gnu/dist/groff/src/libs/libbib/index.cpp

   1 /*      $NetBSD$        */
   2
   3 // -*- C++ -*-
   4 /* Copyright (C) 1989, 1990, 1991, 1992, 2001, 2004
   5    Free Software Foundation, Inc.
   6      Written by James Clark (jjc@jclark.com)
   7
   8 This file is part of groff.
   9
  10 groff is free software; you can redistribute it and/or modify it under
  11 the terms of the GNU General Public License as published by the Free
  12 Software Foundation; either version 2, or (at your option) any later
  13 version.
  14
  15 groff is distributed in the hope that it will be useful, but WITHOUT ANY
  16 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  17 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  18 for more details.
  19
  20 You should have received a copy of the GNU General Public License along
  21 with groff; see the file COPYING.  If not, write to the Free Software
  22 Foundation, 51 Franklin St - Fifth Floor, Boston, MA 02110-1301, USA. */
  23
  24 #include "lib.h"
  25
  26 #include <stdlib.h>
  27 #include <errno.h>
  28
  29 #include "posix.h"
  30 #include "cset.h"
  31 #include "cmap.h"
  32 #include "errarg.h"
  33 #include "error.h"
  34
  35 #include "refid.h"
  36 #include "search.h"
  37 #include "index.h"
  38 #include "defs.h"
  39
  40 #include "nonposix.h"
  41
  42 // Interface to mmap.
  43 extern "C" {
  44   void *mapread(int fd, int len);
  45   int unmap(void *, int len);
  46 }
  47
  48 #if 0
  49 const
  50 #endif
  51 int minus_one = -1;
  52
  53 int verify_flag = 0;
  54
  55 struct word_list;
  56
  57 class index_search_item : public search_item {
  58   search_item *out_of_date_files;
  59   index_header header;
  60   char *buffer;
  61   void *map_addr;
  62   int map_len;
  63   tag *tags;
  64   int *table;
  65   int *lists;
  66   char *pool;
  67   char *key_buffer;
  68   char *filename_buffer;
  69   int filename_buflen;
  70   char **common_words_table;
  71   int common_words_table_size;
  72   const char *ignore_fields;
  73   time_t mtime;
  74
  75   const char *do_verify();
  76   const int *search1(const char **pp, const char *end);
  77   const int *search(const char *ptr, int length, int **temp_listp);
  78   const char *munge_filename(const char *);
  79   void read_common_words_file();
  80   void add_out_of_date_file(int fd, const char *filename, int fid);
  81 public:
  82   index_search_item(const char *, int);
  83   ~index_search_item();
  84   int load(int fd);
  85   search_item_iterator *make_search_item_iterator(const char *);
  86   int verify();
  87   void check_files();
  88   int next_filename_id() const;
  89   friend class index_search_item_iterator;
  90 };
  91
  92 class index_search_item_iterator : public search_item_iterator {
  93   index_search_item *indx;
  94   search_item_iterator *out_of_date_files_iter;
  95   search_item *next_out_of_date_file;
  96   const int *found_list;
  97   int *temp_list;
  98   char *buf;
  99   int buflen;
 100   linear_searcher searcher;
 101   char *query;
 102   int get_tag(int tagno, const linear_searcher &, const char **, int *,
 103               reference_id *);
 104 public:
 105   index_search_item_iterator(index_search_item *, const char *);
 106   ~index_search_item_iterator();
 107   int next(const linear_searcher &, const char **, int *, reference_id *);
 108 };
 109
 110
 111 index_search_item::index_search_item(const char *filename, int fid)
 112 : search_item(filename, fid), out_of_date_files(0), buffer(0), map_addr(0),
 113   map_len(0), key_buffer(0), filename_buffer(0), filename_buflen(0),
 114   common_words_table(0)
 115 {
 116 }
 117
 118 index_search_item::~index_search_item()
 119 {
 120   if (buffer)
 121     free(buffer);
 122   if (map_addr) {
 123     if (unmap(map_addr, map_len) < 0)
 124       error("unmap: %1", strerror(errno));
 125   }
 126   while (out_of_date_files) {
 127     search_item *tem = out_of_date_files;
 128     out_of_date_files = out_of_date_files->next;
 129     delete tem;
 130   }
 131   a_delete filename_buffer;
 132   a_delete key_buffer;
 133   if (common_words_table) {
 134     for (int i = 0; i < common_words_table_size; i++)
 135       a_delete common_words_table[i];
 136     a_delete common_words_table;
 137   }
 138 }
 139
 140 class file_closer {
 141   int *fdp;
 142 public:
 143   file_closer(int &fd) : fdp(&fd) { }
 144   ~file_closer() { close(*fdp); }
 145 };
 146
 147 // Tell the compiler that a variable is intentionally unused.
 148 inline void unused(void *) { }
 149
 150 int index_search_item::load(int fd)
 151 {
 152   file_closer fd_closer(fd);    // close fd on return
 153   unused(&fd_closer);
 154   struct stat sb;
 155   if (fstat(fd, &sb) < 0) {
 156     error("can't fstat `%1': %2", name, strerror(errno));
 157     return 0;
 158   }
 159   if (!S_ISREG(sb.st_mode)) {
 160     error("`%1' is not a regular file", name);
 161     return 0;
 162   }
 163   mtime = sb.st_mtime;
 164   int size = int(sb.st_size);
 165   char *addr;
 166   map_addr = mapread(fd, size);
 167   if (map_addr) {
 168     addr = (char *)map_addr;
 169     map_len = size;
 170   }
 171   else {
 172     addr = buffer = (char *)malloc(size);
 173     if (buffer == 0) {
 174       error("can't allocate buffer for `%1'", name);
 175       return 0;
 176     }
 177     char *ptr = buffer;
 178     int bytes_to_read = size;
 179     while (bytes_to_read > 0) {
 180       int nread = read(fd, ptr, bytes_to_read);
 181       if (nread == 0) {
 182         error("unexpected EOF on `%1'", name);
 183         return 0;
 184       }
 185       if (nread < 0) {
 186         error("read error on `%1': %2", name, strerror(errno));
 187         return 0;
 188       }
 189       bytes_to_read -= nread;
 190       ptr += nread;
 191     }
 192   }
 193   header = *(index_header *)addr;
 194   if (header.magic != INDEX_MAGIC) {
 195     error("`%1' is not an index file: wrong magic number", name);
 196     return 0;
 197   }
 198   if (header.version != INDEX_VERSION) {
 199     error("version number in `%1' is wrong: was %2, should be %3",
 200           name, header.version, INDEX_VERSION);
 201     return 0;
 202   }
 203   int sz = (header.tags_size * sizeof(tag)
 204             + header.lists_size * sizeof(int)
 205             + header.table_size * sizeof(int)
 206             + header.strings_size
 207             + sizeof(header));
 208   if (sz != size) {
 209     error("size of `%1' is wrong: was %2, should be %3",
 210           name, size, sz);
 211     return 0;
 212   }
 213   tags = (tag *)(addr + sizeof(header));
 214   lists = (int *)(tags + header.tags_size);
 215   table = (int *)(lists + header.lists_size);
 216   pool = (char *)(table + header.table_size);
 217   ignore_fields = strchr(strchr(pool, '\0') + 1, '\0') + 1;
 218   key_buffer = new char[header.truncate];
 219   read_common_words_file();
 220   return 1;
 221 }
 222
 223 const char *index_search_item::do_verify()
 224 {
 225   if (tags == 0)
 226     return "not loaded";
 227   if (lists[header.lists_size - 1] >= 0)
 228     return "last list element not negative";
 229   int i;
 230   for (i = 0; i < header.table_size; i++) {
 231     int li = table[i];
 232     if (li >= header.lists_size)
 233       return "bad list index";
 234     if (li >= 0) {
 235       for (int *ptr = lists + li; *ptr >= 0; ptr++) {
 236         if (*ptr >= header.tags_size)
 237           return "bad tag index";
 238         if (*ptr >= ptr[1] && ptr[1] >= 0)
 239           return "list not ordered";
 240       }
 241     }
 242   }
 243   for (i = 0; i < header.tags_size; i++) {
 244     if (tags[i].filename_index >= header.strings_size)
 245       return "bad index in tags";
 246     if (tags[i].length < 0)
 247       return "bad length in tags";
 248     if (tags[i].start < 0)
 249       return "bad start in tags";
 250   }
 251   if (pool[header.strings_size - 1] != '\0')
 252     return "last character in pool not nul";
 253   return 0;
 254 }
 255
 256 int index_search_item::verify()
 257 {
 258   const char *reason = do_verify();
 259   if (!reason)
 260     return 1;
 261   error("`%1' is bad: %2", name, reason);
 262   return 0;
 263 }
 264
 265 int index_search_item::next_filename_id() const
 266 {
 267   return filename_id + header.strings_size + 1;
 268 }
 269
 270 search_item_iterator *index_search_item::make_search_item_iterator(
 271   const char *query)
 272 {
 273   return new index_search_item_iterator(this, query);
 274 }
 275
 276 search_item *make_index_search_item(const char *filename, int fid)
 277 {
 278   char *index_filename = new char[strlen(filename) + sizeof(INDEX_SUFFIX)];
 279   strcpy(index_filename, filename);
 280   strcat(index_filename, INDEX_SUFFIX);
 281   int fd = open(index_filename, O_RDONLY | O_BINARY);
 282   if (fd < 0)
 283     return 0;
 284   index_search_item *item = new index_search_item(index_filename, fid);
 285   a_delete index_filename;
 286   if (!item->load(fd)) {
 287     close(fd);
 288     delete item;
 289     return 0;
 290   }
 291   else if (verify_flag && !item->verify()) {
 292     delete item;
 293     return 0;
 294   }
 295   else {
 296     item->check_files();
 297     return item;
 298   }
 299 }
 300
 301
 302 index_search_item_iterator::index_search_item_iterator(index_search_item *ind,
 303                                                        const char *q)
 304 : indx(ind), out_of_date_files_iter(0), next_out_of_date_file(0), temp_list(0),
 305   buf(0), buflen(0),
 306   searcher(q, strlen(q), ind->ignore_fields, ind->header.truncate),
 307   query(strsave(q))
 308 {
 309   found_list = indx->search(q, strlen(q), &temp_list);
 310   if (!found_list) {
 311     found_list = &minus_one;
 312     warning("all keys would have been discarded in constructing index `%1'",
 313             indx->name);
 314   }
 315 }
 316
 317 index_search_item_iterator::~index_search_item_iterator()
 318 {
 319   a_delete temp_list;
 320   a_delete buf;
 321   a_delete query;
 322   delete out_of_date_files_iter;
 323 }
 324
 325 int index_search_item_iterator::next(const linear_searcher &,
 326                                      const char **pp, int *lenp,
 327                                      reference_id *ridp)
 328 {
 329   if (found_list) {
 330     for (;;) {
 331       int tagno = *found_list;
 332       if (tagno == -1)
 333         break;
 334       found_list++;
 335       if (get_tag(tagno, searcher, pp, lenp, ridp))
 336         return 1;
 337     }
 338     found_list = 0;
 339     next_out_of_date_file = indx->out_of_date_files;
 340   }
 341   while (next_out_of_date_file) {
 342     if (out_of_date_files_iter == 0)
 343       out_of_date_files_iter
 344         = next_out_of_date_file->make_search_item_iterator(query);
 345     if (out_of_date_files_iter->next(searcher, pp, lenp, ridp))
 346       return 1;
 347     delete out_of_date_files_iter;
 348     out_of_date_files_iter = 0;
 349     next_out_of_date_file = next_out_of_date_file->next;
 350   }
 351   return 0;
 352 }
 353
 354 int index_search_item_iterator::get_tag(int tagno,
 355                                         const linear_searcher &searchr,
 356                                         const char **pp, int *lenp,
 357                                         reference_id *ridp)
 358 {
 359   if (tagno < 0 || tagno >= indx->header.tags_size) {
 360     error("bad tag number");
 361     return 0;
 362   }
 363   tag *tp = indx->tags + tagno;
 364   const char *filename = indx->munge_filename(indx->pool + tp->filename_index);
 365   int fd = open(filename, O_RDONLY | O_BINARY);
 366   if (fd < 0) {
 367     error("can't open `%1': %2", filename, strerror(errno));
 368     return 0;
 369   }
 370   struct stat sb;
 371   if (fstat(fd, &sb) < 0) {
 372     error("can't fstat: %1", strerror(errno));
 373     close(fd);
 374     return 0;
 375   }
 376   time_t mtime = sb.st_mtime;
 377   if (mtime > indx->mtime) {
 378     indx->add_out_of_date_file(fd, filename,
 379                                indx->filename_id + tp->filename_index);
 380     return 0;
 381   }
 382   int res = 0;
 383   FILE *fp = fdopen(fd, FOPEN_RB);
 384   if (!fp) {
 385     error("fdopen failed");
 386     close(fd);
 387     return 0;
 388   }
 389   if (tp->start != 0 && fseek(fp, long(tp->start), 0) < 0)
 390     error("can't seek on `%1': %2", filename, strerror(errno));
 391   else {
 392     int length = tp->length;
 393     int err = 0;
 394     if (length == 0) {
 395       if (fstat(fileno(fp), &sb) < 0) {
 396         error("can't stat `%1': %2", filename, strerror(errno));
 397         err = 1;
 398       }
 399       else if (!S_ISREG(sb.st_mode)) {
 400         error("`%1' is not a regular file", filename);
 401         err = 1;
 402       }
 403       else
 404         length = int(sb.st_size);
 405     }
 406     if (!err) {
 407       if (length + 2 > buflen) {
 408         a_delete buf;
 409         buflen = length + 2;
 410         buf = new char[buflen];
 411       }
 412       if (fread(buf + 1, 1, length, fp) != (size_t)length)
 413         error("fread on `%1' failed: %2", filename, strerror(errno));
 414       else {
 415         buf[0] = '\n';
 416         // Remove the CR characters from CRLF pairs.
 417         int sidx = 1, didx = 1;
 418         for ( ; sidx < length + 1; sidx++, didx++)
 419           {
 420             if (buf[sidx] == '\r')
 421               {
 422                 if (buf[++sidx] != '\n')
 423                   buf[didx++] = '\r';
 424                 else
 425                   length--;
 426               }
 427             if (sidx != didx)
 428               buf[didx] = buf[sidx];
 429           }
 430         buf[length + 1] = '\n';
 431         res = searchr.search(buf + 1, buf + 2 + length, pp, lenp);
 432         if (res && ridp)
 433           *ridp = reference_id(indx->filename_id + tp->filename_index,
 434                                tp->start);
 435       }
 436     }
 437   }
 438   fclose(fp);
 439   return res;
 440 }
 441
 442 const char *index_search_item::munge_filename(const char *filename)
 443 {
 444   if (IS_ABSOLUTE(filename))
 445     return filename;
 446   const char *cwd = pool;
 447   int need_slash = (cwd[0] != 0
 448                     && strchr(DIR_SEPS, strchr(cwd, '\0')[-1]) == 0);
 449   int len = strlen(cwd) + strlen(filename) + need_slash + 1;
 450   if (len > filename_buflen) {
 451     a_delete filename_buffer;
 452     filename_buflen = len;
 453     filename_buffer = new char[len];
 454   }
 455   strcpy(filename_buffer, cwd);
 456   if (need_slash)
 457     strcat(filename_buffer, "/");
 458   strcat(filename_buffer, filename);
 459   return filename_buffer;
 460 }
 461
 462 const int *index_search_item::search1(const char **pp, const char *end)
 463 {
 464   while (*pp < end && !csalnum(**pp))
 465     *pp += 1;
 466   if (*pp >= end)
 467     return 0;
 468   const char *start = *pp;
 469   while (*pp < end && csalnum(**pp))
 470     *pp += 1;
 471   int len = *pp - start;
 472   if (len < header.shortest)
 473     return 0;
 474   if (len > header.truncate)
 475     len = header.truncate;
 476   int is_number = 1;
 477   for (int i = 0; i < len; i++)
 478     if (csdigit(start[i]))
 479       key_buffer[i] = start[i];
 480     else {
 481       key_buffer[i] = cmlower(start[i]);
 482       is_number = 0;
 483     }
 484   if (is_number && !(len == 4 && start[0] == '1' && start[1] == '9'))
 485     return 0;
 486   unsigned hc = hash(key_buffer, len);
 487   if (common_words_table) {
 488     for (int h = hc % common_words_table_size;
 489          common_words_table[h];
 490          --h) {
 491       if (strlen(common_words_table[h]) == (size_t)len
 492           && memcmp(common_words_table[h], key_buffer, len) == 0)
 493         return 0;
 494       if (h == 0)
 495         h = common_words_table_size;
 496     }
 497   }
 498   int li = table[int(hc % header.table_size)];
 499   return li < 0 ? &minus_one : lists + li;
 500 }
 501
 502 static void merge(int *result, const int *s1, const int *s2)
 503 {
 504   for (; *s1 >= 0; s1++) {
 505     while (*s2 >= 0 && *s2 < *s1)
 506       s2++;
 507     if (*s2 == *s1)
 508       *result++ = *s2;
 509   }
 510   *result++ = -1;
 511 }
 512
 513 const int *index_search_item::search(const char *ptr, int length,
 514                                      int **temp_listp)
 515 {
 516   const char *end = ptr + length;
 517   if (*temp_listp) {
 518     a_delete *temp_listp;
 519     *temp_listp = 0;
 520   }
 521   const int *first_list = 0;
 522   while (ptr < end && (first_list = search1(&ptr, end)) == 0)
 523     ;
 524   if (!first_list)
 525     return 0;
 526   if (*first_list < 0)
 527     return first_list;
 528   const int *second_list = 0;
 529   while (ptr < end && (second_list = search1(&ptr, end)) == 0)
 530     ;
 531   if (!second_list)
 532     return first_list;
 533   if (*second_list < 0)
 534     return second_list;
 535   const int *p;
 536   for (p = first_list; *p >= 0; p++)
 537     ;
 538   int len = p - first_list;
 539   for (p = second_list; *p >= 0; p++)
 540     ;
 541   if (p - second_list < len)
 542     len = p - second_list;
 543   int *matches = new int[len + 1];
 544   merge(matches, first_list, second_list);
 545   while (ptr < end) {
 546     const int *list = search1(&ptr, end);
 547     if (list != 0) {
 548       if (*list < 0) {
 549         a_delete matches;
 550         return list;
 551       }
 552       merge(matches, matches, list);
 553       if (*matches < 0) {
 554         a_delete matches;
 555         return &minus_one;
 556       }
 557     }
 558   }
 559   *temp_listp = matches;
 560   return matches;
 561 }
 562
 563 void index_search_item::read_common_words_file()
 564 {
 565   if (header.common <= 0)
 566     return;
 567   const char *common_words_file = munge_filename(strchr(pool, '\0') + 1);
 568   errno = 0;
 569   FILE *fp = fopen(common_words_file, "r");
 570   if (!fp) {
 571     error("can't open `%1': %2", common_words_file, strerror(errno));
 572     return;
 573   }
 574   common_words_table_size = 2*header.common + 1;
 575   while (!is_prime(common_words_table_size))
 576     common_words_table_size++;
 577   common_words_table = new char *[common_words_table_size];
 578   for (int i = 0; i < common_words_table_size; i++)
 579     common_words_table[i] = 0;
 580   int count = 0;
 581   int key_len = 0;
 582   for (;;) {
 583     int c = getc(fp);
 584     while (c != EOF && !csalnum(c))
 585       c = getc(fp);
 586     if (c == EOF)
 587       break;
 588     do {
 589       if (key_len < header.truncate)
 590         key_buffer[key_len++] = cmlower(c);
 591       c = getc(fp);
 592     } while (c != EOF && csalnum(c));
 593     if (key_len >= header.shortest) {
 594       int h = hash(key_buffer, key_len) % common_words_table_size;
 595       while (common_words_table[h]) {
 596         if (h == 0)
 597           h = common_words_table_size;
 598         --h;
 599       }
 600       common_words_table[h] = new char[key_len + 1];
 601       memcpy(common_words_table[h], key_buffer, key_len);
 602       common_words_table[h][key_len] = '\0';
 603     }
 604     if (++count >= header.common)
 605       break;
 606     key_len = 0;
 607     if (c == EOF)
 608       break;
 609   }
 610   fclose(fp);
 611 }
 612
 613 void index_search_item::add_out_of_date_file(int fd, const char *filename,
 614                                              int fid)
 615 {
 616   search_item **pp;
 617   for (pp = &out_of_date_files; *pp; pp = &(*pp)->next)
 618     if ((*pp)->is_named(filename))
 619       return;
 620   *pp = make_linear_search_item(fd, filename, fid);
 621   warning("`%1' modified since `%2' created", filename, name);
 622 }
 623
 624 void index_search_item::check_files()
 625 {
 626   const char *pool_end = pool + header.strings_size;
 627   for (const char *ptr = strchr(ignore_fields, '\0') + 1;
 628        ptr < pool_end;
 629        ptr = strchr(ptr, '\0') + 1) {
 630     const char *path = munge_filename(ptr);
 631     struct stat sb;
 632     if (stat(path, &sb) < 0)
 633       error("can't stat `%1': %2", path, strerror(errno));
 634     else if (sb.st_mtime > mtime) {
 635       int fd = open(path, O_RDONLY | O_BINARY);
 636       if (fd < 0)
 637         error("can't open `%1': %2", path, strerror(errno));
 638       else
 639         add_out_of_date_file(fd, path, filename_id + (ptr - pool));
 640     }
 641   }
 642 }