ImageLoader/Wikipedia: load images from wikipedia article

src/ImageLoader.js

@@ -1,4 +1,5 @@
 var wikidata = require('./wikidata')
+var wikipedia = require('./wikipedia')
 var cache = {}
 
 function ImageLoader (data) {
@@ -55,6 +56,13 @@ ImageLoader.prototype.parseObject = function (data) {
     }
   }
 
+  if (data.object.tags.wikipedia) {
+    this.sources.push({
+      type: 'wikipedia',
+      value: data.object.tags.wikipedia
+    })
+  }
+
   if (data.object.tags.wikidata) {
     this.sources.push({
       type: 'wikidata',
@@ -142,6 +150,28 @@ ImageLoader.prototype.loadWikimediaCommons = function (src, callback) {
   }
 }
 
+ImageLoader.prototype.loadWikipedia = function (src, callback) {
+  var value = src.value
+
+  wikipedia.getImages(value, function (err, result) {
+    if (err) {
+      return callback(err, null)
+    }
+
+    result.forEach(function (d) {
+      if (this.found.indexOf(d) === -1) {
+        this.found.push(d)
+        this.data[d] = {
+          id: d,
+          type: 'wikimedia'
+        }
+      }
+    }.bind(this))
+
+    callback(null)
+  }.bind(this))
+}
+
 ImageLoader.prototype.handlePending = function () {
   var pending = this.pendingCallbacks
   delete this.pendingCallbacks
@@ -169,6 +199,8 @@ ImageLoader.prototype.callbackCurrent = function (index, options, callback) {
       this.loadWikimediaCommons(src, this.handlePending.bind(this))
     } else if (src.type === 'wikidata') {
       this.loadWikidata(src, this.handlePending.bind(this))
+    } else if (src.type === 'wikipedia') {
+      this.loadWikipedia(src, this.handlePending.bind(this))
     }
 
     return

src/wikipedia.js

@@ -283,3 +283,38 @@ function showWikipedia (tagValue, dom, callback) {
     callback(err)
   })
 }
+
+function getImages (tagValue, callback) {
+  get(tagValue, function (err, result) {
+    if (err) {
+      return callback(err, null)
+    }
+
+    var imgs = result.div.getElementsByTagName('img')
+    var result = []
+
+    for (i = 0; i < imgs.length; i++) {
+      var img = imgs[i]
+
+      // ignore icons
+      if (img.width <= 64 && img.height <= 64) {
+        continue
+      }
+
+      img.removeAttribute('width')
+      img.removeAttribute('height')
+
+      var m = img.src.match(/^https?:\/\/upload.wikimedia.org\/wikipedia\/commons\/thumb\/\w+\/\w+\/([^\/]+)/)
+      if (m) {
+        var file = decodeURIComponent(m[1]).replace(/_/g, ' ')
+        result.push(file)
+      }
+    }
+
+    callback(null, result)
+  })
+}
+
+module.exports = {
+  getImages: getImages
+}