Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
| 2 | /* |
| 3 | * Utility routines. |
| 4 | * |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 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 |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 8 | * |
Denys Vlasenko | 0ef64bd | 2010-08-16 20:14:46 +0200 | [diff] [blame^] | 9 | * Licensed under GPLv2 or later, see file LICENSE in this source tree. |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 10 | */ |
| 11 | |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 12 | #include "libbb.h" |
| 13 | |
Denis Vlasenko | f7996f3 | 2007-01-11 17:20:00 +0000 | [diff] [blame] | 14 | /* This function reads an entire line from a text file, up to a newline |
Denis Vlasenko | abee3d0 | 2007-12-26 20:44:45 +0000 | [diff] [blame] | 15 | * 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 Vlasenko | 69f4f9a | 2008-08-09 17:16:40 +0000 | [diff] [blame] | 17 | * 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. */ |
| 22 | char* FAST_FUNC bb_get_chunk_with_continuation(FILE *file, int *end, int *lineno) |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 23 | { |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 24 | int ch; |
| 25 | int idx = 0; |
| 26 | char *linebuf = NULL; |
| 27 | int linebufsz = 0; |
| 28 | |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 29 | while ((ch = getc(file)) != EOF) { |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 30 | /* grow the line buffer as necessary */ |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 31 | if (idx >= linebufsz) { |
Denis Vlasenko | 69f4f9a | 2008-08-09 17:16:40 +0000 | [diff] [blame] | 32 | linebufsz += 256; |
Denis Vlasenko | 51742f4 | 2007-04-12 00:32:05 +0000 | [diff] [blame] | 33 | linebuf = xrealloc(linebuf, linebufsz); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 34 | } |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 35 | linebuf[idx++] = (char) ch; |
Denis Vlasenko | 69f4f9a | 2008-08-09 17:16:40 +0000 | [diff] [blame] | 36 | if (!ch) |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 37 | break; |
Denis Vlasenko | 69f4f9a | 2008-08-09 17:16:40 +0000 | [diff] [blame] | 38 | 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 Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 46 | } |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 47 | if (end) |
| 48 | *end = idx; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 49 | if (linebuf) { |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 50 | // huh, does fgets discard prior data on error like this? |
Denis Vlasenko | ab24e18 | 2006-11-30 16:41:15 +0000 | [diff] [blame] | 51 | // I don't think so.... |
| 52 | //if (ferror(file)) { |
| 53 | // free(linebuf); |
| 54 | // return NULL; |
| 55 | //} |
Denis Vlasenko | deeed59 | 2008-07-08 05:14:36 +0000 | [diff] [blame] | 56 | linebuf = xrealloc(linebuf, idx + 1); |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 57 | linebuf[idx] = '\0'; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 58 | } |
Eric Andersen | aad1a88 | 2001-03-16 22:47:14 +0000 | [diff] [blame] | 59 | return linebuf; |
| 60 | } |
| 61 | |
Denis Vlasenko | 69f4f9a | 2008-08-09 17:16:40 +0000 | [diff] [blame] | 62 | char* 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 Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 67 | /* Get line, including trailing \n if any */ |
Denis Vlasenko | defc1ea | 2008-06-27 02:52:20 +0000 | [diff] [blame] | 68 | char* FAST_FUNC xmalloc_fgets(FILE *file) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 69 | { |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 70 | int i; |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 71 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 72 | return bb_get_chunk_from_file(file, &i); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 73 | } |
Denis Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 74 | /* Get line. Remove trailing \n */ |
Denis Vlasenko | defc1ea | 2008-06-27 02:52:20 +0000 | [diff] [blame] | 75 | char* FAST_FUNC xmalloc_fgetline(FILE *file) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 76 | { |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 77 | int i; |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 78 | char *c = bb_get_chunk_from_file(file, &i); |
| 79 | |
| 80 | if (i && c[--i] == '\n') |
Denis Vlasenko | 9a7cef9 | 2006-12-20 02:46:48 +0000 | [diff] [blame] | 81 | c[i] = '\0'; |
Bernhard Reutner-Fischer | 2d1a6e7 | 2006-06-10 11:04:43 +0000 | [diff] [blame] | 82 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 83 | return c; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 84 | } |
Denis Vlasenko | 2132e02 | 2008-07-15 10:33:12 +0000 | [diff] [blame] | 85 | |
Denis Vlasenko | 5e476ba | 2008-07-15 21:29:44 +0000 | [diff] [blame] | 86 | #if 0 |
Denis Vlasenko | 3fd15e1 | 2008-08-09 16:15:14 +0000 | [diff] [blame] | 87 | /* GNUism getline() should be faster (not tested) than a loop with fgetc */ |
| 88 | |
| 89 | /* Get line, including trailing \n if any */ |
| 90 | char* 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 */ |
| 103 | char* 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 Vlasenko | 2132e02 | 2008-07-15 10:33:12 +0000 | [diff] [blame] | 124 | /* 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 Vlasenko | 5e476ba | 2008-07-15 21:29:44 +0000 | [diff] [blame] | 128 | * Ironically, grep can't use it because of NUL issue. |
Denis Vlasenko | 2132e02 | 2008-07-15 10:33:12 +0000 | [diff] [blame] | 129 | * We sorely need C lib to provide fgets which reports size! |
Denis Vlasenko | 5e476ba | 2008-07-15 21:29:44 +0000 | [diff] [blame] | 130 | * |
| 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 Vlasenko | 2132e02 | 2008-07-15 10:33:12 +0000 | [diff] [blame] | 141 | */ |
| 142 | |
| 143 | static 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 */ |
| 176 | char* 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 Vlasenko | 2132e02 | 2008-07-15 10:33:12 +0000 | [diff] [blame] | 185 | char* 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 */ |
| 192 | char* 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 |