refined gallery regex

mmdc_modules.py

@@ -33,9 +33,15 @@ def api_page(pageid, query):
         response = api_request('action=query&titles=File:{}&prop=imageinfo&iiprop=url&iiurlwidth=500', pagename)  # iiurlwidht dermines with of thumbnail         
     return response
 
+def api_img_url(filename): # get full-size's image url
+    '''get url of image'''
+   page_content_dict = api_page(filename, 'fullimage')   
+   if 'imageinfo' in page_content_dict.keys():
+       thumburl = ((page_content_dict.get('imageinfo'))[0].get('thumburl'))
+       return thumburl
 
 def api_thumb_url(filename):
-   '''get thumbnail url of image'''
+    '''get thumbnail url of image'''
    page_content_dict = api_page(filename, 'imageinfo')   
    if 'imageinfo' in page_content_dict.keys():
        thumburl = ((page_content_dict.get('imageinfo'))[0].get('thumburl'))

mmdc_pages.py

@@ -15,7 +15,8 @@ pages_path = 'web/work'
 #def generate_xml():
 gallery_exp=re.compile('<gallery>.*?</gallery>')
 file_exp=re.compile('File:(.*?)(?=File:|<\/gallery>)')
-img_exp=re.compile('(^.*?\.(gif|jpg|jpeg|png))', re.I)
+#img_exp=re.compile('(^.*?\.(gif|jpg|jpeg|png))', re.I)
+img_exp=re.compile('File:((.*?)\.(gif|jpg|jpeg|png))(?=\||File:|<\/gallery>)', re.I)
 
 def replace_gallery(content):
     # from <gallery>.*</gallery> imgs, return list of img ET elements
@@ -26,17 +27,16 @@ def replace_gallery(content):
     print 'gallery_found', gallery_found
     for gallery in gallery_found: # in case there is more than 1 <gallery>
         print 'GALLERY', gallery
-        allfiles =re.findall(file_exp, gallery)
+        allfiles =re.findall(img_exp, gallery)
         print 'ALLFILES', allfiles
         for imgfile in allfiles:
-            img = ((re.search(img_exp, imgfile)).group(0))#.decode('utf-8')
-            imgsrc = api_thumb_url(img)
+            imgfile = imgfile[0]
+            imgsrc = api_thumb_url(imgfile) # search for original image
+            print imgfile, imgsrc
             img_el = ET.Element('img', attrib={'src': imgsrc})
             gallery_imgs.append(img_el)
-            print 'gallery img', img, ET.tostring(img_el)
+#            print 'gallery img', imgfile, ET.tostring(img_el)
     return content, gallery_imgs
-        # need to return a list of images that is the gallery
-        # need to replace <gallery>.*</gallery> with ''
     
 def create_workpage( allworks_dict, work_key): # replace text content in dict with html nodes, holding the content    
     for key in allworks_dict.keys():