index/extract.c

   1 /* $Id: extract.c,v 1.252 2007-03-14 11:48:32 adam Exp $
   2    Copyright (C) 1995-2007
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 #include <stdio.h>
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #ifdef WIN32
  27 #include <io.h>
  28 #endif
  29 #if HAVE_UNISTD_H
  30 #include <unistd.h>
  31 #endif
  32 #include <fcntl.h>
  33
  34 #include "index.h"
  35 #include "orddict.h"
  36 #include <direntz.h>
  37 #include <charmap.h>
  38
  39 static int log_level_extract = 0;
  40 static int log_level_details = 0;
  41 static int log_level_initialized = 0;
  42
  43 static void zebra_init_log_level(void)
  44 {
  45     if (!log_level_initialized)
  46     {
  47         log_level_initialized = 1;
  48
  49         log_level_extract = yaz_log_module_level("extract");
  50         log_level_details = yaz_log_module_level("indexdetails");
  51     }
  52 }
  53
  54 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
  55                                       int cmd, zebra_rec_keys_t reckeys,
  56                                       zint staticrank);
  57 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  58                                     int cmd, zebra_rec_keys_t skp);
  59 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
  60 static void extract_token_add (RecWord *p);
  61
  62 static void check_log_limit(ZebraHandle zh)
  63 {
  64     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  65     {
  66         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  67                 zh->m_file_verbose_limit);
  68     }
  69 }
  70
  71 static void logRecord (ZebraHandle zh)
  72 {
  73     check_log_limit(zh);
  74     ++zh->records_processed;
  75     if (!(zh->records_processed % 1000))
  76     {
  77         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  78                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  79                 zh->records_processed, zh->records_inserted,
  80                 zh->records_updated, zh->records_deleted);
  81     }
  82 }
  83
  84 static void extract_add_index_string (RecWord *p,
  85                                       zinfo_index_category_t cat,
  86                                       const char *str, int length);
  87
  88 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
  89
  90 static void extract_init(struct recExtractCtrl *p, RecWord *w)
  91 {
  92     w->seqno = 1;
  93     w->index_name = "any";
  94     w->index_type = 'w';
  95     w->extractCtrl = p;
  96     w->record_id = 0;
  97     w->section_id = 0;
  98     w->segment = 0;
  99 }
 100
 101 static void searchRecordKey(ZebraHandle zh,
 102                             zebra_rec_keys_t reckeys,
 103                             const char *index_name,
 104                             const char **ws, int ws_length)
 105 {
 106     int i;
 107     int ch = -1;
 108     zinfo_index_category_t cat = zinfo_index_category_index;
 109
 110     for (i = 0; i<ws_length; i++)
 111         ws[i] = NULL;
 112
 113     if (ch < 0)
 114         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
 115     if (ch < 0)
 116         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
 117     if (ch < 0)
 118         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
 119
 120     if (ch < 0)
 121         return ;
 122
 123     if (zebra_rec_keys_rewind(reckeys))
 124     {
 125         zint startSeq = -1;
 126         const char *str;
 127         size_t slen;
 128         struct it_key key;
 129         zint seqno;
 130         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 131         {
 132             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 133
 134             seqno = key.mem[key.len-1];
 135
 136             if (key.mem[0] == ch)
 137             {
 138                 zint woff;
 139
 140                 if (startSeq == -1)
 141                     startSeq = seqno;
 142                 woff = seqno - startSeq;
 143                 if (woff >= 0 && woff < ws_length)
 144                     ws[woff] = str;
 145             }
 146         }
 147     }
 148 }
 149
 150 #define FILE_MATCH_BLANK "\t "
 151
 152 static char *get_match_from_spec(ZebraHandle zh,
 153                           zebra_rec_keys_t reckeys,
 154                           const char *fname, const char *spec)
 155 {
 156     static char dstBuf[2048];      /* static here ??? */
 157     char *dst = dstBuf;
 158     const char *s = spec;
 159
 160     while (1)
 161     {
 162         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 163             ;
 164         if (!*s)
 165             break;
 166         if (*s == '(')
 167         {
 168             const char *ws[32];
 169             char attset_str[64], attname_str[64];
 170             int i;
 171             int first = 1;
 172
 173             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 174                 ;
 175             for (i = 0; *s && *s != ',' && *s != ')' &&
 176                      !strchr(FILE_MATCH_BLANK, *s); s++)
 177                 if (i+1 < sizeof(attset_str))
 178                     attset_str[i++] = *s;
 179             attset_str[i] = '\0';
 180
 181             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 182                 ;
 183             if (*s != ',')
 184                 strcpy(attname_str, attset_str);
 185             else
 186             {
 187                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 188                     ;
 189                 for (i = 0; *s && *s != ')' &&
 190                          !strchr(FILE_MATCH_BLANK, *s); s++)
 191                     if (i+1 < sizeof(attname_str))
 192                         attname_str[i++] = *s;
 193                 attname_str[i] = '\0';
 194             }
 195
 196             searchRecordKey (zh, reckeys, attname_str, ws, 32);
 197
 198             if (*s != ')')
 199             {
 200                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
 201                       spec, zh->m_group ? zh->m_group : "none");
 202                 return NULL;
 203             }
 204             s++;
 205
 206             for (i = 0; i<32; i++)
 207                 if (ws[i])
 208                 {
 209                     if (first)
 210                     {
 211                         *dst++ = ' ';
 212                         first = 0;
 213                     }
 214                     strcpy (dst, ws[i]);
 215                     dst += strlen(ws[i]);
 216                 }
 217             if (first)
 218             {
 219                 yaz_log (YLOG_WARN, "Record didn't contain match"
 220                       " fields in (%s,%s)", attset_str, attname_str);
 221                 return NULL;
 222             }
 223         }
 224         else if (*s == '$')
 225         {
 226             int spec_len;
 227             char special[64];
 228             const char *spec_src = NULL;
 229             const char *s1 = ++s;
 230             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 231                 s1++;
 232
 233             spec_len = s1 - s;
 234             if (spec_len > sizeof(special)-1)
 235                 spec_len = sizeof(special)-1;
 236             memcpy (special, s, spec_len);
 237             special[spec_len] = '\0';
 238             s = s1;
 239
 240             if (!strcmp (special, "group"))
 241                 spec_src = zh->m_group;
 242             else if (!strcmp (special, "database"))
 243                 spec_src = zh->basenames[0];
 244             else if (!strcmp (special, "filename")) {
 245                 spec_src = fname;
 246             }
 247             else if (!strcmp (special, "type"))
 248                 spec_src = zh->m_record_type;
 249             else
 250                 spec_src = NULL;
 251             if (spec_src)
 252             {
 253                 strcpy (dst, spec_src);
 254                 dst += strlen (spec_src);
 255             }
 256         }
 257         else if (*s == '\"' || *s == '\'')
 258         {
 259             int stopMarker = *s++;
 260             char tmpString[64];
 261             int i = 0;
 262
 263             while (*s && *s != stopMarker)
 264             {
 265                 if (i+1 < sizeof(tmpString))
 266                     tmpString[i++] = *s++;
 267             }
 268             if (*s)
 269                 s++;
 270             tmpString[i] = '\0';
 271             strcpy (dst, tmpString);
 272             dst += strlen (tmpString);
 273         }
 274         else
 275         {
 276             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
 277                   spec, zh->m_group ? zh->m_group : "none");
 278             return NULL;
 279         }
 280         *dst++ = 1;
 281     }
 282     if (dst == dstBuf)
 283     {
 284         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
 285               fname, zh->m_group ? zh->m_group : "none");
 286         return NULL;
 287     }
 288     *dst = '\0';
 289     return dstBuf;
 290 }
 291
 292 struct recordLogInfo {
 293     const char *fname;
 294     int recordOffset;
 295     struct recordGroup *rGroup;
 296 };
 297
 298 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
 299 {
 300     int i;
 301     for (i = 0; i<256; i++)
 302     {
 303         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
 304             ctrl->seqno[i] = 1;
 305         else
 306             ctrl->seqno[i] = 0;
 307     }
 308     ctrl->flagShowRecords = !zh->m_flag_rw;
 309 }
 310
 311 static void all_matches_add(struct recExtractCtrl *ctrl)
 312 {
 313     RecWord word;
 314     extract_init(ctrl, &word);
 315     word.index_name = "_ALLRECORDS";
 316     word.index_type = 'w';
 317     word.seqno = 1;
 318     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
 319                               "", 0);
 320 }
 321
 322 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 323                                        struct ZebraRecStream *stream,
 324                                        enum zebra_recctrl_action_t action,
 325                                        int test_mode,
 326                                        const char *recordType,
 327                                        zint *sysno,
 328                                        const char *match_criteria,
 329                                        const char *fname,
 330                                        RecType recType,
 331                                        void *recTypeClientData);
 332
 333
 334 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 335                              int deleteFlag)
 336 {
 337     ZEBRA_RES r = ZEBRA_OK;
 338     int i, fd;
 339     char gprefix[128];
 340     char ext[128];
 341     char ext_res[128];
 342     struct file_read_info *fi = 0;
 343     const char *original_record_type = 0;
 344     RecType recType;
 345     void *recTypeClientData;
 346     struct ZebraRecStream stream, *streamp;
 347
 348     zebra_init_log_level();
 349
 350     if (!zh->m_group || !*zh->m_group)
 351         *gprefix = '\0';
 352     else
 353         sprintf (gprefix, "%s.", zh->m_group);
 354
 355     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 356
 357     /* determine file extension */
 358     *ext = '\0';
 359     for (i = strlen(fname); --i >= 0; )
 360         if (fname[i] == '/')
 361             break;
 362         else if (fname[i] == '.')
 363         {
 364             strcpy (ext, fname+i+1);
 365             break;
 366         }
 367     /* determine file type - depending on extension */
 368     original_record_type = zh->m_record_type;
 369     if (!zh->m_record_type)
 370     {
 371         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
 372         zh->m_record_type = res_get (zh->res, ext_res);
 373     }
 374     if (!zh->m_record_type)
 375     {
 376         check_log_limit(zh);
 377         if (zh->records_processed + zh->records_skipped
 378             < zh->m_file_verbose_limit)
 379             yaz_log (YLOG_LOG, "? %s", fname);
 380         zh->records_skipped++;
 381         return 0;
 382     }
 383     /* determine match criteria */
 384     if (!zh->m_record_id)
 385     {
 386         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
 387         zh->m_record_id = res_get (zh->res, ext_res);
 388     }
 389
 390     if (!(recType =
 391           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
 392                           &recTypeClientData)))
 393     {
 394         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 395         return ZEBRA_FAIL;
 396     }
 397
 398     switch(recType->version)
 399     {
 400     case 0:
 401         break;
 402     default:
 403         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 404     }
 405     if (sysno && deleteFlag)
 406     {
 407         streamp = 0;
 408         fi = 0;
 409     }
 410     else
 411     {
 412         char full_rep[1024];
 413
 414         if (zh->path_reg && !yaz_is_abspath (fname))
 415         {
 416             strcpy (full_rep, zh->path_reg);
 417             strcat (full_rep, "/");
 418             strcat (full_rep, fname);
 419         }
 420         else
 421             strcpy (full_rep, fname);
 422
 423         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
 424         {
 425             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 426             zh->m_record_type = original_record_type;
 427             return ZEBRA_FAIL;
 428         }
 429         streamp = &stream;
 430         zebra_create_stream_fd(streamp, fd, 0);
 431     }
 432     r = zebra_extract_records_stream(zh, streamp,
 433                                      deleteFlag ?
 434                                      action_delete : action_update,
 435                                      0, /* tst_mode */
 436                                      zh->m_record_type,
 437                                      sysno,
 438                                      0, /*match_criteria */
 439                                      fname,
 440                                      recType, recTypeClientData);
 441     if (streamp)
 442         stream.destroy(streamp);
 443     zh->m_record_type = original_record_type;
 444     return r;
 445 }
 446
 447 /*
 448   If sysno is provided, then it's used to identify the reocord.
 449   If not, and match_criteria is provided, then sysno is guessed
 450   If not, and a record is provided, then sysno is got from there
 451
 452  */
 453
 454 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 455                                       const char *buf, size_t buf_size,
 456                                       enum zebra_recctrl_action_t action,
 457                                       int test_mode,
 458                                       const char *recordType,
 459                                       zint *sysno,
 460                                       const char *match_criteria,
 461                                       const char *fname)
 462 {
 463     struct ZebraRecStream stream;
 464     ZEBRA_RES res;
 465     void *clientData;
 466     RecType recType = 0;
 467
 468     if (recordType && *recordType)
 469     {
 470         yaz_log(log_level_extract,
 471                 "Record type explicitly specified: %s", recordType);
 472         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
 473                                   &clientData);
 474     }
 475     else
 476     {
 477         if (!(zh->m_record_type))
 478         {
 479             yaz_log (YLOG_WARN, "No such record type defined");
 480             return ZEBRA_FAIL;
 481         }
 482         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 483                 zh->m_record_type);
 484         recType = recType_byName (zh->reg->recTypes, zh->res,
 485                                   zh->m_record_type, &clientData);
 486         recordType = zh->m_record_type;
 487     }
 488
 489     if (!recType)
 490     {
 491         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
 492         return ZEBRA_FAIL;
 493     }
 494
 495     zebra_create_stream_mem(&stream, buf, buf_size);
 496
 497     res = zebra_extract_records_stream(zh, &stream,
 498                                        action,
 499                                        test_mode,
 500                                        recordType,
 501                                        sysno,
 502                                        match_criteria,
 503                                        fname,
 504                                        recType, clientData);
 505     stream.destroy(&stream);
 506     return res;
 507 }
 508
 509 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 510                                        struct ZebraRecStream *stream,
 511                                        enum zebra_recctrl_action_t action,
 512                                        int test_mode,
 513                                        const char *recordType,
 514                                        zint *sysno,
 515                                        const char *match_criteria,
 516                                        const char *fname,
 517                                        RecType recType,
 518                                        void *recTypeClientData)
 519 {
 520     ZEBRA_RES res = ZEBRA_OK;
 521     while (1)
 522     {
 523         int more = 0;
 524         res = zebra_extract_record_stream(zh, stream,
 525                                           action,
 526                                           test_mode,
 527                                           recordType,
 528                                           sysno,
 529                                           match_criteria,
 530                                           fname,
 531                                           recType, recTypeClientData, &more);
 532         if (!more)
 533         {
 534             res = ZEBRA_OK;
 535             break;
 536         }
 537         if (res != ZEBRA_OK)
 538             break;
 539         if (sysno)
 540             break;
 541     }
 542     return res;
 543 }
 544
 545
 546 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 547                                       struct ZebraRecStream *stream,
 548                                       enum zebra_recctrl_action_t action,
 549                                       int test_mode,
 550                                       const char *recordType,
 551                                       zint *sysno,
 552                                       const char *match_criteria,
 553                                       const char *fname,
 554                                       RecType recType,
 555                                       void *recTypeClientData,
 556                                       int *more)
 557
 558 {
 559     zint sysno0 = 0;
 560     RecordAttr *recordAttr;
 561     struct recExtractCtrl extractCtrl;
 562     int r;
 563     const char *matchStr = 0;
 564     Record rec;
 565     off_t start_offset = 0, end_offset = 0;
 566     const char *pr_fname = fname;  /* filename to print .. */
 567     int show_progress = zh->records_processed + zh->records_skipped
 568         < zh->m_file_verbose_limit ? 1:0;
 569
 570     zebra_init_log_level();
 571
 572     if (!pr_fname)
 573         pr_fname = "<no file>";  /* make it printable if file is omitted */
 574
 575     zebra_rec_keys_reset(zh->reg->keys);
 576     zebra_rec_keys_reset(zh->reg->sortKeys);
 577
 578     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
 579     {
 580         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
 581                                       zh->m_explain_database))
 582             return ZEBRA_FAIL;
 583     }
 584
 585     if (stream)
 586     {
 587         off_t null_offset = 0;
 588         extractCtrl.stream = stream;
 589
 590         start_offset = stream->tellf(stream);
 591
 592         extractCtrl.first_record = start_offset ? 0 : 1;
 593
 594         stream->endf(stream, &null_offset);;
 595
 596         extractCtrl.init = extract_init;
 597         extractCtrl.tokenAdd = extract_token_add;
 598         extractCtrl.schemaAdd = extract_schema_add;
 599         extractCtrl.dh = zh->reg->dh;
 600         extractCtrl.handle = zh;
 601         extractCtrl.match_criteria[0] = '\0';
 602         extractCtrl.staticrank = 0;
 603
 604         init_extractCtrl(zh, &extractCtrl);
 605
 606         extract_set_store_data_prepare(&extractCtrl);
 607
 608         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 609
 610         switch (r)
 611         {
 612         case RECCTRL_EXTRACT_EOF:
 613             return ZEBRA_FAIL;
 614         case RECCTRL_EXTRACT_ERROR_GENERIC:
 615             /* error occured during extraction ... */
 616             yaz_log (YLOG_WARN, "extract error: generic");
 617             return ZEBRA_FAIL;
 618         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 619             /* error occured during extraction ... */
 620             yaz_log (YLOG_WARN, "extract error: no such filter");
 621             return ZEBRA_FAIL;
 622         case RECCTRL_EXTRACT_SKIP:
 623             if (show_progress)
 624                 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 625                          recordType, pr_fname, (zint) start_offset);
 626             *more = 1;
 627
 628             end_offset = stream->endf(stream, 0);
 629             if (end_offset)
 630                 stream->seekf(stream, end_offset);
 631
 632             return ZEBRA_OK;
 633         case RECCTRL_EXTRACT_OK:
 634             break;
 635         default:
 636             yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
 637             return ZEBRA_FAIL;
 638         }
 639         end_offset = stream->endf(stream, 0);
 640         if (end_offset)
 641             stream->seekf(stream, end_offset);
 642         else
 643             end_offset = stream->tellf(stream);
 644
 645         all_matches_add(&extractCtrl);
 646
 647         if (extractCtrl.match_criteria[0])
 648             match_criteria = extractCtrl.match_criteria;
 649     }
 650
 651     *more = 1;
 652     if (!sysno)
 653     {
 654         sysno = &sysno0;
 655
 656         if (match_criteria && *match_criteria) {
 657             matchStr = match_criteria;
 658         } else {
 659             if (zh->m_record_id && *zh->m_record_id) {
 660                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 661                                                zh->m_record_id);
 662                 if (!matchStr)
 663                 {
 664                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 665                              pr_fname, (zint) start_offset);
 666                     return ZEBRA_FAIL;
 667                 }
 668             }
 669         }
 670         if (matchStr)
 671         {
 672             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 673             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 674                                           matchStr);
 675             if (rinfo)
 676             {
 677                 assert(*rinfo == sizeof(*sysno));
 678                 memcpy (sysno, rinfo+1, sizeof(*sysno));
 679             }
 680         }
 681     }
 682     if (zebra_rec_keys_empty(zh->reg->keys))
 683     {
 684         /* the extraction process returned no information - the record
 685            is probably empty - unless flagShowRecords is in use */
 686         if (test_mode)
 687             return ZEBRA_OK;
 688     }
 689
 690     if (! *sysno)
 691     {
 692         /* new record */
 693         if (action == action_delete)
 694         {
 695             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 696                          pr_fname, (zint) start_offset);
 697             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
 698             return ZEBRA_FAIL;
 699         }
 700         else if (action == action_replace)
 701         {
 702             yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 703                          pr_fname, (zint) start_offset);
 704             yaz_log (YLOG_WARN, "cannot update record above (seems new)");
 705             return ZEBRA_FAIL;
 706         }
 707         if (show_progress)
 708             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 709                      (zint) start_offset);
 710         rec = rec_new (zh->reg->records);
 711
 712         *sysno = rec->sysno;
 713
 714         recordAttr = rec_init_attr (zh->reg->zei, rec);
 715         if (extractCtrl.staticrank < 0)
 716         {
 717             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
 718             extractCtrl.staticrank = 0;
 719         }
 720         recordAttr->staticrank = extractCtrl.staticrank;
 721
 722         if (matchStr)
 723         {
 724             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 725             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
 726                             sizeof(*sysno), sysno);
 727         }
 728
 729         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 730         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 731                                   recordAttr->staticrank);
 732         zh->records_inserted++;
 733     }
 734     else
 735     {
 736         /* record already exists */
 737         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 738         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
 739         if (action == action_insert)
 740         {
 741             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
 742                          recordType, pr_fname, (zint) start_offset);
 743             logRecord(zh);
 744             return ZEBRA_FAIL;
 745         }
 746
 747         rec = rec_get (zh->reg->records, *sysno);
 748         assert (rec);
 749
 750         recordAttr = rec_init_attr (zh->reg->zei, rec);
 751
 752         /* decrease total size */
 753         zebraExplain_recordBytesIncrement (zh->reg->zei,
 754                                            - recordAttr->recordSize);
 755
 756         zebra_rec_keys_set_buf(delkeys,
 757                                rec->info[recInfo_delKeys],
 758                                rec->size[recInfo_delKeys],
 759                                0);
 760         zebra_rec_keys_set_buf(sortKeys,
 761                                rec->info[recInfo_sortKeys],
 762                                rec->size[recInfo_sortKeys],
 763                                0);
 764
 765         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
 766         extract_flush_record_keys(zh, *sysno, 0, delkeys,
 767                                   recordAttr->staticrank);
 768         if (action == action_delete)
 769         {
 770             /* record going to be deleted */
 771             if (zebra_rec_keys_empty(delkeys))
 772             {
 773                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 774                         pr_fname, (zint) start_offset);
 775                 yaz_log(YLOG_WARN, "cannot delete file above, "
 776                         "storeKeys false (3)");
 777             }
 778             else
 779             {
 780                 if (show_progress)
 781                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 782                             pr_fname, (zint) start_offset);
 783                 zh->records_deleted++;
 784                 if (matchStr)
 785                 {
 786                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 787                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
 788                 }
 789                 rec_del (zh->reg->records, &rec);
 790             }
 791             rec_free(&rec);
 792             logRecord(zh);
 793             return ZEBRA_OK;
 794         }
 795         else
 796         {   /* update or special_update */
 797             if (show_progress)
 798                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 799                         pr_fname, (zint) start_offset);
 800             recordAttr->staticrank = extractCtrl.staticrank;
 801             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 802             extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 803                                       recordAttr->staticrank);
 804             zh->records_updated++;
 805         }
 806         zebra_rec_keys_close(delkeys);
 807         zebra_rec_keys_close(sortKeys);
 808     }
 809     /* update file type */
 810     xfree (rec->info[recInfo_fileType]);
 811     rec->info[recInfo_fileType] =
 812         rec_strdup (recordType, &rec->size[recInfo_fileType]);
 813
 814     /* update filename */
 815     xfree (rec->info[recInfo_filename]);
 816     rec->info[recInfo_filename] =
 817         rec_strdup (fname, &rec->size[recInfo_filename]);
 818
 819     /* update delete keys */
 820     xfree (rec->info[recInfo_delKeys]);
 821     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
 822     {
 823         zebra_rec_keys_get_buf(zh->reg->keys,
 824                                &rec->info[recInfo_delKeys],
 825                                &rec->size[recInfo_delKeys]);
 826     }
 827     else
 828     {
 829         rec->info[recInfo_delKeys] = NULL;
 830         rec->size[recInfo_delKeys] = 0;
 831     }
 832     /* update sort keys */
 833     xfree (rec->info[recInfo_sortKeys]);
 834
 835     zebra_rec_keys_get_buf(zh->reg->sortKeys,
 836                            &rec->info[recInfo_sortKeys],
 837                            &rec->size[recInfo_sortKeys]);
 838
 839     if (stream)
 840     {
 841         recordAttr->recordSize = end_offset - start_offset;
 842         zebraExplain_recordBytesIncrement(zh->reg->zei,
 843                                           recordAttr->recordSize);
 844     }
 845
 846     /* set run-number for this record */
 847     recordAttr->runNumber =
 848         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
 849
 850     /* update store data */
 851     xfree (rec->info[recInfo_storeData]);
 852
 853     /* update store data */
 854     if (zh->store_data_buf)
 855     {
 856         rec->size[recInfo_storeData] = zh->store_data_size;
 857         rec->info[recInfo_storeData] = zh->store_data_buf;
 858         zh->store_data_buf = 0;
 859         recordAttr->recordSize = zh->store_data_size;
 860     }
 861     else if (zh->m_store_data)
 862     {
 863         off_t cur_offset = stream->tellf(stream);
 864
 865         rec->size[recInfo_storeData] = recordAttr->recordSize;
 866         rec->info[recInfo_storeData] = (char *)
 867             xmalloc (recordAttr->recordSize);
 868         stream->seekf(stream, start_offset);
 869         stream->readf(stream, rec->info[recInfo_storeData],
 870                       recordAttr->recordSize);
 871         stream->seekf(stream, cur_offset);
 872     }
 873     else
 874     {
 875         rec->info[recInfo_storeData] = NULL;
 876         rec->size[recInfo_storeData] = 0;
 877     }
 878     /* update database name */
 879     xfree (rec->info[recInfo_databaseName]);
 880     rec->info[recInfo_databaseName] =
 881         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
 882
 883     /* update offset */
 884     recordAttr->recordOffset = start_offset;
 885
 886     /* commit this record */
 887     rec_put (zh->reg->records, &rec);
 888     logRecord(zh);
 889     return ZEBRA_OK;
 890 }
 891
 892 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
 893 {
 894     ZebraHandle zh = (ZebraHandle) handle;
 895     struct recExtractCtrl extractCtrl;
 896
 897     if (zebraExplain_curDatabase (zh->reg->zei,
 898                                   rec->info[recInfo_databaseName]))
 899     {
 900         abort();
 901         if (zebraExplain_newDatabase (zh->reg->zei,
 902                                       rec->info[recInfo_databaseName], 0))
 903             abort ();
 904     }
 905
 906     zebra_rec_keys_reset(zh->reg->keys);
 907     zebra_rec_keys_reset(zh->reg->sortKeys);
 908
 909     extractCtrl.init = extract_init;
 910     extractCtrl.tokenAdd = extract_token_add;
 911     extractCtrl.schemaAdd = extract_schema_add;
 912     extractCtrl.dh = zh->reg->dh;
 913
 914     init_extractCtrl(zh, &extractCtrl);
 915
 916     extractCtrl.flagShowRecords = 0;
 917     extractCtrl.match_criteria[0] = '\0';
 918     extractCtrl.staticrank = 0;
 919     extractCtrl.handle = handle;
 920     extractCtrl.first_record = 1;
 921
 922     extract_set_store_data_prepare(&extractCtrl);
 923
 924     if (n)
 925         grs_extract_tree(&extractCtrl, n);
 926
 927     if (rec->size[recInfo_delKeys])
 928     {
 929         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 930
 931         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
 932
 933         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
 934                                rec->size[recInfo_delKeys],
 935                                0);
 936         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
 937         zebra_rec_keys_close(delkeys);
 938
 939         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
 940                                rec->size[recInfo_sortKeys],
 941                                0);
 942
 943         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
 944         zebra_rec_keys_close(sortkeys);
 945     }
 946     extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
 947     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
 948
 949     xfree (rec->info[recInfo_delKeys]);
 950     zebra_rec_keys_get_buf(zh->reg->keys,
 951                            &rec->info[recInfo_delKeys],
 952                            &rec->size[recInfo_delKeys]);
 953
 954     xfree (rec->info[recInfo_sortKeys]);
 955     zebra_rec_keys_get_buf(zh->reg->sortKeys,
 956                            &rec->info[recInfo_sortKeys],
 957                            &rec->size[recInfo_sortKeys]);
 958     return ZEBRA_OK;
 959 }
 960
 961 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
 962                           zebra_rec_keys_t reckeys,
 963                           int level)
 964 {
 965     if (zebra_rec_keys_rewind(reckeys))
 966     {
 967         size_t slen;
 968         const char *str;
 969         struct it_key key;
 970         NMEM nmem = nmem_create();
 971
 972         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
 973         {
 974             char keystr[200]; /* room for zints to print */
 975             char *dst_term = 0;
 976             int ord = CAST_ZINT_TO_INT(key.mem[0]);
 977             int index_type, i;
 978             const char *string_index;
 979
 980             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
 981                                     0/* db */, &string_index);
 982             assert(index_type);
 983             zebra_term_untrans_iconv(zh, nmem, index_type,
 984                                      &dst_term, str);
 985             *keystr = '\0';
 986             for (i = 0; i<key.len; i++)
 987             {
 988                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
 989             }
 990
 991             if (*str < CHR_BASE_CHAR)
 992             {
 993                 int i;
 994                 char dst_buf[200]; /* room for special chars */
 995
 996                 strcpy(dst_buf , "?");
 997
 998                 if (!strcmp(str, ""))
 999                     strcpy(dst_buf, "alwaysmatches");
1000                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1001                     strcpy(dst_buf, "firstinfield");
1002                 else if (!strcmp(str, CHR_UNKNOWN))
1003                     strcpy(dst_buf, "unknown");
1004                 else if (!strcmp(str, CHR_SPACE))
1005                     strcpy(dst_buf, "space");
1006
1007                 for (i = 0; i<slen; i++)
1008                 {
1009                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1010                 }
1011                 yaz_log(level, "%s%c %s %s", keystr, index_type,
1012                         string_index, dst_buf);
1013
1014             }
1015             else
1016                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1017                         string_index, dst_term);
1018
1019             nmem_reset(nmem);
1020         }
1021         nmem_destroy(nmem);
1022     }
1023 }
1024
1025 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1026                              zebra_rec_keys_t reckeys)
1027 {
1028     ZebraExplainInfo zei = zh->reg->zei;
1029     struct ord_stat {
1030         int no;
1031         int ord;
1032         struct ord_stat *next;
1033     };
1034
1035     if (zebra_rec_keys_rewind(reckeys))
1036     {
1037         struct ord_stat *ord_list = 0;
1038         struct ord_stat *p;
1039         size_t slen;
1040         const char *str;
1041         struct it_key key_in;
1042         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1043         {
1044             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1045
1046             for (p = ord_list; p ; p = p->next)
1047                 if (p->ord == ord)
1048                 {
1049                     p->no++;
1050                     break;
1051                 }
1052             if (!p)
1053             {
1054                 p = xmalloc(sizeof(*p));
1055                 p->no = 1;
1056                 p->ord = ord;
1057                 p->next = ord_list;
1058                 ord_list = p;
1059             }
1060         }
1061
1062         p = ord_list;
1063         while (p)
1064         {
1065             struct ord_stat *p1 = p;
1066
1067             if (is_insert)
1068                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1069             else
1070                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1071             p = p->next;
1072             xfree(p1);
1073         }
1074     }
1075 }
1076
1077 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1078                                zebra_rec_keys_t reckeys,
1079                                zint staticrank)
1080 {
1081     ZebraExplainInfo zei = zh->reg->zei;
1082
1083     extract_rec_keys_adjust(zh, cmd, reckeys);
1084
1085     if (log_level_details)
1086     {
1087         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1088                 sysno, cmd ? "insert" : "delete");
1089         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1090     }
1091
1092     if (!zh->reg->key_block)
1093     {
1094         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1095         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1096         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1097         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1098     }
1099     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1100
1101     if (zebra_rec_keys_rewind(reckeys))
1102     {
1103         size_t slen;
1104         const char *str;
1105         struct it_key key_in;
1106         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1107         {
1108             key_block_write(zh->reg->key_block, sysno,
1109                             &key_in, cmd, str, slen,
1110                             staticrank, zh->m_staticrank);
1111         }
1112     }
1113 }
1114
1115
1116 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1117                                      zebra_rec_keys_t reckeys,
1118                                      zebra_snippets *snippets)
1119 {
1120     NMEM nmem = nmem_create();
1121     if (zebra_rec_keys_rewind(reckeys))
1122     {
1123         const char *str;
1124         size_t slen;
1125         struct it_key key;
1126         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1127         {
1128             char *dst_term = 0;
1129             int ord;
1130             zint seqno;
1131             int index_type;
1132
1133             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1134             seqno = key.mem[key.len-1];
1135             ord = CAST_ZINT_TO_INT(key.mem[0]);
1136
1137             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1138                                     0/* db */, 0 /* string_index */);
1139             assert(index_type);
1140             zebra_term_untrans_iconv(zh, nmem, index_type,
1141                                      &dst_term, str);
1142             zebra_snippets_append(snippets, seqno, ord, dst_term);
1143             nmem_reset(nmem);
1144         }
1145     }
1146     nmem_destroy(nmem);
1147     return ZEBRA_OK;
1148 }
1149
1150 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1151 {
1152     yaz_log(YLOG_LOG, "print_rec_keys");
1153     if (zebra_rec_keys_rewind(reckeys))
1154     {
1155         const char *str;
1156         size_t slen;
1157         struct it_key key;
1158         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1159         {
1160             char dst_buf[IT_MAX_WORD];
1161             zint seqno;
1162             int index_type;
1163             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1164             const char *db = 0;
1165             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1166
1167             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1168
1169             seqno = key.mem[key.len-1];
1170
1171             zebra_term_untrans(zh, index_type, dst_buf, str);
1172
1173             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1174                     " term=%s", ord, seqno, dst_buf);
1175         }
1176     }
1177 }
1178
1179 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1180                                      const char *str, int length)
1181 {
1182     struct it_key key;
1183     ZebraHandle zh = p->extractCtrl->handle;
1184     ZebraExplainInfo zei = zh->reg->zei;
1185     int ch, i;
1186
1187     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1188     if (ch < 0)
1189         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1190
1191     i = 0;
1192     key.mem[i++] = ch;
1193     key.mem[i++] = p->record_id;
1194     key.mem[i++] = p->section_id;
1195
1196     if (zh->m_segment_indexing)
1197         key.mem[i++] = p->segment;
1198     key.mem[i++] = p->seqno;
1199     key.len = i;
1200
1201     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1202 }
1203
1204 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1205 {
1206     struct it_key key;
1207     ZebraHandle zh = p->extractCtrl->handle;
1208     ZebraExplainInfo zei = zh->reg->zei;
1209     int ch;
1210     zinfo_index_category_t cat = zinfo_index_category_sort;
1211
1212     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1213     if (ch < 0)
1214         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1215     key.len = 2;
1216     key.mem[0] = ch;
1217     key.mem[1] = p->record_id;
1218
1219     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1220 }
1221
1222 static void extract_add_staticrank_string(RecWord *p,
1223                                           const char *str, int length)
1224 {
1225     char valz[40];
1226     struct recExtractCtrl *ctrl = p->extractCtrl;
1227
1228     if (length > sizeof(valz)-1)
1229         length = sizeof(valz)-1;
1230
1231     memcpy(valz, str, length);
1232     valz[length] = '\0';
1233     ctrl->staticrank = atozint(valz);
1234 }
1235
1236 static void extract_add_string(RecWord *p, const char *string, int length)
1237 {
1238     ZebraHandle zh = p->extractCtrl->handle;
1239     assert (length > 0);
1240
1241     if (!p->index_name)
1242         return;
1243
1244     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1245     {
1246         extract_add_index_string(p, zinfo_index_category_index,
1247                                  string, length);
1248         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1249         {
1250             RecWord word;
1251             memcpy(&word, p, sizeof(word));
1252
1253             word.seqno = 1;
1254             extract_add_index_string(
1255                 &word, zinfo_index_category_alwaysmatches, "", 0);
1256         }
1257     }
1258     else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1259     {
1260         extract_add_sort_string(p, string, length);
1261     }
1262     else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1263     {
1264         extract_add_staticrank_string(p, string, length);
1265     }
1266 }
1267
1268 static void extract_add_incomplete_field(RecWord *p)
1269 {
1270     ZebraHandle zh = p->extractCtrl->handle;
1271     const char *b = p->term_buf;
1272     int remain = p->term_len;
1273     int first = 1;
1274     const char **map = 0;
1275
1276     if (remain > 0)
1277         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1278
1279     while (map)
1280     {
1281         char buf[IT_MAX_WORD+1];
1282         int i, remain;
1283
1284         /* Skip spaces */
1285         while (map && *map && **map == *CHR_SPACE)
1286         {
1287             remain = p->term_len - (b - p->term_buf);
1288             if (remain > 0)
1289                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1290                                        remain, 0);
1291             else
1292                 map = 0;
1293         }
1294         if (!map)
1295             break;
1296         i = 0;
1297         while (map && *map && **map != *CHR_SPACE)
1298         {
1299             const char *cp = *map;
1300
1301             while (i < IT_MAX_WORD && *cp)
1302                 buf[i++] = *(cp++);
1303             remain = p->term_len - (b - p->term_buf);
1304             if (remain > 0)
1305                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1306             else
1307                 map = 0;
1308         }
1309         if (!i)
1310             return;
1311
1312         if (first)
1313         {
1314             first = 0;
1315             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1316             {
1317                 /* first in field marker */
1318                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1319                 p->seqno++;
1320             }
1321         }
1322         extract_add_string (p, buf, i);
1323         p->seqno++;
1324     }
1325 }
1326
1327 static void extract_add_complete_field (RecWord *p)
1328 {
1329     ZebraHandle zh = p->extractCtrl->handle;
1330     const char *b = p->term_buf;
1331     char buf[IT_MAX_WORD+1];
1332     const char **map = 0;
1333     int i = 0, remain = p->term_len;
1334
1335     if (remain > 0)
1336         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1337
1338     while (remain > 0 && i < IT_MAX_WORD)
1339     {
1340         while (map && *map && **map == *CHR_SPACE)
1341         {
1342             remain = p->term_len - (b - p->term_buf);
1343
1344             if (remain > 0)
1345             {
1346                 int first = i ? 0 : 1;  /* first position */
1347                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1348             }
1349             else
1350                 map = 0;
1351         }
1352         if (!map)
1353             break;
1354
1355         if (i && i < IT_MAX_WORD)
1356             buf[i++] = *CHR_SPACE;
1357         while (map && *map && **map != *CHR_SPACE)
1358         {
1359             const char *cp = *map;
1360
1361             if (**map == *CHR_CUT)
1362             {
1363                 i = 0;
1364             }
1365             else
1366             {
1367                 if (i >= IT_MAX_WORD)
1368                     break;
1369                 while (i < IT_MAX_WORD && *cp)
1370                     buf[i++] = *(cp++);
1371             }
1372             remain = p->term_len  - (b - p->term_buf);
1373             if (remain > 0)
1374             {
1375                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1376                                         remain, 0);
1377             }
1378             else
1379                 map = 0;
1380         }
1381     }
1382     if (!i)
1383         return;
1384     extract_add_string (p, buf, i);
1385 }
1386
1387 static void extract_token_add(RecWord *p)
1388 {
1389     ZebraHandle zh = p->extractCtrl->handle;
1390     WRBUF wrbuf;
1391
1392     if (log_level_extract)
1393     {
1394         yaz_log(log_level_extract, "extract_token_add "
1395                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1396                 p->index_type, p->index_name,
1397                 p->seqno, p->term_len, p->term_buf);
1398     }
1399     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1400                                p->term_buf, p->term_len)))
1401     {
1402         p->term_buf = wrbuf_buf(wrbuf);
1403         p->term_len = wrbuf_len(wrbuf);
1404     }
1405     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1406         extract_add_complete_field (p);
1407     else
1408         extract_add_incomplete_field(p);
1409 }
1410
1411 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1412                                       void *buf, size_t sz)
1413 {
1414     ZebraHandle zh = (ZebraHandle) p->handle;
1415
1416     xfree(zh->store_data_buf);
1417     zh->store_data_buf = 0;
1418     zh->store_data_size = 0;
1419     if (buf && sz)
1420     {
1421         zh->store_data_buf = xmalloc(sz);
1422         zh->store_data_size = sz;
1423         memcpy(zh->store_data_buf, buf, sz);
1424     }
1425 }
1426
1427 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1428 {
1429     ZebraHandle zh = (ZebraHandle) p->handle;
1430     xfree(zh->store_data_buf);
1431     zh->store_data_buf = 0;
1432     zh->store_data_size = 0;
1433     p->setStoreData = extract_set_store_data_cb;
1434 }
1435
1436 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1437 {
1438     ZebraHandle zh = (ZebraHandle) p->handle;
1439     zebraExplain_addSchema (zh->reg->zei, oid);
1440 }
1441
1442 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1443                              int cmd, zebra_rec_keys_t reckeys)
1444 {
1445 #if 0
1446     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1447             cmd, sysno);
1448     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1449 #endif
1450
1451     if (zebra_rec_keys_rewind(reckeys))
1452     {
1453         zebra_sort_index_t si = zh->reg->sort_index;
1454         size_t slen;
1455         const char *str;
1456         struct it_key key_in;
1457
1458         zebra_sort_sysno(si, sysno);
1459
1460         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1461         {
1462             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1463
1464             zebra_sort_type(si, ord);
1465             if (cmd == 1)
1466                 zebra_sort_add(si, str, slen);
1467             else
1468                 zebra_sort_delete(si);
1469         }
1470     }
1471 }
1472
1473 /*
1474  * Local variables:
1475  * c-basic-offset: 4
1476  * indent-tabs-mode: nil
1477  * End:
1478  * vim: shiftwidth=4 tabstop=8 expandtab
1479  */
1480