twitwi 0.22.0__tar.gz → 0.22.1__tar.gz

{twitwi-0.22.0/twitwi.egg-info → twitwi-0.22.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: twitwi
-Version: 0.22.0
+Version: 0.22.1
 Summary: A collection of Twitter-related helper functions for python.
 Home-page: http://github.com/medialab/twitwi
 Author: Béatrice Mazoyer, Guillaume Plique, Benjamin Ooghe-Tabanou

{twitwi-0.22.0 → twitwi-0.22.1}/setup.py

@@ -5,7 +5,7 @@ with open("./README.md", "r") as f:
 
 setup(
     name="twitwi",
-    version="0.22.0",
+    version="0.22.1",
     description="A collection of Twitter-related helper functions for python.",
     long_description=long_description,
     long_description_content_type="text/markdown",

{twitwi-0.22.0 → twitwi-0.22.1}/twitwi/bluesky/normalizers.py

@@ -594,11 +594,13 @@ def normalize_post(
 
         # Process quotes
         if quoted_data and "value" in quoted_data:
-            if quoted_data["cid"] != post["quoted_cid"]:
+            # We're checking on the uri as the cid can be different in some cases,
+            # and the uri seems to be unique for each post
+            if quoted_data["uri"] != post["quoted_uri"]: 
                 raise BlueskyPayloadError(
                     post["url"],
-                    "inconsistent quote cid found between record.embed.record.cid & embed.record.cid: %s %s"
-                    % (post["quoted_cid"], quoted_data),
+                    "inconsistent quote uri found between record.embed.record.uri & embed.record.uri: %s %s"
+                    % (post["quoted_uri"], quoted_data),
                 )
 
             quoted_data["record"] = quoted_data["value"]
@@ -706,7 +708,12 @@ def normalize_post(
             repost_data["indexedAt"], locale=locale, source="bluesky"
         )
 
-    post["text"] = text.decode("utf-8")
+    try:
+        post["text"] = text.decode("utf-8")
+    except UnicodeDecodeError as e:
+        raise UnicodeDecodeError(
+            f"Failed to decode post text: {e}\nPost URL: {post['url']}\nOriginal text bytes: {text}"
+        )
 
     if collection_source is not None:
         post["collected_via"] = [collection_source]

{twitwi-0.22.0 → twitwi-0.22.1/twitwi.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: twitwi
-Version: 0.22.0
+Version: 0.22.1
 Summary: A collection of Twitter-related helper functions for python.
 Home-page: http://github.com/medialab/twitwi
 Author: Béatrice Mazoyer, Guillaume Plique, Benjamin Ooghe-Tabanou