busybox/coreutils/expand.c

/* expand - convert tabs to spaces
 * unexpand - convert spaces to tabs
 *
 * Copyright (C) 89, 91, 1995-2006 Free Software Foundation, Inc.
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 *
 * David MacKenzie <djm@gnu.ai.mit.edu>
 *
 * Options for expand:
 * -t num  --tabs=NUM      Convert tabs to num spaces (default 8 spaces).
 * -i      --initial       Only convert initial tabs on each line to spaces.
 *
 * Options for unexpand:
 * -a      --all           Convert all blanks, instead of just initial blanks.
 * -f      --first-only    Convert only leading sequences of blanks (default).
 * -t num  --tabs=NUM      Have tabs num characters apart instead of 8.
 *
 *  Busybox version (C) 2007 by Tito Ragusa <farmatito@tiscali.it>
 *
 *  Caveat: this versions of expand and unexpand don't accept tab lists.
 */
#include "libbb.h"
#include "unicode.h"

enum {
	OPT_INITIAL     = 1 << 0,
	OPT_TABS        = 1 << 1,
	OPT_ALL         = 1 << 2,
};

#if ENABLE_EXPAND
static void expand(FILE *file, unsigned tab_size, unsigned opt)
{
	char *line;

	while ((line = xmalloc_fgets(file)) != NULL) {
		unsigned char c;
		char *ptr;
		char *ptr_strbeg;

		ptr = ptr_strbeg = line;
		while ((c = *ptr) != '\0') {
			if ((opt & OPT_INITIAL) && !isblank(c)) {
				/* not space or tab */
				break;
			}
			if (c == '\t') {
				unsigned len;
				*ptr = '\0';
# if ENABLE_UNICODE_SUPPORT
				{
					uni_stat_t uni_stat;
					printable_string(&uni_stat, ptr_strbeg);
					len = uni_stat.unicode_width;
				}
# else
				len = ptr - ptr_strbeg;
# endif
				len = tab_size - (len % tab_size);
				/*while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once */
				printf("%s%*s", ptr_strbeg, len, "");
				ptr_strbeg = ptr + 1;
			}
			ptr++;
		}
		fputs(ptr_strbeg, stdout);
		free(line);
	}
}
#endif

#if ENABLE_UNEXPAND
static void unexpand(FILE *file, unsigned tab_size, unsigned opt)
{
	char *line;

	while ((line = xmalloc_fgets(file)) != NULL) {
		char *ptr = line;
		unsigned column = 0;

		while (*ptr) {
			unsigned n;
			unsigned len = 0;

			while (*ptr == ' ') {
				ptr++;
				len++;
			}
			column += len;
			if (*ptr == '\t') {
				column += tab_size - (column % tab_size);
				ptr++;
				continue;
			}

			n = column / tab_size;
			if (n) {
				len = column = column % tab_size;
				while (n--)
					putchar('\t');
			}

			if ((opt & OPT_INITIAL) && ptr != line) {
				printf("%*s%s", len, "", ptr);
				break;
			}
			n = strcspn(ptr, "\t ");
			printf("%*s%.*s", len, "", n, ptr);
# if ENABLE_UNICODE_SUPPORT
			{
				char c;
				uni_stat_t uni_stat;
				c = ptr[n];
				ptr[n] = '\0';
				printable_string(&uni_stat, ptr);
				len = uni_stat.unicode_width;
				ptr[n] = c;
			}
# else
			len = n;
# endif
			ptr += n;
			column = (column + len) % tab_size;
		}
		free(line);
	}
}
#endif

int expand_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int expand_main(int argc UNUSED_PARAM, char **argv)
{
	/* Default 8 spaces for 1 tab */
	const char *opt_t = "8";
	FILE *file;
	unsigned tab_size;
	unsigned opt;
	int exit_status = EXIT_SUCCESS;

#if ENABLE_FEATURE_EXPAND_LONG_OPTIONS
	static const char expand_longopts[] ALIGN1 =
		/* name, has_arg, val */
		"initial\0"          No_argument       "i"
		"tabs\0"             Required_argument "t"
	;
#endif
#if ENABLE_FEATURE_UNEXPAND_LONG_OPTIONS
	static const char unexpand_longopts[] ALIGN1 =
		/* name, has_arg, val */
		"first-only\0"       No_argument       "i"
		"tabs\0"             Required_argument "t"
		"all\0"              No_argument       "a"
	;
#endif
	init_unicode();

	if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) {
		IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts);
		opt = getopt32(argv, "it:", &opt_t);
	} else {
		IF_FEATURE_UNEXPAND_LONG_OPTIONS(applet_long_options = unexpand_longopts);
		/* -t NUM sets also -a */
		opt_complementary = "ta";
		opt = getopt32(argv, "ft:a", &opt_t);
		/* -f --first-only is the default */
		if (!(opt & OPT_ALL)) opt |= OPT_INITIAL;
	}
	tab_size = xatou_range(opt_t, 1, UINT_MAX);

	argv += optind;

	if (!*argv) {
		*--argv = (char*)bb_msg_standard_input;
	}
	do {
		file = fopen_or_warn_stdin(*argv);
		if (!file) {
			exit_status = EXIT_FAILURE;
			continue;
		}

		if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e'))
			IF_EXPAND(expand(file, tab_size, opt));
		else
			IF_UNEXPAND(unexpand(file, tab_size, opt));

		/* Check and close the file */
		if (fclose_if_not_stdin(file)) {
			bb_simple_perror_msg(*argv);
			exit_status = EXIT_FAILURE;
		}
		/* If stdin also clear EOF */
		if (file == stdin)
			clearerr(file);
	} while (*++argv);

	/* Now close stdin also */
	/* (if we didn't read from it, it's a no-op) */
	if (fclose(stdin))
		bb_perror_msg_and_die(bb_msg_standard_input);

	fflush_stdout_and_exit(exit_status);
}
1	niro	816	/* expand - convert tabs to spaces
2			* unexpand - convert spaces to tabs
3			*
4			* Copyright (C) 89, 91, 1995-2006 Free Software Foundation, Inc.
5			*
6			* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
7			*
8			* David MacKenzie <djm@gnu.ai.mit.edu>
9			*
10			* Options for expand:
11			* -t num --tabs=NUM Convert tabs to num spaces (default 8 spaces).
12			* -i --initial Only convert initial tabs on each line to spaces.
13			*
14			* Options for unexpand:
15			* -a --all Convert all blanks, instead of just initial blanks.
16			* -f --first-only Convert only leading sequences of blanks (default).
17			* -t num --tabs=NUM Have tabs num characters apart instead of 8.
18			*
19			* Busybox version (C) 2007 by Tito Ragusa <farmatito@tiscali.it>
20			*
21			* Caveat: this versions of expand and unexpand don't accept tab lists.
22			*/
23			#include "libbb.h"
24	niro	984	#include "unicode.h"
25	niro	816
26			enum {
27			OPT_INITIAL = 1 << 0,
28			OPT_TABS = 1 << 1,
29			OPT_ALL = 1 << 2,
30			};
31
32			#if ENABLE_EXPAND
33			static void expand(FILE *file, unsigned tab_size, unsigned opt)
34			{
35			char *line;
36
37	niro	984	while ((line = xmalloc_fgets(file)) != NULL) {
38			unsigned char c;
39			char *ptr;
40			char *ptr_strbeg;
41	niro	816
42	niro	984	ptr = ptr_strbeg = line;
43			while ((c = *ptr) != '\0') {
44			if ((opt & OPT_INITIAL) && !isblank(c)) {
45			/* not space or tab */
46			break;
47	niro	816	}
48	niro	984	if (c == '\t') {
49			unsigned len;
50			*ptr = '\0';
51	niro	1123	# if ENABLE_UNICODE_SUPPORT
52			{
53			uni_stat_t uni_stat;
54			printable_string(&uni_stat, ptr_strbeg);
55			len = uni_stat.unicode_width;
56			}
57	niro	984	# else
58			len = ptr - ptr_strbeg;
59			# endif
60			len = tab_size - (len % tab_size);
61			/while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once /
62			printf("%s%*s", ptr_strbeg, len, "");
63			ptr_strbeg = ptr + 1;
64	niro	816	}
65	niro	984	ptr++;
66	niro	816	}
67	niro	984	fputs(ptr_strbeg, stdout);
68			free(line);
69	niro	816	}
70			}
71			#endif
72
73			#if ENABLE_UNEXPAND
74	niro	984	static void unexpand(FILE *file, unsigned tab_size, unsigned opt)
75	niro	816	{
76			char *line;
77
78			while ((line = xmalloc_fgets(file)) != NULL) {
79	niro	984	char *ptr = line;
80			unsigned column = 0;
81
82			while (*ptr) {
83			unsigned n;
84	niro	1123	unsigned len = 0;
85	niro	984
86			while (*ptr == ' ') {
87			ptr++;
88	niro	1123	len++;
89	niro	816	}
90	niro	1123	column += len;
91	niro	984	if (*ptr == '\t') {
92			column += tab_size - (column % tab_size);
93			ptr++;
94			continue;
95	niro	816	}
96	niro	984
97			n = column / tab_size;
98	niro	1123	if (n) {
99			len = column = column % tab_size;
100			while (n--)
101			putchar('\t');
102			}
103	niro	984
104			if ((opt & OPT_INITIAL) && ptr != line) {
105	niro	1123	printf("%*s%s", len, "", ptr);
106	niro	984	break;
107			}
108			n = strcspn(ptr, "\t ");
109	niro	1123	printf("%s%.s", len, "", n, ptr);
110			# if ENABLE_UNICODE_SUPPORT
111	niro	984	{
112			char c;
113	niro	1123	uni_stat_t uni_stat;
114	niro	984	c = ptr[n];
115			ptr[n] = '\0';
116	niro	1123	printable_string(&uni_stat, ptr);
117			len = uni_stat.unicode_width;
118	niro	984	ptr[n] = c;
119			}
120			# else
121			len = n;
122			# endif
123			ptr += n;
124			column = (column + len) % tab_size;
125	niro	816	}
126	niro	984	free(line);
127	niro	816	}
128			}
129			#endif
130
131			int expand_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
132			int expand_main(int argc UNUSED_PARAM, char **argv)
133			{
134			/* Default 8 spaces for 1 tab */
135			const char *opt_t = "8";
136			FILE *file;
137			unsigned tab_size;
138			unsigned opt;
139			int exit_status = EXIT_SUCCESS;
140
141			#if ENABLE_FEATURE_EXPAND_LONG_OPTIONS
142			static const char expand_longopts[] ALIGN1 =
143			/* name, has_arg, val */
144			"initial\0" No_argument "i"
145			"tabs\0" Required_argument "t"
146			;
147			#endif
148			#if ENABLE_FEATURE_UNEXPAND_LONG_OPTIONS
149			static const char unexpand_longopts[] ALIGN1 =
150			/* name, has_arg, val */
151			"first-only\0" No_argument "i"
152			"tabs\0" Required_argument "t"
153			"all\0" No_argument "a"
154			;
155			#endif
156	niro	984	init_unicode();
157	niro	816
158			if (ENABLE_EXPAND && (!ENABLE_UNEXPAND \|\| applet_name[0] == 'e')) {
159	niro	984	IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts);
160	niro	816	opt = getopt32(argv, "it:", &opt_t);
161			} else {
162	niro	984	IF_FEATURE_UNEXPAND_LONG_OPTIONS(applet_long_options = unexpand_longopts);
163	niro	816	/* -t NUM sets also -a */
164			opt_complementary = "ta";
165			opt = getopt32(argv, "ft:a", &opt_t);
166			/* -f --first-only is the default */
167			if (!(opt & OPT_ALL)) opt \|= OPT_INITIAL;
168			}
169			tab_size = xatou_range(opt_t, 1, UINT_MAX);
170
171			argv += optind;
172
173			if (!*argv) {
174			--argv = (char)bb_msg_standard_input;
175			}
176			do {
177			file = fopen_or_warn_stdin(*argv);
178			if (!file) {
179			exit_status = EXIT_FAILURE;
180			continue;
181			}
182
183			if (ENABLE_EXPAND && (!ENABLE_UNEXPAND \|\| applet_name[0] == 'e'))
184	niro	984	IF_EXPAND(expand(file, tab_size, opt));
185	niro	816	else
186	niro	984	IF_UNEXPAND(unexpand(file, tab_size, opt));
187	niro	816
188			/* Check and close the file */
189			if (fclose_if_not_stdin(file)) {
190			bb_simple_perror_msg(*argv);
191			exit_status = EXIT_FAILURE;
192			}
193			/* If stdin also clear EOF */
194			if (file == stdin)
195			clearerr(file);
196			} while (*++argv);
197
198			/* Now close stdin also */
199			/* (if we didn't read from it, it's a no-op) */
200			if (fclose(stdin))
201			bb_perror_msg_and_die(bb_msg_standard_input);
202
203			fflush_stdout_and_exit(exit_status);
204			}