X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/ec82b688426f48593b1c3af0c6df288cb62310ef..4e23ddd4d595d50de4b7fa0ab6bf3533bcf42723:/test/getcwidth.c?ds=sidebyside

diff --git a/test/getcwidth.c b/test/getcwidth.c
index f24172fe..9e1ba65c 100644
--- a/test/getcwidth.c
+++ b/test/getcwidth.c
@@ -1,7 +1,7 @@
 /*
- * getcwidth - Get the OS's idea of the width of a combining character
+ * getcwidth - Get the OS's idea of the width of Unicode codepoints
  *
- * This code is Copyright (c) 2012, by the authors of nmh.  See the
+ * This code is Copyright (c) 2013, by the authors of nmh.  See the
  * COPYRIGHT file in the root directory of the nmh distribution for
  * complete copyright information.
  */
@@ -20,61 +20,144 @@
 #include <wchar.h>
 #endif
 
+#ifdef MULTIBYTE_SUPPORT
+static void usage(char *);
+static void dumpwidth(void);
+static void getwidth(const char *);
+#endif /* MULTIBYTE_SUPPORT */
+
 int
 main(int argc, char *argv[])
 {
+#ifndef MULTIBYTE_SUPPORT
+	fprintf(stderr, "Nmh was not configured with multibyte support\n");
+	exit(1);
+#else /* MULTIBYTE_SUPPORT */
 	wchar_t c;
-	int charlen;
-	char *p;
+	int i;
+
+	setlocale(LC_ALL, "");
+
+	if (argc < 2)
+		usage(argv[0]);
+
+	if (strcmp(argv[1], "--dump") == 0) {
+		if (argc == 2) {
+			dumpwidth();
+			exit(0);
+		} else {
+			fprintf(stderr, "--dump cannot be combined with "
+				"other arguments\n");
+			exit(1);
+		}
+	}
 
 	/*
-	 * This is the UTF-8 for "n" + U+0308 (Combining Diaeresis)
+	 * Process each argument.  If it begins with "U+", then try to
+	 * convert it to a Unicode codepoint.  Otherwise, take each
+	 * string and get the total width
 	 */
 
-	unsigned char string[] = "n\xcc\x88";
+	for (i = 1; i < argc; i++) {
+		if (strncmp(argv[i], "U+", 2) == 0) {
+			/*
+			 * We're making a big assumption here that
+			 * wchar_t represents a Unicode codepoint.
+			 * That technically isn't valid unless the
+			 * C compiler defines __STDC_ISO_10646__, but
+			 * we're going to assume now that it works.
+			 */
+			errno = 0;
+			c = strtoul(argv[i] + 2, NULL, 16);
+			if (errno) {
+				fprintf(stderr, "Codepoint %s invalid\n",
+					argv[i]);
+				continue;
+			}
+			printf("%d\n", wcwidth(c));
+		} else {
+			getwidth(argv[i]);
+		}
+	}
 
-	setlocale(LC_ALL, "");
+	exit(0);
+}
 
-	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);
-	}
+static void
+usage(char *argv0)
+{
+	fprintf(stderr, "Usage: %s [--dump]\n", argv0);
+	fprintf(stderr, "       %s U+XXXX [...]\n", argv0);
+	fprintf(stderr, "       %s utf-8-sequence [...]\n", argv0);
+	fprintf(stderr, "Returns the column width of a Unicode codepoint "
+		"or UTF-8 character sequence\n");
+	fprintf(stderr, "\t--dump\tDump complete width table\n");
 
-#ifndef MULTIBYTE_SUPPORT
-	fprintf(stderr, "Nmh was not configured with multibyte support\n");
 	exit(1);
-#else
+}
+
+static void
+getwidth(const char *string)
+{
+	wchar_t c;
+	int charlen, charleft = strlen(string);
+	int length = 0;
+
 	/*
-	 * 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.
+	 * In theory we should be able to use wcswidth(), but since we're
+	 * testing out how the format libraries behave we'll do it a character
+	 * at a time.
 	 */
 
 	mbtowc(NULL, NULL, 0);
 
-	charlen = mbtowc(&c, (char *) string, strlen((char *) string));
+	while (charleft > 0) {
+		int clen;
 
-	if (charlen != 1) {
-		fprintf(stderr, "We expected a beginning character length "
-			"of 1, got %d instead\n", charlen);
-		exit(1);
-	}
+		charlen = mbtowc(&c, string, charleft);
 
-	p = (char *) (string + charlen);
+		if (charlen == 0)
+			break;
 
-	charlen = mbtowc(&c, p, strlen(p));
+		if (charlen < 0) {
+			fprintf(stderr, "Unable to convert string \"%s\"\n",
+				string);
+			return;
+		}
 
-	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);
+		if ((clen = wcwidth(c)) < 0) {
+			fprintf(stderr, "U+%04lX non-printable\n",
+				(unsigned long int) c);
+			return;
+		}
+
+		length += clen;
+		string += charlen;
+		charleft -= charlen;
 	}
 
-	printf("%d\n", wcwidth(c));
+	printf("%d\n", length);
+}
 
-	exit(0);
-#endif /* MULTIBYTE_SUPPORT */
+static void
+dumpwidth(void)
+{
+	wchar_t wc, low;
+	int width, lastwidth;
+
+	for (wc = 0, low = 1, lastwidth = wcwidth(1); wc < 0xffff; wc++) {
+		width = wcwidth(wc+1);
+		if (width != lastwidth) {
+			printf("%04lX - %04lX = %d\n", (unsigned long int) low,
+			       (unsigned long int) (wc), lastwidth);
+			low = wc+1;
+		}
+		lastwidth = width;
+	}
+
+	width = wcwidth(wc);
+	if (width == lastwidth)
+		printf("%04lX - %04lX = %d\n", (unsigned long int) low,
+		       (unsigned long int) (wc), width);
 }
+#endif /* MULTIBYTE_SUPPORT */