Incorporate Osmose's feedback

* Replace top level keys in CSS selector data (in ‘fallback_extraction_selectors.js’) with a string composed of supported domains/subdomains. * Update ‘getFeatureInfo’ to split the string into a list of domains/subdomains before checking for a hostname match. * Break out part of ‘extractProduct’ in ‘fallback_extraction.js` for better readability. * Rename some internal variables for more consistency/clarity.
mozilla · Sep 21, 2018 · dc0ca41 · dc0ca41
1 parent d0aece3
commit dc0ca41
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 32 deletions.
diff --git a/src/extraction/fallback_extraction.js b/src/extraction/fallback_extraction.js
@@ -24,16 +24,33 @@ const OPEN_GRAPH_PROPERTY_VALUES = {
  * for the page.
  */
 function getFeatureInfo() {
-  const url = window.location.href;
-  for (const [regExpStr, featureInfo] of Object.entries(extractionData)) {
-    const regExp = new RegExp(regExpStr);
-    if (regExp.test(url)) {
-      return featureInfo;
+  const hostname = new URL(window.location.href).host;
+  for (const [vendorDomainsStr, featureInfo] of Object.entries(extractionData)) {
+    const vendorDomains = vendorDomainsStr.split('_');
+    for (const domain of vendorDomains) {
+      if (hostname.includes(domain)) {
+        return featureInfo;
+      }
     }
   }
   return null;
 }
 
+function findValue(extractors) {
+  for (const [selector, extractionMethod] of extractors) {
+    const element = document.querySelector(selector);
+    if (element) {
+      const value = extractionMethod(element);
+      if (value) {
+        return value;
+      }
+      throw new Error('Element found did not return a valid value for the product feature.');
+    }
+  }
+  // None of the selectors matched an element on the page
+  throw new Error('No elements found with vendor data for the product feature.');
+}
+
 /**
  * Returns any product information available on the page from CSS
  * selectors if they exist, otherwise from Open Graph <meta> tags.
@@ -42,28 +59,14 @@ export default function extractProduct() {
   const extractedProduct = {};
   const featureInfo = getFeatureInfo();
   if (featureInfo) {
-    for (const [feature, routines] of Object.entries(featureInfo)) {
-      for (const routine of routines) {
-        const [selector, extractionMethod] = routine;
-        const element = document.querySelector(selector);
-        if (element) {
-          extractedProduct[feature] = extractionMethod(element);
-          if (extractedProduct[feature]) {
-            break;
-          } else {
-            throw new Error(`Element found did not return a valid product ${feature}.`);
-          }
-        } else if (routine === routines[routines.length - 1]) {
-          // None of the selectors matched an element on the page
-          throw new Error(`No elements found with vendor data for product ${feature}.`);
-        }
-      }
+    for (const [feature, extractors] of Object.entries(featureInfo)) {
+      extractedProduct[feature] = findValue(extractors);
     }
   } else {
-    for (const [key, value] of Object.entries(OPEN_GRAPH_PROPERTY_VALUES)) {
-      const metaEle = document.querySelector(`meta[property='${value}']`);
+    for (const [feature, propertyValue] of Object.entries(OPEN_GRAPH_PROPERTY_VALUES)) {
+      const metaEle = document.querySelector(`meta[property='${propertyValue}']`);
       if (metaEle) {
-        extractedProduct[key] = metaEle.getAttribute('content');
+        extractedProduct[feature] = metaEle.getAttribute('content');
       }
     }
   }

diff --git a/src/extraction/fallback_extraction_selectors.js b/src/extraction/fallback_extraction_selectors.js
@@ -18,11 +18,12 @@ function fromAttribute(attribute) {
 
 
 /**
- * CSS selector data by site (represented by a regular expression), where each selector is paired
- * with a method that extracts the value from the element returned by that selector.
+ * CSS selector data by site (represented by a string of acceptable hostnames), where each
+ * selector is paired with a method that extracts the value from the element returned by
+ * that selector.
  */
 const fallbackExtractionData = {
-  '^https?:\\/\\/([a-zA-Z0-9]+\\.){0,}amazon\\.com': {
+  'amazon.com_www.amazon.com_smile.amazon.com': {
     title: [
       ['#productTitle', fromProperty('innerText')],
       ['.product-title', fromProperty('innerText')],
@@ -44,7 +45,7 @@ const fallbackExtractionData = {
       ['#ebooksImgBlkFront', fromProperty('src')],
     ],
   },
-  '^https?:\\/\\/([a-zA-Z0-9]+\\.){0,}bestbuy\\.com': {
+  'bestbuy.com_www.bestbuy.com': {
     title: [
       ['.sku-title h1', fromProperty('innerText')],
     ],
@@ -55,7 +56,7 @@ const fallbackExtractionData = {
       ['img.primary-image', fromProperty('src')],
     ],
   },
-  '^https?:\\/\\/([a-zA-Z0-9]+\\.){0,}ebay\\.com': {
+  'ebay.com_www.ebay.com': {
     title: [
       ['#itemTitle', fromProperty('innerText')],
       ['.product-title', fromProperty('innerText')],
@@ -71,7 +72,7 @@ const fallbackExtractionData = {
       ['.vi-image-gallery__image.vi-image-gallery__image--absolute-center', fromProperty('src')],
     ],
   },
-  '^https?:\\/\\/([a-zA-Z0-9]+\\.){0,}homedepot\\.com': {
+  'homedepot.com_www.homedepot.com': {
     title: [
       ['h1.product-title__title', fromProperty('innerText')],
     ],
@@ -83,7 +84,7 @@ const fallbackExtractionData = {
       ['#mainImage', fromProperty('src')],
     ],
   },
-  '^https?:\\/\\/([a-zA-Z0-9]+\\.){0,}walmart\\.com': {
+  'walmart.com_www.walmart.com': {
     title: [
       ['h1.prod-ProductTitle', fromAttribute('content')],
       ['h1.prod-ProductTitle', fromProperty('innerText')],
@@ -98,7 +99,7 @@ const fallbackExtractionData = {
       ['.prod-hero-image-carousel-image', fromProperty('src')],
     ],
   },
-  'www.mkelly.me': {
+  'mkelly.me_www.mkelly.me': {
     title: [
       ['#title', fromProperty('innerText')],
     ],