diff mbox series

[v2,4/5] wildmatch: use char instead of uchar

Message ID 20230226115021.1681834-5-masahiroy@kernel.org (mailing list archive)
State New, archived
Headers show
Series Clean up wildmatch.c | expand

Commit Message

Masahiro Yamada Feb. 26, 2023, 11:50 a.m. UTC
dowild() casts (char *) and (uchar *) back-and-forth, which is
ugly.

This file was imported from rsync, which started to use (unsigned char)
since the following commit:

 | commit e11c42511903adc6d27cf1671cc76fa711ea37e5
 | Author: Wayne Davison <wayned@samba.org>
 | Date:   Sun Jul 6 04:33:54 2003 +0000
 |
 |     - Added [:class:] handling to the character-class code.
 |     - Use explicit unsigned characters for proper set checks.
 |     - Made the character-class code honor backslash escapes.
 |     - Accept '^' as a class-negation character in addition to '!'.

Perhaps, it was needed because rsync relies on is*() from <ctypes.h>.

GIT has its own implementations, so the behavior is clear.

In fact, commit 4546738b58a0 ("Unlocalized isspace and friends")
says one of the motivations is "we want the right signed behaviour".

sane_istest() casts the given character to (unsigned char) anyway
before sane_ctype[] table lookup, so dowild() can use 'char'.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
---

(no changes since v1)

 wildmatch.c | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

Comments

Junio C Hamano Feb. 27, 2023, 8:07 p.m. UTC | #1
Masahiro Yamada <masahiroy@kernel.org> writes:

> GIT has its own implementations, so the behavior is clear.
>
> In fact, commit 4546738b58a0 ("Unlocalized isspace and friends")
> says one of the motivations is "we want the right signed behaviour".
>
> sane_istest() casts the given character to (unsigned char) anyway
> before sane_ctype[] table lookup, so dowild() can use 'char'.

Use of values taken from a char/uchar in sane_istest() is designed
to be safe, and between using char*/uchar* to scan pieces of memory
would not make much difference, so changes like ...

>  		case '*':
>  			if (*++p == '*') {
> -				const uchar *prev_p = p - 2;
> +				const char *prev_p = p - 2;

... this is safe, and so is ...

> -				const char *slash = strchr((char*)text, '/');
> +				const char *slash = strchr(text, '/');

... this.

But does the comparison between t_ch_upper and prev_ch behave the
same with and without this patch?

> -						uchar t_ch_upper = toupper(t_ch);
> +						char t_ch_upper = toupper(t_ch);
>  						if (t_ch_upper <= p_ch && t_ch_upper >= prev_ch)

Here t_ch is already known to pass islower(), so t_ch_upper would be
within a reasonable range and "char" should be able to store it
safely without its value wrapping around to negative.  Do we have a
similar guarantee on prev_ch, or can it be more than 128, which
would have caused the original to say "t_ch_upper is smaller than
prev_ch" but now because prev_ch can wrap around to a negative
value, leading the conditional to behave differently, or something?

>  							matched = 1;
>  					}
>  					p_ch = 0; /* This makes "prev_ch" get set to 0. */
>  				} else if (p_ch == '[' && p[1] == ':') {
> -					const uchar *s;
> +					const char *s;
>  					int i;
>  					for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
>  					if (!p_ch)
> @@ -237,5 +235,5 @@ static int dowild(const uchar *p, const uchar *text, unsigned int flags)
>  /* Match the "pattern" against the "text" string. */
>  int wildmatch(const char *pattern, const char *text, unsigned int flags)
>  {
> -	return dowild((const uchar*)pattern, (const uchar*)text, flags);
> +	return dowild(pattern, text, flags);
>  }
diff mbox series

Patch

diff --git a/wildmatch.c b/wildmatch.c
index 93800b8eac..7dffd783cb 100644
--- a/wildmatch.c
+++ b/wildmatch.c
@@ -12,21 +12,19 @@ 
 #include "cache.h"
 #include "wildmatch.h"
 
-typedef unsigned char uchar;
-
 #define CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
 				    && *(class) == *(litmatch) \
-				    && strncmp((char*)class, litmatch, len) == 0)
+				    && strncmp(class, litmatch, len) == 0)
 
 /* Match pattern "p" against "text" */
-static int dowild(const uchar *p, const uchar *text, unsigned int flags)
+static int dowild(const char *p, const char *text, unsigned int flags)
 {
-	uchar p_ch;
-	const uchar *pattern = p;
+	char p_ch;
+	const char *pattern = p;
 
 	for ( ; (p_ch = *p) != '\0'; text++, p++) {
 		int matched, match_slash, negated;
-		uchar t_ch, prev_ch;
+		char t_ch, prev_ch;
 		if ((t_ch = *text) == '\0' && p_ch != '*')
 			return WM_ABORT_ALL;
 		if ((flags & WM_CASEFOLD) && isupper(t_ch))
@@ -50,7 +48,7 @@  static int dowild(const uchar *p, const uchar *text, unsigned int flags)
 			continue;
 		case '*':
 			if (*++p == '*') {
-				const uchar *prev_p = p - 2;
+				const char *prev_p = p - 2;
 				while (*++p == '*') {}
 				if (!(flags & WM_PATHNAME))
 					/* without WM_PATHNAME, '*' == '**' */
@@ -90,10 +88,10 @@  static int dowild(const uchar *p, const uchar *text, unsigned int flags)
 				 * with WM_PATHNAME matches the next
 				 * directory
 				 */
-				const char *slash = strchr((char*)text, '/');
+				const char *slash = strchr(text, '/');
 				if (!slash)
 					return WM_NOMATCH;
-				text = (const uchar*)slash;
+				text = slash;
 				/* the slash is consumed by the top-level for loop */
 				break;
 			}
@@ -160,13 +158,13 @@  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);
+						char 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. */
 				} else if (p_ch == '[' && p[1] == ':') {
-					const uchar *s;
+					const char *s;
 					int i;
 					for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
 					if (!p_ch)
@@ -237,5 +235,5 @@  static int dowild(const uchar *p, const uchar *text, unsigned int flags)
 /* Match the "pattern" against the "text" string. */
 int wildmatch(const char *pattern, const char *text, unsigned int flags)
 {
-	return dowild((const uchar*)pattern, (const uchar*)text, flags);
+	return dowild(pattern, text, flags);
 }