ImportError: cannot import name 'html2text' from partially initialized module 'urlwatch.html2txt' (most likely due to a circular import) (/usr/lib/python3/dist-packages/urlwatch/html2txt.py)

[jpiszcz]

After a recent apt-get dist-upgrade on Debian x86_64, python has been updated to v3.9 from v3.8 and now:

Traceback (most recent call last): File "/usr/lib/python3/dist-packages/urlwatch/handler.py", line 113, in process data = FilterBase.process(filter_kind, subfilter, self, data) File "/usr/lib/python3/dist-packages/urlwatch/filters.py", line 146, in process return filtercls(state.job, state).filter(data, subfilter) File "/usr/lib/python3/dist-packages/urlwatch/filters.py", line 291, in filter from .html2txt import html2text ImportError: cannot import name 'html2text' from partially initialized module 'urlwatch.html2txt' (most likely due to a circular import) (/usr/lib/python3/dist-packages/urlwatch/html2txt.py)

.. $ python3 -m pip install html2text Requirement already satisfied: html2text in /usr/lib/python3/dist-packages (2020.1.16)

[thp]

I cannot reproduce this issue with Python 3.9.0, html2text 2020.1.16 and urlwatch 2.21:

% python3.9 -m pip list | egrep 'html2text|urlwatch'
html2text  2020.1.16
urlwatch   2.21

I used the following urls.yaml for testing:

name: test
url: https://thp.io/
filter:
  - html2text:
    method: pyhtml2text

And then testing the filter using:

python3.9 urlwatch --test-filter 1

Which version of urlwatch do you have installed? Can you diff your local file /usr/lib/python3/dist-packages/urlwatch/html2txt.py to the file in the Git repository to see if there are any differences there?

[jpiszcz]

Hello,

No differences reported: urlwatch-master$ diff -u lib/urlwatch/html2txt.py /usr/lib/python3/dist-packages/urlwatch/html2txt.py urlwatch-master$

$ dpkg -S /usr/lib/python3/dist-packages/urlwatch/html2txt.py urlwatch: /usr/lib/python3/dist-packages/urlwatch/html2txt.py

$ dpkg -l urlwatch Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend |/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad) ||/ Name Version Architecture Description +++-==============-============-============-================================= ii urlwatch 2.21-1 all monitors webpages for you

Here is the recipe/definition that is having an issue (only started after python was upgraded from 3.8->3.9)

---
name: "IAD Non-Stop Destinations"
url: "http://www.flydulles.com/iad/nonstop-destinations"
filter:
 - html2text
 - grep: 'iad_intl_*'
---

[scottmac]

I have some backtraces when I tracked this down in #559 with a previous version of python.

[thp]

@jpiszcz If it's the same issue as @scottmac mentioned, can you please check if the latest commit in the Git master branch fixes this?

[jpiszcz]

Yup—this error only happens once and awhile—I’ve updated my crontab to point to urlwatch from the Gitmaster—will see if the issue recurs.

From: Thomas Perl notifications@github.com Sent: Friday, December 11, 2020 4:54 AM To: thp/urlwatch urlwatch@noreply.github.com Cc: jpiszcz jpiszcz@lucidpixels.com; Mention mention@noreply.github.com Subject: Re: [thp/urlwatch] ImportError: cannot import name 'html2text' from partially initialized module 'urlwatch.html2txt' (most likely due to a circular import) (/usr/lib/python3/dist-packages/urlwatch/html2txt.py) (#601)

@jpiszcz https://github.com/jpiszcz If it's the same issue as @scottmac https://github.com/scottmac mentioned, can you please check if the latest commit in the Git master branch fixes this?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/thp/urlwatch/issues/601#issuecomment-743095146 , or unsubscribe https://github.com/notifications/unsubscribe-auth/ABT5QWEFI2OJY32B3ICN26DSUHT3ZANCNFSM4UUFZGFA . https://github.com/notifications/beacon/ABT5QWCJEYZEYOFBIC4PF3DSUHT3ZA5CNFSM4UUFZGFKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOFRFLW2Q.gif

[polyzen]

Getting this on Arch Linux ARM now that the Python 3.9 rebuild has made it over. Applied the patch to 2.21, and will report back if it happens again.

[jpiszcz]

As of December 14, the error has yet to recur using the latest commit in Git master.

[thp]

Since it has been some days since December 14, I'm going to close this for now. Please feel free to reopen/comment if this happens again.