blob: a0ed9193fa3eed30d34a8996217137a1d7f9d921 [file] [log] [blame]
Eric Andersenaad1a882001-03-16 22:47:14 +00001/* vi: set sw=4 ts=4: */
2/*
3 * Utility routines.
4 *
Rob Landley2b26fd52006-02-24 02:30:39 +00005 * Copyright (C) 2005, 2006 Rob Landley <rob@landley.net>
6 * Copyright (C) 2004 Erik Andersen <andersen@codepoet.org>
7 * Copyright (C) 2001 Matt Krai
Eric Andersenaad1a882001-03-16 22:47:14 +00008 *
Denys Vlasenko0ef64bd2010-08-16 20:14:46 +02009 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
Eric Andersenaad1a882001-03-16 22:47:14 +000010 */
11
Eric Andersenaad1a882001-03-16 22:47:14 +000012#include "libbb.h"
13
Denis Vlasenkof7996f32007-01-11 17:20:00 +000014/* This function reads an entire line from a text file, up to a newline
Denis Vlasenkoabee3d02007-12-26 20:44:45 +000015 * or NUL byte, inclusive. It returns a malloc'ed char * which
16 * must be free'ed by the caller. If end is NULL '\n' isn't considered
Denis Vlasenko69f4f9a2008-08-09 17:16:40 +000017 * end of line. If end isn't NULL, length of the chunk is stored in it.
18 * If lineno is not NULL, *lineno is incremented for each line,
19 * and also trailing '\' is recognized as line continuation.
20 *
21 * Returns NULL if EOF/error. */
22char* FAST_FUNC bb_get_chunk_with_continuation(FILE *file, int *end, int *lineno)
Eric Andersenaad1a882001-03-16 22:47:14 +000023{
Eric Andersenaad1a882001-03-16 22:47:14 +000024 int ch;
25 int idx = 0;
26 char *linebuf = NULL;
27 int linebufsz = 0;
28
Manuel Novoa III cad53642003-03-19 09:13:01 +000029 while ((ch = getc(file)) != EOF) {
Eric Andersenaad1a882001-03-16 22:47:14 +000030 /* grow the line buffer as necessary */
Denis Vlasenko8b22b072006-12-02 17:58:10 +000031 if (idx >= linebufsz) {
Denis Vlasenko69f4f9a2008-08-09 17:16:40 +000032 linebufsz += 256;
Denis Vlasenko51742f42007-04-12 00:32:05 +000033 linebuf = xrealloc(linebuf, linebufsz);
Manuel Novoa III cad53642003-03-19 09:13:01 +000034 }
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000035 linebuf[idx++] = (char) ch;
Denis Vlasenko69f4f9a2008-08-09 17:16:40 +000036 if (!ch)
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000037 break;
Denis Vlasenko69f4f9a2008-08-09 17:16:40 +000038 if (end && ch == '\n') {
39 if (lineno == NULL)
40 break;
41 (*lineno)++;
42 if (idx < 2 || linebuf[idx-2] != '\\')
43 break;
44 idx -= 2;
45 }
Eric Andersenaad1a882001-03-16 22:47:14 +000046 }
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000047 if (end)
48 *end = idx;
Manuel Novoa III cad53642003-03-19 09:13:01 +000049 if (linebuf) {
Denis Vlasenko8b22b072006-12-02 17:58:10 +000050 // huh, does fgets discard prior data on error like this?
Denis Vlasenkoab24e182006-11-30 16:41:15 +000051 // I don't think so....
52 //if (ferror(file)) {
53 // free(linebuf);
54 // return NULL;
55 //}
Denis Vlasenkodeeed592008-07-08 05:14:36 +000056 linebuf = xrealloc(linebuf, idx + 1);
Denis Vlasenko8b22b072006-12-02 17:58:10 +000057 linebuf[idx] = '\0';
Manuel Novoa III cad53642003-03-19 09:13:01 +000058 }
Eric Andersenaad1a882001-03-16 22:47:14 +000059 return linebuf;
60}
61
Denis Vlasenko69f4f9a2008-08-09 17:16:40 +000062char* FAST_FUNC bb_get_chunk_from_file(FILE *file, int *end)
63{
64 return bb_get_chunk_with_continuation(file, end, NULL);
65}
66
Denis Vlasenkob97c9842006-10-01 21:05:12 +000067/* Get line, including trailing \n if any */
Denis Vlasenkodefc1ea2008-06-27 02:52:20 +000068char* FAST_FUNC xmalloc_fgets(FILE *file)
Manuel Novoa III cad53642003-03-19 09:13:01 +000069{
Rob Landley2b26fd52006-02-24 02:30:39 +000070 int i;
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000071
Rob Landley2b26fd52006-02-24 02:30:39 +000072 return bb_get_chunk_from_file(file, &i);
Manuel Novoa III cad53642003-03-19 09:13:01 +000073}
Denis Vlasenkob97c9842006-10-01 21:05:12 +000074/* Get line. Remove trailing \n */
Denis Vlasenkodefc1ea2008-06-27 02:52:20 +000075char* FAST_FUNC xmalloc_fgetline(FILE *file)
Manuel Novoa III cad53642003-03-19 09:13:01 +000076{
Rob Landley2b26fd52006-02-24 02:30:39 +000077 int i;
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000078 char *c = bb_get_chunk_from_file(file, &i);
79
80 if (i && c[--i] == '\n')
Denis Vlasenko9a7cef92006-12-20 02:46:48 +000081 c[i] = '\0';
Bernhard Reutner-Fischer2d1a6e72006-06-10 11:04:43 +000082
Rob Landley2b26fd52006-02-24 02:30:39 +000083 return c;
Manuel Novoa III cad53642003-03-19 09:13:01 +000084}
Denis Vlasenko2132e022008-07-15 10:33:12 +000085
Denis Vlasenko5e476ba2008-07-15 21:29:44 +000086#if 0
Denis Vlasenko3fd15e12008-08-09 16:15:14 +000087/* GNUism getline() should be faster (not tested) than a loop with fgetc */
88
89/* Get line, including trailing \n if any */
90char* FAST_FUNC xmalloc_fgets(FILE *file)
91{
92 char *res_buf = NULL;
93 size_t res_sz;
94
95 if (getline(&res_buf, &res_sz, file) == -1) {
96 free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
97 res_buf = NULL;
98 }
99//TODO: trimming to res_sz?
100 return res_buf;
101}
102/* Get line. Remove trailing \n */
103char* FAST_FUNC xmalloc_fgetline(FILE *file)
104{
105 char *res_buf = NULL;
106 size_t res_sz;
107
108 res_sz = getline(&res_buf, &res_sz, file);
109
110 if ((ssize_t)res_sz != -1) {
111 if (res_buf[res_sz - 1] == '\n')
112 res_buf[--res_sz] = '\0';
113//TODO: trimming to res_sz?
114 } else {
115 free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
116 res_buf = NULL;
117 }
118 return res_buf;
119}
120
121#endif
122
123#if 0
Denis Vlasenko2132e022008-07-15 10:33:12 +0000124/* Faster routines (~twice as fast). +170 bytes. Unused as of 2008-07.
125 *
126 * NB: they stop at NUL byte too.
127 * Performance is important here. Think "grep 50gigabyte_file"...
Denis Vlasenko5e476ba2008-07-15 21:29:44 +0000128 * Ironically, grep can't use it because of NUL issue.
Denis Vlasenko2132e022008-07-15 10:33:12 +0000129 * We sorely need C lib to provide fgets which reports size!
Denis Vlasenko5e476ba2008-07-15 21:29:44 +0000130 *
131 * Update:
132 * Actually, uclibc and glibc have it. man getline. It's GNUism,
133 * but very useful one (if it's as fast as this code).
134 * TODO:
135 * - currently, sed and sort use bb_get_chunk_from_file and heavily
136 * depend on its "stop on \n or \0" behavior, and STILL they fail
137 * to handle all cases with embedded NULs correctly. So:
138 * - audit sed and sort; convert them to getline FIRST.
139 * - THEN ditch bb_get_chunk_from_file, replace it with getline.
140 * - provide getline implementation for non-GNU systems.
Denis Vlasenko2132e022008-07-15 10:33:12 +0000141 */
142
143static char* xmalloc_fgets_internal(FILE *file, int *sizep)
144{
145 int len;
146 int idx = 0;
147 char *linebuf = NULL;
148
149 while (1) {
150 char *r;
151
152 linebuf = xrealloc(linebuf, idx + 0x100);
153 r = fgets(&linebuf[idx], 0x100, file);
154 if (!r) {
155 /* need to terminate in case this is error
156 * (EOF puts NUL itself) */
157 linebuf[idx] = '\0';
158 break;
159 }
160 /* stupid. fgets knows the len, it should report it somehow */
161 len = strlen(&linebuf[idx]);
162 idx += len;
163 if (len != 0xff || linebuf[idx - 1] == '\n')
164 break;
165 }
166 *sizep = idx;
167 if (idx) {
168 /* xrealloc(linebuf, idx + 1) is up to caller */
169 return linebuf;
170 }
171 free(linebuf);
172 return NULL;
173}
174
175/* Get line, remove trailing \n */
176char* FAST_FUNC xmalloc_fgetline_fast(FILE *file)
177{
178 int sz;
179 char *r = xmalloc_fgets_internal(file, &sz);
180 if (r && r[sz - 1] == '\n')
181 r[--sz] = '\0';
182 return r; /* not xrealloc(r, sz + 1)! */
183}
184
Denis Vlasenko2132e022008-07-15 10:33:12 +0000185char* FAST_FUNC xmalloc_fgets(FILE *file)
186{
187 int sz;
188 return xmalloc_fgets_internal(file, &sz);
189}
190
191/* Get line, remove trailing \n */
192char* FAST_FUNC xmalloc_fgetline(FILE *file)
193{
194 int sz;
195 char *r = xmalloc_fgets_internal(file, &sz);
196 if (!r)
197 return r;
198 if (r[sz - 1] == '\n')
199 r[--sz] = '\0';
200 return xrealloc(r, sz + 1);
201}
202#endif