xdiff/xutils.c

   1 /*
   2  *  LibXDiff by Davide Libenzi ( File Differential Library )
   3  *  Copyright (C) 2003  Davide Libenzi
   4  *
   5  *  This library is free software; you can redistribute it and/or
   6  *  modify it under the terms of the GNU Lesser General Public
   7  *  License as published by the Free Software Foundation; either
   8  *  version 2.1 of the License, or (at your option) any later version.
   9  *
  10  *  This library is distributed in the hope that it will be useful,
  11  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  *  Lesser General Public License for more details.
  14  *
  15  *  You should have received a copy of the GNU Lesser General Public
  16  *  License along with this library; if not, write to the Free Software
  17  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  18  *
  19  *  Davide Libenzi <davidel@xmailserver.org>
  20  *
  21  */
  22
  23 #include <limits.h>
  24 #include <assert.h>
  25 #include "xinclude.h"
  26
  27
  28
  29
  30 long xdl_bogosqrt(long n) {
  31         long i;
  32
  33         /*
  34          * Classical integer square root approximation using shifts.
  35          */
  36         for (i = 1; n > 0; n >>= 2)
  37                 i <<= 1;
  38
  39         return i;
  40 }
  41
  42
  43 int xdl_emit_diffrec(char const *rec, long size, char const *pre, long psize,
  44                      xdemitcb_t *ecb) {
  45         int i = 2;
  46         mmbuffer_t mb[3];
  47
  48         mb[0].ptr = (char *) pre;
  49         mb[0].size = psize;
  50         mb[1].ptr = (char *) rec;
  51         mb[1].size = size;
  52         if (size > 0 && rec[size - 1] != '\n') {
  53                 mb[2].ptr = (char *) "\n\\ No newline at end of file\n";
  54                 mb[2].size = strlen(mb[2].ptr);
  55                 i++;
  56         }
  57         if (ecb->outf(ecb->priv, mb, i) < 0) {
  58
  59                 return -1;
  60         }
  61
  62         return 0;
  63 }
  64
  65 void *xdl_mmfile_first(mmfile_t *mmf, long *size)
  66 {
  67         *size = mmf->size;
  68         return mmf->ptr;
  69 }
  70
  71
  72 long xdl_mmfile_size(mmfile_t *mmf)
  73 {
  74         return mmf->size;
  75 }
  76
  77
  78 int xdl_cha_init(chastore_t *cha, long isize, long icount) {
  79
  80         cha->head = cha->tail = NULL;
  81         cha->isize = isize;
  82         cha->nsize = icount * isize;
  83         cha->ancur = cha->sncur = NULL;
  84         cha->scurr = 0;
  85
  86         return 0;
  87 }
  88
  89
  90 void xdl_cha_free(chastore_t *cha) {
  91         chanode_t *cur, *tmp;
  92
  93         for (cur = cha->head; (tmp = cur) != NULL;) {
  94                 cur = cur->next;
  95                 xdl_free(tmp);
  96         }
  97 }
  98
  99
 100 void *xdl_cha_alloc(chastore_t *cha) {
 101         chanode_t *ancur;
 102         void *data;
 103
 104         if (!(ancur = cha->ancur) || ancur->icurr == cha->nsize) {
 105                 if (!(ancur = (chanode_t *) xdl_malloc(sizeof(chanode_t) + cha->nsize))) {
 106
 107                         return NULL;
 108                 }
 109                 ancur->icurr = 0;
 110                 ancur->next = NULL;
 111                 if (cha->tail)
 112                         cha->tail->next = ancur;
 113                 if (!cha->head)
 114                         cha->head = ancur;
 115                 cha->tail = ancur;
 116                 cha->ancur = ancur;
 117         }
 118
 119         data = (char *) ancur + sizeof(chanode_t) + ancur->icurr;
 120         ancur->icurr += cha->isize;
 121
 122         return data;
 123 }
 124
 125 long xdl_guess_lines(mmfile_t *mf, long sample) {
 126         long nl = 0, size, tsize = 0;
 127         char const *data, *cur, *top;
 128
 129         if ((cur = data = xdl_mmfile_first(mf, &size)) != NULL) {
 130                 for (top = data + size; nl < sample && cur < top; ) {
 131                         nl++;
 132                         if (!(cur = memchr(cur, '\n', top - cur)))
 133                                 cur = top;
 134                         else
 135                                 cur++;
 136                 }
 137                 tsize += (long) (cur - data);
 138         }
 139
 140         if (nl && tsize)
 141                 nl = xdl_mmfile_size(mf) / (tsize / nl);
 142
 143         return nl + 1;
 144 }
 145
 146 int xdl_blankline(const char *line, long size, long flags)
 147 {
 148         long i;
 149
 150         if (!(flags & XDF_WHITESPACE_FLAGS))
 151                 return (size <= 1);
 152
 153         for (i = 0; i < size && XDL_ISSPACE(line[i]); i++)
 154                 ;
 155
 156         return (i == size);
 157 }
 158
 159 int xdl_recmatch(const char *l1, long s1, const char *l2, long s2, long flags)
 160 {
 161         int i1, i2;
 162
 163         if (s1 == s2 && !memcmp(l1, l2, s1))
 164                 return 1;
 165         if (!(flags & XDF_WHITESPACE_FLAGS))
 166                 return 0;
 167
 168         i1 = 0;
 169         i2 = 0;
 170
 171         /*
 172          * -w matches everything that matches with -b, and -b in turn
 173          * matches everything that matches with --ignore-space-at-eol.
 174          *
 175          * Each flavor of ignoring needs different logic to skip whitespaces
 176          * while we have both sides to compare.
 177          */
 178         if (flags & XDF_IGNORE_WHITESPACE) {
 179                 goto skip_ws;
 180                 while (i1 < s1 && i2 < s2) {
 181                         if (l1[i1++] != l2[i2++])
 182                                 return 0;
 183                 skip_ws:
 184                         while (i1 < s1 && XDL_ISSPACE(l1[i1]))
 185                                 i1++;
 186                         while (i2 < s2 && XDL_ISSPACE(l2[i2]))
 187                                 i2++;
 188                 }
 189         } else if (flags & XDF_IGNORE_WHITESPACE_CHANGE) {
 190                 while (i1 < s1 && i2 < s2) {
 191                         if (XDL_ISSPACE(l1[i1]) && XDL_ISSPACE(l2[i2])) {
 192                                 /* Skip matching spaces and try again */
 193                                 while (i1 < s1 && XDL_ISSPACE(l1[i1]))
 194                                         i1++;
 195                                 while (i2 < s2 && XDL_ISSPACE(l2[i2]))
 196                                         i2++;
 197                                 continue;
 198                         }
 199                         if (l1[i1++] != l2[i2++])
 200                                 return 0;
 201                 }
 202         } else if (flags & XDF_IGNORE_WHITESPACE_AT_EOL) {
 203                 while (i1 < s1 && i2 < s2 && l1[i1++] == l2[i2++])
 204                         ; /* keep going */
 205         }
 206
 207         /*
 208          * After running out of one side, the remaining side must have
 209          * nothing but whitespace for the lines to match.  Note that
 210          * ignore-whitespace-at-eol case may break out of the loop
 211          * while there still are characters remaining on both lines.
 212          */
 213         if (i1 < s1) {
 214                 while (i1 < s1 && XDL_ISSPACE(l1[i1]))
 215                         i1++;
 216                 if (s1 != i1)
 217                         return 0;
 218         }
 219         if (i2 < s2) {
 220                 while (i2 < s2 && XDL_ISSPACE(l2[i2]))
 221                         i2++;
 222                 return (s2 == i2);
 223         }
 224         return 1;
 225 }
 226
 227 static unsigned long xdl_hash_record_with_whitespace(char const **data,
 228                 char const *top, long flags) {
 229         unsigned long ha = 5381;
 230         char const *ptr = *data;
 231
 232         for (; ptr < top && *ptr != '\n'; ptr++) {
 233                 if (XDL_ISSPACE(*ptr)) {
 234                         const char *ptr2 = ptr;
 235                         int at_eol;
 236                         while (ptr + 1 < top && XDL_ISSPACE(ptr[1])
 237                                         && ptr[1] != '\n')
 238                                 ptr++;
 239                         at_eol = (top <= ptr + 1 || ptr[1] == '\n');
 240                         if (flags & XDF_IGNORE_WHITESPACE)
 241                                 ; /* already handled */
 242                         else if (flags & XDF_IGNORE_WHITESPACE_CHANGE
 243                                  && !at_eol) {
 244                                 ha += (ha << 5);
 245                                 ha ^= (unsigned long) ' ';
 246                         }
 247                         else if (flags & XDF_IGNORE_WHITESPACE_AT_EOL
 248                                  && !at_eol) {
 249                                 while (ptr2 != ptr + 1) {
 250                                         ha += (ha << 5);
 251                                         ha ^= (unsigned long) *ptr2;
 252                                         ptr2++;
 253                                 }
 254                         }
 255                         continue;
 256                 }
 257                 ha += (ha << 5);
 258                 ha ^= (unsigned long) *ptr;
 259         }
 260         *data = ptr < top ? ptr + 1: ptr;
 261
 262         return ha;
 263 }
 264
 265 #ifdef XDL_FAST_HASH
 266
 267 #define REPEAT_BYTE(x)  ((~0ul / 0xff) * (x))
 268
 269 #define ONEBYTES        REPEAT_BYTE(0x01)
 270 #define NEWLINEBYTES    REPEAT_BYTE(0x0a)
 271 #define HIGHBITS        REPEAT_BYTE(0x80)
 272
 273 /* Return the high bit set in the first byte that is a zero */
 274 static inline unsigned long has_zero(unsigned long a)
 275 {
 276         return ((a - ONEBYTES) & ~a) & HIGHBITS;
 277 }
 278
 279 static inline long count_masked_bytes(unsigned long mask)
 280 {
 281         if (sizeof(long) == 8) {
 282                 /*
 283                  * Jan Achrenius on G+: microoptimized version of
 284                  * the simpler "(mask & ONEBYTES) * ONEBYTES >> 56"
 285                  * that works for the bytemasks without having to
 286                  * mask them first.
 287                  */
 288                 /*
 289                  * return mask * 0x0001020304050608 >> 56;
 290                  *
 291                  * Doing it like this avoids warnings on 32-bit machines.
 292                  */
 293                 long a = (REPEAT_BYTE(0x01) / 0xff + 1);
 294                 return mask * a >> (sizeof(long) * 7);
 295         } else {
 296                 /* Carl Chatfield / Jan Achrenius G+ version for 32-bit */
 297                 /* (000000 0000ff 00ffff ffffff) -> ( 1 1 2 3 ) */
 298                 long a = (0x0ff0001 + mask) >> 23;
 299                 /* Fix the 1 for 00 case */
 300                 return a & mask;
 301         }
 302 }
 303
 304 unsigned long xdl_hash_record(char const **data, char const *top, long flags)
 305 {
 306         unsigned long hash = 5381;
 307         unsigned long a = 0, mask = 0;
 308         char const *ptr = *data;
 309         char const *end = top - sizeof(unsigned long) + 1;
 310
 311         if (flags & XDF_WHITESPACE_FLAGS)
 312                 return xdl_hash_record_with_whitespace(data, top, flags);
 313
 314         ptr -= sizeof(unsigned long);
 315         do {
 316                 hash += hash << 5;
 317                 hash ^= a;
 318                 ptr += sizeof(unsigned long);
 319                 if (ptr >= end)
 320                         break;
 321                 a = *(unsigned long *)ptr;
 322                 /* Do we have any '\n' bytes in this word? */
 323                 mask = has_zero(a ^ NEWLINEBYTES);
 324         } while (!mask);
 325
 326         if (ptr >= end) {
 327                 /*
 328                  * There is only a partial word left at the end of the
 329                  * buffer. Because we may work with a memory mapping,
 330                  * we have to grab the rest byte by byte instead of
 331                  * blindly reading it.
 332                  *
 333                  * To avoid problems with masking in a signed value,
 334                  * we use an unsigned char here.
 335                  */
 336                 const char *p;
 337                 for (p = top - 1; p >= ptr; p--)
 338                         a = (a << 8) + *((const unsigned char *)p);
 339                 mask = has_zero(a ^ NEWLINEBYTES);
 340                 if (!mask)
 341                         /*
 342                          * No '\n' found in the partial word.  Make a
 343                          * mask that matches what we read.
 344                          */
 345                         mask = 1UL << (8 * (top - ptr) + 7);
 346         }
 347
 348         /* The mask *below* the first high bit set */
 349         mask = (mask - 1) & ~mask;
 350         mask >>= 7;
 351         hash += hash << 5;
 352         hash ^= a & mask;
 353
 354         /* Advance past the last (possibly partial) word */
 355         ptr += count_masked_bytes(mask);
 356
 357         if (ptr < top) {
 358                 assert(*ptr == '\n');
 359                 ptr++;
 360         }
 361
 362         *data = ptr;
 363
 364         return hash;
 365 }
 366
 367 #else /* XDL_FAST_HASH */
 368
 369 unsigned long xdl_hash_record(char const **data, char const *top, long flags) {
 370         unsigned long ha = 5381;
 371         char const *ptr = *data;
 372
 373         if (flags & XDF_WHITESPACE_FLAGS)
 374                 return xdl_hash_record_with_whitespace(data, top, flags);
 375
 376         for (; ptr < top && *ptr != '\n'; ptr++) {
 377                 ha += (ha << 5);
 378                 ha ^= (unsigned long) *ptr;
 379         }
 380         *data = ptr < top ? ptr + 1: ptr;
 381
 382         return ha;
 383 }
 384
 385 #endif /* XDL_FAST_HASH */
 386
 387 unsigned int xdl_hashbits(unsigned int size) {
 388         unsigned int val = 1, bits = 0;
 389
 390         for (; val < size && bits < CHAR_BIT * sizeof(unsigned int); val <<= 1, bits++);
 391         return bits ? bits: 1;
 392 }
 393
 394
 395 int xdl_num_out(char *out, long val) {
 396         char *ptr, *str = out;
 397         char buf[32];
 398
 399         ptr = buf + sizeof(buf) - 1;
 400         *ptr = '\0';
 401         if (val < 0) {
 402                 *--ptr = '-';
 403                 val = -val;
 404         }
 405         for (; val && ptr > buf; val /= 10)
 406                 *--ptr = "0123456789"[val % 10];
 407         if (*ptr)
 408                 for (; *ptr; ptr++, str++)
 409                         *str = *ptr;
 410         else
 411                 *str++ = '0';
 412         *str = '\0';
 413
 414         return str - out;
 415 }
 416
 417 int xdl_emit_hunk_hdr(long s1, long c1, long s2, long c2,
 418                       const char *func, long funclen, xdemitcb_t *ecb) {
 419         int nb = 0;
 420         mmbuffer_t mb;
 421         char buf[128];
 422
 423         memcpy(buf, "@@ -", 4);
 424         nb += 4;
 425
 426         nb += xdl_num_out(buf + nb, c1 ? s1: s1 - 1);
 427
 428         if (c1 != 1) {
 429                 memcpy(buf + nb, ",", 1);
 430                 nb += 1;
 431
 432                 nb += xdl_num_out(buf + nb, c1);
 433         }
 434
 435         memcpy(buf + nb, " +", 2);
 436         nb += 2;
 437
 438         nb += xdl_num_out(buf + nb, c2 ? s2: s2 - 1);
 439
 440         if (c2 != 1) {
 441                 memcpy(buf + nb, ",", 1);
 442                 nb += 1;
 443
 444                 nb += xdl_num_out(buf + nb, c2);
 445         }
 446
 447         memcpy(buf + nb, " @@", 3);
 448         nb += 3;
 449         if (func && funclen) {
 450                 buf[nb++] = ' ';
 451                 if (funclen > sizeof(buf) - nb - 1)
 452                         funclen = sizeof(buf) - nb - 1;
 453                 memcpy(buf + nb, func, funclen);
 454                 nb += funclen;
 455         }
 456         buf[nb++] = '\n';
 457
 458         mb.ptr = buf;
 459         mb.size = nb;
 460         if (ecb->outf(ecb->priv, &mb, 1) < 0)
 461                 return -1;
 462
 463         return 0;
 464 }
 465
 466 int xdl_fall_back_diff(xdfenv_t *diff_env, xpparam_t const *xpp,
 467                 int line1, int count1, int line2, int count2)
 468 {
 469         /*
 470          * This probably does not work outside Git, since
 471          * we have a very simple mmfile structure.
 472          *
 473          * Note: ideally, we would reuse the prepared environment, but
 474          * the libxdiff interface does not (yet) allow for diffing only
 475          * ranges of lines instead of the whole files.
 476          */
 477         mmfile_t subfile1, subfile2;
 478         xdfenv_t env;
 479
 480         subfile1.ptr = (char *)diff_env->xdf1.recs[line1 - 1]->ptr;
 481         subfile1.size = diff_env->xdf1.recs[line1 + count1 - 2]->ptr +
 482                 diff_env->xdf1.recs[line1 + count1 - 2]->size - subfile1.ptr;
 483         subfile2.ptr = (char *)diff_env->xdf2.recs[line2 - 1]->ptr;
 484         subfile2.size = diff_env->xdf2.recs[line2 + count2 - 2]->ptr +
 485                 diff_env->xdf2.recs[line2 + count2 - 2]->size - subfile2.ptr;
 486         if (xdl_do_diff(&subfile1, &subfile2, xpp, &env) < 0)
 487                 return -1;
 488
 489         memcpy(diff_env->xdf1.rchg + line1 - 1, env.xdf1.rchg, count1);
 490         memcpy(diff_env->xdf2.rchg + line2 - 1, env.xdf2.rchg, count2);
 491
 492         xdl_free_env(&env);
 493
 494         return 0;
 495 }