Handle wide characters and ECMA-48 CSI sequences in the format string when calculating string widths.

2024-12-08 21:38:38 +00:00
parent f495303177
commit 35d5b9efab
12 changed files with 664 additions and 595 deletions
@@ -22,77 +22,27 @@


 /*
- * Return the number of display columns needed to show the given string.
- *
- * To do this, we convert it to a wide character string, and use the wide
- * character display width function "wcswidth()" on it.
+ * Return the number of display columns needed to show the given
+ * null-terminated string.
 *
 * If NLS is disabled, or the string cannot be converted, this is just the
 * same as "strlen()".
 */
 static size_t display_width(const char *string)
 {
-	size_t width;
-
-#if defined(ENABLE_NLS) && defined(HAVE_WCHAR_H)
-	size_t wide_char_count;
-	size_t wide_string_buffer_size;
-	wchar_t *wide_string;
+	size_t bytes;

 	if (NULL == string)
 		return 0;

-	/*@-nullpass@ */
+	bytes = strlen(string);		    /* flawfinder: ignore */
 	/*
-	 * splint note: mbstowcs() manual page on Linux explicitly says it
-	 * takes NULL.
+	 * flawfinder rationale: we have already checked for NULL, and it is
+	 * explicitly required of the caller to provide a null-terminated
+	 * string.
 	 */
-	wide_char_count = mbstowcs(NULL, string, 0);
-	/*@+nullpass@ */
-	if (wide_char_count == (size_t) -1) {
-		debug("%s: %s: %s", "mbstowcs", string, strerror(errno));
-		return strlen(string);	    /* flawfinder: ignore */
-		/*
-		 * flawfinder rationale: we have already checked for NULL,
-		 * and we don't know the size of the originating buffer so
-		 * can't use strnlen(); it is up to the caller to provide a
-		 * terminated string.
-		 */
-	}

-	wide_string_buffer_size = sizeof(*wide_string) * (1 + wide_char_count);
-	wide_string = malloc(wide_string_buffer_size);
-	if (NULL == wide_string) {
-		perror("malloc");
-		return strlen(string);	    /* flawfinder: ignore */
-		/* flawfinder rationale: see above. */
-	}
-	memset(wide_string, 0, wide_string_buffer_size);
-
-	if (mbstowcs(wide_string, string, 1 + wide_char_count) == (size_t) -1) {
-		debug("%s: %s: %s", "mbstowcs", string, strerror(errno));
-		width = strlen(string);	    /* flawfinder: ignore */
-		/* flawfinder rationale: see above. */
-	} else if (NULL != wide_string) {
-		/*@-unrecog @ */
-		/* splint seems unable to see the prototype. */
-		width = wcswidth(wide_string, wide_char_count);
-		/*@+unrecog @ */
-	} else {
-		width = 0;
-	}
-
-	free(wide_string);
-
-#else				/* ! defined(ENABLE_NLS) && defined(HAVE_WCHAR_H) */
-	if (NULL == string)
-		return 0;
-
-	width = strlen(string);		    /* flawfinder: ignore */
-	/* flawfinder rationale: see above. */
-#endif				/* defined(ENABLE_NLS) && defined(HAVE_WCHAR_H) */
-
-	return width;
+	return pv_strwidth(string, bytes);
 }


@@ -134,9 +84,9 @@ static void display_word_wrap_7bit(const char *string, size_t display_width, siz


 /*
- * Output a string to standard output, word wrapping to "display_width"
- * display character positions, and left-padding any new lines after the
- * first one with "left_margin" spaces.
+ * Output a null-terminated string to standard output, word wrapping to
+ * "display_width" display character positions, and left-padding any new
+ * lines after the first one with "left_margin" spaces.
 *
 * Wide characters are handled if NLS is enabled, but if they can't be, this
 * falls back to a version which just counts bytes as characters.
@@ -153,7 +103,7 @@ static void display_word_wrap(const char *string, size_t display_width, size_t l
 		return;

 	/*@-nullpass@ */
-	/* splint note: see earlier mbstowcs() call. */
+	/* splint note: see mbstowcs() call in pv_strwidth(). */
 	wide_char_count = mbstowcs(NULL, string, 0);
 	/*@+nullpass@ */
 	if (wide_char_count == (size_t) -1) {
@@ -181,7 +131,9 @@ static void display_word_wrap(const char *string, size_t display_width, size_t l
 	start_idx = 0;
 	chars_remaining = wide_char_count;

+	/*@-unrecog@ *//* splint seems unable to see the prototype for wcswidth(). */
 	while (chars_remaining > 0 && wcswidth(&(wide_string[start_idx]), chars_remaining) > display_width) {
+		/*@+unrecog@ */
 		size_t next_idx;

 		end_idx = start_idx + display_width;
@@ -1530,6 +1530,9 @@ static void pv__format_init(pvstate_t state, /*@null@ */ const char *format_supp
 			display->format[segment].bytes = str_bytes;
 			display->format[segment].width = pv_strwidth(&(display_format[str_start]), str_bytes);

+			debug("format[%d]:[%.*s], length=%d, width=%d", segment, str_bytes, display_format + str_start,
+			      str_bytes, display->format[segment].width);
+
 		} else {
 			char dummy_buffer[4];	/* flawfinder: ignore - unused. */

@@ -1777,6 +1780,9 @@ bool pv_format(pvstate_t state, /*@null@ */ const char *format_supplied, pvdispl

 		new_display_string_bytes += segment->bytes;
 		new_display_string_width += segment->width;
+
+		debug("segment[%d]: bytes=%d, width=%d: [%.*s]", segment_idx, segment->bytes, segment->width,
+		      segment->bytes, display->display_buffer + display_buffer_offset - segment->bytes);
 	}

 	debug("%s: %d", "new display string length in bytes", (int) new_display_string_bytes);
@@ -13,6 +13,12 @@
 #include <stdarg.h>
 #include <string.h>
 #include <errno.h>
+#if defined(ENABLE_NLS) && defined(HAVE_WCHAR_H)
+#include <wchar.h>
+#if defined(HAVE_WCTYPE_H)
+#include <wctype.h>
+#endif
+#endif


 /*
@@ -161,9 +167,9 @@ char *pv_strdup(const char *original)
 void *pv_memrchr(const void *buffer, int match, size_t length)
 {
 #ifdef HAVE_MEMRCHR
-	/*@-unrecog @ *//* splint doesn't know of memrchr() */
+	/*@-unrecog@ *//* splint doesn't know of memrchr() */
 	return memrchr(buffer, match, length);
-	/*@+unrecog @ */
+	/*@+unrecog@ */
 #else
 	unsigned char *ptr;

@@ -183,14 +189,118 @@ void *pv_memrchr(const void *buffer, int match, size_t length)


 /*
- * Return the displayed width of a string.
+ * Return the number of display columns needed to show the
+ * non-null-terminated string "string" whose length in bytes is "bytes".
+ *
+ * Skips ECMA-48 CSI (ESC [ ...) sequences, but any other control characters
+ * are treated as printable.
+ *
+ * To do this, we convert it to a wide character string, and use the wide
+ * character display width function "wcswidth()" on it.
+ *
+ * If NLS is disabled, or the string cannot be converted, this just returns
+ * the value of "bytes".
+ *
+ * Note that this function uses internal buffers if the string is short
+ * enough, otherwise it has to call malloc() and free(), so it becomes less
+ * efficient with larger strings.
 */
 size_t pv_strwidth(const char *string, size_t bytes)
 {
+	char *allocated_raw = NULL;
+	static char internal_raw[256];	 /* flawfinder: ignore - bounded */
+	char *raw_string = NULL;
+	size_t read_pos, write_pos;
+	size_t raw_bytes, width;
+#if defined(ENABLE_NLS) && defined(HAVE_WCHAR_H)
+	size_t wide_char_count;
+	size_t wide_string_buffer_size;
+	wchar_t *allocated_wide = NULL;
+	static wchar_t internal_wide[256];	/* flawfinder: ignore - bounded */
+	wchar_t *wide_string = NULL;
+#endif				/* defined(ENABLE_NLS) && defined(HAVE_WCHAR_H) */
+
 	if (NULL == string)
 		return 0;
 	if (0 == bytes)
 		return 0;
-	/* TODO: actually calculate displayed width rather than assuming. */
-	return bytes;
+
+	if (bytes < sizeof(internal_raw) - 1) {
+		raw_string = internal_raw;
+	} else {
+		allocated_raw = calloc(1, 1 + bytes);
+		if (NULL == allocated_raw)
+			return bytes;
+		raw_string = allocated_raw;
+	}
+
+	/* Copy the original string, skipping ECMA-48 CSI sequences. */
+	for (read_pos = 0, write_pos = 0; read_pos < bytes; read_pos++) {
+		if ((string[read_pos] != '\033') || (read_pos >= bytes - 1) || (string[read_pos + 1] != '[')) {
+			raw_string[write_pos++] = string[read_pos];
+			continue;
+		}
+		read_pos += 2;
+		while ((read_pos < bytes - 1)
+		       && ((string[read_pos] >= '0' && string[read_pos] <= '9')
+			   || (';' == string[read_pos])
+		       )
+		    ) {
+			read_pos++;
+		}
+	}
+	raw_string[write_pos] = '\0';
+	raw_bytes = write_pos;
+
+	width = raw_bytes;
+
+#if defined(ENABLE_NLS) && defined(HAVE_WCHAR_H)
+	/*@-nullpass@ */
+	/*
+	 * splint note: mbstowcs() manual page on Linux explicitly says it
+	 * takes NULL.
+	 */
+	wide_char_count = mbstowcs(NULL, raw_string, 0);
+	/*@+nullpass@ */
+	if (wide_char_count == (size_t) -1) {
+		debug("%s: %s: %s", "mbstowcs", raw_string, strerror(errno));
+		if (NULL != allocated_raw)
+			free(allocated_raw);
+		return raw_bytes;
+	}
+
+	wide_string_buffer_size = sizeof(*wide_string) * (1 + wide_char_count);
+
+	if (wide_string_buffer_size < sizeof(internal_wide)) {
+		wide_string = internal_wide;
+	} else {
+		allocated_wide = malloc(wide_string_buffer_size);
+		if (NULL == allocated_wide) {
+			perror("malloc");
+			if (NULL != allocated_raw)
+				free(allocated_raw);
+			return raw_bytes;
+		}
+		wide_string = allocated_wide;
+	}
+	memset(wide_string, 0, wide_string_buffer_size);
+
+	if (mbstowcs(wide_string, raw_string, 1 + wide_char_count) == (size_t) -1) {
+		debug("%s: %s: %s", "mbstowcs", raw_string, strerror(errno));
+	} else if (NULL != wide_string) {
+		/*@-unrecog@ *//* splint seems unable to see the prototype. */
+		width = wcswidth(wide_string, wide_char_count);
+		/*@+unrecog@ */
+	} else {
+		width = 0;
+	}
+
+	if (NULL != allocated_wide)
+		free(allocated_wide);
+#endif				/* defined(ENABLE_NLS) && defined(HAVE_WCHAR_H) */
+
+	if (NULL != allocated_raw)
+		free(allocated_raw);
+
+	return width;
 }