new trout from rejs ( https://twitter.com/kingsbookstore/status/1620432807196778496 )

[irc.git] / commands.py

diff --git a/commands.py b/commands.py
index 721089d3912d6e86d466c2be1d65edbf3253f7e9..eec69e7f416dd0d085f60e8b08dda26e1ae3b57b 100755 (executable)
--- a/commands.py
+++ b/commands.py
@@ -2,6 +2,7 @@
 import string, cPickle, random, urllib, sys, time, re, os, twitter, subprocess, datetime, urlparse, hashlib
 from collections import defaultdict
 from irclib import irc_lower, nm_to_n
+import json
 
 try:
     from blame_filter import bfd
@@ -571,13 +572,38 @@ def getTweet(urlstring,twitapi,inclusion=False,recurlvl=0):
         tweeter_name = tweeter_name + " RTing " + status.user.name #.encode('UTF-8', 'replace')
     tweetText = status.full_text
     if status.media:
-        replacements = defaultdict( list )
+        replacements = defaultdict(list)
+
         for medium in status.media:
             replacements[medium.url].append(medium.media_url_https)
 
-        for k,v in replacements.items():
+        # The twitter-api 'conveniently' parses this for you and
+        # throws away the actual video URLs, so we have to take the
+        # JSON and reparse it :sadpanda:
+        # This is particularly annoying because we don't know
+        # for sure that status.media and the JSON 'media' entry
+        # have the same elements in the same order.  Probably they
+        # do but maybe twitter-api randomly reorganised things or
+        # filtered the list or something.  So instead we go through
+        # the JSON and handle the media urls, discarding whatever
+        # unfortunate thing we have put in replacements already.
+        parsed_tweet = json.loads(status.AsJsonString())
+        for medium in parsed_tweet.get('media', []):
+            if medium['type'] == 'video':
+                best = { 'bitrate': -1 }
+                for vt in medium['video_info']['variants']:
+                    if (vt.get('content_type') == 'video/mp4' and
+                        vt.get('bitrate', -1) > best['bitrate']):
+                        best = vt
+                if 'url' in best:
+                    video_url = best['url'].split('?',1)[0]
+                    duration = medium['video_info']['duration_millis']
+                    # ^ duration_millis is a string
+                    duration = "%.1f" % (float(duration)/1000.)
+                    video_desc = "%s (%ss)" % (video_url, duration)
+                    replacements[medium['url']] = [video_desc]
 
-            v = [re.sub(r"/tweet_video_thumb/([\w\-]+).jpg", r"/tweet_video/\1.mp4", link) for link in v]
+        for k,v in replacements.items():
             if len(v) > 1:
                 replacementstring = "[" +  " ; ".join(v) +"]"
             else: