See also: #32316

• closes DOC: Standardize references to pandas in the documentation #32316
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

In #36845 the RST files were updated to normalize usage of the word "pandas". The following tokens were replaced:

• `pandas`
• ``pandas``
• *pandas*
• **pandas**
• Pandas

In that PR, I wrongly said that I caught all patterns. I was wrong, and this PR address that (my apologies).

Misc Note

I'm also working on a rule for code_checks.sh to catch this in the future. I have the logic of the rule, but I'm struggling to get it to output nicely in such a way that it conforms with the rest of the rules in code_checks.sh. I want to leave this information here in case someone has any ideas I can bounce of off, or in case anyone else wants to try their hand at this. I could be spinning my wheels there for awhile and don't want it to hold up getting these changes in.

I used the following rule to find the remaining cases:

    MSG='Check doc/source RST files which format pandas incorrectly'; echo $MSG
    invgrep -R --include=*.rst -Poz '((\*|`)+)pandas\2|((((\.\. code-block:: ipython\s+)(\s +[^\n]*)+)|.\.\ [^\n]+https?[^\n]+|(`+).*\s+<https?[^\n]+>\8(__)?|`.*`_+\s+~+|.*:ref:`.*`)(*SKIP)(?!)|\bPandas\b(?!-))' doc/source/ | sed 's/\x0/\n/g'
    RET=$(($RET + $?)) ; echo $MSG "DONE"

This finds easily identifiable formatting errors (``pandas``, for example), but it also respects article titles and code examples in an RST code-blocks. It uses grep -v to get all text on a single line, but that causes error messages to lose the line number of the file that the error occurred on (everything in file foo.rst is reported to have happened on line 1). Because of this I'm not including the regex in this PR. If anyone is interested in giving me pointers on getting that bit fixed, or even taking my idea and running with it themselves, please do so.

Edit:

Because its desired to retain ``pandas`` in some places where it contextually make sense, regex won't be able to capture such uses. The above regex can instead be:

    MSG='Check doc/source RST files which format pandas incorrectly'; echo $MSG
    invgrep -R --include=*.rst -Poz '(\*+|`)pandas\1|((((\.\. code-block:: ipython\s+)(\s +[^\n]*)+)|.\.\ [^\n]+https?[^\n]+|(`+).*\s+<https?[^\n]+>\7(__)?|`.*`_+\s+~+|.*:ref:`.*`)(*SKIP)(?!)|\bPandas\b(?!-))' doc/source/ | sed 's/\x0/\n/g'
    RET=$(($RET + $?)) ; echo $MSG "DONE"

which will ignore ``pandas`` but catch `pandas`