From a9aa24409f887feb4cf058e9a1dbb32990b08bcd Mon Sep 17 00:00:00 2001
From: tech234a <46801700+tech234a@users.noreply.github.com>
Date: Wed, 23 Sep 2020 16:22:59 -0400
Subject: [PATCH] Fix exception on marking as complete

---
 tracker.py |  2 +-
 worker.py  | 14 +++++++++-----
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/tracker.py b/tracker.py
index 55ea317..90da312 100644
--- a/tracker.py
+++ b/tracker.py
@@ -9,7 +9,7 @@ from os.path import isfile
 from json import loads
 
 # https://github.com/ArchiveTeam/tencent-weibo-grab/blob/9bae5f9747e014db9227821a9c11557267967023/pipeline.py
-VERSION = "20200923.01"
+VERSION = "20200923.02"
 
 TRACKER_ID = "ext-yt-communitycontribs"
 TRACKER_HOST = "trackerproxy.meo.ws"
diff --git a/worker.py b/worker.py
index bb1508d..a134c95 100644
--- a/worker.py
+++ b/worker.py
@@ -21,6 +21,7 @@ from discovery import getmetadata
 from export import subprrun
 
 batchcontent = []
+actualitems = []
 
 HEROKU = False
 if isfile("../Procfile"):
@@ -53,6 +54,7 @@ def batchfunc():
             print("Ignoring item for now", desit)
         
         batchcontent.append(desit.split(":", 1)[1])
+        actualitems.append(desit)
 
 def submitfunc(submitqueue):
     while not submitqueue.empty():
@@ -179,6 +181,7 @@ while not gkiller.kill_now:
         pass
 
     batchcontent.clear()
+    actualitems.clear()
 
     # Get a batch ID
     batchthreads = []
@@ -310,11 +313,12 @@ while not gkiller.kill_now:
                 system("curl -F "+filzip+"=@directory/"+filzip+" "+targetloc)
 
     # Report the batch as complete
-    for itemb in batchcontent:
-        if isfile("directory/"+itemb.split(":", 1)[1]+".zip"):
-            size = getsize("directory/"+itemb.split(":", 1)[1]+".zip")
-        else:
-            size = 0
+    for itemb in actualitems:
+        size = 0
+        if ":" in itemb:
+            if itemb.split(":", 1)[0] == "video":
+                if isfile("directory/"+itemb.split(":", 1)[1]+".zip"):
+                    size = getsize("directory/"+itemb.split(":", 1)[1]+".zip")
         tracker.mark_item_as_done(itemb, size)
 
     # clear the output directories