« back to all changes in this revision
Viewing changes to main.c
- browse files at revision 4
- compare with another revision
- download diff
- download tarball
- view history from revision 4
- Committer: Bazaar Package Importer
- Author(s): Daniel Baumann
- Date: 2005-12-27 17:01:00 UTC
- mfrom: (1.1.1 upstream)
- Revision ID: james.westby@ubuntu.com-20051227170100-nk2hnq0bnlkbk3q3
Tags: 4.17-2
Added patch to fix FTBS on amd64 (Closes: #344909).
- files added:
- Makefile.NeXT
- debian/patches
- save
- files removed:
- agrep.ps.1
- files modified:
- COPYRIGHT
added
removed
main.c
1
/* Copyright (c) 1991 Sun Wu and Udi Manber. All Rights Reserved. */
1
/* Copyright (c) 1994 Sun Wu, Udi Manber, Burra Gopal. All Rights Reserved. */
2
#include <stdio.h>
2
3
#include "agrep.h"
3
#include "checkfile.h"
4
5
unsigned Mask[MAXSYM];
6
unsigned Init1, NO_ERR_MASK, Init[MaxError];
7
unsigned Bit[WORD+1];
8
CHAR buffer[BlockSize+Maxline+1];
9
unsigned Next[MaxNext], Next1[MaxNext];
10
unsigned wildmask, endposition, D_endpos;
11
int REGEX, RE_ERR, FNAME, WHOLELINE, SIMPLEPATTERN;
12
int COUNT, HEAD, TAIL, LINENUM, INVERSE, I, S, DD, AND, SGREP, JUMP;
13
int Num_Pat, PSIZE, num_of_matched, SILENT, NOPROMPT, BESTMATCH, NOUPPER;
14
int NOMATCH, TRUNCATE, FIRST_IN_RE, FIRSTOUTPUT;
15
int WORDBOUND, DELIMITER, D_length;
16
int EATFIRST, OUTTAIL;
17
int FILEOUT;
18
int DNA = 0;
19
int APPROX = 0;
20
int PAT_FILE = 0;
21
int CONSTANT = 0;
22
int total_line = 0; /* used in mgrep */
23
24
CHAR **Textfiles; /* array of filenames to be searched */
25
CHAR old_D_pat[MaxDelimit] = "\n"; /* to hold original D_pattern */
26
CHAR CurrentFileName[MAXNAME];
27
CHAR Progname[MAXNAME];
28
CHAR D_pattern[MaxDelimit] = "\n; "; /* string which delimits records --
29
defaults to newline */
30
int NOFILENAME = 0, /* Boolean flag, set for -h option */
31
FILENAMEONLY = 0,/* Boolean flag, set for -l option */
32
Numfiles = 0; /* indicates how many files in Textfiles */
33
extern int init();
34
int table[WORD][WORD];
35
36
initial_value()
37
{
38
int i;
39
40
JUMP = REGEX = FNAME = BESTMATCH = NOPROMPT = NOUPPER = 0;
41
COUNT = LINENUM = WHOLELINE = SGREP = 0;
42
EATFIRST = INVERSE = AND = TRUNCATE = OUTTAIL = 0;
43
FIRST_IN_RE = NOMATCH = FIRSTOUTPUT = ON;
44
I = DD = S = 1;
45
HEAD = TAIL = ON;
46
D_length = 2;
47
SILENT = Num_Pat = PSIZE = SIMPLEPATTERN = num_of_matched = 0 ;
48
WORDBOUND = DELIMITER = RE_ERR = 0;
49
Bit[WORD] = 1;
50
for (i = WORD - 1; i > 0 ; i--) Bit[i] = Bit[i+1] << 1;
51
for (i=0; i< MAXSYM; i++) Mask[i] = 0;
52
}
53
54
compute_next(M, Next, Next1)
55
int M; unsigned *Next, *Next1;
56
{
57
int i, j=0, n, k, temp;
58
int mid, pp;
59
int MM, base;
60
unsigned V[WORD];
61
62
base = WORD - M;
63
temp = Bit[base]; Bit[base] = 0;
64
for (i=0; i<WORD; i++) V[i] = 0;
65
for (i=1; i<M; i++)
66
{
67
j=0;
68
while (table[i][j] > 0 && j < 10) {
69
V[i] = V[i] | Bit[base + table[i][j++]];
70
}
71
}
72
Bit[base]=temp;
73
if(M <= SHORTREG)
74
{
75
k = exponen(M);
76
pp = 2*k;
77
for(i=k; i<pp ; i++)
78
{ n = i;
79
Next[i]= (k>>1);
80
for(j=M; j>=1; j--)
81
{
82
if(n & Bit[WORD]) Next[i] = Next[i] | V[j];
83
n = (n>>1);
84
}
85
}
86
return;
87
}
88
if(M > MAXREG) fprintf(stderr, "%s: regular expression too long\n", Progname);
89
MM = M;
90
if(M & 1) M=M+1;
91
k = exponen(M/2);
92
pp = 2*k;
93
mid = MM/2;
94
for(i=k; i<pp ; i++)
95
{ n = i;
96
Next[i]= (Bit[base]>>1);
97
for(j=MM; j>mid ; j--)
98
{
99
if(n & Bit[WORD]) Next[i] = Next[i] | V[j-mid];
100
n = (n>>1);
101
}
102
n=i-k;
103
Next1[i-k] = 0;
104
for(j = 0; j<mid; j++)
105
{
106
if(n & Bit[WORD]) Next1[i-k] = Next1[i-k] | V[MM-j];
107
n = (n>>1);
108
}
109
}
110
return;
111
}
112
113
exponen(m)
114
int m;
115
{ int i, ex;
116
ex= 1;
117
for (i=0; i<m; i++) ex= ex*2;
118
return(ex);
119
}
120
121
re1(Text, M, D)
122
int Text, M, D;
123
{
124
register unsigned i, c, r0, r1, r2, r3, CMask, Newline, Init0, r_NO_ERR;
125
register unsigned end;
126
register unsigned hh, LL=0, k; /* Lower part */
127
int FIRST_TIME=ON, num_read , j=0, base;
128
unsigned A[MaxRerror+1], B[MaxRerror+1];
129
unsigned Next[MaxNext], Next1[MaxNext];
130
CHAR buffer[BlockSize+Maxline+1];
131
int FIRST_LOOP = 1;
132
133
r_NO_ERR = NO_ERR_MASK;
134
if(M > 30) {
135
fprintf(stderr, "%s: regular expression too long\n", Progname);
136
exit(2);
137
}
138
base = WORD - M;
139
hh = M/2;
140
for(i=WORD, j=0; j < hh ; i--, j++) LL = LL | Bit[i];
141
if(FIRST_IN_RE) compute_next(M, Next, Next1);
142
/*SUN: try: change to memory allocation */
143
FIRST_IN_RE = 0;
144
Newline = '\n';
145
Init[0] = Bit[base];
146
if(HEAD) Init[0] = Init[0] | Bit[base+1];
147
for(i=1; i<= D; i++) Init[i] = Init[i-1] | Next[Init[i-1]>>hh] | Next1[Init[i-1]&LL];
148
Init1 = Init[0] | 1;
149
Init0 = Init[0];
150
r2 = r3 = Init[0];
151
for(k=0; k<= D; k++) { A[k] = B[k] = Init[k]; }
152
if ( D == 0 )
153
{
154
while ((num_read = read(Text, buffer + Maxline, BlockSize)) > 0)
155
{
156
i=Maxline; end = num_read + Maxline;
157
if((num_read < BlockSize) && buffer[end-1] != '\n') buffer[end] = '\n';
158
if(FIRST_LOOP) { /* if first time in the loop add a newline */
159
buffer[i-1] = '\n'; /* in front the text. */
160
i--;
161
FIRST_LOOP = 0;
162
}
163
while ( i < end )
164
{
165
c = buffer[i++];
166
CMask = Mask[c];
167
if(c != Newline)
168
{ if(CMask != 0) {
169
r1 = Init1 & r3;
170
r2 = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | r1;
171
}
172
else {
173
r2 = r3 & Init1;
174
}
175
}
176
else { j++;
177
r1 = Init1 & r3; /* match against endofline */
178
r2 = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | r1;
179
if(TAIL) r2 = (Next[r2>>hh] | Next1[r2&LL]) | r2; /* epsilon move */
180
if(( r2 & 1 ) ^ INVERSE) {
181
if(FILENAMEONLY) {
182
num_of_matched++;
183
printf("%s\n", CurrentFileName);
184
return;
185
}
186
r_output(buffer, i-1, end, j);
187
}
188
r3 = Init0;
189
r2 = (Next[r3>>hh] | Next1[r3&LL]) & CMask | Init0;
190
/* match begin of line */
191
}
192
c = buffer[i++];
193
CMask = Mask[c];
194
if(c != Newline)
195
{ if(CMask != 0) {
196
r1 = Init1 & r2;
197
r3 = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
198
}
199
else r3 = r2 & Init1;
200
} /* if(NOT Newline) */
201
else { j++;
202
r1 = Init1 & r2; /* match against endofline */
203
r3 = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
204
if(TAIL) r3 = ( Next[r3>>hh] | Next1[r3&LL] ) | r3;
205
/* epsilon move */
206
if(( r3 & 1 ) ^ INVERSE) {
207
if(FILENAMEONLY) {
208
num_of_matched++;
209
printf("%s\n", CurrentFileName);
210
return;
211
}
212
r_output(buffer, i-1, end, j);
213
}
214
r2 = Init0;
215
r3 = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | Init0;
216
/* match begin of line */
217
}
218
} /* while i < end ... */
219
strncpy(buffer, buffer+num_read, Maxline);
220
} /* end while read()... */
221
return;
222
} /* end if (D == 0) */
223
while ((num_read = read(Text, buffer + Maxline, BlockSize)) > 0)
224
{
225
i=Maxline; end = Maxline + num_read;
226
if((num_read < BlockSize) && buffer[end-1] != '\n') buffer[end] = '\n';
227
if(FIRST_TIME) { /* if first time in the loop add a newline */
228
buffer[i-1] = '\n'; /* in front the text. */
229
i--;
230
FIRST_TIME = 0;
231
}
232
while (i < end )
233
{
234
c = buffer[i];
235
CMask = Mask[c];
236
if(c != Newline)
237
{
238
if(CMask != 0) {
239
r2 = B[0];
240
r1 = Init1 & r2;
241
A[0] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
242
r3 = B[1];
243
r1 = Init1 & r3;
244
r0 = r2 | A[0]; /* A[0] | B[0] */
245
A[1] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | (( r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
246
if(D == 1) goto Nextchar;
247
r2 = B[2];
248
r1 = Init1 & r2;
249
r0 = r3 | A[1];
250
A[2] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
251
if(D == 2) goto Nextchar;
252
r3 = B[3];
253
r1 = Init1 & r3;
254
r0 = r2 | A[2];
255
A[3] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
256
if(D == 3) goto Nextchar;
257
r2 = B[4];
258
r1 = Init1 & r2;
259
r0 = r3 | A[3];
260
A[4] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
261
if(D == 4) goto Nextchar;
262
} /* if(CMask) */
263
else {
264
r2 = B[0];
265
A[0] = r2 & Init1;
266
r3 = B[1];
267
r1 = Init1 & r3;
268
r0 = r2 | A[0];
269
A[1] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
270
if(D == 1) goto Nextchar;
271
r2 = B[2];
272
r1 = Init1 & r2;
273
r0 = r3 | A[1];
274
A[2] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
275
if(D == 2) goto Nextchar;
276
r3 = B[3];
277
r1 = Init1 & r3;
278
r0 = r2 | A[2];
279
A[3] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
280
if(D == 3) goto Nextchar;
281
r2 = B[4];
282
r1 = Init1 & r2;
283
r0 = r3 | A[3];
284
A[4] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
285
if(D == 4) goto Nextchar;
286
}
287
}
288
else { j++;
289
r1 = Init1 & B[D]; /* match against endofline */
290
A[D] = ((Next[B[D]>>hh] | Next1[B[D]&LL]) & CMask) | r1;
291
if(TAIL) A[D] = ( Next[A[D]>>hh] | Next1[A[D]&LL] ) | A[D];
292
/* epsilon move */
293
if(( A[D] & 1 ) ^ INVERSE) {
294
if(FILENAMEONLY) {
295
num_of_matched++;
296
printf("%s\n", CurrentFileName);
297
return;
298
}
299
r_output(buffer, i, end, j);
300
}
301
for(k=0; k<=D; k++) B[k] = Init[0];
302
r1 = Init1 & B[0];
303
A[0] = (( Next[B[0]>>hh] | Next1[B[0]&LL]) & CMask) | r1;
304
for(k=1; k<=D; k++) {
305
r3 = B[k];
306
r1 = Init1 & r3;
307
r2 = A[k-1] | B[k-1];
308
A[k] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((B[k-1] | Next[r2>>hh] | Next1[r2&LL]) & r_NO_ERR) | r1;
309
}
310
}
311
Nextchar: i=i+1;
312
c = buffer[i];
313
CMask = Mask[c];
314
if(c != Newline)
315
{
316
if(CMask != 0) {
317
r2 = A[0];
318
r1 = Init1 & r2;
319
B[0] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
320
r3 = A[1];
321
r1 = Init1 & r3;
322
r0 = B[0] | r2;
323
B[1] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL]) & r_NO_ERR) | r1 ;
324
if(D == 1) goto Nextchar1;
325
r2 = A[2];
326
r1 = Init1 & r2;
327
r0 = B[1] | r3;
328
B[2] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
329
if(D == 2) goto Nextchar1;
330
r3 = A[3];
331
r1 = Init1 & r3;
332
r0 = B[2] | r2;
333
B[3] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
334
if(D == 3) goto Nextchar1;
335
r2 = A[4];
336
r1 = Init1 & r2;
337
r0 = B[3] | r3;
338
B[4] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
339
if(D == 4) goto Nextchar1;
340
} /* if(CMask) */
341
else {
342
r2 = A[0];
343
B[0] = r2 & Init1;
344
r3 = A[1];
345
r1 = Init1 & r3;
346
r0 = B[0] | r2;
347
B[1] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
348
if(D == 1) goto Nextchar1;
349
r2 = A[2];
350
r1 = Init1 & r2;
351
r0 = B[1] | r3;
352
B[2] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
353
if(D == 2) goto Nextchar1;
354
r3 = A[3];
355
r1 = Init1 & r3;
356
r0 = B[2] | r2;
357
B[3] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
358
if(D == 3) goto Nextchar1;
359
r2 = A[4];
360
r1 = Init1 & r2;
361
r0 = B[3] | r3;
362
B[4] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;
363
if(D == 4) goto Nextchar1;
364
}
365
} /* if(NOT Newline) */
366
else { j++;
367
r1 = Init1 & A[D]; /* match against endofline */
368
B[D] = ((Next[A[D]>>hh] | Next1[A[D]&LL]) & CMask) | r1;
369
if(TAIL) B[D] = ( Next[B[D]>>hh] | Next1[B[D]&LL] ) | B[D];
370
/* epsilon move */
371
if(( B[D] & 1 ) ^ INVERSE) {
372
if(FILENAMEONLY) {
373
num_of_matched++;
374
printf("%s\n", CurrentFileName);
375
return;
376
}
377
r_output(buffer, i, end, j);
378
}
379
for(k=0; k<=D; k++) A[k] = Init0;
380
r1 = Init1 & A[0];
381
B[0] = ((Next[A[0]>>hh] | Next1[A[0]&LL]) & CMask) | r1;
382
for(k=1; k<=D; k++) {
383
r3 = A[k];
384
r1 = Init1 & r3;
385
r2 = A[k-1] | B[k-1];
386
B[k] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((A[k-1] | Next[r2>>hh] | Next1[r2&LL]) & r_NO_ERR) | r1;
387
}
388
}
389
Nextchar1: i=i+1;
390
} /* while */
391
strncpy(buffer, buffer+num_read, Maxline);
392
} /* while */
393
return;
394
} /* re1 */
395
396
re(Text, M, D)
397
int Text, M, D;
398
{
399
register unsigned i, c, r1, r2, r3, CMask, k, Newline, Init0, Init1, end;
400
register unsigned r_even, r_odd, r_NO_ERR ;
401
unsigned RMask[MAXSYM];
402
unsigned A[MaxRerror+1], B[MaxRerror+1];
403
int num_read, j=0, bp, lasti, base, ResidueSize;
404
int FIRST_TIME; /* Flag */
405
base = WORD - M;
406
k = 2*exponen(M);
407
if(FIRST_IN_RE) {
408
compute_next(M, Next, Next1);
409
FIRST_IN_RE = 0; }
410
for(i=0; i< MAXSYM; i++) RMask[i] = Mask[i];
411
r_NO_ERR = NO_ERR_MASK;
412
Newline = '\n';
413
lasti = Maxline;
414
Init0 = Init[0] = Bit[base];
415
if(HEAD) Init0 = Init[0] = Init0 | Bit[base+1] ;
416
for(i=1; i<= D; i++) Init[i] = Init[i-1] | Next[Init[i-1]]; /* can be out? */
417
Init1 = Init0 | 1;
418
r2 = r3 = Init0;
419
for(k=0; k<= D; k++) { A[k] = B[k] = Init[0]; } /* can be out? */
420
FIRST_TIME = ON;
421
if ( D == 0 )
422
{
423
while ((num_read = read(Text, buffer + Maxline, BlockSize)) > 0)
424
{
425
i=Maxline; end = Maxline + num_read ;
426
if((num_read < BlockSize)&&buffer[end-1] != '\n') buffer[end] = '\n';
427
if(FIRST_TIME) {
428
buffer[i-1] = '\n';
429
i--;
430
FIRST_TIME = 0;
431
}
432
while (i < end)
433
{
434
c = buffer[i++];
435
CMask = RMask[c];
436
if(c != Newline)
437
{
438
r1 = Init1 & r3;
439
r2 = (Next[r3] & CMask) | r1;
440
}
441
else {
442
r1 = Init1 & r3; /* match against '\n' */
443
r2 = Next[r3] & CMask | r1;
444
j++;
445
if(TAIL) r2 = Next[r2] | r2 ; /* epsilon move */
446
if(( r2 & 1) ^ INVERSE) {
447
if(FILENAMEONLY) {
448
num_of_matched++;
449
printf("%s\n", CurrentFileName);
450
return;
451
}
452
r_output(buffer, i-1, end, j);
453
}
454
lasti = i - 1;
455
r3 = Init0;
456
r2 = (Next[r3] & CMask) | Init0;
457
}
458
c = buffer[i++];
459
CMask = RMask[c];
460
if(c != Newline)
461
{
462
r1 = Init1 & r2;
463
r3 = (Next[r2] & CMask) | r1;
464
}
465
else { j++;
466
r1 = Init1 & r2; /* match against endofline */
467
r3 = Next[r2] & CMask | r1;
468
if(TAIL) r3 = Next[r3] | r3;
469
if(( r3 & 1) ^ INVERSE) {
470
if(FILENAMEONLY) {
471
num_of_matched++;
472
printf("%s\n", CurrentFileName);
473
return;
474
}
475
r_output(buffer, i-1, end, j);
476
}
477
lasti = i - 1;
478
r2 = Init0;
479
r3 = (Next[r2] & CMask) | Init0; /* match the newline */
480
}
481
} /* while */
482
ResidueSize = Maxline + num_read - lasti;
483
if(ResidueSize > Maxline) {
484
ResidueSize = Maxline; }
485
strncpy(buffer+Maxline-ResidueSize, buffer+lasti, ResidueSize);
486
lasti = Maxline - ResidueSize;
487
} /* while */
488
return;
489
} /* end if(D==0) */
490
while ((num_read = read(Text, buffer + Maxline, BlockSize)) > 0)
491
{
492
i=Maxline; end = Maxline+num_read;
493
if((num_read < BlockSize) && buffer[end-1] != '\n') buffer[end] = '\n';
494
if(FIRST_TIME) {
495
buffer[i-1] = '\n';
496
i--;
497
FIRST_TIME = 0;
498
}
499
while (i < end)
500
{ c = buffer[i++];
501
CMask = RMask[c];
502
if (c != Newline)
503
{
504
r_even = B[0];
505
r1 = Init1 & r_even;
506
A[0] = (Next[r_even] & CMask) | r1;
507
r_odd = B[1];
508
r1 = Init1 & r_odd;
509
r2 = (r_even | Next[r_even|A[0]]) &r_NO_ERR;
510
A[1] = (Next[r_odd] & CMask) | r2 | r1 ;
511
if(D == 1) goto Nextchar;
512
r_even = B[2];
513
r1 = Init1 & r_even;
514
r2 = (r_odd | Next[r_odd|A[1]]) &r_NO_ERR;
515
A[2] = (Next[r_even] & CMask) | r2 | r1 ;
516
if(D == 2) goto Nextchar;
517
r_odd = B[3];
518
r1 = Init1 & r_odd;
519
r2 = (r_even | Next[r_even|A[2]]) &r_NO_ERR;
520
A[3] = (Next[r_odd] & CMask) | r2 | r1 ;
521
if(D == 3) goto Nextchar;
522
r_even = B[4];
523
r1 = Init1 & r_even;
524
r2 = (r_odd | Next[r_odd|A[3]]) &r_NO_ERR;
525
A[4] = (Next[r_even] & CMask) | r2 | r1 ;
526
goto Nextchar;
527
} /* if NOT Newline */
528
else { j++;
529
r1 = Init1 & B[D]; /* match endofline */
530
A[D] = (Next[B[D]] & CMask) | r1;
531
if(TAIL) A[D] = Next[A[D]] | A[D];
532
if((A[D] & 1) ^ INVERSE ) {
533
if(FILENAMEONLY) {
534
num_of_matched++;
535
printf("%s\n", CurrentFileName);
536
return;
537
}
538
r_output(buffer, i-1, end, j);
539
}
540
for(k=0; k<= D; k++) { A[k] = B[k] = Init[k]; }
541
r1 = Init1 & B[0];
542
A[0] = (Next[B[0]] & CMask) | r1;
543
for(k=1; k<= D; k++) {
544
r1 = Init1 & B[k];
545
r2 = (B[k-1] | Next[A[k-1]|B[k-1]]) &r_NO_ERR;
546
A[k] = (Next[B[k]] & CMask) | r1 | r2;
547
}
548
}
549
Nextchar:
550
c = buffer[i];
551
CMask = RMask[c];
552
if(c != Newline)
553
{
554
r1 = Init1 & A[0];
555
B[0] = (Next[A[0]] & CMask) | r1;
556
r1 = Init1 & A[1];
557
B[1] = (Next[A[1]] & CMask) | ((A[0] | Next[A[0] | B[0]]) & r_NO_ERR) | r1 ;
558
if(D == 1) goto Nextchar1;
559
r1 = Init1 & A[2];
560
B[2] = (Next[A[2]] & CMask) | ((A[1] | Next[A[1] | B[1]]) &r_NO_ERR) | r1 ;
561
if(D == 2) goto Nextchar1;
562
r1 = Init1 & A[3];
563
B[3] = (Next[A[3]] & CMask) | ((A[2] | Next[A[2] | B[2]])&r_NO_ERR) | r1 ;
564
if(D == 3) goto Nextchar1;
565
r1 = Init1 & A[4];
566
B[4] = (Next[A[4]] & CMask) | ((A[3] | Next[A[3] | B[3]])&r_NO_ERR) | r1 ;
567
goto Nextchar1;
568
} /* if(NOT Newline) */
569
else { j++;
570
r1 = Init1 & A[D]; /* match endofline */
571
B[D] = (Next[A[D]] & CMask) | r1;
572
if(TAIL) B[D] = Next[B[D]] | B[D];
573
if((B[D] & 1) ^ INVERSE ) {
574
if(FILENAMEONLY) {
575
num_of_matched++;
576
printf("%s\n", CurrentFileName);
577
return;
578
}
579
r_output(buffer, i, end, j);
580
}
581
for(k=0; k<= D; k++) { A[k] = B[k] = Init[k]; }
582
r1 = Init1 & A[0];
583
B[0] = (Next[A[0]] & CMask) | r1;
584
for(k=1; k<= D; k++) {
585
r1 = Init1 & A[k];
586
r2 = (A[k-1] | Next[A[k-1]|B[k-1]])&r_NO_ERR;
587
B[k] = (Next[A[k]] & CMask) | r1 | r2;
588
}
589
}
590
Nextchar1: i++;
591
} /* while i < end */
592
strncpy(buffer, buffer+num_read, Maxline);
593
} /* while read() */
594
return;
595
} /* re */
596
597
598
r_output (buffer, i, end, j)
599
int i, end, j;
600
CHAR *buffer;
601
{
602
int bp;
603
if(i >= end) return;
604
num_of_matched++;
605
if(COUNT) return;
606
if(FNAME) printf("%s: ", CurrentFileName);
607
bp = i-1;
608
while ((buffer[bp] != '\n') && (bp > 0)) bp--;
609
if(LINENUM) printf("%d: ", j);
610
if(buffer[bp] != '\n') bp = Maxline-1;
611
bp++;
612
while (bp <= i ) putchar(buffer[bp++]);
613
}
614
4
#if ISO_CHAR_SET
5
#include <locale.h> /* support for 8bit character set: ew@senate.be */
6
#endif
7
8
#if MEASURE_TIMES
9
extern int INFILTER_ms, OUTFILTER_ms, FILTERALGO_ms;
10
#endif /*MEASURE_TIMES*/
11
12
extern char Pattern[MAXPAT];
13
extern int EXITONERROR;
14
15
#include "dummysyscalls.c"
16
17
int
615
18
main(argc, argv)
616
int argc; char *argv[];
617
{
618
int N, M, D=0, fp, fd, i, j;
619
char c;
620
int filetype;
621
unsigned char Pattern[MAXPAT], OldPattern[MAXPAT], temp[MAXPAT];
622
623
initial_value();
624
strcpy(Progname, argv[0]);
625
if (argc < 2) usage();
626
Pattern[0] = '\0';
627
while(--argc > 0 && (*++argv)[0] == '-') {
628
c = *(argv[0]+1);
629
switch(c) {
630
case 'c' : COUNT = ON; /* output the # of matched */
631
break;
632
case 's' : SILENT = ON; /* silent mode */
633
break;
634
case 'p' : I = 0; /* insertion cost is 0 */
635
break;
636
case 'x' : WHOLELINE = ON; /* match the whole line */
637
if(WORDBOUND) {
638
fprintf(stderr, "%s: illegal option combination\n", Progname);
639
exit(2);
640
}
641
break;
642
case 'L' : break;
643
case 'd' : DELIMITER = ON; /* user defines delimiter */
644
if(argc <= 1) usage();
645
if (argv[0][2] == '\0') {/* space after -d option */
646
argv++;
647
if ((D_length = strlen(argv[0])) > MaxDelimit) {
648
fprintf(stderr, "%s: delimiter pattern too long\n", Progname);
649
exit(2);
650
}
651
D_pattern[0] = '<';
652
strcpy(D_pattern+1, argv[0]);
653
argc--;
654
} else {
655
if ((D_length = strlen(argv[0]+2)) > MaxDelimit) {
656
fprintf(stderr, "%s: delimiter pattern too long\n", Progname);
657
exit(2);
658
}
659
D_pattern[0] = '<';
660
strcpy(D_pattern+1, argv[0]+2);
661
} /* else */
662
strcat(D_pattern, ">; ");
663
D_length++; /* to count ';' as one */
664
break;
665
case 'e' : argc--;
666
if(argc == 0) {
667
fprintf(stderr, "%s: the pattern should immediately follow the -e option\n", Progname);
668
usage();
669
}
670
if((++argv)[0][0] == '-') {
671
Pattern[0] = '\\';
672
strcat(Pattern, (argv)[0]);
673
}
674
else strcat(Pattern, argv[0]);
675
break;
676
case 'f' : PAT_FILE = ON;
677
argv++;
678
argc--;
679
if((fp = open(argv[0], 0)) < 0) {
680
fprintf(stderr, "%s: Can't open pattern file %s\n", Progname, argv[0]);
681
exit(2);
682
}
683
break;
684
case 'h' : NOFILENAME = ON;
685
break;
686
case 'i' : NOUPPER = ON;
687
break;
688
case 'k' : argc--;
689
if(argc == 0) {
690
fprintf(stderr, "%s: the pattern should immediately follow the -k option\n", Progname);
691
usage();
692
}
693
CONSTANT = ON;
694
argv++;
695
strcat(Pattern, argv[0]);
696
if(argc > 1 && argv[1][0] == '-') {
697
fprintf(stderr, "%s: -k should be the last option in the command\n", Progname);
698
exit(2);
699
}
700
break;
701
case 'l' : FILENAMEONLY = ON;
702
break;
703
case 'n' : LINENUM = ON; /* output prefixed by line no*/
704
break;
705
case 'v' : INVERSE = ON; /* output no-matched lines */
706
break;
707
case 't' : OUTTAIL = ON; /* output from tail of delimiter */
708
break;
709
case 'B' : BESTMATCH = ON;
710
break;
711
case 'w' : WORDBOUND = ON;/* match to words */
712
if(WHOLELINE) {
713
fprintf(stderr, "%s: illegal option combination\n", Progname);
714
exit(2);
715
}
716
break;
717
case 'y' : NOPROMPT = ON;
718
break;
719
case 'I' : I = atoi(argv[0]+2); /* Insertion Cost */
720
JUMP = ON;
721
break;
722
case 'S' : S = atoi(argv[0]+2); /* Substitution Cost */
723
JUMP = ON;
724
break;
725
case 'D' : DD = atoi(argv[0]+2); /* Deletion Cost */
726
JUMP = ON;
727
break;
728
case 'G' : FILEOUT = ON;
729
COUNT = ON;
730
break;
731
default : if (isdigit(c)) {
732
APPROX = ON;
733
D = atoi(argv[0]+1);
734
if (D > MaxError) {
735
fprintf(stderr,"%s: the maximum number of errors is %d \n", Progname, MaxError);
736
exit(2);
737
}
738
} else {
739
fprintf(stderr, "%s: illegal option -%c\n",Progname, c);
740
usage();
741
}
742
} /* switch(c) */
743
} /* while (--argc > 0 && (*++argv)[0] == '-') */
744
745
if (FILENAMEONLY && NOFILENAME) {
746
fprintf(stderr, "%s: -h and -l options are mutually exclusive\n",Progname);
747
}
748
if (COUNT && (FILENAMEONLY || NOFILENAME)) {
749
FILENAMEONLY = OFF;
750
if(!FILEOUT) NOFILENAME = OFF;
751
}
752
if (!(PAT_FILE) && Pattern[0] == '\0') { /* Pattern not set with -e option */
753
if (argc == 0) usage();
754
strcpy(Pattern, *argv);
755
argc--;
756
argv++;
757
}
758
Numfiles = 0;
759
fd = 3; /* make sure it's not 0 */
760
if (argc == 0) /* check Pattern against stdin */
761
fd = 0;
762
else {
763
if (!(Textfiles = (CHAR **)malloc(argc * sizeof(CHAR *) ))) {
764
fprintf(stderr, "%s: malloc failure (you probably don't have enough memory)\n", Progname);
765
exit(2);
766
}
767
while (argc--) { /* one or more filenames on command line -- put
768
the valid filenames in a array of strings */
769
/*
770
if ((filetype = check_file(*argv)) != ISASCIIFILE) {
771
if(filetype == NOSUCHFILE) fprintf(stderr,"%s: %s: no such file or directory\n",Progname,*argv);
772
argv++;
773
*/
774
775
if ((filetype = check_file(*argv)) == NOSUCHFILE) {
776
if(filetype == NOSUCHFILE) fprintf(stderr,"%s: %s: no such file or directory\n",Progname,*argv);
777
argv++;
778
} else { /* file is ascii*/
779
if (!(Textfiles[Numfiles] = (CHAR *)malloc((strlen(*argv)+1)))) {
780
fprintf(stderr, "%s: malloc failure (you probably don't have enough memory)\n", Progname);
781
exit(2);
782
}
783
strcpy(Textfiles[Numfiles++], *argv++);
784
} /* else */
785
} /* while (argc--) */
786
} /* else */
787
checksg(Pattern, D); /* check if the pattern is simple */
788
strcpy(OldPattern, Pattern);
789
if (SGREP == 0) {
790
preprocess(D_pattern, Pattern);
791
strcpy(old_D_pat, D_pattern);
792
M = maskgen(Pattern, D);
793
}
794
else M = strlen(OldPattern);
795
if (PAT_FILE) prepf(fp);
796
if (Numfiles > 1) FNAME = ON;
797
if (NOFILENAME) FNAME = 0;
798
num_of_matched = 0;
799
compat(); /* check compatibility between options */
800
if (fd == 0) {
801
if(FILENAMEONLY) {
802
fprintf(stderr, "%s: -l option is not compatible with standard input\n", Progname);
803
exit(2);
804
}
805
if(PAT_FILE) mgrep(fd);
806
else {
807
if(SGREP) sgrep(OldPattern, strlen(OldPattern), fd, D);
808
else bitap(old_D_pat, Pattern, fd, M, D);
809
}
810
if (COUNT) {
811
if(INVERSE && PAT_FILE) printf("%d\n", total_line-num_of_matched);
812
else printf("%d\n", num_of_matched);
813
}
814
}
815
else {
816
for (i = 0; i < Numfiles; i++, close(fd), num_of_matched = 0) {
817
strcpy(CurrentFileName, Textfiles[i]);
818
if ((fd = open(Textfiles[i], 0)) <= 0) {
819
fprintf(stderr, "%s: can't open file %s\n",Progname, Textfiles[i]);
820
}
821
else {
822
if(PAT_FILE) mgrep(fd);
823
else {
824
if(SGREP) sgrep(OldPattern, strlen(OldPattern), fd, D);
825
else bitap(old_D_pat, Pattern, fd, M, D);
826
}
827
if (num_of_matched) NOMATCH = OFF;
828
if (COUNT && !FILEOUT) {
829
if(INVERSE && PAT_FILE) {
830
if(FNAME) printf("%s: %d\n", CurrentFileName, total_line - num_of_matched);
831
else printf("%d\n", total_line - num_of_matched);
832
}
833
else {
834
if(FNAME) printf("%s: %d\n", CurrentFileName, num_of_matched);
835
else printf("%d\n", num_of_matched);
836
}
837
} /* if COUNT */
838
if(FILEOUT && num_of_matched) {
839
file_out(CurrentFileName);
840
}
841
} /* else */
842
} /* for i < Numfiles */
843
if(NOMATCH && BESTMATCH) {
844
if(WORDBOUND || WHOLELINE || LINENUM || INVERSE) {
845
SGREP = 0;
846
preprocess(D_pattern, Pattern);
847
strcpy(old_D_pat, D_pattern);
848
M = maskgen(Pattern, D);
849
}
850
COUNT=ON; D=1;
851
while(D<M && D<=MaxError && num_of_matched == 0) {
852
for (i = 0; i < Numfiles; i++, close(fd)) {
853
strcpy(CurrentFileName, Textfiles[i]);
854
if ((fd = open(Textfiles[i], 0)) > 0) {
855
if(PAT_FILE) mgrep(fd);
856
else {
857
if(SGREP) sgrep(OldPattern,strlen(OldPattern),fd,D);
858
else bitap(old_D_pat,Pattern,fd,M,D);
859
}
860
}
861
} /* for i < Numfiles */
862
D++;
863
} /* while */
864
if(num_of_matched > 0) {
865
D--; COUNT = 0;
866
if(NOPROMPT) goto GO_AHEAD;
867
if(D==1) fprintf(stderr, "best match has 1 error, ");
868
else fprintf(stderr, "best match has %d errors, ", D);
869
fflush(stderr);
870
if(num_of_matched == 1) fprintf(stderr,"there is 1 match, output it? (y/n)");
871
else fprintf(stderr,"there are %d matches, output them? (y/n)", num_of_matched);
872
scanf("%c",&c);
873
if(c != 'y') goto CONT;
874
GO_AHEAD:
875
for (i = 0; i < Numfiles; i++, close(fd)) {
876
strcpy(CurrentFileName, Textfiles[i]);
877
if ((fd = open(Textfiles[i], 0)) > 0) {
878
if(PAT_FILE) mgrep(fd);
879
else {
880
if(SGREP) sgrep(OldPattern,strlen(OldPattern),fd,D);
881
else bitap(old_D_pat,Pattern,fd,M,D);
882
}
883
}
884
} /* for i < Numfiles */
885
NOMATCH = 0;
886
}
887
}
888
}
889
CONT:
890
if(EATFIRST) {
891
printf("\n");
892
EATFIRST = OFF;
893
}
894
if(num_of_matched) NOMATCH = OFF;
895
if(NOMATCH) exit(1);
896
exit(0);
897
} /* end of main() */
898
899
900
file_out(fname)
901
char *fname;
902
{
903
int num_read;
904
int fd;
905
int i, len;
906
CHAR buf[4097];
907
if(FNAME) {
908
len = strlen(fname);
909
putchar('\n');
910
for(i=0; i< len; i++) putchar(':');
911
putchar('\n');
912
printf("%s\n", CurrentFileName);
913
len = strlen(fname);
914
for(i=0; i< len; i++) putchar(':');
915
putchar('\n');
916
fflush(stdout);
917
}
918
fd = open(fname, 0);
919
while((num_read = read(fd, buf, 4096)) > 0)
920
write(1, buf, num_read);
921
}
922
923
924
usage()
925
{
926
fprintf(stderr, "usage: %s [-#cdehiklnpstvwxBDGIS] [-f patternfile] pattern [files]\n", Progname);
927
printf("\n");
928
fprintf(stderr, "summary of frequently used options:\n");
929
fprintf(stderr, "-#: find matches with at most # errors\n");
930
fprintf(stderr, "-c: output the number of matched records\n");
931
fprintf(stderr, "-d: define record delimiter\n");
932
fprintf(stderr, "-h: do not output file names\n");
933
fprintf(stderr, "-i: case-insensitive search, e.g., 'a' = 'A'\n");
934
fprintf(stderr, "-l: output the names of files that contain a match\n");
935
fprintf(stderr, "-n: output record prefixed by record number\n");
936
fprintf(stderr, "-v: output those records containing no matches\n");
937
fprintf(stderr, "-w: pattern has to match as a word, e.g., 'win' will not match 'wind'\n");
938
fprintf(stderr, "-B: best match mode. find the closest matches to the pattern\n");
939
fprintf(stderr, "-G: output the files that contain a match\n");
940
printf("\n");
941
942
exit(2);
943
}
944
945
checksg(Pattern, D)
946
CHAR *Pattern; int D;
947
{
948
char c;
949
int i, m;
950
m = strlen(Pattern);
951
if(!(PAT_FILE) && m <= D) {
952
fprintf(stderr, "%s: size of pattern must be greater than number of errors\n", Progname);
953
exit(2);
954
}
955
SIMPLEPATTERN = ON;
956
for (i=0; i < m; i++)
957
{
958
switch(Pattern[i])
959
{
960
case ';' : SIMPLEPATTERN = OFF; break;
961
case ',' : SIMPLEPATTERN = OFF; break;
962
case '.' : SIMPLEPATTERN = OFF; break;
963
case '*' : SIMPLEPATTERN = OFF; break;
964
case '-' : SIMPLEPATTERN = OFF; break;
965
case '[' : SIMPLEPATTERN = OFF; break;
966
case ']' : SIMPLEPATTERN = OFF; break;
967
case '(' : SIMPLEPATTERN = OFF; break;
968
case ')' : SIMPLEPATTERN = OFF; break;
969
case '<' : SIMPLEPATTERN = OFF; break;
970
case '>' : SIMPLEPATTERN = OFF; break;
971
case '^' : if(D > 0) SIMPLEPATTERN = OFF;
972
break;
973
case '$' : if(D > 0) SIMPLEPATTERN = OFF;
974
break;
975
case '|' : SIMPLEPATTERN = OFF; break;
976
case '#' : SIMPLEPATTERN = OFF; break;
977
case '\\' : SIMPLEPATTERN = OFF; break;
978
default : break;
979
}
980
}
981
if (CONSTANT) SIMPLEPATTERN = ON;
982
if (SIMPLEPATTERN == OFF) return;
983
if (NOUPPER && D) return;
984
if (JUMP == ON) return;
985
if (I == 0) return;
986
if (LINENUM) return;
987
if (DELIMITER) return;
988
if (INVERSE) return;
989
if (WORDBOUND && D > 0) return;
990
if (WHOLELINE && D > 0) return;
991
if (SILENT) return; /* REMINDER: to be removed */
992
SGREP = ON;
993
if(m >= 16) DNA = ON;
994
for(i=0; i<m; i++) {
995
c = Pattern[i];
996
if(c == 'a' || c == 'c' || c == 't' || c == 'g' ) ;
997
else DNA = OFF;
998
}
999
return;
1000
}
1001
1002
output (buffer, i1, i2, j)
1003
register CHAR *buffer; int i1, i2, j;
1004
{
1005
register CHAR *bp, *outend;
1006
if(i1 > i2) return;
1007
num_of_matched++;
1008
if(COUNT) return;
1009
if(SILENT) return;
1010
if(OUTTAIL) {
1011
i1 = i1 + D_length;
1012
i2 = i2 + D_length;
1013
}
1014
if(DELIMITER) j = j+1;
1015
if(FIRSTOUTPUT) {
1016
if (buffer[i1] == '\n') {
1017
i1++;
1018
EATFIRST = ON;
1019
}
1020
FIRSTOUTPUT = 0;
1021
}
1022
if(TRUNCATE) {
1023
fprintf(stderr, "WARNING!!! some lines have been truncated in output record #%d\n", num_of_matched-1);
1024
}
1025
while(buffer[i1] == '\n' && i1 <= i2) {
1026
printf("\n");
1027
i1++;
1028
}
1029
if(FNAME == ON) printf("%s: ", CurrentFileName);
1030
if(LINENUM) printf("%d: ", j-1);
1031
bp = buffer + i1;
1032
outend = buffer + i2;
1033
while(bp <= outend) putchar(*bp++);
1034
}
1035
1036
/* end of main.c */
19
int argc;
20
char *argv[];
21
{
22
int ret;
23
24
#if ISO_CHAR_SET
25
setlocale(LC_ALL,""); /* support for 8bit character set: ew@senate.be, Henrik.Martin@eua.ericsson.se, "O.Bartunov" <megera@sai.msu.su>, S.Nazin (leng@sai.msu.su) */
26
#endif
27
28
EXITONERROR = 1; /* the only place where it is set to 1 */
29
ret = fileagrep(argc, argv, 0, stdout);
30
31
#if MEASURE_TIMES
32
fprintf(stderr, "ret = %d infilter = %d ms\toutfilter = %d ms\tfilteralgo = %d ms\n", ret, INFILTER_ms, OUTFILTER_ms, FILTERALGO_ms);
33
#endif /*MEASURE_TIMES*/
34
if(ret<0) exit(2);
35
if(ret==0) exit(1);
36
exit(0);
37
}
Loggerhead is a web-based interface for Breezy
Version: 2.0.1