Handle the case in the test suite where wcwidth() can return different

values for combining Unicode codepoints on different operating systems.

diff --git a/.gitignore b/.gitignore
index 0be1cc289c10fa2465882aacca9c568d3d9b9e12..af04fac2f15bd4b796ac7dee3f728523e509452f 100644 (file)
--- a/.gitignore
+++ b/.gitignore
@@ -103,6 +103,7 @@ a.out.dSYM/
 /test/fakepop
 /test/getfullname
 /test/getfqdn
 /test/fakepop
 /test/getfullname
 /test/getfqdn

+/test/getcwidth

 
 # Removed by mostlyclean:
 *.o
 
 # Removed by mostlyclean:
 *.o

diff --git a/Makefile.am b/Makefile.am
index 953a9b7723b27df7ef1b8d46935c310dd65e5ae0..39ec8467fb1e21d4865802bd86de45793c2a6562 100644 (file)
--- a/Makefile.am
+++ b/Makefile.am
@@ -84,7 +84,8 @@ TESTS = test/ali/test-ali test/anno/test-anno \
        test/cleanup ## The "cleanup" test should always be last.
 
 check_SCRIPTS = test/common.sh
        test/cleanup ## The "cleanup" test should always be last.
 
 check_SCRIPTS = test/common.sh

-check_PROGRAMS = test/getfullname test/getfqdn test/fakepop test/fakesmtp
+check_PROGRAMS = test/getfullname test/getfqdn test/fakepop test/fakesmtp \
+                test/getcwidth

 DISTCHECK_CONFIGURE_FLAGS = DISABLE_SETGID_MAIL=1
 
 ##
 DISTCHECK_CONFIGURE_FLAGS = DISABLE_SETGID_MAIL=1
 
 ##

diff --git a/test/getcwidth.c b/test/getcwidth.c
new file mode 100644 (file)
index 0000000..2df307c
--- /dev/null
+++ b/test/getcwidth.c
@@ -0,0 +1,80 @@
+/*
+ * getcwidth - Get the OS's idea of the width of a combining character
+ *
+ * This code is Copyright (c) 2012, by the authors of nmh.  See the
+ * COPYRIGHT file in the root directory of the nmh distribution for
+ * complete copyright information.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#ifdef MULTIBYTE_SUPPORT
+#include <locale.h>
+#include <wchar.h>
+#endif
+
+int
+main(int argc, char *argv[])
+{
+       wchar_t c;
+       int charlen;
+       char *p;
+
+       /*
+        * This is the UTF-8 for "n" + U+0308 (Combining Diaeresis)
+        */
+
+       unsigned char string[] = "n\xcc\x88";
+
+       setlocale(LC_ALL, "");
+
+       if (argc != 1) {
+               fprintf(stderr, "Usage: %s\n", argv[0]);
+               fprintf(stderr, "Returns the column width of a UTF-8 "
+                       "multibyte character\n");
+               exit(1);
+       }
+
+#ifndef MULTIBYTE_SUPPORT
+       fprintf(stderr, "Nmh was not configured with multibyte support\n");
+       exit(1);
+#else
+       /*
+        * It's not clear to me that we can just call mbtowc() with a
+        * combining character; just to be safe, feed it in a base
+        * character first.
+        */
+
+       mbtowc(NULL, NULL, 0);
+
+       charlen = mbtowc(&c, string, strlen(string));
+
+       if (charlen != 1) {
+               fprintf(stderr, "We expected a beginning character length "
+                       "of 1, got %d instead\n", charlen);
+               exit(1);
+       }
+
+       p = string + charlen;
+
+       charlen = mbtowc(&c, p, strlen(p));
+
+       if (charlen != 2) {
+               fprintf(stderr, "We expected a multibyte character length "
+                       "of 2, got %d instead\n", charlen);
+               fprintf(stderr, "Are you using a UTF-8 locale?\n");
+               exit(1);
+       }
+
+       printf("%d\n", wcwidth(c));
+
+       exit(0);
+#endif /* MULTIBYTE_SUPPORT */
+}

diff --git a/test/scan/test-scan-multibyte b/test/scan/test-scan-multibyte
index 39f22240213411660172479935e32a0d09821380..1e9349a09163476e360554be40aae00b856a5bbc 100755 (executable)
--- a/test/scan/test-scan-multibyte
+++ b/test/scan/test-scan-multibyte
@@ -31,22 +31,43 @@ LC_ALL=en_US.UTF-8; export LC_ALL
 #
 # Create a test message with RFC 2047 headers we can scan
 #
 #
 # Create a test message with RFC 2047 headers we can scan
 #

+# In this Subject header in this message is a "n" with a Combining Diaeresis
+# (U+0308).  There is different interpretation of this character with respect
+# to wcwidth() (which is supposed to return the column width of a character).
+# We use a test program to determine what the output width of U+0308 is
+# and adjust our test output appropriately.
+#

 
 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
 From: David =?utf-8?q?=EF=AC=86?= Hubbins <hubbins@example.com>
 To: Sir Denis =?utf-8?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
 Date: Friday, 2 Mar 1984 00:00:00
 
 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
 From: David =?utf-8?q?=EF=AC=86?= Hubbins <hubbins@example.com>
 To: Sir Denis =?utf-8?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
 Date: Friday, 2 Mar 1984 00:00:00

-Subject: =?utf-8?q?Spin=CC=88al_Tap_=E2=86=92_Tap_into_America!?=
+Subject: =?utf-8?q?Sp=C4=B1n=CC=88al_Tap_=E2=86=92_Tap_into_America!?=

 
 Things are looking great!
 EOF
 
 
 Things are looking great!
 EOF
 

+width=`${MH_OBJ_DIR}/test/getcwidth`
+if test $? -ne 0; then
+    echo "getcwidth failed to run"
+    exit 1
+fi
+

 expected="$MH_TEST_DIR/$$.expected"
 actual="$MH_TEST_DIR/$$.actual"
 
 expected="$MH_TEST_DIR/$$.expected"
 actual="$MH_TEST_DIR/$$.actual"
 

+if test "$width" -eq 1; then

 cat > "$expected" <<EOF
 cat > "$expected" <<EOF

-  11  03/02 David st Hubbins    Spin̈al Tap → Tap into America!<<Things are look
+  11  03/02 David st Hubbins    Spın̈al Tap → Tap into America!<<Things are look

 EOF
 EOF

+elif test "$width" -eq 0; then
+cat > "$expected" <<EOF
+  11  03/02 David st Hubbins    Spın̈al Tap → Tap into America!<<Things are looki
+EOF
+else
+    echo "Unsupported width for combining diaeresis: $width"
+    exit 1
+fi

 
 scan -width 80 +inbox 11 > $actual || exit 1
 check "$expected" "$actual"
 
 scan -width 80 +inbox 11 > $actual || exit 1
 check "$expected" "$actual"