|
Line 0
Link Here
|
|
|
1 |
#include <sys/types.h> |
| 2 |
#include <stdio.h> |
| 3 |
#include <regex.h> |
| 4 |
#include <string.h> |
| 5 |
#include <stdlib.h> |
| 6 |
#include <errno.h> |
| 7 |
|
| 8 |
typedef struct { |
| 9 |
regex_t re; |
| 10 |
char *regexp; |
| 11 |
} mawk_re_t; |
| 12 |
|
| 13 |
static mawk_re_t *last_used_regexp = NULL; |
| 14 |
static int err_code = 0; |
| 15 |
|
| 16 |
void prepare_regexp (regexp) |
| 17 |
char *regexp; |
| 18 |
{ |
| 19 |
int bs = 0; |
| 20 |
char *tail = regexp; |
| 21 |
char ch; |
| 22 |
|
| 23 |
while ((ch = *regexp++) != 0){ |
| 24 |
if (bs){ |
| 25 |
switch (ch){ |
| 26 |
case 'n': *tail++ = '\n'; break; |
| 27 |
case 't': *tail++ = '\t'; break; |
| 28 |
case 'f': *tail++ = '\f'; break; |
| 29 |
case 'b': *tail++ = '\b'; break; |
| 30 |
case 'r': *tail++ = '\r'; break; |
| 31 |
case 'a': *tail++ = '\07'; break; |
| 32 |
case 'v': *tail++ = '\013'; break; |
| 33 |
default: *tail++ = '\\'; *tail++ = ch; |
| 34 |
} |
| 35 |
|
| 36 |
bs = 0; |
| 37 |
}else{ |
| 38 |
if (ch == '\\'){ |
| 39 |
bs = 1; |
| 40 |
}else{ |
| 41 |
*tail++ = ch; |
| 42 |
} |
| 43 |
} |
| 44 |
} |
| 45 |
|
| 46 |
*tail = 0; |
| 47 |
} |
| 48 |
|
| 49 |
void * REcompile(regexp) |
| 50 |
char *regexp; |
| 51 |
{ |
| 52 |
mawk_re_t *re = (mawk_re_t*) malloc (sizeof (mawk_re_t)); |
| 53 |
size_t len = strlen (regexp); |
| 54 |
char *new_regexp = (char *) malloc (len + 3); |
| 55 |
/* fprintf (stderr, "REcompile: %s\n", regexp); */ |
| 56 |
|
| 57 |
if (!re || !new_regexp) |
| 58 |
return NULL; |
| 59 |
|
| 60 |
new_regexp [0] = '('; |
| 61 |
memcpy (new_regexp + 1, regexp, len); |
| 62 |
new_regexp [len+1] = ')'; |
| 63 |
new_regexp [len+2] = 0; |
| 64 |
|
| 65 |
prepare_regexp (new_regexp); |
| 66 |
|
| 67 |
last_used_regexp = re; |
| 68 |
|
| 69 |
memset (re, 0, sizeof (mawk_re_t)); |
| 70 |
re -> regexp = strdup (new_regexp); |
| 71 |
err_code = regcomp (&re->re, new_regexp, REG_EXTENDED | REG_NEWLINE); |
| 72 |
|
| 73 |
free (new_regexp); |
| 74 |
|
| 75 |
if (err_code) |
| 76 |
return NULL; |
| 77 |
|
| 78 |
return re; |
| 79 |
} |
| 80 |
|
| 81 |
int |
| 82 |
REtest(str, re) |
| 83 |
char *str ; |
| 84 |
mawk_re_t* re ; |
| 85 |
{ |
| 86 |
/* fprintf (stderr, "REtest: \"%s\" ~ /%s/", str, re -> regexp); */ |
| 87 |
|
| 88 |
last_used_regexp = re; |
| 89 |
|
| 90 |
if (regexec (&re->re, str, 0, NULL, 0)){ |
| 91 |
/* fprintf (stderr, "=1\n"); */ |
| 92 |
return 0; |
| 93 |
}else{ |
| 94 |
/* fprintf (stderr, "=0\n"); */ |
| 95 |
return 1; |
| 96 |
} |
| 97 |
} |
| 98 |
|
| 99 |
char *REmatch(str, re, lenp) |
| 100 |
char *str ; |
| 101 |
mawk_re_t* re ; |
| 102 |
unsigned *lenp ; |
| 103 |
{ |
| 104 |
regmatch_t match [100]; |
| 105 |
/* fprintf (stderr, "REmatch: \"%s\" ~ /%s/", str, re -> regexp); */ |
| 106 |
|
| 107 |
last_used_regexp = re; |
| 108 |
|
| 109 |
if (!regexec (&re->re, str, 100, match, 0)){ |
| 110 |
*lenp = match [0].rm_eo - match[0].rm_so; |
| 111 |
/* fprintf (stderr, "=%i/%i\n", match [0].rm_so, *lenp); */ |
| 112 |
return str + match [0].rm_so; |
| 113 |
}else{ |
| 114 |
/* fprintf (stderr, "=0\n"); */ |
| 115 |
return NULL; |
| 116 |
} |
| 117 |
} |
| 118 |
|
| 119 |
void REmprint(m, f) |
| 120 |
void * m ; |
| 121 |
FILE *f ; |
| 122 |
{ |
| 123 |
/* no debugging code available */ |
| 124 |
abort (); |
| 125 |
} |
| 126 |
|
| 127 |
static char error_buffer [2048]; |
| 128 |
|
| 129 |
char *REerror () |
| 130 |
{ |
| 131 |
size_t len; |
| 132 |
if (last_used_regexp){ |
| 133 |
len = regerror (err_code, &last_used_regexp -> re, |
| 134 |
error_buffer, sizeof (error_buffer)); |
| 135 |
return error_buffer; |
| 136 |
}else{ |
| 137 |
snprintf (error_buffer, sizeof (error_buffer), "malloc failed: %s", |
| 138 |
strerror (errno)); |
| 139 |
} |
| 140 |
} |