ignore words become case-insensitive if it is only lowercase.

[fujitatomoya]

according to https://github.com/codespell-project/codespell?tab=readme-ov-file#ignoring-words,

When ignoring false positives, note that spelling errors are case-insensitive but words to ignore are case-sensitive.

but the following example tells me that is not implemented when word to ignore is only with lowercase.

tomoyafujita@~/DVT/codespell >cat test.md 
ros2 should not be detected.
ROS2 should be detected.
tomoyafujita@~/DVT/codespell >cat dictionary.txt 
ROS2->ROS 2
tomoyafujita@~/DVT/codespell >codespell -D dictionary.txt,- test.md
test.md:1: ros2 ==> ros 2
test.md:2: ROS2 ==> ROS 2
tomoyafujita@~/DVT/codespell >codespell --ignore-words-list ros2 -D dictionary.txt,- test.md
tomoyafujita@~/DVT/codespell >codespell --ignore-words-list ROS2 -D dictionary.txt,- test.md
test.md:1: ros2 ==> ros 2

i believe this is unexpected behavior, at least it is not consistent with documentation above.

[fujitatomoya]

#2323 could be related.

[DimitriPapadopoulos]

The relevant code was introduced by #3272:

codespell/codespell_lib/_codespell.py

Lines 699 to 707 in 3f0532c

	def process_ignore_words(
	words: Iterable[str], ignore_words: Set[str], ignore_words_cased: Set[str]
	) -> None:
	for word in words:
	word = word.strip()
	if word == word.lower():
	ignore_words.add(word)
	else:
	ignore_words_cased.add(word)

I seem to recall that the spirit of that change is that lowercase words to ignore are handled in a case-insensitive way (ignore_words), while words with uppercase letters (ignore_words_cased) are handled in a case-sensitive way. Does this make sense to you?

In any case, you're right: at the very least the documentation should have been updated.

[DimitriPapadopoulos]

While #2323 is similar, I don't think it's directly related.

[fujitatomoya]

@DimitriPapadopoulos thanks for checking on this, really appreciate that.

I seem to recall that the spirit of that change is that lowercase words to ignore are handled in a case-insensitive way (ignore_words), while words with uppercase letters (ignore_words_cased) are handled in a case-sensitive way. Does this make sense to you?

unfortunately no, it does not work for our use case.
what we want to do is that ignoring the words in a case-sensitive way including lower cases only word. in this particular case, we want to ignore ros2 only, but ROS2, RoS2 or Ros2.

i think this is inconsistent behavior for user perspective.

tomoyafujita@~/DVT/codespell >cat test.md 
ros2 should not be detected.
ROS2 should be detected.
tomoyafujita@~/DVT/codespell >cat dictionary.txt 
ROS2->ROS 2
tomoyafujita@~/DVT/codespell >codespell -D dictionary.txt,- test.md
test.md:1: ros2 ==> ros 2
test.md:2: ROS2 ==> ROS 2
tomoyafujita@~/DVT/codespell >codespell --ignore-words-list ros2 -D dictionary.txt,- test.md
tomoyafujita@~/DVT/codespell >codespell --ignore-words-list ROS2 -D dictionary.txt,- test.md
test.md:1: ros2 ==> ros 2

When ignoring false positives, note that spelling errors are case-insensitive but words to ignore are case-sensitive. makes sense to me, that gives the flexibility for the user to filter out the specific words in a case-sensitive way?

[fujitatomoya]

@DimitriPapadopoulos

could you take a look at #3658?

since this is the 1st time for me to make PR against codespell-project, i would be missing something. but i did went through code and documentation, and i checked local verification is passing.