jsutterfield · August 29, 2015 14:01 · May 14, 2014 · May 14, 2014 · May 14, 2014 · May 14, 2014
diff --git a/wikia_video_stats → wikia_video_stats.py b/wikia_video_stats → wikia_video_stats.py
diff --git a/wikia_video_stats b/wikia_video_stats
@@ -51,20 +51,5 @@ def main():
     print "======================="
     print len(sorted_vids_wiki)
 
-
-    # Get number of videos contained in top 5% of wikis with videos
-    i = 1
-    top_count = 0
-    threshold = int(len(sorted_vids_wiki) * .05)
-    for vid in sorted_vids_wiki:
-        if i == threshold:
-            break
-        top_count += vid[1]
-        i += 1
-
-    print "\nTotal videos in top 5% of wikis with videos"
-    print "==========================================="
-    print top_count
-
 if __name__ == "__main__":
     main()
diff --git a/wikia_video_stats b/wikia_video_stats
@@ -40,8 +40,7 @@ def main():
     print "=========="
     print sum_total
 
-    # Top 10 wikis with most videos (skipping the first 2 which are
-    # video and community)
+    # Top 10 wikis with most videos (skipping the first 2 which are video and community)
     print "\nTop 10 wikis"
     print "============"
     for i in range(2, 12):

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -22,7 +22,7 @@ def main():
     fp.close()
 
     # Sort both dictionaries based on number of videos creating a list of tuples
-    # eg [('minecraftinfinity\n', 1), ('lyinggame\n', 1), ('dope101\n', 1)...]
+    # eg [('minecraftinfinity', 1), ('lyinggame', 1), ('dope101', 1)...]
     sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)
     # eg [('youtube', 748344), ('ign', 100159), ('screenplay', 60724)...]
     sorted_vids_provider = sorted(vids_by_provider.iteritems(), key=operator.itemgetter(1), reverse=True)

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -21,8 +21,10 @@ def main():
 
     fp.close()
 
-    # Sort both dictionaries based on number of videos
+    # Sort both dictionaries based on number of videos creating a list of tuples
+    # eg [('minecraftinfinity\n', 1), ('lyinggame\n', 1), ('dope101\n', 1)...]
     sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)
+    # eg [('youtube', 748344), ('ign', 100159), ('screenplay', 60724)...]
     sorted_vids_provider = sorted(vids_by_provider.iteritems(), key=operator.itemgetter(1), reverse=True)
 
     # Total number of videos, grouped by provider

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -18,6 +18,8 @@ def main():
         # Keep a tally of total videos by wiki, and by provider
         vids_by_provider[vals[0]] += int(vals[1])
         vids_by_wiki[vals[2]] += int(vals[1])
+
+    fp.close()
 
     # Sort both dictionaries based on number of videos
     sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -2,7 +2,7 @@
 import operator
 
 def main():
-    fp = open("reportTotalVideosOnWikiaNonPrem.csv", "r")
+    fp = open("reportTotalVideosOnWikiaAll.csv", "r")
     vids_by_provider = {}
     vids_by_wiki = {}
 

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -19,7 +19,7 @@ def main():
         vids_by_provider[vals[0]] += int(vals[1])
         vids_by_wiki[vals[2]] += int(vals[1])
 
-    # Sort both dictionary based on number of videos
+    # Sort both dictionaries based on number of videos
     sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)
     sorted_vids_provider = sorted(vids_by_provider.iteritems(), key=operator.itemgetter(1), reverse=True)
 

diff --git a/wikia_video_stats b/wikia_video_stats
@@ -0,0 +1,67 @@
+#!/bin/python
+import operator
+
+def main():
+    fp = open("reportTotalVideosOnWikiaNonPrem.csv", "r")
+    vids_by_provider = {}
+    vids_by_wiki = {}
+
+    # Lines are in the csv form provider,count,wiki
+    for line in fp.readlines():
+        vals = line.split(",")
+
+        if vals[0] not in vids_by_provider:
+            vids_by_provider[vals[0]] = 0
+        if vals[2] not in vids_by_wiki:
+            vids_by_wiki[vals[2]] = 0
+
+        # Keep a tally of total videos by wiki, and by provider
+        vids_by_provider[vals[0]] += int(vals[1])
+        vids_by_wiki[vals[2]] += int(vals[1])
+
+    # Sort both dictionary based on number of videos
+    sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)
+    sorted_vids_provider = sorted(vids_by_provider.iteritems(), key=operator.itemgetter(1), reverse=True)
+
+    # Total number of videos, grouped by provider
+    print "\nVideos By Provider"
+    print "================="
+    sum_total = 0
+    for vid in sorted_vids_provider:
+        sum_total += vid[1]
+        print "%s: %s" % (vid[0], vid[1])
+
+    # Total number of videos overall
+    print "\nTotal vids"
+    print "=========="
+    print sum_total
+
+    # Top 10 wikis with most videos (skipping the first 2 which are
+    # video and community)
+    print "\nTop 10 wikis"
+    print "============"
+    for i in range(2, 12):
+        print "%s: %s" % (sorted_vids_wiki[i][0].strip(), sorted_vids_wiki[i][1])
+
+
+    print "\nTotal wikis with videos"
+    print "======================="
+    print len(sorted_vids_wiki)
+
+
+    # Get number of videos contained in top 5% of wikis with videos
+    i = 1
+    top_count = 0
+    threshold = int(len(sorted_vids_wiki) * .05)
+    for vid in sorted_vids_wiki:
+        if i == threshold:
+            break
+        top_count += vid[1]
+        i += 1
+
+    print "\nTotal videos in top 5% of wikis with videos"
+    print "==========================================="
+    print top_count
+
+if __name__ == "__main__":
+    main()
No results found