Re: [PATCH v5] wildmatch: properly fold case everywhere

[Junio C Hamano <gitster@pobox.com>]

Anthony Ramine <n.oxyde@gmail.com> writes:

> ase folding is not done correctly when matching against the [:upper:]
> character class and uppercased character ranges (e.g. A-Z).
> Specifically, an uppercase letter fails to match against any of them
> when case folding is requested because plain characters in the pattern
> and the whole string are preemptively lowercased to handle the base case
> fast.
>
> That optimization is kept and ISLOWER() is used in the [:upper:] case
> when case folding is requested, while matching against a character range
> is retried with toupper() if the character was lowercase, as the bounds
> of the range itself cannot be modified (in a case-insensitive context,
> [A-_] is not equivalent to [a-_]).
>
> Signed-off-by: Anthony Ramine <n.oxyde@gmail.com>
> Reviewed-by: Duy Nguyen <pclouds@gmail.com>
> ---

Thanks.

>  t/t3070-wildmatch.sh | 55 ++++++++++++++++++++++++++++++++++++++++++++++------
>  wildmatch.c          |  7 +++++++
>  2 files changed, 56 insertions(+), 6 deletions(-)
>
> I added Duy as reviewer and fixed a typo in the commit message reported by
> Eric Sunshine.
>
> diff --git a/t/t3070-wildmatch.sh b/t/t3070-wildmatch.sh
> index 4c37057..38446a0 100755
> --- a/t/t3070-wildmatch.sh
> +++ b/t/t3070-wildmatch.sh
> @@ -6,20 +6,20 @@ test_description='wildmatch tests'
>  
>  match() {
>      if [ $1 = 1 ]; then
> -	test_expect_success "wildmatch:    match '$3' '$4'" "
> +	test_expect_success "wildmatch:     match '$3' '$4'" "
>  	    test-wildmatch wildmatch '$3' '$4'
>  	"
>      else
> -	test_expect_success "wildmatch: no match '$3' '$4'" "
> +	test_expect_success "wildmatch:  no match '$3' '$4'" "
>  	    ! test-wildmatch wildmatch '$3' '$4'
>  	"
>      fi
>      if [ $2 = 1 ]; then
> -	test_expect_success "fnmatch:      match '$3' '$4'" "
> +	test_expect_success "fnmatch:       match '$3' '$4'" "
>  	    test-wildmatch fnmatch '$3' '$4'
>  	"
>      elif [ $2 = 0 ]; then
> -	test_expect_success "fnmatch:   no match '$3' '$4'" "
> +	test_expect_success "fnmatch:    no match '$3' '$4'" "
>  	    ! test-wildmatch fnmatch '$3' '$4'
>  	"
>  #    else

Is the above about aligning $3/$4 across checks of different types
(i.e. purely cosmetic)?  I am not complaining; just making sure if
there is nothing deeper going on.

It is outside the scope of this change, but the shell style of this
script (most notably use of [] instead of test) needs to be fixed
someday, preferrably soon, including the commented out else clause
at the end of the hunk.

> @@ -235,4 +247,35 @@ pathmatch 1 abcXdefXghi '*X*i'
>  pathmatch 1 ab/cXd/efXg/hi '*/*X*/*/*i'
>  pathmatch 1 ab/cXd/efXg/hi '*Xg*i'
>  
> +# Case-sensitivy features
> +match 0 x 'a' '[A-Z]'
> +match 1 x 'A' '[A-Z]'
> +match 0 x 'A' '[a-z]'
> +match 1 x 'a' '[a-z]'
> +match 0 x 'a' '[[:upper:]]'
> +match 1 x 'A' '[[:upper:]]'
> +match 0 x 'A' '[[:lower:]]'
> +match 1 x 'a' '[[:lower:]]'
> +match 0 x 'A' '[B-Za]'
> +match 1 x 'a' '[B-Za]'
> +match 0 x 'A' '[B-a]'
> +match 1 x 'a' '[B-a]'
> +match 0 x 'z' '[Z-y]'
> +match 1 x 'Z' '[Z-y]'
> +
> +imatch 1 'a' '[A-Z]'

Do we want "# Case-insensitivity features" commment here as well?

> +imatch 1 'A' '[A-Z]'
> +imatch 1 'A' '[a-z]'
> +imatch 1 'a' '[a-z]'
> +imatch 1 'a' '[[:upper:]]'
> +imatch 1 'A' '[[:upper:]]'
> +imatch 1 'A' '[[:lower:]]'
> +imatch 1 'a' '[[:lower:]]'
> +imatch 1 'A' '[B-Za]'
> +imatch 1 'a' '[B-Za]'
> +imatch 1 'A' '[B-a]'
> +imatch 1 'a' '[B-a]'
> +imatch 1 'z' '[Z-y]'
> +imatch 1 'Z' '[Z-y]'

> +
>  test_done
> diff --git a/wildmatch.c b/wildmatch.c
> index 7192bdc..f91ba99 100644
> --- a/wildmatch.c
> +++ b/wildmatch.c
> @@ -196,6 +196,11 @@ static int dowild(const uchar *p, const uchar *text, unsigned int flags)
>  					}
>  					if (t_ch <= p_ch && t_ch >= prev_ch)
>  						matched = 1;
> +					else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch)) {
> +						uchar t_ch_upper = toupper(t_ch);
> +						if (t_ch_upper <= p_ch && t_ch_upper >= prev_ch)
> +							matched = 1;
> +					}
>  					p_ch = 0; /* This makes "prev_ch" get set to 0. */

Hmm, this looks somewhat strange.

 * At the beginning of the outermost "per characters in the text"
   loop, we seem to downcase t_ch when WM_CASEFOLD is set.
 * Also at the same place, we also seem to downcase p_ch under the
   same condition.

which makes me wonder why the fix is not like this:

	+	if (flags & WM_CASEFOLD) {
        +		if (ISUPPER(p_ch))
        +			p_ch = tolower(p_ch);
        +		if (prev_ch && ISUPPER(prev_ch))
        +			prev_ch = tolower(prev_ch);
	+	}
		if (t_ch <= p_ch && t_ch >= prev_ch)
			matched = 1;
		p_ch = 0; /* This sets "prev_ch" to 0 */


Ahh, OK, the "seemingly strange" construct is about handling a range
like "[Z-y]"; we do not want to upcase or downcase the p_ch/prev_ch
make the range "[z-y]" (empty) which would exclude bytes like "^",
"_" or even "Z".

And it is also OK to downcase p_ch in a single-letter case, not the
characters in a range, at the beginning of the outermost loop,
because we always compare for equality against t_ch (which is
downcased) in that case.

> @@ -245,6 +250,8 @@ static int dowild(const uchar *p, const uchar *text, unsigned int flags)
>  					} else if (CC_EQ(s,i, "upper")) {
>  						if (ISUPPER(t_ch))
>  							matched = 1;
> +						else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch))
> +							matched = 1;

This also looks somewhat strange but correct in that t_ch is already
downcased so we do not need a corresponding change for CC_EQ("lower")
codepath.

Interesting.  Will apply.

Thanks.