rewrite: tw: improve twitter rewrite to force mp4 for videos in embed…

…ded tweets (#761)
webrecorder · Sep 1, 2022 · 8ef4ff1 · 8ef4ff1
1 parent 16135d9
commit 8ef4ff1
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 1 deletion.
diff --git a/pywb/rewrite/rewrite_dash.py b/pywb/rewrite/rewrite_dash.py
@@ -90,17 +90,25 @@ def rewrite_tw_dash(string, *args):
     try:
         best_variant = None
         best_bitrate = 0
+        best_src = ""
         max_bitrate = 5000000
 
         data = json.loads(string)
         for variant in data["variants"]:
-            if variant["content_type"] != "video/mp4":
+            if (("content_type" in variant and variant["content_type"] != "video/mp4") or
+                ("type" in variant and variant["type"] != "video/mp4")):
                 continue
 
             bitrate = variant.get("bitrate")
+            src = variant.get("src")
+
             if bitrate and bitrate > best_bitrate and bitrate <= max_bitrate:
                 best_variant = variant
                 best_bitrate = bitrate
+            # just compare src strings with dimensions
+            elif src and src > best_src:
+                best_variant = variant
+                best_src = src
 
         if best_variant:
             data["variants"] = [best_variant]

diff --git a/pywb/rules.yaml b/pywb/rules.yaml
@@ -61,6 +61,7 @@ rules:
 
     # twitter rules
     #=================================================================
+
     - url_prefix: 'com,twitter)/i/profiles/show/'
 
       fuzzy_lookup: '/profiles/show/.*with_replies\?.*(max_id=[^&]+)'
@@ -84,6 +85,14 @@ rules:
               function: 'pywb.rewrite.rewrite_dash:rewrite_tw_dash'
 
 
+    - url_prefix: ['com,twimg,syndication,cdn)/tweet-result']
+
+      rewrite:
+        js_regexs:
+            - match: 'video":(.*?viewCount":\d+})'
+              group: 1
+              function: 'pywb.rewrite.rewrite_dash:rewrite_tw_dash'
+
 
 
     # facebook rules