To latex position #35284
To latex position #35284
Conversation
WillAyd
left a comment
WillAyd
left a comment
There was a problem hiding this comment.
Thanks for the PR. Minor comment on test but otherwise I think this is OK
| assert result_l == expected_l | ||
|
|
||
| # test when only the label is provided | ||
| result_p = df.to_latex(position=the_position) |
There was a problem hiding this comment.
Can you create this as its own test? Same comment with one below
|
Hello @WillAyd , thanks for the review ! I did the changes you asked for, but one test seems to have failed, I'm not sure it's related to what I've written Would you also want tests when the caption, the label and the position is given ? |
|
Just restarted the failure let's see |
toobaz
left a comment
toobaz
left a comment
There was a problem hiding this comment.
Mostly looks good, two small comments (but I wouldn't object to solving the self._table_float one in a separate PR).
pandas/io/formats/latex.py
Outdated
| or self.position is not None | ||
| ): | ||
| # then write output in a nested table/tabular environment | ||
| buf.write(f"\\begin{{table}}") |
There was a problem hiding this comment.
Any obvious reason why this was moved up here? I would find buf.write(f"\\begin{{table}}{position_}") more readable that the write below.
There was a problem hiding this comment.
The reason I did this was only because the line was getting too long, but I can write it as it was for sure.
Maybe we should unify the way it's done in _write_tabular_begin and _write_longtable_begin ? But it may be nitpicking
There was a problem hiding this comment.
The reason I did this was only because the line was getting too long, but I can write it as it was for sure.
I do find it more readable - but if you want to break the line, you could do so at \n, so after the position.
Maybe we should unify the way it's done in
_write_tabular_beginand_write_longtable_begin?
Wow... indeed the difference between _write_tabular_begin and _write_longtable_begin seems to amount to the environment name: unless I'm mistaken, the merge is definitely worth doing (same for _end). By the way, if the function are unified feel free to move back the any([p is not None... down there.
pandas/io/formats/latex.py
Outdated
| self.caption is not None | ||
| or self.label is not None | ||
| or self.position is not None | ||
| ): |
There was a problem hiding this comment.
As this test grows, it would be better to have self._table_float = any([p is not None for p in (caption, label, position)]) inside __init__.py, to be used both here and in _write_tabular_end.
|
@toobaz I tried to match the two functions
|
pandas/io/formats/latex.py
Outdated
| if self.position is None: | ||
| position_ = "" | ||
| else: | ||
| position_ = f"[{self.position}]" |
My suggestion was actually to entirely suppress But this is not a strict necessity, at least in this PR. So if you prefer we can merge this PR after solving my last comment above. |
OK @toobaz I hadn't understood ! I tried doing something to resolve this, but I suggest doing it on a separate PR because I think it will get a bit messy. This work is currently on a branch of mine and we can look at it once this PR is merged |
|
Looks ready to merge to me. @WillAyd any idea on the two checks still failing? |
|
Thanks @SylvainLan |
4 participants
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff