[git commit master] fold: unicode support. Based on a patch by Tomas Heinrich <heinrich.tomas at gmail.com>

Denys Vlasenko vda.linux at googlemail.com
Mon Jan 4 19:49:58 UTC 2010


commit: http://git.busybox.net/busybox/commit/?id=28055028a709020ba7eb44f9e5037d0a952b51d6
branch: http://git.busybox.net/busybox/commit/?id=refs/heads/master

General Unicode support is tweaked to expose unicode_status.

function                                             old     new   delta
init_unicode                                           -      77     +77
write2stdout                                           -      19     +19
adjust_column                                         68      71      +3
unicode_status                                         -       1      +1
unicode_is_enabled                                     1       -      -1
grep_main                                            780     773      -7
fold_main                                            619     552     -67
check_unicode_in_env                                  77       -     -77
------------------------------------------------------------------------------
(add/remove: 3/2 grow/shrink: 1/2 up/down: 100/-152)          Total: -52 bytes

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 coreutils/df.c                |    2 +-
 coreutils/expand.c            |    2 +-
 coreutils/fold.c              |  161 ++++++++++++++++++++++-------------------
 coreutils/ls.c                |    2 +-
 include/unicode.h             |   18 ++++-
 libbb/lineedit.c              |    2 +-
 libbb/progress.c              |    2 +-
 libbb/unicode.c               |   46 +++++++-----
 modutils/lsmod.c              |    2 +-
 networking/udhcp/dumpleases.c |    2 +-
 testsuite/fold.tests          |   49 ++++++++++++-
 11 files changed, 183 insertions(+), 105 deletions(-)

diff --git a/coreutils/df.c b/coreutils/df.c
index 83794ad..bcde783 100644
--- a/coreutils/df.c
+++ b/coreutils/df.c
@@ -58,7 +58,7 @@ int df_main(int argc UNUSED_PARAM, char **argv)
 	const char *disp_units_hdr = NULL;
 	char *chp;
 
-	check_unicode_in_env();
+	init_unicode();
 
 #if ENABLE_FEATURE_HUMAN_READABLE && ENABLE_FEATURE_DF_FANCY
 	opt_complementary = "k-mB:m-Bk:B-km";
diff --git a/coreutils/expand.c b/coreutils/expand.c
index e08eb1d..649b4c1 100644
--- a/coreutils/expand.c
+++ b/coreutils/expand.c
@@ -144,7 +144,7 @@ int expand_main(int argc UNUSED_PARAM, char **argv)
 		"all\0"              No_argument       "a"
 	;
 #endif
-	check_unicode_in_env();
+	init_unicode();
 
 	if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) {
 		IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts);
diff --git a/coreutils/fold.c b/coreutils/fold.c
index 56a3466..cbea31f 100644
--- a/coreutils/fold.c
+++ b/coreutils/fold.c
@@ -9,8 +9,8 @@
 
    Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
 */
-
 #include "libbb.h"
+#include "unicode.h"
 
 /* Must match getopt32 call */
 #define FLAG_COUNT_BYTES        1
@@ -20,39 +20,53 @@
 /* Assuming the current column is COLUMN, return the column that
    printing C will move the cursor to.
    The first column is 0. */
-static int adjust_column(int column, char c)
+static int adjust_column(unsigned column, char c)
 {
-	if (!(option_mask32 & FLAG_COUNT_BYTES)) {
-		if (c == '\b') {
-			if (column > 0)
-				column--;
-		} else if (c == '\r')
+	if (option_mask32 & FLAG_COUNT_BYTES)
+		return ++column;
+
+	if (c == '\t')
+		return column + 8 - column % 8;
+
+	if (c == '\b') {
+		if ((int)--column < 0)
 			column = 0;
-		else if (c == '\t')
-			column = column + 8 - column % 8;
-		else			/* if (isprint(c)) */
+	}
+	else if (c == '\r')
+		column = 0;
+	else { /* just a printable char */
+		if (unicode_status != UNICODE_ON /* every byte is a new char */
+		 || (c & 0xc0) != 0x80 /* it isn't a 2nd+ byte of a Unicode char */
+		) {
 			column++;
-	} else
-		column++;
+		}
+	}
 	return column;
 }
 
+/* Note that this function can write NULs, unlike fputs etc. */
+static void write2stdout(const void *buf, unsigned size)
+{
+	fwrite(buf, 1, size, stdout);
+}
+
 int fold_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 int fold_main(int argc UNUSED_PARAM, char **argv)
 {
 	char *line_out = NULL;
-	int allocated_out = 0;
-	char *w_opt;
-	int width = 80;
-	int i;
-	int errs = 0;
+	const char *w_opt = "80";
+	unsigned width;
+	smallint exitcode = EXIT_SUCCESS;
+
+	init_unicode();
 
 	if (ENABLE_INCLUDE_SUSv2) {
 		/* Turn any numeric options into -w options.  */
+		int i;
 		for (i = 1; argv[i]; i++) {
-			char const *a = argv[i];
-
-			if (*a++ == '-') {
+			const char *a = argv[i];
+			if (*a == '-') {
+				a++;
 				if (*a == '-' && !a[1]) /* "--" */
 					break;
 				if (isdigit(*a))
@@ -62,8 +76,7 @@ int fold_main(int argc UNUSED_PARAM, char **argv)
 	}
 
 	getopt32(argv, "bsw:", &w_opt);
-	if (option_mask32 & FLAG_WIDTH)
-		width = xatoul_range(w_opt, 1, 10000);
+	width = xatou_range(w_opt, 1, 10000);
 
 	argv += optind;
 	if (!*argv)
@@ -72,79 +85,81 @@ int fold_main(int argc UNUSED_PARAM, char **argv)
 	do {
 		FILE *istream = fopen_or_warn_stdin(*argv);
 		int c;
-		int column = 0;		/* Screen column where next char will go. */
-		int offset_out = 0;	/* Index in 'line_out' for next char. */
+		unsigned column = 0;     /* Screen column where next char will go */
+		unsigned offset_out = 0; /* Index in 'line_out' for next char */
 
 		if (istream == NULL) {
-			errs |= EXIT_FAILURE;
+			exitcode = EXIT_FAILURE;
 			continue;
 		}
 
 		while ((c = getc(istream)) != EOF) {
-			if (offset_out + 1 >= allocated_out) {
-				allocated_out += 1024;
-				line_out = xrealloc(line_out, allocated_out);
+			/* We grow line_out in chunks of 0x1000 bytes */
+			if ((offset_out & 0xfff) == 0) {
+				line_out = xrealloc(line_out, offset_out + 0x1000);
 			}
-
+ rescan:
+			line_out[offset_out] = c;
 			if (c == '\n') {
-				line_out[offset_out++] = c;
-				fwrite(line_out, sizeof(char), (size_t) offset_out, stdout);
+				write2stdout(line_out, offset_out + 1);
 				column = offset_out = 0;
 				continue;
 			}
- rescan:
 			column = adjust_column(column, c);
+			if (column <= width || offset_out == 0) {
+				/* offset_out == 0 case happens
+				 * with small width (say, 1) and tabs.
+				 * The very first tab already goes to column 8,
+				 * but we must not wrap it */
+				offset_out++;
+				continue;
+			}
 
-			if (column > width) {
-				/* This character would make the line too long.
-				   Print the line plus a newline, and make this character
-				   start the next line. */
-				if (option_mask32 & FLAG_BREAK_SPACES) {
-					/* Look for the last blank. */
-					int logical_end;
-
-					for (logical_end = offset_out - 1; logical_end >= 0; logical_end--) {
-						if (isblank(line_out[logical_end])) {
-							break;
-						}
+			/* This character would make the line too long.
+			 * Print the line plus a newline, and make this character
+			 * start the next line */
+			if (option_mask32 & FLAG_BREAK_SPACES) {
+				unsigned i;
+				unsigned logical_end;
+
+				/* Look for the last blank. */
+				for (logical_end = offset_out - 1; (int)logical_end >= 0; logical_end--) {
+					if (!isblank(line_out[logical_end]))
+						continue;
+
+					/* Found a space or tab.
+					 * Output up to and including it, and start a new line */
+					logical_end++;
+					/*line_out[logical_end] = '\n'; - NO! this nukes one buffered character */
+					write2stdout(line_out, logical_end);
+					putchar('\n');
+					/* Move the remainder to the beginning of the next line.
+					 * The areas being copied here might overlap. */
+					memmove(line_out, line_out + logical_end, offset_out - logical_end);
+					offset_out -= logical_end;
+					for (column = i = 0; i < offset_out; i++) {
+						column = adjust_column(column, line_out[i]);
 					}
-					if (logical_end >= 0) {
-						/* Found a blank.  Don't output the part after it. */
-						logical_end++;
-						fwrite(line_out, sizeof(char), (size_t) logical_end, stdout);
-						bb_putchar('\n');
-						/* Move the remainder to the beginning of the next line.
-						   The areas being copied here might overlap. */
-						memmove(line_out, line_out + logical_end, offset_out - logical_end);
-						offset_out -= logical_end;
-						for (column = i = 0; i < offset_out; i++) {
-							column = adjust_column(column, line_out[i]);
-						}
-						goto rescan;
-					}
-				}
-				if (offset_out == 0) {
-					line_out[offset_out++] = c;
-					continue;
+					goto rescan;
 				}
-				line_out[offset_out++] = '\n';
-				fwrite(line_out, sizeof(char), (size_t) offset_out, stdout);
-				column = offset_out = 0;
-				goto rescan;
+				/* No blank found, wrap will split the overlong word */
 			}
-
-			line_out[offset_out++] = c;
-		}
+			/* Output what we accumulated up to now, and start a new line */
+			line_out[offset_out] = '\n';
+			write2stdout(line_out, offset_out + 1);
+			column = offset_out = 0;
+			goto rescan;
+		} /* while (not EOF) */
 
 		if (offset_out) {
-			fwrite(line_out, sizeof(char), (size_t) offset_out, stdout);
+			write2stdout(line_out, offset_out);
 		}
 
 		if (fclose_if_not_stdin(istream)) {
-			bb_simple_perror_msg(*argv);	/* Avoid multibyte problems. */
-			errs |= EXIT_FAILURE;
+			bb_simple_perror_msg(*argv);
+			exitcode = EXIT_FAILURE;
 		}
 	} while (*++argv);
 
-	fflush_stdout_and_exit(errs);
+	fflush_stdout_and_exit(exitcode);
 }
diff --git a/coreutils/ls.c b/coreutils/ls.c
index 13c863c..5ea3a0b 100644
--- a/coreutils/ls.c
+++ b/coreutils/ls.c
@@ -945,7 +945,7 @@ int ls_main(int argc UNUSED_PARAM, char **argv)
 
 	INIT_G();
 
-	check_unicode_in_env();
+	init_unicode();
 
 	all_fmt = LIST_SHORT |
 		(ENABLE_FEATURE_LS_SORTFILES * (SORT_NAME | SORT_FORWARD));
diff --git a/include/unicode.h b/include/unicode.h
index e006147..9f27657 100644
--- a/include/unicode.h
+++ b/include/unicode.h
@@ -5,10 +5,17 @@
 #ifndef UNICODE_H
 #define UNICODE_H 1
 
+enum {
+	UNICODE_UNKNOWN = 0,
+	UNICODE_OFF = 1,
+	UNICODE_ON = 2,
+};
+
 #if !ENABLE_FEATURE_ASSUME_UNICODE
 
 # define bb_mbstrlen(string) strlen(string)
-# define check_unicode_in_env() ((void)0)
+# define unicode_status UNICODE_OFF
+# define init_unicode() ((void)0)
 
 #else
 
@@ -18,16 +25,19 @@ size_t bb_mbstrlen(const char *string) FAST_FUNC;
 
 #  include <wchar.h>
 #  include <wctype.h>
-#  define check_unicode_in_env() ((void)0)
+extern uint8_t unicode_status;
+void init_unicode(void) FAST_FUNC;
 
 # else
 
 /* Crude "locale support" which knows only C and Unicode locales */
 
 #  if !ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
-#   define check_unicode_in_env() ((void)0)
+#   define unicode_status UNICODE_ON
+#   define init_unicode() ((void)0)
 #  else
-void check_unicode_in_env(void) FAST_FUNC;
+extern uint8_t unicode_status;
+void init_unicode(void) FAST_FUNC;
 #  endif
 
 #  undef MB_CUR_MAX
diff --git a/libbb/lineedit.c b/libbb/lineedit.c
index c73e6b7..b73f1d6 100644
--- a/libbb/lineedit.c
+++ b/libbb/lineedit.c
@@ -1763,7 +1763,7 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
 		return len;
 	}
 
-	check_unicode_in_env();
+	init_unicode();
 
 // FIXME: audit & improve this
 	if (maxsize > MAX_LINELEN)
diff --git a/libbb/progress.c b/libbb/progress.c
index f6f2692..3a245ae 100644
--- a/libbb/progress.c
+++ b/libbb/progress.c
@@ -78,7 +78,7 @@ void FAST_FUNC bb_progress_update(bb_progress_t *p,
 	}
 
 #if ENABLE_FEATURE_ASSUME_UNICODE
-	check_unicode_in_env();
+	init_unicode();
 	/* libbb candidate? */
 	{
 		wchar_t wbuf21[21];
diff --git a/libbb/unicode.c b/libbb/unicode.c
index 544528a..9d316df 100644
--- a/libbb/unicode.c
+++ b/libbb/unicode.c
@@ -7,7 +7,9 @@
  * Licensed under GPL version 2, see file LICENSE in this tarball for details.
  */
 #include "libbb.h"
-# include "unicode.h"
+#include "unicode.h"
+
+uint8_t unicode_status;
 
 size_t FAST_FUNC bb_mbstrlen(const char *string)
 {
@@ -17,32 +19,38 @@ size_t FAST_FUNC bb_mbstrlen(const char *string)
 	return width;
 }
 
-#if !ENABLE_LOCALE_SUPPORT
+#if ENABLE_LOCALE_SUPPORT
+
+/* Unicode support using libc */
+
+void FAST_FUNC init_unicode(void)
+{
+	/* In unicode, this is a one character string */
+	static const char unicode_0x394[] = { 0xce, 0x94, 0 };
+
+	if (unicode_status != UNICODE_UNKNOWN)
+		return;
+
+	unicode_status = bb_mbstrlen(unicode_0x394) == 1 ? UNICODE_ON : UNICODE_OFF;
+}
+
+#else
 
 /* Crude "locale support" which knows only C and Unicode locales */
 
-/* unicode_is_enabled:
- * 0: not known yet,
- * 1: not unicode (IOW: assuming one char == one byte)
- * 2: unicode
- */
-# if !ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
-#  define unicode_is_enabled 2
-# else
-static smallint unicode_is_enabled;
-void FAST_FUNC check_unicode_in_env(void)
+# if ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
+void FAST_FUNC init_unicode(void)
 {
 	char *lang;
 
-	if (unicode_is_enabled)
+	if (unicode_status != UNICODE_UNKNOWN)
 		return;
-	unicode_is_enabled = 1;
 
+	unicode_status = UNICODE_OFF;
 	lang = getenv("LANG");
 	if (!lang || !(strstr(lang, ".utf") || strstr(lang, ".UTF")))
 		return;
-
-	unicode_is_enabled = 2;
+	unicode_status = UNICODE_ON;
 }
 # endif
 
@@ -85,7 +93,7 @@ static size_t wcrtomb_internal(char *s, wchar_t wc)
 
 size_t FAST_FUNC wcrtomb(char *s, wchar_t wc, mbstate_t *ps UNUSED_PARAM)
 {
-	if (unicode_is_enabled != 2) {
+	if (unicode_status != UNICODE_ON) {
 		*s = wc;
 		return 1;
 	}
@@ -97,7 +105,7 @@ size_t FAST_FUNC wcstombs(char *dest, const wchar_t *src, size_t n)
 {
 	size_t org_n = n;
 
-	if (unicode_is_enabled != 2) {
+	if (unicode_status != UNICODE_ON) {
 		while (n) {
 			wchar_t c = *src++;
 			*dest++ = c;
@@ -137,7 +145,7 @@ size_t FAST_FUNC mbstowcs(wchar_t *dest, const char *src, size_t n)
 {
 	size_t org_n = n;
 
-	if (unicode_is_enabled != 2) {
+	if (unicode_status != UNICODE_ON) {
 		while (n) {
 			unsigned char c = *src++;
 
diff --git a/modutils/lsmod.c b/modutils/lsmod.c
index 582503b..20d9cf1 100644
--- a/modutils/lsmod.c
+++ b/modutils/lsmod.c
@@ -49,7 +49,7 @@ int lsmod_main(int argc UNUSED_PARAM, char **argv UNUSED_PARAM)
 # if ENABLE_FEATURE_ASSUME_UNICODE
 	size_t name_len;
 # endif
-	check_unicode_in_env();
+	init_unicode();
 
 	printf("%-24sSize  Used by", "Module");
 	check_tainted();
diff --git a/networking/udhcp/dumpleases.c b/networking/udhcp/dumpleases.c
index 017c801..97605f0 100644
--- a/networking/udhcp/dumpleases.c
+++ b/networking/udhcp/dumpleases.c
@@ -43,7 +43,7 @@ int dumpleases_main(int argc UNUSED_PARAM, char **argv)
 
 	applet_long_options = dumpleases_longopts;
 #endif
-	check_unicode_in_env();
+	init_unicode();
 
 	opt_complementary = "=0:a--r:r--a";
 	opt = getopt32(argv, "arf:", &file);
diff --git a/testsuite/fold.tests b/testsuite/fold.tests
index ca0e511..17721a1 100755
--- a/testsuite/fold.tests
+++ b/testsuite/fold.tests
@@ -7,8 +7,53 @@
 # testing "test name" "options" "expected result" "file input" "stdin"
 
 testing "fold -s" "fold -w 7 -s" \
-	"123456\n\t\nasdf" \
+        "123456\n\t\nasdf" \
+        "" \
+        "123456\tasdf" \
+
+testing "fold -w1" "fold -w1" \
+	"q\nq\n \nw\n \ne\ne\ne\n \nr\n \nt\nt\nt\nt\n \ny" \
+	"" \
+	"qq w eee r tttt y" \
+
+testing "fold with NULs" "fold -sw22" \
+	"\
+The NUL is here:>\0< \n\
+and another one is \n\
+here:>\0< - they must \n\
+be preserved
+" \
+	"" \
+	"The NUL is here:>\0< and another one \
+is here:>\0< - they must be preserved
+" \
+
+# The text was taken from English and Ukrainian wikipedia pages
+testing "fold -sw66 with unicode input" "fold -sw66" \
+	"\
+The Andromeda Galaxy (pronounced /ænˈdrɒmədə/, also known as \n\
+Messier 31, M31, or NGC224; often referred to as the Great \n\
+Andromeda Nebula in older texts) is a spiral galaxy approximately \n\
+2,500,000 light-years (1.58×10^11 AU) away in the constellation \n\
+Andromeda. It is the nearest spiral galaxy to our own, the Milky \n\
+Way.\n\
+Галактика або Туманність Андромеди (також відома як M31 за \n\
+каталогом Мессьє та NGC224 за Новим загальним каталогом) — \n\
+спіральна галактика, що знаходиться на відстані приблизно у 2,5 \n\
+мільйони світлових років від нашої планети у сузір'ї Андромеди. \n\
+На початку ХХІ ст. в центрі галактики виявлено чорну дірку." \
 	"" \
-	"123456\tasdf" \
+	"\
+The Andromeda Galaxy (pronounced /ænˈdrɒmədə/, also known as \
+Messier 31, M31, or NGC224; often referred to as the Great \
+Andromeda Nebula in older texts) is a spiral galaxy approximately \
+2,500,000 light-years (1.58×10^11 AU) away in the constellation \
+Andromeda. It is the nearest spiral galaxy to our own, the Milky \
+Way.
+Галактика або Туманність Андромеди (також відома як M31 за \
+каталогом Мессьє та NGC224 за Новим загальним каталогом) — \
+спіральна галактика, що знаходиться на відстані приблизно у 2,5 \
+мільйони світлових років від нашої планети у сузір'ї Андромеди. \
+На початку ХХІ ст. в центрі галактики виявлено чорну дірку." \
 
 exit $FAILCOUNT
-- 
1.6.3.3



More information about the busybox-cvs mailing list