1 | /* 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 source tree.
|
---|
10 | */
|
---|
11 |
|
---|
12 | #include "libbb.h"
|
---|
13 |
|
---|
14 | char* FAST_FUNC bb_get_chunk_from_file(FILE *file, int *end)
|
---|
15 | {
|
---|
16 | int ch;
|
---|
17 | unsigned idx = 0;
|
---|
18 | char *linebuf = NULL;
|
---|
19 |
|
---|
20 | while ((ch = getc(file)) != EOF) {
|
---|
21 | /* grow the line buffer as necessary */
|
---|
22 | if (!(idx & 0xff))
|
---|
23 | linebuf = xrealloc(linebuf, idx + 0x100);
|
---|
24 | linebuf[idx++] = (char) ch;
|
---|
25 | if (ch == '\0')
|
---|
26 | break;
|
---|
27 | if (end && ch == '\n')
|
---|
28 | break;
|
---|
29 | }
|
---|
30 | if (end)
|
---|
31 | *end = idx;
|
---|
32 | if (linebuf) {
|
---|
33 | // huh, does fgets discard prior data on error like this?
|
---|
34 | // I don't think so....
|
---|
35 | //if (ferror(file)) {
|
---|
36 | // free(linebuf);
|
---|
37 | // return NULL;
|
---|
38 | //}
|
---|
39 | linebuf = xrealloc(linebuf, idx + 1);
|
---|
40 | linebuf[idx] = '\0';
|
---|
41 | }
|
---|
42 | return linebuf;
|
---|
43 | }
|
---|
44 |
|
---|
45 | /* Get line, including trailing \n if any */
|
---|
46 | char* FAST_FUNC xmalloc_fgets(FILE *file)
|
---|
47 | {
|
---|
48 | int i;
|
---|
49 |
|
---|
50 | return bb_get_chunk_from_file(file, &i);
|
---|
51 | }
|
---|
52 | /* Get line. Remove trailing \n */
|
---|
53 | char* FAST_FUNC xmalloc_fgetline(FILE *file)
|
---|
54 | {
|
---|
55 | int i;
|
---|
56 | char *c = bb_get_chunk_from_file(file, &i);
|
---|
57 |
|
---|
58 | if (i && c[--i] == '\n')
|
---|
59 | c[i] = '\0';
|
---|
60 |
|
---|
61 | return c;
|
---|
62 | }
|
---|
63 |
|
---|
64 | #if 0
|
---|
65 | /* GNUism getline() should be faster (not tested) than a loop with fgetc */
|
---|
66 |
|
---|
67 | /* Get line, including trailing \n if any */
|
---|
68 | char* FAST_FUNC xmalloc_fgets(FILE *file)
|
---|
69 | {
|
---|
70 | char *res_buf = NULL;
|
---|
71 | size_t res_sz;
|
---|
72 |
|
---|
73 | if (getline(&res_buf, &res_sz, file) == -1) {
|
---|
74 | free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
|
---|
75 | res_buf = NULL;
|
---|
76 | }
|
---|
77 | //TODO: trimming to res_sz?
|
---|
78 | return res_buf;
|
---|
79 | }
|
---|
80 | /* Get line. Remove trailing \n */
|
---|
81 | char* FAST_FUNC xmalloc_fgetline(FILE *file)
|
---|
82 | {
|
---|
83 | char *res_buf = NULL;
|
---|
84 | size_t res_sz;
|
---|
85 |
|
---|
86 | res_sz = getline(&res_buf, &res_sz, file);
|
---|
87 |
|
---|
88 | if ((ssize_t)res_sz != -1) {
|
---|
89 | if (res_buf[res_sz - 1] == '\n')
|
---|
90 | res_buf[--res_sz] = '\0';
|
---|
91 | //TODO: trimming to res_sz?
|
---|
92 | } else {
|
---|
93 | free(res_buf); /* uclibc allocates a buffer even on EOF. WTF? */
|
---|
94 | res_buf = NULL;
|
---|
95 | }
|
---|
96 | return res_buf;
|
---|
97 | }
|
---|
98 |
|
---|
99 | #endif
|
---|
100 |
|
---|
101 | #if 0
|
---|
102 | /* Faster routines (~twice as fast). +170 bytes. Unused as of 2008-07.
|
---|
103 | *
|
---|
104 | * NB: they stop at NUL byte too.
|
---|
105 | * Performance is important here. Think "grep 50gigabyte_file"...
|
---|
106 | * Ironically, grep can't use it because of NUL issue.
|
---|
107 | * We sorely need C lib to provide fgets which reports size!
|
---|
108 | *
|
---|
109 | * Update:
|
---|
110 | * Actually, uclibc and glibc have it. man getline. It's GNUism,
|
---|
111 | * but very useful one (if it's as fast as this code).
|
---|
112 | * TODO:
|
---|
113 | * - currently, sed and sort use bb_get_chunk_from_file and heavily
|
---|
114 | * depend on its "stop on \n or \0" behavior, and STILL they fail
|
---|
115 | * to handle all cases with embedded NULs correctly. So:
|
---|
116 | * - audit sed and sort; convert them to getline FIRST.
|
---|
117 | * - THEN ditch bb_get_chunk_from_file, replace it with getline.
|
---|
118 | * - provide getline implementation for non-GNU systems.
|
---|
119 | */
|
---|
120 |
|
---|
121 | static char* xmalloc_fgets_internal(FILE *file, int *sizep)
|
---|
122 | {
|
---|
123 | int len;
|
---|
124 | int idx = 0;
|
---|
125 | char *linebuf = NULL;
|
---|
126 |
|
---|
127 | while (1) {
|
---|
128 | char *r;
|
---|
129 |
|
---|
130 | linebuf = xrealloc(linebuf, idx + 0x100);
|
---|
131 | r = fgets(&linebuf[idx], 0x100, file);
|
---|
132 | if (!r) {
|
---|
133 | /* need to terminate in case this is error
|
---|
134 | * (EOF puts NUL itself) */
|
---|
135 | linebuf[idx] = '\0';
|
---|
136 | break;
|
---|
137 | }
|
---|
138 | /* stupid. fgets knows the len, it should report it somehow */
|
---|
139 | len = strlen(&linebuf[idx]);
|
---|
140 | idx += len;
|
---|
141 | if (len != 0xff || linebuf[idx - 1] == '\n')
|
---|
142 | break;
|
---|
143 | }
|
---|
144 | *sizep = idx;
|
---|
145 | if (idx) {
|
---|
146 | /* xrealloc(linebuf, idx + 1) is up to caller */
|
---|
147 | return linebuf;
|
---|
148 | }
|
---|
149 | free(linebuf);
|
---|
150 | return NULL;
|
---|
151 | }
|
---|
152 |
|
---|
153 | /* Get line, remove trailing \n */
|
---|
154 | char* FAST_FUNC xmalloc_fgetline_fast(FILE *file)
|
---|
155 | {
|
---|
156 | int sz;
|
---|
157 | char *r = xmalloc_fgets_internal(file, &sz);
|
---|
158 | if (r && r[sz - 1] == '\n')
|
---|
159 | r[--sz] = '\0';
|
---|
160 | return r; /* not xrealloc(r, sz + 1)! */
|
---|
161 | }
|
---|
162 |
|
---|
163 | char* FAST_FUNC xmalloc_fgets(FILE *file)
|
---|
164 | {
|
---|
165 | int sz;
|
---|
166 | return xmalloc_fgets_internal(file, &sz);
|
---|
167 | }
|
---|
168 |
|
---|
169 | /* Get line, remove trailing \n */
|
---|
170 | char* FAST_FUNC xmalloc_fgetline(FILE *file)
|
---|
171 | {
|
---|
172 | int sz;
|
---|
173 | char *r = xmalloc_fgets_internal(file, &sz);
|
---|
174 | if (!r)
|
---|
175 | return r;
|
---|
176 | if (r[sz - 1] == '\n')
|
---|
177 | r[--sz] = '\0';
|
---|
178 | return xrealloc(r, sz + 1);
|
---|
179 | }
|
---|
180 | #endif
|
---|