Year category intersection implemented

README

@@ -5,4 +5,4 @@ The process of creating a Website for MMD&C on the backend entails 4 steps:
 * create JSON dictionary
 * create index
 * create pages
-* (parse pages)
+* (parse pages

mmdc_create_json.py

@@ -74,17 +74,37 @@ def parse_work_page(title, content):
                     print 'THUMB:', thumburl
                 work_dict[key]=val
             return work_dict
+
+def api_PageCategories(pageid):
+    '''Find all the categories, and their parent category of a page '''
+    query = 'action=query&pageids={}&prop=categories'.format(pageid)
+    url = endpoint + query
+    request = urllib2.urlopen(url)
+    jsonp = json.loads(request.read())    
+    json_dic = jsonp['query']['pages']
+    page_id =  json_dic.keys()[0]
+    page_categories = json_dic[page_id][u'categories']
+    all_cats = [ entry[u'title'].encode('utf-8') for entry in page_categories ] #.replace('Category:', '')
+    return all_cats
+
         
 def api_category(category, year): #Find all pages incategory and add to allworks dictionary
-    ''' TODO: category intersection; With SirrusSearch'''
     category =  category.replace(' ', '_')
-    if year:
-        api_url = endpoint + 'action=query&list=categorymembers&cmlimit=500&cmtitle=Category:{}&cmtitle=Category:{}'.format(category, year)
-    else:
-        api_url = endpoint + 'action=query&list=categorymembers&cmlimit=500&cmtitle=Category:{}'.format(category)    
-    request = urllib2.urlopen(api_url)
+    apiCatMembers = endpoint + 'action=query&list=categorymembers&cmlimit=1000&cmtitle=Category:{}'.format(category)    
+    request = urllib2.urlopen(apiCatMembers)
     jsonp = json.loads(request.read())    
-    for page in  jsonp['query']['categorymembers']:
+    Graduation_work_Members = jsonp['query']['categorymembers']
+    intersectCatMembers = []
+    if year:          
+        for member in Graduation_work_Members:
+            page_cats = api_PageCategories(member['pageid'])
+            if ('Category:{}'.format(year)) in page_cats:
+                print year, 'in', page_cats
+                intersectCatMembers.append(member)# add member to intersectCatMembers
+    else:
+        intersectCatMembers =  Graduation_work_Members
+        
+    for page in  intersectCatMembers: 
         title = ((page['title']).encode('utf-8') ).replace(" ", "_") #snakecase for page titles
         pageid = page['pageid']
         article = api_page(pageid, 'content')