Media command

CHANGES.rst

@@ -1,8 +1,8 @@
 Changes
 =======
 
-1.5.0 (in development)
-----------------------
+1.5.0
+-----
 
 * Python 2 is not supported anymore.
 * Use the ``full_text`` field to retrieve tweet's text, fall back to ``text`` if
@@ -16,6 +16,7 @@ Changes
 * The ``Tweet.bounding_box`` property is introduced, it is always a polygon.
 * The ``--filters`` option for ``filter`` to define what filters are used.
 * Refactored communication with Twitter and internal stream handling.
+* New ``media`` command.
 
 1.3.0
 -----

README.rst

@@ -3,3 +3,5 @@ Poultry
 
 :documentation: http://poultry.readthedocs.org
 :homepage: https://github.com/dimazest/poultry
+
+Poultry is a tweet collection manager.

poultry/consumers.py

@@ -4,7 +4,7 @@
 import logging
 import sys
 import time
-import json
+import csv
 
 try:
     from Queue import Full
@@ -417,3 +417,28 @@ def extract_retweets(target):
             target.send(retweeted_status)
 
         target.send(raw_tweet)
+
+
+@consumer
+def print_media(output=None):
+    """Print media items."""
+    if output is None:
+        output = sys.stdout
+
+    field_names = 'tweet_id', 'index', 'media_id', 'type', 'media_url'
+    writer = csv.DictWriter(output, fieldnames=field_names)
+
+    while True:
+        tweet = yield
+        media = tweet.parsed.get('extended_entities', {}).get('media', [])
+
+        for i, m in enumerate(media):
+            writer.writerow(
+                {
+                    'tweet_id': tweet.id,
+                    'index': i,
+                    'media_id': m['id'],
+                    'type': m['type'],
+                    'media_url': m['media_url'],
+                }
+            )

poultry/main.py

@@ -104,3 +104,13 @@ def timeline(producer, window=('w', '%Y-%m-%d-%H', '')):
             ),
         ),
     )
+
+
+@command()
+def media(producer, output):
+    """Retrieve media urls."""
+    producer(
+        consumers.to_tweet(
+            consumers.print_media(output=output)
+        )
+    )