busybox/libbb/get_line_from_file.c

/* vi: set sw=4 ts=4: */
/*
 * Utility routines.
 *
 * Copyright (C) 2005, 2006 Rob Landley <rob@landley.net>
 * Copyright (C) 2004 Erik Andersen <andersen@codepoet.org>
 * Copyright (C) 2001 Matt Krai
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 */

/* for getline() [GNUism]
#ifndef _GNU_SOURCE
#define _GNU_SOURCE 1
#endif
*/
#include "libbb.h"

/* This function reads an entire line from a text file, up to a newline
 * or NUL byte, inclusive.  It returns a malloc'ed char * which
 * must be free'ed by the caller.  If end is NULL '\n' isn't considered
 * end of line.  If end isn't NULL, length of the chunk is stored in it.
 * If lineno is not NULL, *lineno is incremented for each line,
 * and also trailing '\' is recognized as line continuation.
 *
 * Returns NULL if EOF/error. */
char* FAST_FUNC bb_get_chunk_with_continuation(FILE *file, int *end, int *lineno)
{
	int ch;
	int idx = 0;
	char *linebuf = NULL;
	int linebufsz = 0;

	while ((ch = getc(file)) != EOF) {
		/* grow the line buffer as necessary */
		if (idx >= linebufsz) {
			linebufsz += 256;
			linebuf = xrealloc(linebuf, linebufsz);
		}
		linebuf[idx++] = (char) ch;
		if (!ch)
			break;
		if (end && ch == '\n') {
			if (lineno == NULL)
				break;
			(*lineno)++;
			if (idx < 2 || linebuf[idx-2] != '\\')
				break;
			idx -= 2;
		}
	}
	if (end)
		*end = idx;
	if (linebuf) {
		// huh, does fgets discard prior data on error like this?
		// I don't think so....
		//if (ferror(file)) {
		//	free(linebuf);
		//	return NULL;
		//}
		linebuf = xrealloc(linebuf, idx + 1);
		linebuf[idx] = '\0';
	}
	return linebuf;
}

char* FAST_FUNC bb_get_chunk_from_file(FILE *file, int *end)
{
	return bb_get_chunk_with_continuation(file, end, NULL);
}

/* Get line, including trailing \n if any */
char* FAST_FUNC xmalloc_fgets(FILE *file)
{
	int i;

	return bb_get_chunk_from_file(file, &i);
}
/* Get line.  Remove trailing \n */
char* FAST_FUNC xmalloc_fgetline(FILE *file)
{
	int i;
	char *c = bb_get_chunk_from_file(file, &i);

	if (i && c[--i] == '\n')
		c[i] = '\0';

	return c;
}

#if 0
/* GNUism getline() should be faster (not tested) than a loop with fgetc */

/* Get line, including trailing \n if any */
char* FAST_FUNC xmalloc_fgets(FILE *file)
{
	char *res_buf = NULL;
	size_t res_sz;

	if (getline(&res_buf, &res_sz, file) == -1) {
		free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
		res_buf = NULL;
	}
//TODO: trimming to res_sz?
	return res_buf;
}
/* Get line.  Remove trailing \n */
char* FAST_FUNC xmalloc_fgetline(FILE *file)
{
	char *res_buf = NULL;
	size_t res_sz;

	res_sz = getline(&res_buf, &res_sz, file);

	if ((ssize_t)res_sz != -1) {
		if (res_buf[res_sz - 1] == '\n')
			res_buf[--res_sz] = '\0';
//TODO: trimming to res_sz?
	} else {
		free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
		res_buf = NULL;
	}
	return res_buf;
}

#endif

#if 0
/* Faster routines (~twice as fast). +170 bytes. Unused as of 2008-07.
 *
 * NB: they stop at NUL byte too.
 * Performance is important here. Think "grep 50gigabyte_file"...
 * Ironically, grep can't use it because of NUL issue.
 * We sorely need C lib to provide fgets which reports size!
 *
 * Update:
 * Actually, uclibc and glibc have it. man getline. It's GNUism,
 *   but very useful one (if it's as fast as this code).
 * TODO:
 * - currently, sed and sort use bb_get_chunk_from_file and heavily
 *   depend on its "stop on \n or \0" behavior, and STILL they fail
 *   to handle all cases with embedded NULs correctly. So:
 * - audit sed and sort; convert them to getline FIRST.
 * - THEN ditch bb_get_chunk_from_file, replace it with getline.
 * - provide getline implementation for non-GNU systems.
 */

static char* xmalloc_fgets_internal(FILE *file, int *sizep)
{
	int len;
	int idx = 0;
	char *linebuf = NULL;

	while (1) {
		char *r;

		linebuf = xrealloc(linebuf, idx + 0x100);
		r = fgets(&linebuf[idx], 0x100, file);
		if (!r) {
			/* need to terminate in case this is error
			 * (EOF puts NUL itself) */
			linebuf[idx] = '\0';
			break;
		}
		/* stupid. fgets knows the len, it should report it somehow */
		len = strlen(&linebuf[idx]);
		idx += len;
		if (len != 0xff || linebuf[idx - 1] == '\n')
			break;
	}
	*sizep = idx;
	if (idx) {
		/* xrealloc(linebuf, idx + 1) is up to caller */
		return linebuf;
	}
	free(linebuf);
	return NULL;
}

/* Get line, remove trailing \n */
char* FAST_FUNC xmalloc_fgetline_fast(FILE *file)
{
	int sz;
	char *r = xmalloc_fgets_internal(file, &sz);
	if (r && r[sz - 1] == '\n')
		r[--sz] = '\0';
	return r; /* not xrealloc(r, sz + 1)! */
}

char* FAST_FUNC xmalloc_fgets(FILE *file)
{
	int sz;
	return xmalloc_fgets_internal(file, &sz);
}

/* Get line, remove trailing \n */
char* FAST_FUNC xmalloc_fgetline(FILE *file)
{
	int sz;
	char *r = xmalloc_fgets_internal(file, &sz);
	if (!r)
		return r;
	if (r[sz - 1] == '\n')
		r[--sz] = '\0';
	return xrealloc(r, sz + 1);
}
#endif
1	niro	532	/* vi: set sw=4 ts=4: */
2			/*
3			* Utility routines.
4			*
5			* Copyright (C) 2005, 2006 Rob Landley <rob@landley.net>
6			* Copyright (C) 2004 Erik Andersen <andersen@codepoet.org>
7			* Copyright (C) 2001 Matt Krai
8			*
9			* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
10			*/
11
12	niro	816	/* for getline() [GNUism]
13			#ifndef _GNU_SOURCE
14			#define _GNU_SOURCE 1
15			#endif
16			*/
17	niro	532	#include "libbb.h"
18
19			/* This function reads an entire line from a text file, up to a newline
20	niro	816	* or NUL byte, inclusive. It returns a malloc'ed char * which
21			* must be free'ed by the caller. If end is NULL '\n' isn't considered
22			* end of line. If end isn't NULL, length of the chunk is stored in it.
23			* If lineno is not NULL, *lineno is incremented for each line,
24			* and also trailing '\' is recognized as line continuation.
25			*
26			* Returns NULL if EOF/error. */
27			char* FAST_FUNC bb_get_chunk_with_continuation(FILE file, int end, int *lineno)
28	niro	532	{
29			int ch;
30			int idx = 0;
31			char *linebuf = NULL;
32			int linebufsz = 0;
33
34			while ((ch = getc(file)) != EOF) {
35			/* grow the line buffer as necessary */
36			if (idx >= linebufsz) {
37	niro	816	linebufsz += 256;
38			linebuf = xrealloc(linebuf, linebufsz);
39	niro	532	}
40			linebuf[idx++] = (char) ch;
41	niro	816	if (!ch)
42	niro	532	break;
43	niro	816	if (end && ch == '\n') {
44			if (lineno == NULL)
45			break;
46			(*lineno)++;
47			if (idx < 2 \|\| linebuf[idx-2] != '\\')
48			break;
49			idx -= 2;
50			}
51	niro	532	}
52			if (end)
53			*end = idx;
54			if (linebuf) {
55			// huh, does fgets discard prior data on error like this?
56			// I don't think so....
57			//if (ferror(file)) {
58			// free(linebuf);
59			// return NULL;
60			//}
61	niro	816	linebuf = xrealloc(linebuf, idx + 1);
62	niro	532	linebuf[idx] = '\0';
63			}
64			return linebuf;
65			}
66
67	niro	816	char* FAST_FUNC bb_get_chunk_from_file(FILE file, int end)
68			{
69			return bb_get_chunk_with_continuation(file, end, NULL);
70			}
71
72	niro	532	/* Get line, including trailing \n if any */
73	niro	816	char* FAST_FUNC xmalloc_fgets(FILE *file)
74	niro	532	{
75			int i;
76
77			return bb_get_chunk_from_file(file, &i);
78			}
79			/* Get line. Remove trailing \n */
80	niro	816	char* FAST_FUNC xmalloc_fgetline(FILE *file)
81	niro	532	{
82			int i;
83			char *c = bb_get_chunk_from_file(file, &i);
84
85			if (i && c[--i] == '\n')
86			c[i] = '\0';
87
88			return c;
89			}
90	niro	816
91			#if 0
92			/* GNUism getline() should be faster (not tested) than a loop with fgetc */
93
94			/* Get line, including trailing \n if any */
95			char* FAST_FUNC xmalloc_fgets(FILE *file)
96			{
97			char *res_buf = NULL;
98			size_t res_sz;
99
100			if (getline(&res_buf, &res_sz, file) == -1) {
101			free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
102			res_buf = NULL;
103			}
104			//TODO: trimming to res_sz?
105			return res_buf;
106			}
107			/* Get line. Remove trailing \n */
108			char* FAST_FUNC xmalloc_fgetline(FILE *file)
109			{
110			char *res_buf = NULL;
111			size_t res_sz;
112
113			res_sz = getline(&res_buf, &res_sz, file);
114
115			if ((ssize_t)res_sz != -1) {
116			if (res_buf[res_sz - 1] == '\n')
117			res_buf[--res_sz] = '\0';
118			//TODO: trimming to res_sz?
119			} else {
120			free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
121			res_buf = NULL;
122			}
123			return res_buf;
124			}
125
126			#endif
127
128			#if 0
129			/* Faster routines (~twice as fast). +170 bytes. Unused as of 2008-07.
130			*
131			* NB: they stop at NUL byte too.
132			* Performance is important here. Think "grep 50gigabyte_file"...
133			* Ironically, grep can't use it because of NUL issue.
134			* We sorely need C lib to provide fgets which reports size!
135			*
136			* Update:
137			* Actually, uclibc and glibc have it. man getline. It's GNUism,
138			* but very useful one (if it's as fast as this code).
139			* TODO:
140			* - currently, sed and sort use bb_get_chunk_from_file and heavily
141			* depend on its "stop on \n or \0" behavior, and STILL they fail
142			* to handle all cases with embedded NULs correctly. So:
143			* - audit sed and sort; convert them to getline FIRST.
144			* - THEN ditch bb_get_chunk_from_file, replace it with getline.
145			* - provide getline implementation for non-GNU systems.
146			*/
147
148			static char* xmalloc_fgets_internal(FILE file, int sizep)
149			{
150			int len;
151			int idx = 0;
152			char *linebuf = NULL;
153
154			while (1) {
155			char *r;
156
157			linebuf = xrealloc(linebuf, idx + 0x100);
158			r = fgets(&linebuf[idx], 0x100, file);
159			if (!r) {
160			/* need to terminate in case this is error
161			* (EOF puts NUL itself) */
162			linebuf[idx] = '\0';
163			break;
164			}
165			/* stupid. fgets knows the len, it should report it somehow */
166			len = strlen(&linebuf[idx]);
167			idx += len;
168			if (len != 0xff \|\| linebuf[idx - 1] == '\n')
169			break;
170			}
171			*sizep = idx;
172			if (idx) {
173			/* xrealloc(linebuf, idx + 1) is up to caller */
174			return linebuf;
175			}
176			free(linebuf);
177			return NULL;
178			}
179
180			/* Get line, remove trailing \n */
181			char* FAST_FUNC xmalloc_fgetline_fast(FILE *file)
182			{
183			int sz;
184			char *r = xmalloc_fgets_internal(file, &sz);
185			if (r && r[sz - 1] == '\n')
186			r[--sz] = '\0';
187			return r; /* not xrealloc(r, sz + 1)! */
188			}
189
190			char* FAST_FUNC xmalloc_fgets(FILE *file)
191			{
192			int sz;
193			return xmalloc_fgets_internal(file, &sz);
194			}
195
196			/* Get line, remove trailing \n */
197			char* FAST_FUNC xmalloc_fgetline(FILE *file)
198			{
199			int sz;
200			char *r = xmalloc_fgets_internal(file, &sz);
201			if (!r)
202			return r;
203			if (r[sz - 1] == '\n')
204			r[--sz] = '\0';
205			return xrealloc(r, sz + 1);
206			}
207			#endif