Converting from raw in code citations to bibtex (1/N).

[zivy]

First PR out of N, converting raw inline citations with various styles to using a bibliography database and the doxygen cite command. Will take multiple PRs to fully complete #3662.

[zivy]

@dave3d is there a way to have the spellchecker ignore the bibliography tags which are used by the doxygen \cite command? Other than listing all the tags in the itk_dict.txt file.

[dave3d]

@dave3d is there a way to have the spellchecker ignore the bibliography tags which are used by the doxygen \cite command? Other than listing all the tags in the itk_dict.txt file.

So do you mean don't spell check the word after the "\cite" command? I'll have a look at it.

[zivy]

Yes. If there's the "\cite" directive ignore the next word which is the bibtex tag for the reference. Otherwise we'll have to add these exceptions for every reference because the tags are arbitrary and generally not "real" words. Thanks.

Another option is to use the bibtexparser package. The spellchecker will have bibliography files that it then parses and auto-magically adds the tags to the ignore list. This is more specific than ignore all words that appear after the "\cite" directive which is particular to doxygen and may not be the right thing to do in a different setting. Some relevant code:

import bibtexparser

bib_file = "my.bib"

with open(bib_file) as biblatex_file:
    bib_database = bibtexparser.load(biblatex_file)

for k in bib_database.get_entry_dict().keys():
    print(k)

[dzenanz]

Already great progress!

[jhlegarreta]

Looks great ❤️ 🚀. Thanks for doing this @zivy.

Two minor comments:

• Remove the dot from the subject message.
• Use the imperative mode.

Some comments that would require more work, as consistency would be important in this file:

• Use the same style for names, e.g. First Name Last Name and etc.
• Use brackets consistency for the field values. Use double brackets for all titles so that the cases are respected.
• Do not add whitespaces, indentations or add them systematically and consistently.
• Add first all articles, the books, then bookchapters, etc.: alphabetically.
• Sort the entries alphabetically within each group.
• Sort the field for each entry type the same way.
• Define and document the keys, e.g. lastnameYYYY and how collisions are resolved in the ITK documentation.

I know this is a lot of work, but the earlier is done, the easier it is to do for the upcoming cases as it sets a clear example.

Have you used some tool to gather this, if yes, please describe it in the commit message.

Is it still a draft or is it ready to be merged according to you @zivy ?

[hjmjohnson]

@zivy Very nice. Perhaps just bulk adding these citations to the itk dictionary would be the easiest.

[zivy]

@jhlegarreta I modified the commit message per your request (first two points). I don't have a style preference and all the variations found in the bib file were obtained from the online source of each of the publications, so there is inherent variability. I think this is ready for merge, but it won't pass the spellchecking due to the use of bibtex tags.

Sadly this is a tedious manual process because the inline citations are free-form, sometimes incomplete and sometimes partially incorrect (detecting all of them by assuming the year will be listed so searching for what looks like a year in all *.h files).

@hjmjohnson Will do a bulk dictionary update if @dave3d doesn't add this functionality in the next couple of days.

For now I think this PR can be merged.

[jhlegarreta]

@jhlegarreta I modified the commit message per your request (first two points). I don't have a style preference and all the variations found in the bib file were obtained from the online source of each of the publications, so there is inherent variability. I think this is ready for merge, but it won't pass the spellchecking due to the use of bibtex tags.

Sadly this is a tedious manual process because the inline citations are free-form, sometimes incomplete and sometimes partially incorrect (detecting all of them by assuming the year will be listed so searching for what looks like a year in all *.h files).

I know. I understand. Thanks for this work @zivy 💯.

[jhlegarreta]

Sorry, closed inadvertently.

[zivy]

@jhlegarreta It's been a long day. I am stepping away from the computer before I do more harm than good 😪

[zivy]

Some additional updates:

1. Per @albert-github's comment, populated the url field in the bibtex file using the doi information. This enables hyperlink creation using the default/current doxygen setup.
2. Per @jhlegarreta comment above, used the bibtex-tidy tool, the bibliography file has a more consistent look-and-feel (even though this doesn't matter in terms of functionality). Not added as a pre-commit hook because it currently does not support a check only setup.

[jhlegarreta]

Per @jhlegarreta comment above, used the bibtex-tidy tool, the bibliography file has a more consistent look-and-feel (even though this doesn't matter in terms of functionality). Not added as a pre-commit hook because it currently does not support a check only setup.

Simply great @zivy 💯. Thanks for doing that !