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_FEATURE_ASSUME_UNICODE
				len = unicode_strlen(ptr_strbeg);
# 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;

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

			n = column / tab_size;
			column = column % tab_size;
			while (n--)
				putchar('\t');

			if ((opt & OPT_INITIAL) && ptr != line) {
				printf("%*s%s", column, "", ptr);
				break;
			}
			n = strcspn(ptr, "\t ");
			printf("%*s%.*s", column, "", n, ptr);
# if ENABLE_FEATURE_ASSUME_UNICODE
			{
				char c;
				c = ptr[n];
				ptr[n] = '\0';
				len = unicode_strlen(ptr);
				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	/* 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	#include "unicode.h"
25
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	while ((line = xmalloc_fgets(file)) != NULL) {
38	unsigned char c;
39	char *ptr;
40	char *ptr_strbeg;
41
42	ptr = ptr_strbeg = line;
43	while ((c = *ptr) != '\0') {
44	if ((opt & OPT_INITIAL) && !isblank(c)) {
45	/* not space or tab */
46	break;
47	}
48	if (c == '\t') {
49	unsigned len;
50	*ptr = '\0';
51	# if ENABLE_FEATURE_ASSUME_UNICODE
52	len = unicode_strlen(ptr_strbeg);
53	# else
54	len = ptr - ptr_strbeg;
55	# endif
56	len = tab_size - (len % tab_size);
57	/while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once /
58	printf("%s%*s", ptr_strbeg, len, "");
59	ptr_strbeg = ptr + 1;
60	}
61	ptr++;
62	}
63	fputs(ptr_strbeg, stdout);
64	free(line);
65	}
66	}
67	#endif
68
69	#if ENABLE_UNEXPAND
70	static void unexpand(FILE *file, unsigned tab_size, unsigned opt)
71	{
72	char *line;
73
74	while ((line = xmalloc_fgets(file)) != NULL) {
75	char *ptr = line;
76	unsigned column = 0;
77
78	while (*ptr) {
79	unsigned n;
80	unsigned len;
81
82	while (*ptr == ' ') {
83	column++;
84	ptr++;
85	}
86	if (*ptr == '\t') {
87	column += tab_size - (column % tab_size);
88	ptr++;
89	continue;
90	}
91
92	n = column / tab_size;
93	column = column % tab_size;
94	while (n--)
95	putchar('\t');
96
97	if ((opt & OPT_INITIAL) && ptr != line) {
98	printf("%*s%s", column, "", ptr);
99	break;
100	}
101	n = strcspn(ptr, "\t ");
102	printf("%s%.s", column, "", n, ptr);
103	# if ENABLE_FEATURE_ASSUME_UNICODE
104	{
105	char c;
106	c = ptr[n];
107	ptr[n] = '\0';
108	len = unicode_strlen(ptr);
109	ptr[n] = c;
110	}
111	# else
112	len = n;
113	# endif
114	ptr += n;
115	column = (column + len) % tab_size;
116	}
117	free(line);
118	}
119	}
120	#endif
121
122	int expand_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
123	int expand_main(int argc UNUSED_PARAM, char **argv)
124	{
125	/* Default 8 spaces for 1 tab */
126	const char *opt_t = "8";
127	FILE *file;
128	unsigned tab_size;
129	unsigned opt;
130	int exit_status = EXIT_SUCCESS;
131
132	#if ENABLE_FEATURE_EXPAND_LONG_OPTIONS
133	static const char expand_longopts[] ALIGN1 =
134	/* name, has_arg, val */
135	"initial\0" No_argument "i"
136	"tabs\0" Required_argument "t"
137	;
138	#endif
139	#if ENABLE_FEATURE_UNEXPAND_LONG_OPTIONS
140	static const char unexpand_longopts[] ALIGN1 =
141	/* name, has_arg, val */
142	"first-only\0" No_argument "i"
143	"tabs\0" Required_argument "t"
144	"all\0" No_argument "a"
145	;
146	#endif
147	init_unicode();
148
149	if (ENABLE_EXPAND && (!ENABLE_UNEXPAND \|\| applet_name[0] == 'e')) {
150	IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts);
151	opt = getopt32(argv, "it:", &opt_t);
152	} else {
153	IF_FEATURE_UNEXPAND_LONG_OPTIONS(applet_long_options = unexpand_longopts);
154	/* -t NUM sets also -a */
155	opt_complementary = "ta";
156	opt = getopt32(argv, "ft:a", &opt_t);
157	/* -f --first-only is the default */
158	if (!(opt & OPT_ALL)) opt \|= OPT_INITIAL;
159	}
160	tab_size = xatou_range(opt_t, 1, UINT_MAX);
161
162	argv += optind;
163
164	if (!*argv) {
165	--argv = (char)bb_msg_standard_input;
166	}
167	do {
168	file = fopen_or_warn_stdin(*argv);
169	if (!file) {
170	exit_status = EXIT_FAILURE;
171	continue;
172	}
173
174	if (ENABLE_EXPAND && (!ENABLE_UNEXPAND \|\| applet_name[0] == 'e'))
175	IF_EXPAND(expand(file, tab_size, opt));
176	else
177	IF_UNEXPAND(unexpand(file, tab_size, opt));
178
179	/* Check and close the file */
180	if (fclose_if_not_stdin(file)) {
181	bb_simple_perror_msg(*argv);
182	exit_status = EXIT_FAILURE;
183	}
184	/* If stdin also clear EOF */
185	if (file == stdin)
186	clearerr(file);
187	} while (*++argv);
188
189	/* Now close stdin also */
190	/* (if we didn't read from it, it's a no-op) */
191	if (fclose(stdin))
192	bb_perror_msg_and_die(bb_msg_standard_input);
193
194	fflush_stdout_and_exit(exit_status);
195	}