rename meta attributes, add 2ch.hk support, version bump

post_process.py

@@ -3,31 +3,20 @@ import re
 LINK_RE = re.compile(r"(https?://[\w\-_.]+\.[a-z]{2,4}([^\s<'\"]*|$))")
 
 
-def post_process(thing, board, helper):
-    thing["v"] = 1.1
-    thing["_id"] = int(thing["no"])
+def post_process(item, board, helper):
+    item["_v"] = 1.2
+    item["_id"] = helper.item_id(item)
 
-    thing["board"] = board
-    thing["chan"] = helper.db_id
+    item["_board"] = board
+    item["_chan"] = helper.db_id
 
-    if "com" in thing and thing["com"]:
-        thing["urls"] = get_links_from_body(thing["com"])
-    elif "sub" in thing and thing["sub"]:
-        thing["urls"] = get_links_from_body(thing["sub"])
-    if "fsize" in thing and thing["fsize"]:
-        url = helper.image_url(board, thing["tim"], thing["ext"])
-        if "urls" in thing:
-            thing["urls"].append(url)
-        else:
-            thing["urls"] = [url]
-    if "urls" not in thing:
-        thing["urls"] = []
+    item["_urls"] = helper.item_urls(item, board)
 
-    return thing
+    return item
 
 
 def get_links_from_body(body):
-    result = set()
+    result = []
 
     body = body \
         .replace("<wbr>", "") \
@@ -37,9 +26,9 @@ def get_links_from_body(body):
     for match in LINK_RE.finditer(body):
         url = match.group(1)
         if is_external(url):
-            result.add(url)
+            result.append(url)
 
-    return list(result)
+    return result
 
 
 def is_external(url):