feedparser - npm Package Compare versions

Comparing version 0.9.12 to 0.9.13

main.js

@@ -21,164 +21,172 @@ /**********************************************************************
 
-function handleMeta (node, type) {
+function handleMeta (node, type, options) {
   if (!type || !node) return {};
 
-  var meta = {};
-  ['title','description','date', 'pubdate', 'pubDate','link', 'xmlurl', 'xmlUrl','author','language','favicon','copyright','generator'].forEach(function (property){
-    meta[property] = null;
-  });
-  meta.image = {};
-  meta.categories = [];
+  var meta = {}
+    , normalize = !options || (options && options.normalize)
+    ;
 
+  if (normalize) {
+    ['title','description','date', 'pubdate', 'pubDate','link', 'xmlurl', 'xmlUrl','author','language','favicon','copyright','generator'].forEach(function (property){
+      meta[property] = null;
+    });
+    meta.image = {};
+    meta.categories = [];
+  }
+
   Object.keys(node).forEach(function(name){
     var el = node[name];
-    switch(name){
-    case('title'):
-      meta.title = utils.get(el);
-      break;
-    case('description'):
-    case('subtitle'):
-      meta.description = utils.get(el);
-      break;
-    case('pubdate'):
-    case('lastbuilddate'):
-    case('published'):
-    case('modified'):
-    case('updated'):
-    case('dc:date'):
-      var date = utils.get(el) ? new Date(el['#']) : null;
-      if (!date) break;
-      if (meta.pubdate === null || name == 'pubdate' || name == 'published')
-        meta.pubdate = meta.pubDate = date;
-      if (meta.date === null || name == 'lastbuilddate' || name == 'modified' || name == 'updated')
-        meta.date = date;
-      break;
-    case('link'):
-    case('atom:link'):
-    case('atom10:link'):
-      if (Array.isArray(el)) {
-        el.forEach(function (link){
-          if (link['@']['href']) { // Atom
-            if (utils.get(link['@'], 'rel')) {
-              if (link['@']['rel'] == 'alternate') meta.link = link['@']['href'];
-              else if (link['@']['rel'] == 'self') meta.xmlurl = meta.xmlUrl = link['@']['href'];
+    if (normalize) {
+      switch(name){
+      case('title'):
+        meta.title = utils.get(el);
+        break;
+      case('description'):
+      case('subtitle'):
+        meta.description = utils.get(el);
+        break;
+      case('pubdate'):
+      case('lastbuilddate'):
+      case('published'):
+      case('modified'):
+      case('updated'):
+      case('dc:date'):
+        var date = utils.get(el) ? new Date(el['#']) : null;
+        if (!date) break;
+        if (meta.pubdate === null || name == 'pubdate' || name == 'published')
+          meta.pubdate = meta.pubDate = date;
+        if (meta.date === null || name == 'lastbuilddate' || name == 'modified' || name == 'updated')
+          meta.date = date;
+        break;
+      case('link'):
+      case('atom:link'):
+      case('atom10:link'):
+        if (Array.isArray(el)) {
+          el.forEach(function (link){
+            if (link['@']['href']) { // Atom
+              if (utils.get(link['@'], 'rel')) {
+                if (link['@']['rel'] == 'alternate') meta.link = link['@']['href'];
+                else if (link['@']['rel'] == 'self') meta.xmlurl = meta.xmlUrl = link['@']['href'];
+              } else {
+                meta.link = link['@']['href'];
+              }
+            } else if (Object.keys(link['@']).length === 0) { // RSS
+              if (!meta.link) meta.link = utils.get(link);
+            }
+          });
+        } else {
+          if (el['@']['href']) { // Atom
+            if (utils.get(el['@'], 'rel')) {
+              if (el['@']['rel'] == 'alternate') meta.link = el['@']['href'];
+              else if (el['@']['rel'] == 'self') meta.xmlurl = meta.xmlUrl = el['@']['href'];
             } else {
-              meta.link = link['@']['href'];
+              meta.link = el['@']['href'];
             }
-          } else if (Object.keys(link['@']).length === 0) { // RSS
-            if (!meta.link) meta.link = utils.get(link);
+          } else if (Object.keys(el['@']).length === 0) { // RSS
+            if (!meta.link) meta.link = utils.get(el);
           }
-        });
-      } else {
-        if (el['@']['href']) { // Atom
-          if (utils.get(el['@'], 'rel')) {
-            if (el['@']['rel'] == 'alternate') meta.link = el['@']['href'];
-            else if (el['@']['rel'] == 'self') meta.xmlurl = meta.xmlUrl = el['@']['href'];
-          } else {
-            meta.link = el['@']['href'];
-          }
-        } else if (Object.keys(el['@']).length === 0) { // RSS
-          if (!meta.link) meta.link = utils.get(el);
         }
-      }
-      break;
-    case('managingeditor'):
-    case('webmaster'):
-    case('author'):
-      if (meta.author === null || name == 'managingeditor')
-        meta.author = utils.get(el);
-      if (name == 'author')
-        meta.author = utils.get(el.name) || utils.get(el.email) || utils.get(el.uri);
-      break;
-    case('language'):
-      meta.language = utils.get(el);
-      break;
-    case('image'):
-    case('logo'):
-      if (el.url)
-        meta.image.url = utils.get(el.url);
-      if (el.title)
-        meta.image.title = utils.get(el.title);
-      else meta.image.url = utils.get(el);
-      break;
-    case('icon'):
-      meta.favicon = utils.get(el);
-      break;
-    case('copyright'):
-    case('rights'):
-    case('dc:rights'):
-      meta.copyright = utils.get(el);
-      break;
-    case('generator'):
-      meta.generator = utils.get(el);
-      if (utils.get(el['@'], 'version'))
-        meta.generator += (meta.generator ? ' ' : '') + 'v' + el['@'].version;
-      if (utils.get(el['@'], 'uri'))
-        meta.generator += meta.generator ? ' (' + el['@'].uri + ')' : el['@'].uri;
-      break;
-    case('category'):
-    case('dc:subject'):
-    case('itunes:category'):
-    case('media:category'):
-      /* We handle all the kinds of categories within the switch loop because meta.categories
-       * is an array, unlike the other properties, and therefore can handle multiple values
-       */
-      if (Array.isArray(el)) {
-        el.forEach(function (category){
+        break;
+      case('managingeditor'):
+      case('webmaster'):
+      case('author'):
+        if (meta.author === null || name == 'managingeditor')
+          meta.author = utils.get(el);
+        if (name == 'author')
+          meta.author = utils.get(el.name) || utils.get(el.email) || utils.get(el.uri);
+        break;
+      case('language'):
+        meta.language = utils.get(el);
+        break;
+      case('image'):
+      case('logo'):
+        if (el.url)
+          meta.image.url = utils.get(el.url);
+        if (el.title)
+          meta.image.title = utils.get(el.title);
+        else meta.image.url = utils.get(el);
+        break;
+      case('icon'):
+        meta.favicon = utils.get(el);
+        break;
+      case('copyright'):
+      case('rights'):
+      case('dc:rights'):
+        meta.copyright = utils.get(el);
+        break;
+      case('generator'):
+        meta.generator = utils.get(el);
+        if (utils.get(el['@'], 'version'))
+          meta.generator += (meta.generator ? ' ' : '') + 'v' + el['@'].version;
+        if (utils.get(el['@'], 'uri'))
+          meta.generator += meta.generator ? ' (' + el['@'].uri + ')' : el['@'].uri;
+        break;
+      case('category'):
+      case('dc:subject'):
+      case('itunes:category'):
+      case('media:category'):
+        /* We handle all the kinds of categories within the switch loop because meta.categories
+         * is an array, unlike the other properties, and therefore can handle multiple values
+         */
+        var _category = ''
+          , _categories = []
+          ;
+        if (Array.isArray(el)) {
+          el.forEach(function (category){
+            if ('category' == name && 'atom' == type) {
+              if (category['@'] && utils.get(category['@'], 'term')) meta.categories.push(utils.get(category['@'], 'term'));
+            } else if ('category' == name && utils.get(category) && 'rss' == type) {
+              _categories = utils.get(category).split(',').map(function (cat){ return cat.trim(); });
+              if (_categories.length) meta.categories = meta.categories.concat(_categories);
+            } else if ('dc:subject' == name && utils.get(category)) {
+              _categories = utils.get(category).split(' ').map(function (cat){ return cat.trim(); });
+              if (_categories.length) meta.categories = meta.categories.concat(_categories);
+            } else if ('itunes:category' == name) {
+              if (category['@'] && utils.get(category['@'], 'text')) _category = utils.get(category['@'], 'text');
+              if (category[name]) {
+                if (Array.isArray(category[name])) {
+                  category[name].forEach(function (subcategory){
+                    if (subcategory['@'] && utils.get(subcategory['@'], 'text')) meta.categories.push(_category + '/' + utils.get(subcategory['@'], 'text'));
+                  });
+                } else {
+                  if (category[name]['@'] && utils.get(category[name]['@'], 'text'))
+                    meta.categories.push(_category + '/' + utils.get(category[name]['@'], 'text'));
+                }
+              } else {
+                meta.categories.push(_category);
+              }
+            } else if ('media:category' == name) {
+              meta.categories.push(utils.get(category));
+            }
+          });
+        } else {
           if ('category' == name && 'atom' == type) {
-            if (category['@'] && utils.get(category['@'], 'term')) meta.categories.push(utils.get(category['@'], 'term'));
-          } else if ('category' == name && utils.get(category) && 'rss' == type) {
-            var categories = utils.get(category).split(',').map(function (cat){ return cat.trim(); });
-            if (categories.length) meta.categories = meta.categories.concat(categories);
-          } else if ('dc:subject' == name && utils.get(category)) {
-            var categories = utils.get(category).split(' ').map(function (cat){ return cat.trim(); });
-            if (categories.length) meta.categories = meta.categories.concat(categories);
+            if (utils.get(el['@'], 'term')) meta.categories.push(utils.get(el['@'], 'term'));
+          } else if ('category' == name && utils.get(el) && 'rss' == type) {
+            _categories = utils.get(el).split(',').map(function (cat){ return cat.trim(); });
+            if (_categories.length) meta.categories = meta.categories.concat(_categories);
+          } else if ('dc:subject' == name && utils.get(el)) {
+            _categories = utils.get(el).split(' ').map(function (cat){ return cat.trim(); });
+            if (_categories.length) meta.categories = meta.categories.concat(_categories);
           } else if ('itunes:category' == name) {
-            var cat;
-            if (category['@'] && utils.get(category['@'], 'text')) cat = utils.get(category['@'], 'text');
-            if (category[name]) {
-              if (Array.isArray(category[name])) {
-                category[name].forEach(function (subcategory){
-                  if (subcategory['@'] && utils.get(subcategory['@'], 'text')) meta.categories.push(cat + '/' + utils.get(subcategory['@'], 'text'));
+            if (el['@'] && utils.get(el['@'], 'text')) _category = utils.get(el['@'], 'text');
+            if (el[name]) {
+              if (Array.isArray(el[name])) {
+                el[name].forEach(function (subcategory){
+                  if (subcategory['@'] && utils.get(subcategory['@'], 'text')) meta.categories.push(_category + '/' + utils.get(subcategory['@'], 'text'));
                 });
               } else {
-                if (category[name]['@'] && utils.get(category[name]['@'], 'text'))
-                  meta.categories.push(cat + '/' + utils.get(category[name]['@'], 'text'));
+                if (el[name]['@'] && utils.get(el[name]['@'], 'text'))
+                  meta.categories.push(_category + '/' + utils.get(el[name]['@'], 'text'));
               }
             } else {
-              meta.categories.push(cat);
+              meta.categories.push(_category);
             }
           } else if ('media:category' == name) {
-            meta.categories.push(utils.get(category));
+            meta.categories.push(utils.get(el));
           }
-        });
-      } else {
-        if ('category' == name && 'atom' == type) {
-          if (utils.get(el['@'], 'term')) meta.categories.push(utils.get(el['@'], 'term'));
-        } else if ('category' == name && utils.get(el) && 'rss' == type) {
-          var categories = utils.get(el).split(',').map(function (cat){ return cat.trim(); });
-          if (categories.length) meta.categories = meta.categories.concat(categories);
-        } else if ('dc:subject' == name && utils.get(el)) {
-          var categories = utils.get(el).split(' ').map(function (cat){ return cat.trim(); });
-          if (categories.length) meta.categories = meta.categories.concat(categories);
-        } else if ('itunes:category' == name) {
-          var cat;
-          if (el['@'] && utils.get(el['@'], 'text')) cat = utils.get(el['@'], 'text');
-          if (el[name]) {
-            if (Array.isArray(el[name])) {
-              el[name].forEach(function (subcategory){
-                if (subcategory['@'] && utils.get(subcategory['@'], 'text')) meta.categories.push(cat + '/' + utils.get(subcategory['@'], 'text'));
-              });
-            } else {
-              if (el[name]['@'] && utils.get(el[name]['@'], 'text'))
-                meta.categories.push(cat + '/' + utils.get(el[name]['@'], 'text'));
-            }
-          } else {
-            meta.categories.push(cat);
-          }
-        } else if ('media:category' == name) {
-          meta.categories.push(utils.get(el));
         }
-      }
-      break;
-    } // switch end
+        break;
+      } // switch end
+    }
     // Fill with all native other namespaced properties
@@ -190,246 +198,256 @@ if (name.indexOf('#') !== 0) {
   }); // forEach end
-  if (!meta.description) {
-    if (node['itunes:summary']) meta.description = utils.get(node['itunes:summary']);
-    else if (node['tagline']) meta.description = utils.get(node['tagline']);
-  }
-  if (!meta.author) {
-    if (node['itunes:author']) meta.author = utils.get(node['itunes:author']);
-    else if (node['itunes:owner'] && node['itunes:owner']['itunes:name']) meta.author = utils.get(node['itunes:owner']['itunes:name']);
-    else if (node['dc:creator']) meta.author = utils.get(node['dc:creator']);
-    else if (node['dc:publisher']) meta.author = utils.get(node['dc:publisher']);
-  }
-  if (!meta.language) {
-    if (node['@'] && node['@']['xml:lang']) meta.language = utils.get(node['@'], 'xml:lang');
-    else if (node['dc:language']) meta.language = utils.get(node['dc:language']);
-  }
-  if (!meta.image.url) {
-    if (node['itunes:image']) meta.image.url = utils.get(node['itunes:image']['@'], 'href');
-    else if (node['media:thumbnail']) meta.image.url = utils.get(node['media:thumbnail']['@'], 'url');
-  }
-  if (!meta.copyright) {
-    if (node['media:copyright']) meta.copyright = utils.get(node['media:copyright']);
-    else if (node['dc:rights']) meta.copyright = utils.get(node['dc:rights']);
-    else if (node['creativecommons:license']) meta.copyright = utils.get(node['creativecommons:license']);
-    else if (node['cc:license']) {
-      if (Array.isArray(node['cc:license']) && node['cc:license'][0]['@'] && node['cc:license'][0]['@']['rdf:resource']) {
-        meta.copyright = utils.get(node['cc:license'][0]['@'], 'rdf:resource');
-      } else if (node['cc:license']['@'] && node['cc:license']['@']['rdf:resource']) {
-        meta.copyright = utils.get(node['cc:license']['@'], 'rdf:resource');
+
+  if (normalize) {
+    if (!meta.description) {
+      if (node['itunes:summary']) meta.description = utils.get(node['itunes:summary']);
+      else if (node['tagline']) meta.description = utils.get(node['tagline']);
+    }
+    if (!meta.author) {
+      if (node['itunes:author']) meta.author = utils.get(node['itunes:author']);
+      else if (node['itunes:owner'] && node['itunes:owner']['itunes:name']) meta.author = utils.get(node['itunes:owner']['itunes:name']);
+      else if (node['dc:creator']) meta.author = utils.get(node['dc:creator']);
+      else if (node['dc:publisher']) meta.author = utils.get(node['dc:publisher']);
+    }
+    if (!meta.language) {
+      if (node['@'] && node['@']['xml:lang']) meta.language = utils.get(node['@'], 'xml:lang');
+      else if (node['dc:language']) meta.language = utils.get(node['dc:language']);
+    }
+    if (!meta.image.url) {
+      if (node['itunes:image']) meta.image.url = utils.get(node['itunes:image']['@'], 'href');
+      else if (node['media:thumbnail']) meta.image.url = utils.get(node['media:thumbnail']['@'], 'url');
+    }
+    if (!meta.copyright) {
+      if (node['media:copyright']) meta.copyright = utils.get(node['media:copyright']);
+      else if (node['dc:rights']) meta.copyright = utils.get(node['dc:rights']);
+      else if (node['creativecommons:license']) meta.copyright = utils.get(node['creativecommons:license']);
+      else if (node['cc:license']) {
+        if (Array.isArray(node['cc:license']) && node['cc:license'][0]['@'] && node['cc:license'][0]['@']['rdf:resource']) {
+          meta.copyright = utils.get(node['cc:license'][0]['@'], 'rdf:resource');
+        } else if (node['cc:license']['@'] && node['cc:license']['@']['rdf:resource']) {
+          meta.copyright = utils.get(node['cc:license']['@'], 'rdf:resource');
+        }
       }
     }
-  }
-  if (!meta.generator) {
-    if (node['admin:generatoragent']) {
-      if (Array.isArray(node['admin:generatoragent']) && node['admin:generatoragent'][0]['@'] && node['admin:generatoragent'][0]['@']['rdf:resource']) {
-        meta.generator = utils.get(node['admin:generatoragent'][0]['@'], 'rdf:resource');
-      } else if (node['admin:generatoragent']['@'] && node['admin:generatoragent']['@']['rdf:resource']) {
-        meta.generator = utils.get(node['admin:generatoragent']['@'], 'rdf:resource');
+    if (!meta.generator) {
+      if (node['admin:generatoragent']) {
+        if (Array.isArray(node['admin:generatoragent']) && node['admin:generatoragent'][0]['@'] && node['admin:generatoragent'][0]['@']['rdf:resource']) {
+          meta.generator = utils.get(node['admin:generatoragent'][0]['@'], 'rdf:resource');
+        } else if (node['admin:generatoragent']['@'] && node['admin:generatoragent']['@']['rdf:resource']) {
+          meta.generator = utils.get(node['admin:generatoragent']['@'], 'rdf:resource');
+        }
       }
     }
+    if (meta.categories.length)
+      meta.categories = utils.unique(meta.categories);
   }
-  if (meta.categories.length)
-    meta.categories = utils.unique(meta.categories);
   return meta;
 }
 
-function handleItem (node, type){
+function handleItem (node, type, options){
   if (!type || !node) return {};
 
-  var item = {};
-  ['title','description','summary','date','pubdate','pubDate','link','guid','author','comments', 'origlink'].forEach(function (property){
-    item[property] = null;
-  });
-  item.image = {};
-  item.source = {};
-  item.categories = [];
-  item.enclosures = [];
+  var item = {}
+    , normalize = !options || (options && options.normalize)
+    ;
 
+  if (normalize) {
+    ['title','description','summary','date','pubdate','pubDate','link','guid','author','comments', 'origlink'].forEach(function (property){
+      item[property] = null;
+    });
+    item.image = {};
+    item.source = {};
+    item.categories = [];
+    item.enclosures = [];
+  }
+
   Object.keys(node).forEach(function(name){
     var el = node[name];
-    switch(name){
-    case('title'):
-      item.title = utils.get(el);
-      break;
-    case('description'):
-    case('summary'):
-      item.summary = utils.get(el);
-      if (!item.description) item.description = utils.get(el);
-      break;
-    case('content'):
-    case('content:encoded'):
-      item.description = utils.get(el);
-      break;
-    case('pubdate'):
-    case('published'):
-    case('issued'):
-    case('modified'):
-    case('updated'):
-    case('dc:date'):
-      var date = utils.get(el) ? new Date(el['#']) : null;
-      if (!date) break;
-      if (item.pubdate === null || name == 'pubdate' || name == 'published' || name == 'issued')
-        item.pubdate = item.pubDate = date;
-      if (item.date === null || name == 'modified' || name == 'updated')
-        item.date = date;
-      break;
-    case('link'):
-      if (Array.isArray(el)) {
-        el.forEach(function (link){
-          if (link['@']['href']) { // Atom
-            if (utils.get(link['@'], 'rel')) {
-              if (link['@']['rel'] == 'canonical') item.origlink = link['@']['href'];
-              if (link['@']['rel'] == 'alternate') item.link = link['@']['href'];
-              if (link['@']['rel'] == 'replies') item.comments = link['@']['href'];
-              if (link['@']['rel'] == 'enclosure') {
+    if (normalize) {
+      switch(name){
+      case('title'):
+        item.title = utils.get(el);
+        break;
+      case('description'):
+      case('summary'):
+        item.summary = utils.get(el);
+        if (!item.description) item.description = utils.get(el);
+        break;
+      case('content'):
+      case('content:encoded'):
+        item.description = utils.get(el);
+        break;
+      case('pubdate'):
+      case('published'):
+      case('issued'):
+      case('modified'):
+      case('updated'):
+      case('dc:date'):
+        var date = utils.get(el) ? new Date(el['#']) : null;
+        if (!date) break;
+        if (item.pubdate === null || name == 'pubdate' || name == 'published' || name == 'issued')
+          item.pubdate = item.pubDate = date;
+        if (item.date === null || name == 'modified' || name == 'updated')
+          item.date = date;
+        break;
+      case('link'):
+        if (Array.isArray(el)) {
+          el.forEach(function (link){
+            if (link['@']['href']) { // Atom
+              if (utils.get(link['@'], 'rel')) {
+                if (link['@']['rel'] == 'canonical') item.origlink = link['@']['href'];
+                if (link['@']['rel'] == 'alternate') item.link = link['@']['href'];
+                if (link['@']['rel'] == 'replies') item.comments = link['@']['href'];
+                if (link['@']['rel'] == 'enclosure') {
+                  var enclosure = {};
+                  enclosure.url = link['@']['href'];
+                  enclosure.type = utils.get(link['@'], 'type');
+                  enclosure.length = utils.get(link['@'], 'length');
+                  item.enclosures.push(enclosure);
+                }
+              } else {
+                item.link = link['@']['href'];
+              }
+            } else if (Object.keys(link['@']).length === 0) { // RSS
+              if (!item.link) item.link = utils.get(link);
+            }
+          });
+        } else {
+          if (el['@']['href']) { // Atom
+            if (utils.get(el['@'], 'rel')) {
+              if (el['@']['rel'] == 'canonical') item.origlink = el['@']['href'];
+              if (el['@']['rel'] == 'alternate') item.link = el['@']['href'];
+              if (el['@']['rel'] == 'replies') item.comments = el['@']['href'];
+              if (el['@']['rel'] == 'enclosure') {
                 var enclosure = {};
-                enclosure.url = link['@']['href'];
-                enclosure.type = utils.get(link['@'], 'type');
-                enclosure.length = utils.get(link['@'], 'length');
+                enclosure.url = el['@']['href'];
+                enclosure.type = utils.get(el['@'], 'type');
+                enclosure.length = utils.get(el['@'], 'length');
                 item.enclosures.push(enclosure);
               }
             } else {
-              item.link = link['@']['href'];
+              item.link = el['@']['href'];
             }
-          } else if (Object.keys(link['@']).length === 0) { // RSS
-            if (!item.link) item.link = utils.get(link);
+          } else if (Object.keys(el['@']).length === 0) { // RSS
+            if (!item.link) item.link = utils.get(el);
           }
-        });
-      } else {
-        if (el['@']['href']) { // Atom
-          if (utils.get(el['@'], 'rel')) {
-            if (el['@']['rel'] == 'canonical') item.origlink = el['@']['href'];
-            if (el['@']['rel'] == 'alternate') item.link = el['@']['href'];
-            if (el['@']['rel'] == 'replies') item.comments = el['@']['href'];
-            if (el['@']['rel'] == 'enclosure') {
-              var enclosure = {};
-              enclosure.url = el['@']['href'];
-              enclosure.type = utils.get(el['@'], 'type');
-              enclosure.length = utils.get(el['@'], 'length');
-              item.enclosures.push(enclosure);
+        }
+        if (!item.guid) item.guid = item.link;
+        break;
+      case('guid'):
+      case('id'):
+        item.guid = utils.get(el);
+        break;
+      case('author'):
+        item.author = utils.get(el.name) || utils.get(el.email) || utils.get(el.uri);
+        break;
+      case('dc:creator'):
+        item.author = utils.get(el);
+        break;
+      case('comments'):
+        item.comments = utils.get(el);
+        break;
+      case('source'):
+        if ('rss' == type) {
+          item.source['title'] = utils.get(el);
+          item.source['url'] = utils.get(el['@'], 'url');
+        } else if ('atom' == type) {
+          if (el.title && utils.get(el.title))
+            item.source['title'] = utils.get(el.title);
+          if (el.link && utils.get(el.link['@'], 'href'))
+          item.source['url'] = utils.get(el.link['@'], 'href');
+        }
+        break;
+      case('enclosure'):
+      case('media:content'):
+        var _enclosure = {};
+        if (Array.isArray(el)) {
+          el.forEach(function (enc){
+            _enclosure.url = utils.get(enc['@'], 'url');
+            _enclosure.type = utils.get(enc['@'], 'type') || utils.get(enc['@'], 'medium');
+            _enclosure.length = utils.get(enc['@'], 'length') || utils.get(enc['@'], 'filesize');
+            item.enclosures.push(_enclosure);
+          });
+        } else {
+          _enclosure.url = utils.get(el['@'], 'url');
+          _enclosure.type = utils.get(el['@'], 'type') || utils.get(el['@'], 'medium');
+          _enclosure.length = utils.get(el['@'], 'length') || utils.get(el['@'], 'filesize');
+          item.enclosures.push(_enclosure);
+        }
+        break;
+      case('enc:enclosure'): // Can't find this in use for an example to debug. Only example found does not comply with the spec -- can't code THAT!
+        break;
+      case('category'):
+      case('dc:subject'):
+      case('itunes:category'):
+      case('media:category'):
+        /* We handle all the kinds of categories within the switch loop because item.categories
+         * is an array, unlike the other properties, and therefore can handle multiple values
+         */
+        var _category = ''
+          , _categories = []
+          ;
+        if (Array.isArray(el)) {
+          el.forEach(function (category){
+            if ('category' == name && 'atom' == type) {
+              if (category['@'] && utils.get(category['@'], 'term')) item.categories.push(utils.get(category['@'], 'term'));
+            } else if ('category' == name && utils.get(category) && 'rss' == type) {
+              _categories = utils.get(category).split(',').map(function (cat){ return cat.trim(); });
+              if (_categories.length) item.categories = item.categories.concat(_categories);
+            } else if ('dc:subject' == name && utils.get(category)) {
+              _categories = utils.get(category).split(' ').map(function (cat){ return cat.trim(); });
+              if (_categories.length) item.categories = item.categories.concat(_categories);
+            } else if ('itunes:category' == name) {
+              if (category['@'] && utils.get(category['@'], 'text')) _category = utils.get(category['@'], 'text');
+              if (category[name]) {
+                if (Array.isArray(category[name])) {
+                  category[name].forEach(function (subcategory){
+                    if (subcategory['@'] && utils.get(subcategory['@'], 'text')) item.categories.push(_category + '/' + utils.get(subcategory['@'], 'text'));
+                  });
+                } else {
+                  if (category[name]['@'] && utils.get(category[name]['@'], 'text'))
+                    item.categories.push(_category + '/' + utils.get(category[name]['@'], 'text'));
+                }
+              } else {
+                item.categories.push(_category);
+              }
+            } else if ('media:category' == name) {
+              item.categories.push(utils.get(category));
             }
-          } else {
-            item.link = el['@']['href'];
-          }
-        } else if (Object.keys(el['@']).length === 0) { // RSS
-          if (!item.link) item.link = utils.get(el);
-        }
-      }
-      if (!item.guid) item.guid = item.link;
-      break;
-    case('guid'):
-    case('id'):
-      item.guid = utils.get(el);
-      break;
-    case('author'):
-      item.author = utils.get(el.name) || utils.get(el.email) || utils.get(el.uri);
-      break;
-    case('dc:creator'):
-      item.author = utils.get(el);
-      break;
-    case('comments'):
-      item.comments = utils.get(el);
-      break;
-    case('source'):
-      if ('rss' == type) {
-        item.source['title'] = utils.get(el);
-        item.source['url'] = utils.get(el['@'], 'url');
-      } else if ('atom' == type) {
-        if (el.title && utils.get(el.title))
-          item.source['title'] = utils.get(el.title);
-        if (el.link && utils.get(el.link['@'], 'href'))
-        item.source['url'] = utils.get(el.link['@'], 'href');
-      }
-      break;
-    case('enclosure'):
-    case('media:content'):
-      if (Array.isArray(el)) {
-        el.forEach(function (enc){
-          var enclosure = {};
-          enclosure.url = utils.get(enc['@'], 'url');
-          enclosure.type = utils.get(enc['@'], 'type') || utils.get(enc['@'], 'medium');
-          enclosure.length = utils.get(enc['@'], 'length') || utils.get(enc['@'], 'filesize');
-          item.enclosures.push(enclosure);
-        });
-      } else {
-        var enclosure = {};
-        enclosure.url = utils.get(el['@'], 'url');
-        enclosure.type = utils.get(el['@'], 'type') || utils.get(el['@'], 'medium');
-        enclosure.length = utils.get(el['@'], 'length') || utils.get(el['@'], 'filesize');
-        item.enclosures.push(enclosure);
-      }
-      break;
-    case('enc:enclosure'): // Can't find this in use for an example to debug. Only example found does not comply with the spec -- can't code THAT!
-      break;
-    case('category'):
-    case('dc:subject'):
-    case('itunes:category'):
-    case('media:category'):
-      /* We handle all the kinds of categories within the switch loop because item.categories
-       * is an array, unlike the other properties, and therefore can handle multiple values
-       */
-      if (Array.isArray(el)) {
-        el.forEach(function (category){
+          });
+        } else {
           if ('category' == name && 'atom' == type) {
-            if (category['@'] && utils.get(category['@'], 'term')) item.categories.push(utils.get(category['@'], 'term'));
-          } else if ('category' == name && utils.get(category) && 'rss' == type) {
-            var categories = utils.get(category).split(',').map(function (cat){ return cat.trim(); });
-            if (categories.length) item.categories = item.categories.concat(categories);
-          } else if ('dc:subject' == name && utils.get(category)) {
-            var categories = utils.get(category).split(' ').map(function (cat){ return cat.trim(); });
-            if (categories.length) item.categories = item.categories.concat(categories);
+            if (utils.get(el['@'], 'term')) item.categories.push(utils.get(el['@'], 'term'));
+          } else if ('category' == name && utils.get(el) && 'rss' == type) {
+            _categories = utils.get(el).split(',').map(function (cat){ return cat.trim(); });
+            if (_categories.length) item.categories = item.categories.concat(_categories);
+          } else if ('dc:subject' == name && utils.get(el)) {
+            _categories = utils.get(el).split(' ').map(function (cat){ return cat.trim(); });
+            if (_categories.length) item.categories = item.categories.concat(_categories);
           } else if ('itunes:category' == name) {
-            var cat;
-            if (category['@'] && utils.get(category['@'], 'text')) cat = utils.get(category['@'], 'text');
-            if (category[name]) {
-              if (Array.isArray(category[name])) {
-                category[name].forEach(function (subcategory){
-                  if (subcategory['@'] && utils.get(subcategory['@'], 'text')) item.categories.push(cat + '/' + utils.get(subcategory['@'], 'text'));
+            if (el['@'] && utils.get(el['@'], 'text')) _category = utils.get(el['@'], 'text');
+            if (el[name]) {
+              if (Array.isArray(el[name])) {
+                el[name].forEach(function (subcategory){
+                  if (subcategory['@'] && utils.get(subcategory['@'], 'text')) item.categories.push(_category + '/' + utils.get(subcategory['@'], 'text'));
                 });
               } else {
-                if (category[name]['@'] && utils.get(category[name]['@'], 'text'))
-                  item.categories.push(cat + '/' + utils.get(category[name]['@'], 'text'));
+                if (el[name]['@'] && utils.get(el[name]['@'], 'text'))
+                  item.categories.push(_category + '/' + utils.get(el[name]['@'], 'text'));
               }
             } else {
-              item.categories.push(cat);
+              item.categories.push(_category);
             }
           } else if ('media:category' == name) {
-            item.categories.push(utils.get(category));
+            item.categories.push(utils.get(el));
           }
-        });
-      } else {
-        if ('category' == name && 'atom' == type) {
-          if (utils.get(el['@'], 'term')) item.categories.push(utils.get(el['@'], 'term'));
-        } else if ('category' == name && utils.get(el) && 'rss' == type) {
-          var categories = utils.get(el).split(',').map(function (cat){ return cat.trim(); });
-          if (categories.length) item.categories = item.categories.concat(categories);
-        } else if ('dc:subject' == name && utils.get(el)) {
-          var categories = utils.get(el).split(' ').map(function (cat){ return cat.trim(); });
-          if (categories.length) item.categories = item.categories.concat(categories);
-        } else if ('itunes:category' == name) {
-          var cat;
-          if (el['@'] && utils.get(el['@'], 'text')) cat = utils.get(el['@'], 'text');
-          if (el[name]) {
-            if (Array.isArray(el[name])) {
-              el[name].forEach(function (subcategory){
-                if (subcategory['@'] && utils.get(subcategory['@'], 'text')) item.categories.push(cat + '/' + utils.get(subcategory['@'], 'text'));
-              });
-            } else {
-              if (el[name]['@'] && utils.get(el[name]['@'], 'text'))
-                item.categories.push(cat + '/' + utils.get(el[name]['@'], 'text'));
-            }
-          } else {
-            item.categories.push(cat);
-          }
-        } else if ('media:category' == name) {
-          item.categories.push(utils.get(el));
         }
-      }
-      break;
-    case('feedburner:origlink'):
-    case('pheedo:origlink'):
-      if (!item.origlink) {
-        item.origlink = utils.get(el);
-      }
-      break;
-    } // switch end
+        break;
+      case('feedburner:origlink'):
+      case('pheedo:origlink'):
+        if (!item.origlink) {
+          item.origlink = utils.get(el);
+        }
+        break;
+      } // switch end
+    }
     // Fill with all native other namespaced properties
@@ -441,19 +459,22 @@ if (name.indexOf('#') !== 0) {
   }); // forEach end
-  if (!item.description) {
-    if (node['itunes:summary']) item.description = utils.get(node['itunes:summary']);
+
+  if (normalize) {
+    if (!item.description) {
+      if (node['itunes:summary']) item.description = utils.get(node['itunes:summary']);
+    }
+    if (!item.author) {
+      if (node['itunes:author']) item.author = utils.get(node['itunes:author']);
+      else if (node['itunes:owner'] && node['itunes:owner']['itunes:name']) item.author = utils.get(node['itunes:owner']['itunes:name']);
+      else if (node['dc:publisher']) item.author = utils.get(node['dc:publisher']);
+    }
+    if (!item.image.url) {
+      if (node['itunes:image']) item.image.url = utils.get(node['itunes:image']['@'], 'href');
+      else if (node['media:thumbnail']) item.image.url = utils.get(node['media:thumbnail']['@'], 'url');
+      else if (node['media:content'] && node['media:content']['media:thumbnail']) item.image.url = utils.get(node['media:content']['media:thumbnail']['@'], 'url');
+      else if (node['media:group'] && node['media:group']['media:thumbnail']) item.image.url = utils.get(node['media:group']['media:thumbnail']['@'], 'url');
+      else if (node['media:group'] && node['media:group']['media:content'] && node['media:group']['media:content']['media:thumbnail']) item.image.url = utils.get(node['media:group']['media:content']['media:thumbnail']['@'], 'url');
+    }
+    if (item.categories.length)
+      item.categories = utils.unique(item.categories);
   }
-  if (!item.author) {
-    if (node['itunes:author']) item.author = utils.get(node['itunes:author']);
-    else if (node['itunes:owner'] && node['itunes:owner']['itunes:name']) item.author = utils.get(node['itunes:owner']['itunes:name']);
-    else if (node['dc:publisher']) item.author = utils.get(node['dc:publisher']);
-  }
-  if (!item.image.url) {
-    if (node['itunes:image']) item.image.url = utils.get(node['itunes:image']['@'], 'href');
-    else if (node['media:thumbnail']) item.image.url = utils.get(node['media:thumbnail']['@'], 'url');
-    else if (node['media:content'] && node['media:content']['media:thumbnail']) item.image.url = utils.get(node['media:content']['media:thumbnail']['@'], 'url');
-    else if (node['media:group'] && node['media:group']['media:thumbnail']) item.image.url = utils.get(node['media:group']['media:thumbnail']['@'], 'url');
-    else if (node['media:group'] && node['media:group']['media:content'] && node['media:group']['media:content']['media:thumbnail']) item.image.url = utils.get(node['media:group']['media:content']['media:thumbnail']['@'], 'url');
-  }
-  if (item.categories.length)
-    item.categories = utils.unique(item.categories);
   return item;
@@ -467,12 +488,16 @@ }
  */
-function FeedParser () {
+function FeedParser (options) {
+
   var parser = this;
+  parser.options = options || {};
+  if (!('normalize' in parser.options)) parser.options.normalize = true;
+  if (!('addMetaToItems' in parser.options)) parser.options.addMetaToItems = true;
   parser._reset();
   parser.stream = sax.createStream(false /* strict mode - no */, {lowercase: true}); // https://github.com/isaacs/sax-js
-  parser.stream.on('error', function (e){ parser.handleSaxError(e, parser) });
-  parser.stream.on('opentag', function (n){ parser.handleOpenTag(n, parser) });
-  parser.stream.on('closetag', function (el){ parser.handleCloseTag(el, parser) });
-  parser.stream.on('text', function (text){ parser.handleText(text, parser) });
-  parser.stream.on('cdata', function (text){ parser.handleText(text, parser) });
-  parser.stream.on('end', function (){ parser.handleEnd(parser) });
+  parser.stream.on('error', function (e){ parser.handleSaxError(e, parser); });
+  parser.stream.on('opentag', function (n){ parser.handleOpenTag(n, parser); });
+  parser.stream.on('closetag', function (el){ parser.handleCloseTag(el, parser); });
+  parser.stream.on('text', function (text){ parser.handleText(text, parser); });
+  parser.stream.on('cdata', function (text){ parser.handleText(text, parser); });
+  parser.stream.on('end', function (){ parser.handleEnd(parser); });
   EventEmitter.call(parser);
@@ -589,3 +614,3 @@ }
  *
- * @param {String} fully qualified uri or a parsed url object from url.parse()
+ * @param {Readable Stream}
  * @param {Function} callback
@@ -668,3 +693,3 @@ * @api public
     return attrs;
-  };
+  }
 
@@ -682,3 +707,3 @@ if (Object.keys(node.attributes).length) {
     parser.xhtml['#'] += '>';
-  } else if (parser.stack.length == 0 &&
+  } else if (parser.stack.length === 0 &&
             (n['#name'] == 'rss' || n['#name'] == 'rdf:rdf' || n['#name'] == 'feed')) {
@@ -690,3 +715,3 @@ parser.meta['#ns'] = [];
       o[name] = n['@'][name];
-      if (name.indexOf('xmlns') == 0) {
+      if (name.indexOf('xmlns') === 0) {
         parser.meta['#ns'].push(o);
@@ -762,7 +787,9 @@ } else if (name != 'version') {
     if (!parser.meta.title) { // We haven't yet parsed all the metadata
-      utils.merge(parser.meta, handleMeta(parser.stack[0], parser.meta['#type']));
+      utils.merge(parser.meta, handleMeta(parser.stack[0], parser.meta['#type'], parser.options));
       parser.emit('meta', parser.meta);
     }
-    item = handleItem(n, parser.meta['#type']);
-    item.meta = parser.meta;
+    item = handleItem(n, parser.meta['#type'], parser.options);
+    if (parser.options.addMetaToItems) {
+      item.meta = parser.meta;
+    }
     if (parser.meta.author && !item.author) item.author = parser.meta.author;
@@ -772,3 +799,3 @@ parser.emit('article', item);
   } else if ((el == 'channel' || el == 'feed') && !parser.meta.title) { // We haven't yet parsed all the metadata
-    utils.merge(parser.meta, handleMeta(n, parser.meta['#type']));
+    utils.merge(parser.meta, handleMeta(n, parser.meta['#type'], parser.options));
     parser.emit('meta', parser.meta);
@@ -817,8 +844,8 @@ }
   this.callback = undefined;
-}
+};
 
 FeedParser.prototype._setCallback = function (callback){
   this.callback = ('function' == typeof callback) ? callback : undefined;
-}
+};
 
 exports = module.exports = FeedParser;

package.json

 { "name" : "feedparser"
 , "author" : "Dan MacTough <[e-mail]>"
 , "description" : "Robust RSS Atom and RDF feed parsing using sax js"
-, "version": "0.9.12"
+, "version": "0.9.13"
 , "keywords" : ["rss", "feed", "atom", "rdf", "xml", "syndication"]
@@ -6,0 +6,0 @@ , "homepage" : "http://github.com/danmactough/node-feedparser"

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

43246

increased by0.15%

Lines of code

890

increased by2.53%

Worsened metrics

Number of package files

6

decreased by-14.29%

No dependency changes