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