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'url', pageUrl, testTag));
 });
 
 
@@ -62,7 +62,7 @@ describe('Icon Rule Tests', function() {
     ['mask-icon', `<link rel="mask-icon" href="${pageIcon}" />`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.icon_url, pageIcon, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'icon_url', pageIcon, testTag));
 });
 
 
@@ -76,7 +76,7 @@ describe('Image Rule Tests', function() {
     ['img', `<img src="${pageImage}" />`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.image_url, pageImage, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'image_url', pageImage, testTag));
 });
 
 
@@ -88,7 +88,7 @@ describe('Description Rule Tests', function() {
     ['description', `<meta name="description" content="${pageDescription}" />`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.description, pageDescription, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'description', pageDescription, testTag));
 });
 
 
@@ -99,5 +99,5 @@ describe('Type Rule Tests', function() {
     ['og:type', `<meta property="og:type" content="${pageType}" />`],
   ];
 
-  ruleTests.map(([testName, testTag]) => ruleTest(testName, metadataRules.type, pageType, testTag));
+  ruleTests.map(([testName, testTag]) => ruleTest(testName, 'type', pageType, testTag));
 });