1 /* $NetBSD: io.c,v 1.1.1.1 2016/01/13 03:15:30 christos Exp $ */
2
3 /* File I/O for GNU DIFF.
4
5 Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1998, 2001, 2002
6 Free Software Foundation, Inc.
7
8 This file is part of GNU DIFF.
9
10 GNU DIFF is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 2, or (at your option)
13 any later version.
14
15 GNU DIFF is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
19
20 You should have received a copy of the GNU General Public License
21 along with this program; see the file COPYING.
22 If not, write to the Free Software Foundation,
23 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
24
25 #include "diff.h"
26 #include <cmpbuf.h>
27 #include <regex.h>
28 #include <setmode.h>
29 #include <xalloc.h>
30
31 /* Rotate an unsigned value to the left. */
32 #define ROL(v, n) ((v) << (n) | (v) >> (sizeof (v) * CHAR_BIT - (n)))
33
34 /* Given a hash value and a new character, return a new hash value. */
35 #define HASH(h, c) ((c) + ROL (h, 7))
36
37 /* The type of a hash value. */
38 typedef size_t hash_value;
39 verify (hash_value_is_unsigned, ! TYPE_SIGNED (hash_value));
40
41 /* Lines are put into equivalence classes of lines that match in lines_differ.
42 Each equivalence class is represented by one of these structures,
43 but only while the classes are being computed.
44 Afterward, each class is represented by a number. */
45 struct equivclass
46 {
47 lin next; /* Next item in this bucket. */
48 hash_value hash; /* Hash of lines in this class. */
49 char const *line; /* A line that fits this class. */
50 size_t length; /* That line's length, not counting its newline. */
51 };
52
53 /* Hash-table: array of buckets, each being a chain of equivalence classes.
54 buckets[-1] is reserved for incomplete lines. */
55 static lin *buckets;
56
57 /* Number of buckets in the hash table array, not counting buckets[-1]. */
58 static size_t nbuckets;
59
60 /* Array in which the equivalence classes are allocated.
61 The bucket-chains go through the elements in this array.
62 The number of an equivalence class is its index in this array. */
63 static struct equivclass *equivs;
64
65 /* Index of first free element in the array `equivs'. */
66 static lin equivs_index;
67
68 /* Number of elements allocated in the array `equivs'. */
69 static lin equivs_alloc;
70
71 /* Read a block of data into a file buffer, checking for EOF and error. */
72
73 void
file_block_read(struct file_data * current,size_t size)74 file_block_read (struct file_data *current, size_t size)
75 {
76 if (size && ! current->eof)
77 {
78 size_t s = block_read (current->desc,
79 FILE_BUFFER (current) + current->buffered, size);
80 if (s == SIZE_MAX)
81 pfatal_with_name (current->name);
82 current->buffered += s;
83 current->eof = s < size;
84 }
85 }
86
87 /* Check for binary files and compare them for exact identity. */
88
89 /* Return 1 if BUF contains a non text character.
90 SIZE is the number of characters in BUF. */
91
92 #define binary_file_p(buf, size) (memchr (buf, 0, size) != 0)
93
94 /* Get ready to read the current file.
95 Return nonzero if SKIP_TEST is zero,
96 and if it appears to be a binary file. */
97
98 static bool
sip(struct file_data * current,bool skip_test)99 sip (struct file_data *current, bool skip_test)
100 {
101 /* If we have a nonexistent file at this stage, treat it as empty. */
102 if (current->desc < 0)
103 {
104 /* Leave room for a sentinel. */
105 current->bufsize = sizeof (word);
106 current->buffer = xmalloc (current->bufsize);
107 }
108 else
109 {
110 current->bufsize = buffer_lcm (sizeof (word),
111 STAT_BLOCKSIZE (current->stat),
112 PTRDIFF_MAX - 2 * sizeof (word));
113 current->buffer = xmalloc (current->bufsize);
114
115 if (! skip_test)
116 {
117 /* Check first part of file to see if it's a binary file. */
118
119 bool was_binary = set_binary_mode (current->desc, 1);
120 off_t buffered;
121 file_block_read (current, current->bufsize);
122 buffered = current->buffered;
123
124 if (! was_binary)
125 {
126 /* Revert to text mode and seek back to the beginning to
127 reread the file. Use relative seek, since file
128 descriptors like stdin might not start at offset
129 zero. */
130
131 if (lseek (current->desc, - buffered, SEEK_CUR) == -1)
132 pfatal_with_name (current->name);
133 set_binary_mode (current->desc, 0);
134 current->buffered = 0;
135 current->eof = 0;
136 }
137
138 return binary_file_p (current->buffer, buffered);
139 }
140 }
141
142 current->buffered = 0;
143 current->eof = 0;
144 return 0;
145 }
146
147 /* Slurp the rest of the current file completely into memory. */
148
149 static void
slurp(struct file_data * current)150 slurp (struct file_data *current)
151 {
152 size_t cc;
153
154 if (current->desc < 0)
155 {
156 /* The file is nonexistent. */
157 return;
158 }
159
160 if (S_ISREG (current->stat.st_mode))
161 {
162 /* It's a regular file; slurp in the rest all at once. */
163
164 /* Get the size out of the stat block.
165 Allocate just enough room for appended newline plus word sentinel,
166 plus word-alignment since we want the buffer word-aligned. */
167 size_t file_size = current->stat.st_size;
168 cc = file_size + 2 * sizeof (word) - file_size % sizeof (word);
169 if (file_size != current->stat.st_size || cc < file_size
170 || PTRDIFF_MAX <= cc)
171 xalloc_die ();
172
173 if (current->bufsize < cc)
174 {
175 current->bufsize = cc;
176 current->buffer = xrealloc (current->buffer, cc);
177 }
178
179 /* Try to read at least 1 more byte than the size indicates, to
180 detect whether the file is growing. This is a nicety for
181 users who run 'diff' on files while they are changing. */
182
183 if (current->buffered <= file_size)
184 {
185 file_block_read (current, file_size + 1 - current->buffered);
186 if (current->buffered <= file_size)
187 return;
188 }
189 }
190
191 /* It's not a regular file, or it's a growing regular file; read it,
192 growing the buffer as needed. */
193
194 file_block_read (current, current->bufsize - current->buffered);
195
196 if (current->buffered)
197 {
198 while (current->buffered == current->bufsize)
199 {
200 if (PTRDIFF_MAX / 2 - sizeof (word) < current->bufsize)
201 xalloc_die ();
202 current->bufsize *= 2;
203 current->buffer = xrealloc (current->buffer, current->bufsize);
204 file_block_read (current, current->bufsize - current->buffered);
205 }
206
207 /* Allocate just enough room for appended newline plus word
208 sentinel, plus word-alignment. */
209 cc = current->buffered + 2 * sizeof (word);
210 current->bufsize = cc - cc % sizeof (word);
211 current->buffer = xrealloc (current->buffer, current->bufsize);
212 }
213 }
214
215 /* Split the file into lines, simultaneously computing the equivalence
216 class for each line. */
217
218 static void
find_and_hash_each_line(struct file_data * current)219 find_and_hash_each_line (struct file_data *current)
220 {
221 hash_value h;
222 unsigned char const *p = (unsigned char const *) current->prefix_end;
223 unsigned char c;
224 lin i, *bucket;
225 size_t length;
226
227 /* Cache often-used quantities in local variables to help the compiler. */
228 char const **linbuf = current->linbuf;
229 lin alloc_lines = current->alloc_lines;
230 lin line = 0;
231 lin linbuf_base = current->linbuf_base;
232 lin *cureqs = xmalloc (alloc_lines * sizeof *cureqs);
233 struct equivclass *eqs = equivs;
234 lin eqs_index = equivs_index;
235 lin eqs_alloc = equivs_alloc;
236 char const *suffix_begin = current->suffix_begin;
237 char const *bufend = FILE_BUFFER (current) + current->buffered;
238 bool diff_length_compare_anyway =
239 ignore_white_space != IGNORE_NO_WHITE_SPACE;
240 bool same_length_diff_contents_compare_anyway =
241 diff_length_compare_anyway | ignore_case;
242
243 while ((char const *) p < suffix_begin)
244 {
245 char const *ip = (char const *) p;
246
247 h = 0;
248
249 /* Hash this line until we find a newline. */
250 if (ignore_case)
251 switch (ignore_white_space)
252 {
253 case IGNORE_ALL_SPACE:
254 while ((c = *p++) != '\n')
255 if (! ISSPACE (c))
256 h = HASH (h, TOLOWER (c));
257 break;
258
259 case IGNORE_SPACE_CHANGE:
260 while ((c = *p++) != '\n')
261 {
262 if (ISSPACE (c))
263 {
264 do
265 if ((c = *p++) == '\n')
266 goto hashing_done;
267 while (ISSPACE (c));
268
269 h = HASH (h, ' ');
270 }
271
272 /* C is now the first non-space. */
273 h = HASH (h, TOLOWER (c));
274 }
275 break;
276
277 case IGNORE_TAB_EXPANSION:
278 {
279 size_t column = 0;
280 while ((c = *p++) != '\n')
281 {
282 int repetitions = 1;
283
284 switch (c)
285 {
286 case '\b':
287 column -= 0 < column;
288 break;
289
290 case '\t':
291 c = ' ';
292 repetitions = TAB_WIDTH - column % TAB_WIDTH;
293 column += repetitions;
294 break;
295
296 case '\r':
297 column = 0;
298 break;
299
300 default:
301 c = TOLOWER (c);
302 column++;
303 break;
304 }
305
306 do
307 h = HASH (h, c);
308 while (--repetitions != 0);
309 }
310 }
311 break;
312
313 default:
314 while ((c = *p++) != '\n')
315 h = HASH (h, TOLOWER (c));
316 break;
317 }
318 else
319 switch (ignore_white_space)
320 {
321 case IGNORE_ALL_SPACE:
322 while ((c = *p++) != '\n')
323 if (! ISSPACE (c))
324 h = HASH (h, c);
325 break;
326
327 case IGNORE_SPACE_CHANGE:
328 while ((c = *p++) != '\n')
329 {
330 if (ISSPACE (c))
331 {
332 do
333 if ((c = *p++) == '\n')
334 goto hashing_done;
335 while (ISSPACE (c));
336
337 h = HASH (h, ' ');
338 }
339
340 /* C is now the first non-space. */
341 h = HASH (h, c);
342 }
343 break;
344
345 case IGNORE_TAB_EXPANSION:
346 {
347 size_t column = 0;
348 while ((c = *p++) != '\n')
349 {
350 int repetitions = 1;
351
352 switch (c)
353 {
354 case '\b':
355 column -= 0 < column;
356 break;
357
358 case '\t':
359 c = ' ';
360 repetitions = TAB_WIDTH - column % TAB_WIDTH;
361 column += repetitions;
362 break;
363
364 case '\r':
365 column = 0;
366 break;
367
368 default:
369 column++;
370 break;
371 }
372
373 do
374 h = HASH (h, c);
375 while (--repetitions != 0);
376 }
377 }
378 break;
379
380 default:
381 while ((c = *p++) != '\n')
382 h = HASH (h, c);
383 break;
384 }
385
386 hashing_done:;
387
388 bucket = &buckets[h % nbuckets];
389 length = (char const *) p - ip - 1;
390
391 if ((char const *) p == bufend
392 && current->missing_newline
393 && ROBUST_OUTPUT_STYLE (output_style))
394 {
395 /* This line is incomplete. If this is significant,
396 put the line into buckets[-1]. */
397 if (ignore_white_space < IGNORE_SPACE_CHANGE)
398 bucket = &buckets[-1];
399
400 /* Omit the inserted newline when computing linbuf later. */
401 p--;
402 bufend = suffix_begin = (char const *) p;
403 }
404
405 for (i = *bucket; ; i = eqs[i].next)
406 if (!i)
407 {
408 /* Create a new equivalence class in this bucket. */
409 i = eqs_index++;
410 if (i == eqs_alloc)
411 {
412 if (PTRDIFF_MAX / (2 * sizeof *eqs) <= eqs_alloc)
413 xalloc_die ();
414 eqs_alloc *= 2;
415 eqs = xrealloc (eqs, eqs_alloc * sizeof *eqs);
416 }
417 eqs[i].next = *bucket;
418 eqs[i].hash = h;
419 eqs[i].line = ip;
420 eqs[i].length = length;
421 *bucket = i;
422 break;
423 }
424 else if (eqs[i].hash == h)
425 {
426 char const *eqline = eqs[i].line;
427
428 /* Reuse existing class if lines_differ reports the lines
429 equal. */
430 if (eqs[i].length == length)
431 {
432 /* Reuse existing equivalence class if the lines are identical.
433 This detects the common case of exact identity
434 faster than lines_differ would. */
435 if (memcmp (eqline, ip, length) == 0)
436 break;
437 if (!same_length_diff_contents_compare_anyway)
438 continue;
439 }
440 else if (!diff_length_compare_anyway)
441 continue;
442
443 if (! lines_differ (eqline, ip))
444 break;
445 }
446
447 /* Maybe increase the size of the line table. */
448 if (line == alloc_lines)
449 {
450 /* Double (alloc_lines - linbuf_base) by adding to alloc_lines. */
451 if (PTRDIFF_MAX / 3 <= alloc_lines
452 || PTRDIFF_MAX / sizeof *cureqs <= 2 * alloc_lines - linbuf_base
453 || PTRDIFF_MAX / sizeof *linbuf <= alloc_lines - linbuf_base)
454 xalloc_die ();
455 alloc_lines = 2 * alloc_lines - linbuf_base;
456 cureqs = xrealloc (cureqs, alloc_lines * sizeof *cureqs);
457 linbuf += linbuf_base;
458 linbuf = xrealloc (linbuf,
459 (alloc_lines - linbuf_base) * sizeof *linbuf);
460 linbuf -= linbuf_base;
461 }
462 linbuf[line] = ip;
463 cureqs[line] = i;
464 ++line;
465 }
466
467 current->buffered_lines = line;
468
469 for (i = 0; ; i++)
470 {
471 /* Record the line start for lines in the suffix that we care about.
472 Record one more line start than lines,
473 so that we can compute the length of any buffered line. */
474 if (line == alloc_lines)
475 {
476 /* Double (alloc_lines - linbuf_base) by adding to alloc_lines. */
477 if (PTRDIFF_MAX / 3 <= alloc_lines
478 || PTRDIFF_MAX / sizeof *cureqs <= 2 * alloc_lines - linbuf_base
479 || PTRDIFF_MAX / sizeof *linbuf <= alloc_lines - linbuf_base)
480 xalloc_die ();
481 alloc_lines = 2 * alloc_lines - linbuf_base;
482 linbuf += linbuf_base;
483 linbuf = xrealloc (linbuf,
484 (alloc_lines - linbuf_base) * sizeof *linbuf);
485 linbuf -= linbuf_base;
486 }
487 linbuf[line] = (char const *) p;
488
489 if ((char const *) p == bufend)
490 break;
491
492 if (context <= i && no_diff_means_no_output)
493 break;
494
495 line++;
496
497 while (*p++ != '\n')
498 continue;
499 }
500
501 /* Done with cache in local variables. */
502 current->linbuf = linbuf;
503 current->valid_lines = line;
504 current->alloc_lines = alloc_lines;
505 current->equivs = cureqs;
506 equivs = eqs;
507 equivs_alloc = eqs_alloc;
508 equivs_index = eqs_index;
509 }
510
511 /* Prepare the text. Make sure the text end is initialized.
512 Make sure text ends in a newline,
513 but remember that we had to add one.
514 Strip trailing CRs, if that was requested. */
515
516 static void
prepare_text(struct file_data * current)517 prepare_text (struct file_data *current)
518 {
519 size_t buffered = current->buffered;
520 char *p = FILE_BUFFER (current);
521 char *dst;
522
523 if (buffered == 0 || p[buffered - 1] == '\n')
524 current->missing_newline = 0;
525 else
526 {
527 p[buffered++] = '\n';
528 current->missing_newline = 1;
529 }
530
531 if (!p)
532 return;
533
534 /* Don't use uninitialized storage when planting or using sentinels. */
535 memset (p + buffered, 0, sizeof (word));
536
537 if (strip_trailing_cr && (dst = memchr (p, '\r', buffered)))
538 {
539 char const *src = dst;
540 char const *srclim = p + buffered;
541
542 do
543 dst += ! ((*dst = *src++) == '\r' && *src == '\n');
544 while (src < srclim);
545
546 buffered -= src - dst;
547 }
548
549 current->buffered = buffered;
550 }
551
552 /* We have found N lines in a buffer of size S; guess the
553 proportionate number of lines that will be found in a buffer of
554 size T. However, do not guess a number of lines so large that the
555 resulting line table might cause overflow in size calculations. */
556 static lin
guess_lines(lin n,size_t s,size_t t)557 guess_lines (lin n, size_t s, size_t t)
558 {
559 size_t guessed_bytes_per_line = n < 10 ? 32 : s / (n - 1);
560 lin guessed_lines = MAX (1, t / guessed_bytes_per_line);
561 return MIN (guessed_lines, PTRDIFF_MAX / (2 * sizeof (char *) + 1) - 5) + 5;
562 }
563
564 /* Given a vector of two file_data objects, find the identical
565 prefixes and suffixes of each object. */
566
567 static void
find_identical_ends(struct file_data filevec[])568 find_identical_ends (struct file_data filevec[])
569 {
570 word *w0, *w1;
571 char *p0, *p1, *buffer0, *buffer1;
572 char const *end0, *beg0;
573 char const **linbuf0, **linbuf1;
574 lin i, lines;
575 size_t n0, n1;
576 lin alloc_lines0, alloc_lines1;
577 lin buffered_prefix, prefix_count, prefix_mask;
578 lin middle_guess, suffix_guess;
579
580 slurp (&filevec[0]);
581 prepare_text (&filevec[0]);
582 if (filevec[0].desc != filevec[1].desc)
583 {
584 slurp (&filevec[1]);
585 prepare_text (&filevec[1]);
586 }
587 else
588 {
589 filevec[1].buffer = filevec[0].buffer;
590 filevec[1].bufsize = filevec[0].bufsize;
591 filevec[1].buffered = filevec[0].buffered;
592 filevec[1].missing_newline = filevec[0].missing_newline;
593 }
594
595 /* Find identical prefix. */
596
597 w0 = filevec[0].buffer;
598 w1 = filevec[1].buffer;
599 p0 = buffer0 = (char *) w0;
600 p1 = buffer1 = (char *) w1;
601 n0 = filevec[0].buffered;
602 n1 = filevec[1].buffered;
603
604 if (p0 == p1)
605 /* The buffers are the same; sentinels won't work. */
606 p0 = p1 += n1;
607 else
608 {
609 /* Insert end sentinels, in this case characters that are guaranteed
610 to make the equality test false, and thus terminate the loop. */
611
612 if (n0 < n1)
613 p0[n0] = ~p1[n0];
614 else
615 p1[n1] = ~p0[n1];
616
617 /* Loop until first mismatch, or to the sentinel characters. */
618
619 /* Compare a word at a time for speed. */
620 while (*w0 == *w1)
621 w0++, w1++;
622
623 /* Do the last few bytes of comparison a byte at a time. */
624 p0 = (char *) w0;
625 p1 = (char *) w1;
626 while (*p0 == *p1)
627 p0++, p1++;
628
629 /* Don't mistakenly count missing newline as part of prefix. */
630 if (ROBUST_OUTPUT_STYLE (output_style)
631 && ((buffer0 + n0 - filevec[0].missing_newline < p0)
632 !=
633 (buffer1 + n1 - filevec[1].missing_newline < p1)))
634 p0--, p1--;
635 }
636
637 /* Now P0 and P1 point at the first nonmatching characters. */
638
639 /* Skip back to last line-beginning in the prefix,
640 and then discard up to HORIZON_LINES lines from the prefix. */
641 i = horizon_lines;
642 while (p0 != buffer0 && (p0[-1] != '\n' || i--))
643 p0--, p1--;
644
645 /* Record the prefix. */
646 filevec[0].prefix_end = p0;
647 filevec[1].prefix_end = p1;
648
649 /* Find identical suffix. */
650
651 /* P0 and P1 point beyond the last chars not yet compared. */
652 p0 = buffer0 + n0;
653 p1 = buffer1 + n1;
654
655 if (! ROBUST_OUTPUT_STYLE (output_style)
656 || filevec[0].missing_newline == filevec[1].missing_newline)
657 {
658 end0 = p0; /* Addr of last char in file 0. */
659
660 /* Get value of P0 at which we should stop scanning backward:
661 this is when either P0 or P1 points just past the last char
662 of the identical prefix. */
663 beg0 = filevec[0].prefix_end + (n0 < n1 ? 0 : n0 - n1);
664
665 /* Scan back until chars don't match or we reach that point. */
666 for (; p0 != beg0; p0--, p1--)
667 if (*p0 != *p1)
668 {
669 /* Point at the first char of the matching suffix. */
670 beg0 = p0;
671 break;
672 }
673
674 /* Are we at a line-beginning in both files? If not, add the rest of
675 this line to the main body. Discard up to HORIZON_LINES lines from
676 the identical suffix. Also, discard one extra line,
677 because shift_boundaries may need it. */
678 i = horizon_lines + !((buffer0 == p0 || p0[-1] == '\n')
679 &&
680 (buffer1 == p1 || p1[-1] == '\n'));
681 while (i-- && p0 != end0)
682 while (*p0++ != '\n')
683 continue;
684
685 p1 += p0 - beg0;
686 }
687
688 /* Record the suffix. */
689 filevec[0].suffix_begin = p0;
690 filevec[1].suffix_begin = p1;
691
692 /* Calculate number of lines of prefix to save.
693
694 prefix_count == 0 means save the whole prefix;
695 we need this for options like -D that output the whole file,
696 or for enormous contexts (to avoid worrying about arithmetic overflow).
697 We also need it for options like -F that output some preceding line;
698 at least we will need to find the last few lines,
699 but since we don't know how many, it's easiest to find them all.
700
701 Otherwise, prefix_count != 0. Save just prefix_count lines at start
702 of the line buffer; they'll be moved to the proper location later.
703 Handle 1 more line than the context says (because we count 1 too many),
704 rounded up to the next power of 2 to speed index computation. */
705
706 if (no_diff_means_no_output && ! function_regexp.fastmap
707 && context < LIN_MAX / 4 && context < n0)
708 {
709 middle_guess = guess_lines (0, 0, p0 - filevec[0].prefix_end);
710 suffix_guess = guess_lines (0, 0, buffer0 + n0 - p0);
711 for (prefix_count = 1; prefix_count <= context; prefix_count *= 2)
712 continue;
713 alloc_lines0 = (prefix_count + middle_guess
714 + MIN (context, suffix_guess));
715 }
716 else
717 {
718 prefix_count = 0;
719 alloc_lines0 = guess_lines (0, 0, n0);
720 }
721
722 prefix_mask = prefix_count - 1;
723 lines = 0;
724 linbuf0 = xmalloc (alloc_lines0 * sizeof *linbuf0);
725 p0 = buffer0;
726
727 /* If the prefix is needed, find the prefix lines. */
728 if (! (no_diff_means_no_output
729 && filevec[0].prefix_end == p0
730 && filevec[1].prefix_end == p1))
731 {
732 end0 = filevec[0].prefix_end;
733 while (p0 != end0)
734 {
735 lin l = lines++ & prefix_mask;
736 if (l == alloc_lines0)
737 {
738 if (PTRDIFF_MAX / (2 * sizeof *linbuf0) <= alloc_lines0)
739 xalloc_die ();
740 alloc_lines0 *= 2;
741 linbuf0 = xrealloc (linbuf0, alloc_lines0 * sizeof *linbuf0);
742 }
743 linbuf0[l] = p0;
744 while (*p0++ != '\n')
745 continue;
746 }
747 }
748 buffered_prefix = prefix_count && context < lines ? context : lines;
749
750 /* Allocate line buffer 1. */
751
752 middle_guess = guess_lines (lines, p0 - buffer0, p1 - filevec[1].prefix_end);
753 suffix_guess = guess_lines (lines, p0 - buffer0, buffer1 + n1 - p1);
754 alloc_lines1 = buffered_prefix + middle_guess + MIN (context, suffix_guess);
755 if (alloc_lines1 < buffered_prefix
756 || PTRDIFF_MAX / sizeof *linbuf1 <= alloc_lines1)
757 xalloc_die ();
758 linbuf1 = xmalloc (alloc_lines1 * sizeof *linbuf1);
759
760 if (buffered_prefix != lines)
761 {
762 /* Rotate prefix lines to proper location. */
763 for (i = 0; i < buffered_prefix; i++)
764 linbuf1[i] = linbuf0[(lines - context + i) & prefix_mask];
765 for (i = 0; i < buffered_prefix; i++)
766 linbuf0[i] = linbuf1[i];
767 }
768
769 /* Initialize line buffer 1 from line buffer 0. */
770 for (i = 0; i < buffered_prefix; i++)
771 linbuf1[i] = linbuf0[i] - buffer0 + buffer1;
772
773 /* Record the line buffer, adjusted so that
774 linbuf[0] points at the first differing line. */
775 filevec[0].linbuf = linbuf0 + buffered_prefix;
776 filevec[1].linbuf = linbuf1 + buffered_prefix;
777 filevec[0].linbuf_base = filevec[1].linbuf_base = - buffered_prefix;
778 filevec[0].alloc_lines = alloc_lines0 - buffered_prefix;
779 filevec[1].alloc_lines = alloc_lines1 - buffered_prefix;
780 filevec[0].prefix_lines = filevec[1].prefix_lines = lines;
781 }
782
783 /* If 1 < k, then (2**k - prime_offset[k]) is the largest prime less
784 than 2**k. This table is derived from Chris K. Caldwell's list
785 <http://www.utm.edu/research/primes/lists/2small/>. */
786
787 static unsigned char const prime_offset[] =
788 {
789 0, 0, 1, 1, 3, 1, 3, 1, 5, 3, 3, 9, 3, 1, 3, 19, 15, 1, 5, 1, 3, 9, 3,
790 15, 3, 39, 5, 39, 57, 3, 35, 1, 5, 9, 41, 31, 5, 25, 45, 7, 87, 21,
791 11, 57, 17, 55, 21, 115, 59, 81, 27, 129, 47, 111, 33, 55, 5, 13, 27,
792 55, 93, 1, 57, 25
793 };
794
795 /* Verify that this host's size_t is not too wide for the above table. */
796
797 verify (enough_prime_offsets,
798 sizeof (size_t) * CHAR_BIT <= sizeof prime_offset);
799
800 /* Given a vector of two file_data objects, read the file associated
801 with each one, and build the table of equivalence classes.
802 Return nonzero if either file appears to be a binary file.
803 If PRETEND_BINARY is nonzero, pretend they are binary regardless. */
804
805 bool
read_files(struct file_data filevec[],bool pretend_binary)806 read_files (struct file_data filevec[], bool pretend_binary)
807 {
808 int i;
809 bool skip_test = text | pretend_binary;
810 bool appears_binary = pretend_binary | sip (&filevec[0], skip_test);
811
812 if (filevec[0].desc != filevec[1].desc)
813 appears_binary |= sip (&filevec[1], skip_test | appears_binary);
814 else
815 {
816 filevec[1].buffer = filevec[0].buffer;
817 filevec[1].bufsize = filevec[0].bufsize;
818 filevec[1].buffered = filevec[0].buffered;
819 }
820 if (appears_binary)
821 {
822 set_binary_mode (filevec[0].desc, 1);
823 set_binary_mode (filevec[1].desc, 1);
824 return 1;
825 }
826
827 find_identical_ends (filevec);
828
829 equivs_alloc = filevec[0].alloc_lines + filevec[1].alloc_lines + 1;
830 if (PTRDIFF_MAX / sizeof *equivs <= equivs_alloc)
831 xalloc_die ();
832 equivs = xmalloc (equivs_alloc * sizeof *equivs);
833 /* Equivalence class 0 is permanently safe for lines that were not
834 hashed. Real equivalence classes start at 1. */
835 equivs_index = 1;
836
837 /* Allocate (one plus) a prime number of hash buckets. Use a prime
838 number between 1/3 and 2/3 of the value of equiv_allocs,
839 approximately. */
840 for (i = 9; (size_t) 1 << i < equivs_alloc / 3; i++)
841 continue;
842 nbuckets = ((size_t) 1 << i) - prime_offset[i];
843 if (PTRDIFF_MAX / sizeof *buckets <= nbuckets)
844 xalloc_die ();
845 buckets = zalloc ((nbuckets + 1) * sizeof *buckets);
846 buckets++;
847
848 for (i = 0; i < 2; i++)
849 find_and_hash_each_line (&filevec[i]);
850
851 filevec[0].equiv_max = filevec[1].equiv_max = equivs_index;
852
853 free (equivs);
854 free (buckets - 1);
855
856 return 0;
857 }
858