BUG: Add warning if rows have more columns than expected #33782
Conversation
|
Multiple tests are not passing due to the new warning. |
pandas/io/parsers.py
Outdated
| content = content[1:] | ||
|
|
||
| alldata = self._rows_to_cols(content) | ||
| if len(columns) != len(alldata) and notna(alldata[len(columns) :]).any(): |
There was a problem hiding this comment.
Is there a reason we need the notna check?
There was a problem hiding this comment.
In example mentioned in linked issue additional comma was added in one row. I assumed that additional commas are common and hence we might ignore them and don't raise a warning.
I'm using notna to check if data that won't be included contains only NaN values.
|
@mproszewska : Sorry for the long wait here! Overall, the solution is on the right track. |
|
Hello @mproszewska! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found: There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻 Comment last updated at 2020-10-08 19:27:23 UTC |
|
I have no idea how to check where this ipython directive error comes from. |
|
can you rebase and ill take a look at the ipython thing |
I rebased it |
|
on the docbuild, it looks like the following is issuing a warning under this PR, is issuing a warning here the correct thing to do? If so, then an |
I think so. First row has 3 values and the rest - 4. Where in in.rst should :okwarning: be added? maybe there's another way to do that. It shouldn't be a common warning. |
|
conceptually this is ok. pls merge master and will re-look (and yes we would have to either fix the warnings or assert_produces_warning, though prob should fix the incorrect usages). |
|
hmm this looks like overlapping with #38587 |
|
closing in favor of #38587 |
5 participants
read_csvwhen given an additional value on the first row of CSV file #33037black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff