methods/rred.cc

   1 // Copyright (c) 2014 Anthony Towns
   2 //
   3 // This program is free software; you can redistribute it and/or modify
   4 // it under the terms of the GNU General Public License as published by
   5 // the Free Software Foundation; either version 2 of the License, or
   6 // (at your option) any later version.
   7
   8 #include <config.h>
   9
  10 #include <apt-pkg/fileutl.h>
  11 #include <apt-pkg/mmap.h>
  12 #include <apt-pkg/error.h>
  13 #include <apt-pkg/acquire-method.h>
  14 #include <apt-pkg/strutl.h>
  15 #include <apt-pkg/hashes.h>
  16 #include <apt-pkg/configuration.h>
  17
  18 #include <string>
  19 #include <list>
  20 #include <vector>
  21 #include <iterator>
  22
  23 #include <assert.h>
  24 #include <stdio.h>
  25 #include <stdlib.h>
  26 #include <string.h>
  27 #include <sys/stat.h>
  28 #include <utime.h>
  29
  30 #include <apti18n.h>
  31
  32 #define BLOCK_SIZE (512*1024)
  33
  34 class MemBlock {
  35    char *start;
  36    size_t size;
  37    char *free;
  38    struct MemBlock *next;
  39
  40    MemBlock(size_t size)
  41    {
  42       free = start = new char[size];
  43       size = size;
  44       next = NULL;
  45    }
  46
  47    size_t avail(void) { return size - (free - start); }
  48
  49    public:
  50
  51    MemBlock(void) {
  52       free = start = new char[BLOCK_SIZE];
  53       size = BLOCK_SIZE;
  54       next = NULL;
  55    }
  56
  57    ~MemBlock() {
  58       delete [] start;
  59       delete next;
  60    }
  61
  62    void clear(void) {
  63       free = start;
  64       if (next)
  65          next->clear();
  66    }
  67
  68    char *add_easy(char *src, size_t len, char *last)
  69    {
  70       if (last) {
  71          for (MemBlock *k = this; k; k = k->next) {
  72             if (k->free == last) {
  73                if (len <= k->avail()) {
  74                   char *n = k->add(src, len);
  75                   assert(last == n);
  76                   if (last == n)
  77                      return NULL;
  78                   return n;
  79                } else {
  80                   break;
  81                }
  82             } else if (last >= start && last < free) {
  83                break;
  84             }
  85          }
  86       }
  87       return add(src, len);
  88    }
  89
  90    char *add(char *src, size_t len) {
  91       if (len > avail()) {
  92          if (!next) {
  93             if (len > BLOCK_SIZE)  {
  94                next = new MemBlock(len);
  95             } else {
  96                next = new MemBlock;
  97             }
  98          }
  99          return next->add(src, len);
 100       }
 101       char *dst = free;
 102       free += len;
 103       memcpy(dst, src, len);
 104       return dst;
 105    }
 106 };
 107
 108 struct Change {
 109    /* Ordering:
 110     *
 111     *   1. write out <offset> lines unchanged
 112     *   2. skip <del_cnt> lines from source
 113     *   3. write out <add_cnt> lines (<add>/<add_len>)
 114     */
 115    size_t offset;
 116    size_t del_cnt;
 117    size_t add_cnt; /* lines */
 118    size_t add_len; /* bytes */
 119    char *add;
 120
 121    Change(int off)
 122    {
 123       offset = off;
 124       del_cnt = add_cnt = add_len = 0;
 125       add = NULL;
 126    }
 127
 128    /* actually, don't write <lines> lines from <add> */
 129    void skip_lines(size_t lines)
 130    {
 131       while (lines > 0) {
 132          char *s = (char*) memchr(add, '\n', add_len);
 133          assert(s != NULL);
 134          s++;
 135          add_len -= (s - add);
 136          add_cnt--;
 137          lines--;
 138          if (add_len == 0) {
 139             add = NULL;
 140             assert(add_cnt == 0);
 141             assert(lines == 0);
 142          } else {
 143             add = s;
 144             assert(add_cnt > 0);
 145          }
 146       }
 147    }
 148 };
 149
 150 class FileChanges {
 151    std::list<struct Change> changes;
 152    std::list<struct Change>::iterator where;
 153    size_t pos; // line number is as far left of iterator as possible
 154
 155    bool pos_is_okay(void)
 156    {
 157 #ifdef POSDEBUG
 158       size_t cpos = 0;
 159       std::list<struct Change>::iterator x;
 160       for (x = changes.begin(); x != where; x++) {
 161          assert(x != changes.end());
 162          cpos += x->offset + x->add_cnt;
 163       }
 164       return cpos == pos;
 165 #else
 166       return true;
 167 #endif
 168    }
 169
 170    public:
 171    FileChanges() {
 172       where = changes.end();
 173       pos = 0;
 174    }
 175
 176    std::list<struct Change>::iterator begin(void) { return changes.begin(); }
 177    std::list<struct Change>::iterator end(void) { return changes.end(); }
 178
 179    std::list<struct Change>::reverse_iterator rbegin(void) { return changes.rbegin(); }
 180    std::list<struct Change>::reverse_iterator rend(void) { return changes.rend(); }
 181
 182    void add_change(Change c) {
 183       assert(pos_is_okay());
 184       go_to_change_for(c.offset);
 185       assert(pos + where->offset == c.offset);
 186       if (c.del_cnt > 0)
 187          delete_lines(c.del_cnt);
 188       assert(pos + where->offset == c.offset);
 189       if (c.add_len > 0) {
 190          assert(pos_is_okay());
 191          if (where->add_len > 0)
 192             new_change();
 193          assert(where->add_len == 0 && where->add_cnt == 0);
 194
 195          where->add_len = c.add_len;
 196          where->add_cnt = c.add_cnt;
 197          where->add = c.add;
 198       }
 199       assert(pos_is_okay());
 200       merge();
 201       assert(pos_is_okay());
 202    }
 203
 204    private:
 205    void merge(void)
 206    {
 207       while (where->offset == 0 && where != changes.begin()) {
 208          left();
 209       }
 210       std::list<struct Change>::iterator next = where;
 211       next++;
 212
 213       while (next != changes.end() && next->offset == 0) {
 214          where->del_cnt += next->del_cnt;
 215          next->del_cnt = 0;
 216          if (next->add == NULL) {
 217             next = changes.erase(next);
 218          } else if (where->add == NULL) {
 219             where->add = next->add;
 220             where->add_len = next->add_len;
 221             where->add_cnt = next->add_cnt;
 222             next = changes.erase(next);
 223          } else {
 224             next++;
 225          }
 226       }
 227    }
 228
 229    void go_to_change_for(size_t line)
 230    {
 231       while(where != changes.end()) {
 232          if (line < pos) {
 233             left();
 234             continue;
 235          }
 236          if (pos + where->offset + where->add_cnt <= line) {
 237             right();
 238             continue;
 239          }
 240          // line is somewhere in this slot
 241          if (line < pos + where->offset) {
 242             break;
 243          } else if (line == pos + where->offset) {
 244             return;
 245          } else {
 246             split(line - pos);
 247             right();
 248             return;
 249          }
 250       }
 251       /* it goes before this patch */
 252       insert(line-pos);
 253    }
 254
 255    void new_change(void) { insert(where->offset); }
 256
 257    void insert(size_t offset)
 258    {
 259       assert(pos_is_okay());
 260       assert(where == changes.end() || offset <= where->offset);
 261       if (where != changes.end())
 262          where->offset -= offset;
 263       changes.insert(where, Change(offset));
 264       where--;
 265       assert(pos_is_okay());
 266    }
 267
 268    void split(size_t offset)
 269    {
 270       assert(pos_is_okay());
 271
 272       assert(where->offset < offset);
 273       assert(offset < where->offset + where->add_cnt);
 274
 275       size_t keep_lines = offset - where->offset;
 276
 277       Change before(*where);
 278
 279       where->del_cnt = 0;
 280       where->offset = 0;
 281       where->skip_lines(keep_lines);
 282
 283       before.add_cnt = keep_lines;
 284       before.add_len -= where->add_len;
 285
 286       changes.insert(where, before);
 287       where--;
 288       assert(pos_is_okay());
 289    }
 290
 291    size_t check_next_offset(size_t max)
 292    {
 293       assert(pos_is_okay());
 294       if (max > 0)
 295       {
 296          where++;
 297          if (where != changes.end()) {
 298             if (where->offset < max)
 299                max = where->offset;
 300          }
 301          where--;
 302          assert(pos_is_okay());
 303       }
 304       return max;
 305    }
 306
 307    void delete_lines(size_t cnt)
 308    {
 309       std::list<struct Change>::iterator x = where;
 310       assert(pos_is_okay());
 311       while (cnt > 0)
 312       {
 313          size_t del;
 314          del = x->add_cnt;
 315          if (del > cnt)
 316             del = cnt;
 317          x->skip_lines(del);
 318          cnt -= del;
 319
 320          x++;
 321          if (x == changes.end()) {
 322             del = cnt;
 323          } else {
 324             del = x->offset;
 325             if (del > cnt)
 326                del = cnt;
 327             x->offset -= del;
 328          }
 329          where->del_cnt += del;
 330          cnt -= del;
 331       }
 332       assert(pos_is_okay());
 333    }
 334
 335    void left(void) {
 336       assert(pos_is_okay());
 337       where--;
 338       pos -= where->offset + where->add_cnt;
 339       assert(pos_is_okay());
 340    }
 341
 342    void right(void) {
 343       assert(pos_is_okay());
 344       pos += where->offset + where->add_cnt;
 345       where++;
 346       assert(pos_is_okay());
 347    }
 348 };
 349
 350 class Patch {
 351    FileChanges filechanges;
 352    MemBlock add_text;
 353
 354    static bool retry_fwrite(char *b, size_t l, FILE *f, Hashes *hash)
 355    {
 356       size_t r = 1;
 357       while (r > 0 && l > 0)
 358       {
 359          r = fwrite(b, 1, l, f);
 360          if (hash)
 361             hash->Add((unsigned char*)b, r);
 362          l -= r;
 363          b += r;
 364       }
 365       return l == 0;
 366    }
 367
 368    static void dump_rest(FILE *o, FILE *i, Hashes *hash)
 369    {
 370       char buffer[BLOCK_SIZE];
 371       size_t l;
 372       while (0 < (l = fread(buffer, 1, sizeof(buffer), i))) {
 373          if (!retry_fwrite(buffer, l, o, hash))
 374             break;
 375       }
 376    }
 377
 378    static void dump_lines(FILE *o, FILE *i, size_t n, Hashes *hash)
 379    {
 380       char buffer[BLOCK_SIZE];
 381       size_t l;
 382       while (n > 0) {
 383          if (fgets(buffer, sizeof(buffer), i) == 0)
 384             buffer[0] = '\0';
 385          l = strlen(buffer);
 386          if (l == 0 || buffer[l-1] == '\n')
 387             n--;
 388          retry_fwrite(buffer, l, o, hash);
 389       }
 390    }
 391
 392    static void skip_lines(FILE *i, int n)
 393    {
 394       char buffer[BLOCK_SIZE];
 395       size_t l;
 396       while (n > 0) {
 397          if (fgets(buffer, sizeof(buffer), i) == 0)
 398             buffer[0] = '\0';
 399          l = strlen(buffer);
 400          if (l == 0 || buffer[l-1] == '\n')
 401             n--;
 402       }
 403    }
 404
 405    static void dump_mem(FILE *o, char *p, size_t s, Hashes *hash) {
 406       retry_fwrite(p, s, o, hash);
 407    }
 408
 409    public:
 410
 411    void read_diff(FileFd &f)
 412    {
 413       char buffer[BLOCK_SIZE];
 414       bool cmdwanted = true;
 415
 416       Change ch(0);
 417       while(f.ReadLine(buffer, sizeof(buffer)))
 418       {
 419          if (cmdwanted) {
 420             char *m, *c;
 421             size_t s, e;
 422             s = strtol(buffer, &m, 10);
 423             if (m == buffer) {
 424                s = e = ch.offset + ch.add_cnt;
 425                c = buffer;
 426             } else if (*m == ',') {
 427                m++;
 428                e = strtol(m, &c, 10);
 429             } else {
 430                e = s;
 431                c = m;
 432             }
 433             switch(*c) {
 434                case 'a':
 435                   cmdwanted = false;
 436                   ch.add = NULL;
 437                   ch.add_cnt = 0;
 438                   ch.add_len = 0;
 439                   ch.offset = s;
 440                   ch.del_cnt = 0;
 441                   break;
 442                case 'c':
 443                   cmdwanted = false;
 444                   ch.add = NULL;
 445                   ch.add_cnt = 0;
 446                   ch.add_len = 0;
 447                   ch.offset = s - 1;
 448                   ch.del_cnt = e - s + 1;
 449                   break;
 450                case 'd':
 451                   ch.offset = s - 1;
 452                   ch.del_cnt = e - s + 1;
 453                   ch.add = NULL;
 454                   ch.add_cnt = 0;
 455                   ch.add_len = 0;
 456                   filechanges.add_change(ch);
 457                   break;
 458             }
 459          } else { /* !cmdwanted */
 460             if (buffer[0] == '.' && buffer[1] == '\n') {
 461                cmdwanted = true;
 462                filechanges.add_change(ch);
 463             } else {
 464                char *last = NULL;
 465                char *add;
 466                size_t l;
 467                if (ch.add)
 468                   last = ch.add + ch.add_len;
 469                l = strlen(buffer);
 470                add = add_text.add_easy(buffer, l, last);
 471                if (!add) {
 472                   ch.add_len += l;
 473                   ch.add_cnt++;
 474                } else {
 475                   if (ch.add) {
 476                      filechanges.add_change(ch);
 477                      ch.del_cnt = 0;
 478                   }
 479                   ch.offset += ch.add_cnt;
 480                   ch.add = add;
 481                   ch.add_len = l;
 482                   ch.add_cnt = 1;
 483                }
 484             }
 485          }
 486       }
 487    }
 488
 489    void write_diff(FILE *f)
 490    {
 491       size_t line = 0;
 492       std::list<struct Change>::reverse_iterator ch;
 493       for (ch = filechanges.rbegin(); ch != filechanges.rend(); ch++) {
 494          line += ch->offset + ch->del_cnt;
 495       }
 496
 497       for (ch = filechanges.rbegin(); ch != filechanges.rend(); ch++) {
 498          std::list<struct Change>::reverse_iterator mg_i, mg_e = ch;
 499          while (ch->del_cnt == 0 && ch->offset == 0)
 500             ch++;
 501          line -= ch->del_cnt;
 502          if (ch->add_cnt > 0) {
 503             if (ch->del_cnt == 0) {
 504                fprintf(f, "%lua\n", line);
 505             } else if (ch->del_cnt == 1) {
 506                fprintf(f, "%luc\n", line+1);
 507             } else {
 508                fprintf(f, "%lu,%luc\n", line+1, line+ch->del_cnt);
 509             }
 510
 511             mg_i = ch;
 512             do {
 513                dump_mem(f, mg_i->add, mg_i->add_len, NULL);
 514             } while (mg_i-- != mg_e);
 515
 516             fprintf(f, ".\n");
 517          } else if (ch->del_cnt == 1) {
 518             fprintf(f, "%lud\n", line+1);
 519          } else if (ch->del_cnt > 1) {
 520             fprintf(f, "%lu,%lud\n", line+1, line+ch->del_cnt);
 521          }
 522          line -= ch->offset;
 523       }
 524    }
 525
 526    void apply_against_file(FILE *out, FILE *in, Hashes *hash = NULL)
 527    {
 528       std::list<struct Change>::iterator ch;
 529       for (ch = filechanges.begin(); ch != filechanges.end(); ch++) {
 530          dump_lines(out, in, ch->offset, hash);
 531          skip_lines(in, ch->del_cnt);
 532          dump_mem(out, ch->add, ch->add_len, hash);
 533       }
 534       dump_rest(out, in, hash);
 535    }
 536 };
 537
 538 class RredMethod : public pkgAcqMethod {
 539    private:
 540       bool Debug;
 541
 542    protected:
 543       virtual bool Fetch(FetchItem *Itm) {
 544          Debug = _config->FindB("Debug::pkgAcquire::RRed", false);
 545          URI Get = Itm->Uri;
 546          std::string Path = Get.Host + Get.Path; // rred:/path - no host
 547
 548          FetchResult Res;
 549          Res.Filename = Itm->DestFile;
 550          if (Itm->Uri.empty())
 551          {
 552             Path = Itm->DestFile;
 553             Itm->DestFile.append(".result");
 554          } else
 555             URIStart(Res);
 556
 557          std::vector<std::string> patchpaths;
 558          Patch patch;
 559
 560          if (FileExists(Path + ".ed") == true)
 561             patchpaths.push_back(Path + ".ed");
 562          else
 563          {
 564             _error->PushToStack();
 565             std::vector<std::string> patches = GetListOfFilesInDir(flNotFile(Path), "gz", true, false);
 566             _error->RevertToStack();
 567
 568             std::string const baseName = Path + ".ed.";
 569             for (std::vector<std::string>::const_iterator p = patches.begin();
 570                   p != patches.end(); ++p)
 571                if (p->compare(0, baseName.length(), baseName) == 0)
 572                   patchpaths.push_back(*p);
 573          }
 574
 575          std::string patch_name;
 576          for (std::vector<std::string>::iterator I = patchpaths.begin();
 577                I != patchpaths.end();
 578                I++)
 579          {
 580             patch_name = *I;
 581             if (Debug == true)
 582                std::clog << "Patching " << Path << " with " << patch_name
 583                   << std::endl;
 584
 585             FileFd p;
 586             // all patches are compressed, even if the name doesn't reflect it
 587             if (p.Open(patch_name, FileFd::ReadOnly, FileFd::Gzip) == false) {
 588                std::cerr << "Could not open patch file " << patch_name << std::endl;
 589                _error->DumpErrors(std::cerr);
 590                abort();
 591             }
 592             patch.read_diff(p);
 593             p.Close();
 594          }
 595
 596          if (Debug == true)
 597             std::clog << "Applying patches against " << Path
 598                << " and writing results to " << Itm->DestFile
 599                << std::endl;
 600
 601          FILE *inp = fopen(Path.c_str(), "r");
 602          FILE *out = fopen(Itm->DestFile.c_str(), "w");
 603
 604          Hashes hash;
 605
 606          patch.apply_against_file(out, inp, &hash);
 607
 608          fclose(out);
 609          fclose(inp);
 610
 611          if (Debug == true) {
 612             std::clog << "rred: finished file patching of " << Path  << "." << std::endl;
 613          }
 614
 615          struct stat bufbase, bufpatch;
 616          if (stat(Path.c_str(), &bufbase) != 0 ||
 617                stat(patch_name.c_str(), &bufpatch) != 0)
 618             return _error->Errno("stat", _("Failed to stat"));
 619
 620          struct utimbuf timebuf;
 621          timebuf.actime = bufbase.st_atime;
 622          timebuf.modtime = bufpatch.st_mtime;
 623          if (utime(Itm->DestFile.c_str(), &timebuf) != 0)
 624             return _error->Errno("utime", _("Failed to set modification time"));
 625
 626          if (stat(Itm->DestFile.c_str(), &bufbase) != 0)
 627             return _error->Errno("stat", _("Failed to stat"));
 628
 629          Res.LastModified = bufbase.st_mtime;
 630          Res.Size = bufbase.st_size;
 631          Res.TakeHashes(hash);
 632          URIDone(Res);
 633
 634          return true;
 635       }
 636
 637    public:
 638       RredMethod() : pkgAcqMethod("2.0",SingleInstance | SendConfig) {}
 639 };
 640
 641 int main(int argc, char **argv)
 642 {
 643    int i;
 644    bool just_diff = true;
 645    Patch patch;
 646
 647    if (argc <= 1) {
 648       RredMethod Mth;
 649       return Mth.Run();
 650    }
 651
 652    if (argc > 1 && strcmp(argv[1], "-f") == 0) {
 653       just_diff = false;
 654       i = 2;
 655    } else {
 656       i = 1;
 657    }
 658
 659    for (; i < argc; i++) {
 660       FileFd p;
 661       if (p.Open(argv[i], FileFd::ReadOnly) == false) {
 662          _error->DumpErrors(std::cerr);
 663          exit(1);
 664       }
 665       patch.read_diff(p);
 666    }
 667
 668    if (just_diff) {
 669       patch.write_diff(stdout);
 670    } else {
 671       FILE *out, *inp;
 672       out = stdout;
 673       inp = stdin;
 674
 675       patch.apply_against_file(out, inp);
 676    }
 677    return 0;
 678 }