Implement support for equivalence classes ([=e=]) when the mapping is

one-to-one (SUSv3)

3 files changed, 45 insertions, 14 deletions

diff --git a/usr.bin/tr/str.c b/usr.bin/tr/str.c
index 46bf340..90f7335 100644
--- a/usr.bin/tr/str.c
+++ b/usr.bin/tr/str.c
@@ -216,14 +216,13 @@ c_class(a, b)
 	return (strcmp(((const CLASS *)a)->name, ((const CLASS *)b)->name));
 }
 
-/*
- * English doesn't have any equivalence classes, so for now
- * we just syntax check and grab the character.
- */
 static void
 genequiv(s)
 	STR *s;
 {
+	int i, p, pri;
+	char src[2], dst[3];
+
 	if (*s->str == '\\') {
 		s->equiv[0] = backslash(s);
 		if (*s->str != '=')
@@ -233,6 +232,28 @@ genequiv(s)
 		if (s->str[1] != '=')
 			errx(1, "misplaced equivalence equals sign");
 	}
+
+	/*
+	 * Calculate the set of all characters in the same equivalence class
+	 * as the specified character (they will have the same primary
+	 * collation weights).
+	 * XXX Knows too much about how strxfrm() is implemented. Assumes
+	 * it fills the string with primary collation weight bytes. Only one-
+	 * to-one mappings are supported.
+	 */
+	src[0] = s->equiv[0];
+	src[1] = '\0';
+	if (strxfrm(dst, src, sizeof(dst)) == 1) {
+		pri = (unsigned char)*dst;
+		for (p = 1, i = 1; i < NCHARS; i++) {
+			*src = i;
+			if (strxfrm(dst, src, sizeof(dst)) == 1 && pri &&
+			    pri == (unsigned char)*dst)
+				s->equiv[p++] = i;
+		}
+		s->equiv[p] = OOBCH;
+	}
+
 	s->str += 2;
 	s->cnt = 0;
 	s->state = SET;
diff --git a/usr.bin/tr/tr.1 b/usr.bin/tr/tr.1
index 2875ccf..b1ac00c 100644
--- a/usr.bin/tr/tr.1
+++ b/usr.bin/tr/tr.1
@@ -200,15 +200,9 @@ in these classes, see
 .Xr ctype 3
 and related manual pages.
 .It [=equiv=]
-Represents all characters or collating (sorting) elements belonging to
-the same equivalence class as
-.Ar equiv .
-If
-there is a secondary ordering within the equivalence class, the characters
-are ordered in ascending sequence.
-Otherwise, they are ordered after their encoded values.
-An example of an equivalence class might be ``c'' and ``ch'' in Spanish;
-English has no equivalence classes.
+Represents all characters belonging to the same equivalence class as
+.Ar equiv ,
+ordered by their encoded values.
 .It [#*n]
 Represents
 .Ar n
@@ -228,6 +222,17 @@ If
 has a leading zero, it is interpreted as an octal value, otherwise,
 it's interpreted as a decimal value.
 .El
+.Sh ENVIRONMENT
+The
+.Ev LANG ,
+.Ev LC_ALL ,
+.Ev LC_CTYPE
+and
+.Ev LC_COLLATE
+environment variables affect the execution of
+.Nm
+as described in
+.Xr environ 7 .
 .Sh DIAGNOSTICS
 .Ex -std
 .Sh EXAMPLES
@@ -245,6 +250,11 @@ Translate the contents of file1 to upper-case.
 Strip out non-printable characters from file1.
 .Pp
 .D1 Li "tr -cd \*q[:print:]\*q < file1"
+.Pp
+Remove diacritical marks from all accented variants of the letter
+.Sq e :
+.Pp
+.Dl "tr \*q[=e=]\*q \*qe\*q"
 .Sh COMPATIBILITY
 System V has historically implemented character ranges using the syntax
 ``[c-c]'' instead of the ``c-c'' used by historic
diff --git a/usr.bin/tr/tr.c b/usr.bin/tr/tr.c
index 359ef05..c74e1d0 100644
--- a/usr.bin/tr/tr.c
+++ b/usr.bin/tr/tr.c
@@ -105,7 +105,7 @@ main(argc, argv)
 	int ch, cnt, lastch, *p;
 	int cflag, dflag, sflag, isstring2;
 
-	(void) setlocale(LC_CTYPE, "");
+	(void)setlocale(LC_ALL, "");
 
 	cflag = dflag = sflag = 0;
 	while ((ch = getopt(argc, argv, "cdsu")) != -1)