Sometimes it is useful to be able to compare two string descriptors
case-insentively, like c_strcasecmp.
2024-10-13 Bruno Haible <br...@clisp.org>
string-desc: New function string_desc_c_casecmp.
* lib/string-desc.h: New declaration.
* lib/string-desc.c: Include <limits.h>, c-ctype.h.
(string_desc_c_casecmp): New function.
* modules/string-desc (Depends-on): Add c-ctype.
* tests/test-string-desc.c (main): Add tests of string_desc_c_casecmp.
diff --git a/lib/string-desc.c b/lib/string-desc.c
index 58a87e1220..78992729bb 100644
--- a/lib/string-desc.c
+++ b/lib/string-desc.c
@@ -25,10 +25,12 @@
/* Specification and inline definitions. */
#include "string-desc.h"
+#include <limits.h>
#include <stdarg.h>
#include <stdlib.h>
#include <string.h>
+#include "c-ctype.h"
#include "ialloc.h"
#include "full-write.h"
@@ -83,6 +85,28 @@ string_desc_cmp (string_desc_t a, string_desc_t b)
}
}
+int
+string_desc_c_casecmp (string_desc_t a, string_desc_t b)
+{
+ /* Don't use memcasecmp here, since it uses the current locale, not the
+ "C" locale. */
+ idx_t an = string_desc_length (a);
+ idx_t bn = string_desc_length (b);
+ const char *ap = string_desc_data (a);
+ const char *bp = string_desc_data (b);
+ idx_t n = (an < bn ? an : bn);
+ idx_t i;
+ for (i = 0; i < n; i++)
+ {
+ int ac = c_tolower ((unsigned char) ap[i]);
+ int bc = c_tolower ((unsigned char) bp[i]);
+ if (ac != bc)
+ return (UCHAR_MAX <= INT_MAX ? ac - bc : _GL_CMP (ac, bc));
+ }
+ /* Here i = n = min (an, bn). */
+ return _GL_CMP (an, bn);
+}
+
ptrdiff_t
string_desc_index (string_desc_t s, char c)
{
diff --git a/lib/string-desc.h b/lib/string-desc.h
index ff9c86d6a0..0b8287121f 100644
--- a/lib/string-desc.h
+++ b/lib/string-desc.h
@@ -102,6 +102,12 @@ extern bool string_desc_endswith (string_desc_t s,
string_desc_t suffix);
'unsigned char'. */
extern int string_desc_cmp (string_desc_t a, string_desc_t b);
+/* Return > 0, == 0, or < 0 if A > B, A == B, A < B.
+ Either A or B must be entirely ASCII.
+ This uses a lexicographic ordering, where the bytes are compared as
+ 'unsigned char', ignoring case, in the "C" locale. */
+extern int string_desc_c_casecmp (string_desc_t a, string_desc_t b);
+
/* Return the index of the first occurrence of C in S,
or -1 if there is none. */
extern ptrdiff_t string_desc_index (string_desc_t s, char c);
diff --git a/modules/string-desc b/modules/string-desc
index 047e2d390c..e3f092dd09 100644
--- a/modules/string-desc
+++ b/modules/string-desc
@@ -9,6 +9,7 @@ lib/string-desc-contains.c
Depends-on:
stdbool
idx
+c-ctype
ialloc
extern-inline
memchr
diff --git a/tests/test-string-desc.c b/tests/test-string-desc.c
index c39077c567..fbe538687b 100644
--- a/tests/test-string-desc.c
+++ b/tests/test-string-desc.c
@@ -94,6 +94,23 @@ main (int argc, char *argv[])
ASSERT (string_desc_cmp (s2, s1) > 0);
ASSERT (string_desc_cmp (s2, s2) == 0);
+ /* Test string_desc_c_casecmp. */
+ ASSERT (string_desc_c_casecmp (s0, s0) == 0);
+ ASSERT (string_desc_c_casecmp (s0, s1) < 0);
+ ASSERT (string_desc_c_casecmp (s0, s2) < 0);
+ ASSERT (string_desc_c_casecmp (s1, s0) > 0);
+ ASSERT (string_desc_c_casecmp (s1, s1) == 0);
+ ASSERT (string_desc_c_casecmp (s1, s2) < 0);
+ ASSERT (string_desc_c_casecmp (s2, s0) > 0);
+ ASSERT (string_desc_c_casecmp (s2, s1) > 0);
+ ASSERT (string_desc_c_casecmp (s2, s2) == 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("acab"),
string_desc_from_c ("AcAB")) == 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("AcAB"),
string_desc_from_c ("acab")) == 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("aca"),
string_desc_from_c ("AcAB")) < 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("AcAB"),
string_desc_from_c ("aca")) > 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("aca"),
string_desc_from_c ("Aca\377")) < 0);
+ ASSERT (string_desc_c_casecmp (string_desc_from_c ("Aca\377"),
string_desc_from_c ("aca")) > 0);
+
/* Test string_desc_index. */
ASSERT (string_desc_index (s0, 'o') == -1);
ASSERT (string_desc_index (s2, 'o') == 12);
