`empty-diff: false` doesn't seem to work, still getting empty reports

[astyagun]

I'm subscribed to an RSS feed:

---
name: ...
url: ...
user_visible_url: ...
max_tries: 5
filter:
  - css:
      selector: 'item > title'
      method: xml
  - html2text
kind: url # Required when `display: {empty-diff: false}`
display:
  empty-diff: false # Don't show any notification when `diff_filter` below results in an empty diff
diff_filter:
  - grep: '^[+]'      # Include all lines starting with "+"
  - grepi: '^[+]{3}'  # Exclude the line starting with "+++"

I want to see notifications only when new items appear in the RSS feed (this works) and don't care about items disappearing from it. But empty-diff: false doesn't seem to work, as I do get notifications like these in Telegram:

=======================================
01. CHANGED: ...
=======================================

---------------------------------------
CHANGED: ... ( ... )
---------------------------------------

---------------------------------------

Is there a way to make it work? Or is this a bug?

[thp]

Which version of urlwatch are you using? empty-diff was added in version 2.25 (released 2022-03-15).

As for the feature, it's implemented like this:

urlwatch/lib/urlwatch/handler.py

Lines 206 to 209 in 0425631

	if (job_state.verb == 'changed'
	and not self.config['display'].get('empty-diff', True)
	and job_state.get_diff() == ''):
	continue

Possibly the diff isn't fully empty, just whitespace? In that case, try patching the above snippet with:

diff --git a/lib/urlwatch/handler.py b/lib/urlwatch/handler.py
index 521f273..9b20a35 100644
--- a/lib/urlwatch/handler.py
+++ b/lib/urlwatch/handler.py
@@ -205,7 +205,7 @@ class Report(object):
                        for verb in ('unchanged', 'new', 'error')):
                 if (job_state.verb == 'changed'
                         and not self.config['display'].get('empty-diff', True)
-                        and job_state.get_diff() == ''):
+                        and job_state.get_diff().strip() == ''):
                     continue
 
                 yield job_state

If that improves the situation for you, we could add that change to the source.

[astyagun]

The version is 2.25.

Will try this change. Thanks.

[astyagun]

Just got en empty notification again 🤔 The .strip() is present in handler.py.

[astyagun]

~$ docker exec urlwatch urlwatch-setup --version
urlwatch 2.25
~$ docker exec urlwatch urlwatch-setup --test-diff-filter 11
=== Filtered diff between state 0 and state 1 ===
+...
+...
+...
+...
=== Filtered diff between state 1 and state 2 ===

~$ docker exec urlwatch grep strip /usr/lib/python3/dist-packages/urlwatch/handler.py
                        and job_state.get_diff().strip() == ''):
~$

[astyagun]

After some debugging and asking an LLM to investigate it seems, that the problem in my case is that per-job display["empty-diff"] setting is not supported. I've moved it to global config and will see if this helps. Will close the issue for now and reopen if that hasn't helped.