index/extract.c

   1 /* $Id: extract.c,v 1.258 2007-05-08 14:27:23 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 /* 1 if we use eliminitate identical delete/insert keys */
  44 /* eventually this the 0-case code will be removed */
  45 #define FLUSH2 1
  46
  47 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  48                                 zebra_rec_keys_t ins_keys,
  49                                 zint ins_rank,
  50                                 zebra_rec_keys_t del_keys,
  51                                 zint del_rank);
  52
  53 static void zebra_init_log_level(void)
  54 {
  55     if (!log_level_initialized)
  56     {
  57         log_level_initialized = 1;
  58
  59         log_level_extract = yaz_log_module_level("extract");
  60         log_level_details = yaz_log_module_level("indexdetails");
  61     }
  62 }
  63
  64 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
  65                                       int cmd, zebra_rec_keys_t reckeys,
  66                                       zint staticrank);
  67 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  68                                     int cmd, zebra_rec_keys_t skp);
  69 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
  70 static void extract_token_add (RecWord *p);
  71
  72 static void check_log_limit(ZebraHandle zh)
  73 {
  74     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  75     {
  76         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  77                 zh->m_file_verbose_limit);
  78     }
  79 }
  80
  81 static void logRecord (ZebraHandle zh)
  82 {
  83     check_log_limit(zh);
  84     ++zh->records_processed;
  85     if (!(zh->records_processed % 1000))
  86     {
  87         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  88                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  89                 zh->records_processed, zh->records_inserted,
  90                 zh->records_updated, zh->records_deleted);
  91     }
  92 }
  93
  94 static void extract_add_index_string (RecWord *p,
  95                                       zinfo_index_category_t cat,
  96                                       const char *str, int length);
  97
  98 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
  99
 100 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 101 {
 102     w->seqno = 1;
 103     w->index_name = "any";
 104     w->index_type = 'w';
 105     w->extractCtrl = p;
 106     w->record_id = 0;
 107     w->section_id = 0;
 108     w->segment = 0;
 109 }
 110
 111 static void searchRecordKey(ZebraHandle zh,
 112                             zebra_rec_keys_t reckeys,
 113                             const char *index_name,
 114                             const char **ws, int ws_length)
 115 {
 116     int i;
 117     int ch = -1;
 118     zinfo_index_category_t cat = zinfo_index_category_index;
 119
 120     for (i = 0; i<ws_length; i++)
 121         ws[i] = NULL;
 122
 123     if (ch < 0)
 124         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
 125     if (ch < 0)
 126         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
 127     if (ch < 0)
 128         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
 129
 130     if (ch < 0)
 131         return ;
 132
 133     if (zebra_rec_keys_rewind(reckeys))
 134     {
 135         zint startSeq = -1;
 136         const char *str;
 137         size_t slen;
 138         struct it_key key;
 139         zint seqno;
 140         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 141         {
 142             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 143
 144             seqno = key.mem[key.len-1];
 145
 146             if (key.mem[0] == ch)
 147             {
 148                 zint woff;
 149
 150                 if (startSeq == -1)
 151                     startSeq = seqno;
 152                 woff = seqno - startSeq;
 153                 if (woff >= 0 && woff < ws_length)
 154                     ws[woff] = str;
 155             }
 156         }
 157     }
 158 }
 159
 160 #define FILE_MATCH_BLANK "\t "
 161
 162 static char *get_match_from_spec(ZebraHandle zh,
 163                           zebra_rec_keys_t reckeys,
 164                           const char *fname, const char *spec)
 165 {
 166     static char dstBuf[2048];      /* static here ??? */
 167     char *dst = dstBuf;
 168     const char *s = spec;
 169
 170     while (1)
 171     {
 172         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 173             ;
 174         if (!*s)
 175             break;
 176         if (*s == '(')
 177         {
 178             const char *ws[32];
 179             char attset_str[64], attname_str[64];
 180             int i;
 181             int first = 1;
 182
 183             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 184                 ;
 185             for (i = 0; *s && *s != ',' && *s != ')' &&
 186                      !strchr(FILE_MATCH_BLANK, *s); s++)
 187                 if (i+1 < sizeof(attset_str))
 188                     attset_str[i++] = *s;
 189             attset_str[i] = '\0';
 190
 191             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 192                 ;
 193             if (*s != ',')
 194                 strcpy(attname_str, attset_str);
 195             else
 196             {
 197                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 198                     ;
 199                 for (i = 0; *s && *s != ')' &&
 200                          !strchr(FILE_MATCH_BLANK, *s); s++)
 201                     if (i+1 < sizeof(attname_str))
 202                         attname_str[i++] = *s;
 203                 attname_str[i] = '\0';
 204             }
 205
 206             searchRecordKey (zh, reckeys, attname_str, ws, 32);
 207
 208             if (*s != ')')
 209             {
 210                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
 211                       spec, zh->m_group ? zh->m_group : "none");
 212                 return NULL;
 213             }
 214             s++;
 215
 216             for (i = 0; i<32; i++)
 217                 if (ws[i])
 218                 {
 219                     if (first)
 220                     {
 221                         *dst++ = ' ';
 222                         first = 0;
 223                     }
 224                     strcpy (dst, ws[i]);
 225                     dst += strlen(ws[i]);
 226                 }
 227             if (first)
 228             {
 229                 yaz_log (YLOG_WARN, "Record didn't contain match"
 230                       " fields in (%s,%s)", attset_str, attname_str);
 231                 return NULL;
 232             }
 233         }
 234         else if (*s == '$')
 235         {
 236             int spec_len;
 237             char special[64];
 238             const char *spec_src = NULL;
 239             const char *s1 = ++s;
 240             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 241                 s1++;
 242
 243             spec_len = s1 - s;
 244             if (spec_len > sizeof(special)-1)
 245                 spec_len = sizeof(special)-1;
 246             memcpy (special, s, spec_len);
 247             special[spec_len] = '\0';
 248             s = s1;
 249
 250             if (!strcmp (special, "group"))
 251                 spec_src = zh->m_group;
 252             else if (!strcmp (special, "database"))
 253                 spec_src = zh->basenames[0];
 254             else if (!strcmp (special, "filename")) {
 255                 spec_src = fname;
 256             }
 257             else if (!strcmp (special, "type"))
 258                 spec_src = zh->m_record_type;
 259             else
 260                 spec_src = NULL;
 261             if (spec_src)
 262             {
 263                 strcpy (dst, spec_src);
 264                 dst += strlen (spec_src);
 265             }
 266         }
 267         else if (*s == '\"' || *s == '\'')
 268         {
 269             int stopMarker = *s++;
 270             char tmpString[64];
 271             int i = 0;
 272
 273             while (*s && *s != stopMarker)
 274             {
 275                 if (i+1 < sizeof(tmpString))
 276                     tmpString[i++] = *s++;
 277             }
 278             if (*s)
 279                 s++;
 280             tmpString[i] = '\0';
 281             strcpy (dst, tmpString);
 282             dst += strlen (tmpString);
 283         }
 284         else
 285         {
 286             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
 287                   spec, zh->m_group ? zh->m_group : "none");
 288             return NULL;
 289         }
 290         *dst++ = 1;
 291     }
 292     if (dst == dstBuf)
 293     {
 294         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
 295               fname, zh->m_group ? zh->m_group : "none");
 296         return NULL;
 297     }
 298     *dst = '\0';
 299     return dstBuf;
 300 }
 301
 302 struct recordLogInfo {
 303     const char *fname;
 304     int recordOffset;
 305     struct recordGroup *rGroup;
 306 };
 307
 308 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
 309 {
 310     int i;
 311     for (i = 0; i<256; i++)
 312     {
 313         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
 314             ctrl->seqno[i] = 1;
 315         else
 316             ctrl->seqno[i] = 0;
 317     }
 318     ctrl->flagShowRecords = !zh->m_flag_rw;
 319 }
 320
 321 static void all_matches_add(struct recExtractCtrl *ctrl)
 322 {
 323     RecWord word;
 324     extract_init(ctrl, &word);
 325     word.index_name = "_ALLRECORDS";
 326     word.index_type = 'w';
 327     word.seqno = 1;
 328     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
 329                               "", 0);
 330 }
 331
 332 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 333                                        struct ZebraRecStream *stream,
 334                                        enum zebra_recctrl_action_t action,
 335                                        int test_mode,
 336                                        const char *recordType,
 337                                        zint *sysno,
 338                                        const char *match_criteria,
 339                                        const char *fname,
 340                                        RecType recType,
 341                                        void *recTypeClientData);
 342
 343
 344 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 345                              int deleteFlag)
 346 {
 347     ZEBRA_RES r = ZEBRA_OK;
 348     int i, fd;
 349     char gprefix[128];
 350     char ext[128];
 351     char ext_res[128];
 352     struct file_read_info *fi = 0;
 353     const char *original_record_type = 0;
 354     RecType recType;
 355     void *recTypeClientData;
 356     struct ZebraRecStream stream, *streamp;
 357
 358     zebra_init_log_level();
 359
 360     if (!zh->m_group || !*zh->m_group)
 361         *gprefix = '\0';
 362     else
 363         sprintf (gprefix, "%s.", zh->m_group);
 364
 365     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 366
 367     /* determine file extension */
 368     *ext = '\0';
 369     for (i = strlen(fname); --i >= 0; )
 370         if (fname[i] == '/')
 371             break;
 372         else if (fname[i] == '.')
 373         {
 374             strcpy (ext, fname+i+1);
 375             break;
 376         }
 377     /* determine file type - depending on extension */
 378     original_record_type = zh->m_record_type;
 379     if (!zh->m_record_type)
 380     {
 381         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
 382         zh->m_record_type = res_get (zh->res, ext_res);
 383     }
 384     if (!zh->m_record_type)
 385     {
 386         check_log_limit(zh);
 387         if (zh->records_processed + zh->records_skipped
 388             < zh->m_file_verbose_limit)
 389             yaz_log (YLOG_LOG, "? %s", fname);
 390         zh->records_skipped++;
 391         return 0;
 392     }
 393     /* determine match criteria */
 394     if (!zh->m_record_id)
 395     {
 396         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
 397         zh->m_record_id = res_get (zh->res, ext_res);
 398     }
 399
 400     if (!(recType =
 401           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
 402                           &recTypeClientData)))
 403     {
 404         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 405         return ZEBRA_FAIL;
 406     }
 407
 408     switch(recType->version)
 409     {
 410     case 0:
 411         break;
 412     default:
 413         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 414     }
 415     if (sysno && deleteFlag)
 416     {
 417         streamp = 0;
 418         fi = 0;
 419     }
 420     else
 421     {
 422         char full_rep[1024];
 423
 424         if (zh->path_reg && !yaz_is_abspath (fname))
 425         {
 426             strcpy (full_rep, zh->path_reg);
 427             strcat (full_rep, "/");
 428             strcat (full_rep, fname);
 429         }
 430         else
 431             strcpy (full_rep, fname);
 432
 433         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
 434         {
 435             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 436             zh->m_record_type = original_record_type;
 437             return ZEBRA_FAIL;
 438         }
 439         streamp = &stream;
 440         zebra_create_stream_fd(streamp, fd, 0);
 441     }
 442     r = zebra_extract_records_stream(zh, streamp,
 443                                      deleteFlag ?
 444                                      action_delete : action_update,
 445                                      0, /* tst_mode */
 446                                      zh->m_record_type,
 447                                      sysno,
 448                                      0, /*match_criteria */
 449                                      fname,
 450                                      recType, recTypeClientData);
 451     if (streamp)
 452         stream.destroy(streamp);
 453     zh->m_record_type = original_record_type;
 454     return r;
 455 }
 456
 457 /*
 458   If sysno is provided, then it's used to identify the reocord.
 459   If not, and match_criteria is provided, then sysno is guessed
 460   If not, and a record is provided, then sysno is got from there
 461
 462  */
 463
 464 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 465                                       const char *buf, size_t buf_size,
 466                                       enum zebra_recctrl_action_t action,
 467                                       int test_mode,
 468                                       const char *recordType,
 469                                       zint *sysno,
 470                                       const char *match_criteria,
 471                                       const char *fname)
 472 {
 473     struct ZebraRecStream stream;
 474     ZEBRA_RES res;
 475     void *clientData;
 476     RecType recType = 0;
 477
 478     if (recordType && *recordType)
 479     {
 480         yaz_log(log_level_extract,
 481                 "Record type explicitly specified: %s", recordType);
 482         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
 483                                   &clientData);
 484     }
 485     else
 486     {
 487         if (!(zh->m_record_type))
 488         {
 489             yaz_log (YLOG_WARN, "No such record type defined");
 490             return ZEBRA_FAIL;
 491         }
 492         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 493                 zh->m_record_type);
 494         recType = recType_byName (zh->reg->recTypes, zh->res,
 495                                   zh->m_record_type, &clientData);
 496         recordType = zh->m_record_type;
 497     }
 498
 499     if (!recType)
 500     {
 501         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
 502         return ZEBRA_FAIL;
 503     }
 504
 505     zebra_create_stream_mem(&stream, buf, buf_size);
 506
 507     res = zebra_extract_records_stream(zh, &stream,
 508                                        action,
 509                                        test_mode,
 510                                        recordType,
 511                                        sysno,
 512                                        match_criteria,
 513                                        fname,
 514                                        recType, clientData);
 515     stream.destroy(&stream);
 516     return res;
 517 }
 518
 519 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 520                                        struct ZebraRecStream *stream,
 521                                        enum zebra_recctrl_action_t action,
 522                                        int test_mode,
 523                                        const char *recordType,
 524                                        zint *sysno,
 525                                        const char *match_criteria,
 526                                        const char *fname,
 527                                        RecType recType,
 528                                        void *recTypeClientData)
 529 {
 530     ZEBRA_RES res = ZEBRA_OK;
 531     while (1)
 532     {
 533         int more = 0;
 534         res = zebra_extract_record_stream(zh, stream,
 535                                           action,
 536                                           test_mode,
 537                                           recordType,
 538                                           sysno,
 539                                           match_criteria,
 540                                           fname,
 541                                           recType, recTypeClientData, &more);
 542         if (!more)
 543         {
 544             res = ZEBRA_OK;
 545             break;
 546         }
 547         if (res != ZEBRA_OK)
 548             break;
 549         if (sysno)
 550             break;
 551     }
 552     return res;
 553 }
 554
 555
 556 static WRBUF wrbuf_hex_str(const char *cstr)
 557 {
 558     size_t i;
 559     WRBUF w = wrbuf_alloc();
 560     for (i = 0; cstr[i]; i++)
 561     {
 562         if (cstr[i] < ' ' || cstr[i] > 126)
 563             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 564         else
 565             wrbuf_putc(w, cstr[i]);
 566     }
 567     return w;
 568 }
 569
 570 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 571                                       struct ZebraRecStream *stream,
 572                                       enum zebra_recctrl_action_t action,
 573                                       int test_mode,
 574                                       const char *recordType,
 575                                       zint *sysno,
 576                                       const char *match_criteria,
 577                                       const char *fname,
 578                                       RecType recType,
 579                                       void *recTypeClientData,
 580                                       int *more)
 581
 582 {
 583     zint sysno0 = 0;
 584     RecordAttr *recordAttr;
 585     struct recExtractCtrl extractCtrl;
 586     int r;
 587     const char *matchStr = 0;
 588     Record rec;
 589     off_t start_offset = 0, end_offset = 0;
 590     const char *pr_fname = fname;  /* filename to print .. */
 591     int show_progress = zh->records_processed + zh->records_skipped
 592         < zh->m_file_verbose_limit ? 1:0;
 593
 594     zebra_init_log_level();
 595
 596     if (!pr_fname)
 597         pr_fname = "<no file>";  /* make it printable if file is omitted */
 598
 599     zebra_rec_keys_reset(zh->reg->keys);
 600     zebra_rec_keys_reset(zh->reg->sortKeys);
 601
 602     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
 603     {
 604         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
 605                                       zh->m_explain_database))
 606             return ZEBRA_FAIL;
 607     }
 608
 609     if (stream)
 610     {
 611         off_t null_offset = 0;
 612         extractCtrl.stream = stream;
 613
 614         start_offset = stream->tellf(stream);
 615
 616         extractCtrl.first_record = start_offset ? 0 : 1;
 617
 618         stream->endf(stream, &null_offset);;
 619
 620         extractCtrl.init = extract_init;
 621         extractCtrl.tokenAdd = extract_token_add;
 622         extractCtrl.schemaAdd = extract_schema_add;
 623         extractCtrl.dh = zh->reg->dh;
 624         extractCtrl.handle = zh;
 625         extractCtrl.match_criteria[0] = '\0';
 626         extractCtrl.staticrank = 0;
 627         extractCtrl.action = action;
 628
 629         init_extractCtrl(zh, &extractCtrl);
 630
 631         extract_set_store_data_prepare(&extractCtrl);
 632
 633         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 634
 635         if (action == action_update)
 636         {
 637             action = extractCtrl.action;
 638         }
 639
 640         switch (r)
 641         {
 642         case RECCTRL_EXTRACT_EOF:
 643             return ZEBRA_FAIL;
 644         case RECCTRL_EXTRACT_ERROR_GENERIC:
 645             /* error occured during extraction ... */
 646             yaz_log (YLOG_WARN, "extract error: generic");
 647             return ZEBRA_FAIL;
 648         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 649             /* error occured during extraction ... */
 650             yaz_log (YLOG_WARN, "extract error: no such filter");
 651             return ZEBRA_FAIL;
 652         case RECCTRL_EXTRACT_SKIP:
 653             if (show_progress)
 654                 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 655                          recordType, pr_fname, (zint) start_offset);
 656             *more = 1;
 657
 658             end_offset = stream->endf(stream, 0);
 659             if (end_offset)
 660                 stream->seekf(stream, end_offset);
 661
 662             return ZEBRA_OK;
 663         case RECCTRL_EXTRACT_OK:
 664             break;
 665         default:
 666             yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
 667             return ZEBRA_FAIL;
 668         }
 669         end_offset = stream->endf(stream, 0);
 670         if (end_offset)
 671             stream->seekf(stream, end_offset);
 672         else
 673             end_offset = stream->tellf(stream);
 674
 675         all_matches_add(&extractCtrl);
 676
 677         if (extractCtrl.match_criteria[0])
 678             match_criteria = extractCtrl.match_criteria;
 679     }
 680
 681     *more = 1;
 682     if (!sysno)
 683     {
 684         sysno = &sysno0;
 685
 686         if (match_criteria && *match_criteria) {
 687             matchStr = match_criteria;
 688         } else {
 689             if (zh->m_record_id && *zh->m_record_id) {
 690                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 691                                                zh->m_record_id);
 692                 if (!matchStr)
 693                 {
 694                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 695                              pr_fname, (zint) start_offset);
 696                     return ZEBRA_FAIL;
 697                 }
 698             }
 699         }
 700         if (matchStr)
 701         {
 702             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 703             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 704                                           matchStr);
 705
 706
 707             if (log_level_extract)
 708             {
 709                 WRBUF w = wrbuf_hex_str(matchStr);
 710                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 711                 wrbuf_destroy(w);
 712             }
 713             if (rinfo)
 714             {
 715                 assert(*rinfo == sizeof(*sysno));
 716                 memcpy (sysno, rinfo+1, sizeof(*sysno));
 717             }
 718        }
 719     }
 720     if (zebra_rec_keys_empty(zh->reg->keys))
 721     {
 722         /* the extraction process returned no information - the record
 723            is probably empty - unless flagShowRecords is in use */
 724         if (test_mode)
 725             return ZEBRA_OK;
 726     }
 727
 728     if (! *sysno)
 729     {
 730         /* new record */
 731         if (action == action_delete)
 732         {
 733             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 734                          pr_fname, (zint) start_offset);
 735             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
 736             return ZEBRA_FAIL;
 737         }
 738         else if (action == action_replace)
 739         {
 740             yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 741                          pr_fname, (zint) start_offset);
 742             yaz_log (YLOG_WARN, "cannot update record above (seems new)");
 743             return ZEBRA_FAIL;
 744         }
 745         if (show_progress)
 746             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 747                      (zint) start_offset);
 748         rec = rec_new (zh->reg->records);
 749
 750         *sysno = rec->sysno;
 751
 752         recordAttr = rec_init_attr (zh->reg->zei, rec);
 753         if (extractCtrl.staticrank < 0)
 754         {
 755             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
 756             extractCtrl.staticrank = 0;
 757         }
 758
 759         if (matchStr)
 760         {
 761             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 762             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
 763                             sizeof(*sysno), sysno);
 764         }
 765
 766         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 767 #if FLUSH2
 768         extract_flush_record_keys2(zh, *sysno,
 769                                    zh->reg->keys, extractCtrl.staticrank,
 770                                    0, recordAttr->staticrank);
 771 #else
 772         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 773                                   extractCtrl.staticrank);
 774 #endif
 775         recordAttr->staticrank = extractCtrl.staticrank;
 776         zh->records_inserted++;
 777     }
 778     else
 779     {
 780         /* record already exists */
 781         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 782         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
 783         if (action == action_insert)
 784         {
 785             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
 786                          recordType, pr_fname, (zint) start_offset);
 787             logRecord(zh);
 788             return ZEBRA_FAIL;
 789         }
 790
 791         rec = rec_get (zh->reg->records, *sysno);
 792         assert (rec);
 793
 794         recordAttr = rec_init_attr (zh->reg->zei, rec);
 795
 796         /* decrease total size */
 797         zebraExplain_recordBytesIncrement (zh->reg->zei,
 798                                            - recordAttr->recordSize);
 799
 800         zebra_rec_keys_set_buf(delkeys,
 801                                rec->info[recInfo_delKeys],
 802                                rec->size[recInfo_delKeys],
 803                                0);
 804         zebra_rec_keys_set_buf(sortKeys,
 805                                rec->info[recInfo_sortKeys],
 806                                rec->size[recInfo_sortKeys],
 807                                0);
 808
 809         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
 810 #if !FLUSH2
 811         extract_flush_record_keys(zh, *sysno, 0, delkeys,
 812                                   recordAttr->staticrank);
 813 #endif
 814         if (action == action_delete)
 815         {
 816             /* record going to be deleted */
 817 #if FLUSH2
 818             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
 819                                        delkeys, recordAttr->staticrank);
 820 #endif
 821             if (zebra_rec_keys_empty(delkeys))
 822             {
 823                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 824                         pr_fname, (zint) start_offset);
 825                 yaz_log(YLOG_WARN, "cannot delete file above, "
 826                         "storeKeys false (3)");
 827             }
 828             else
 829             {
 830                 if (show_progress)
 831                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 832                             pr_fname, (zint) start_offset);
 833                 zh->records_deleted++;
 834                 if (matchStr)
 835                 {
 836                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 837                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
 838                 }
 839                 rec_del (zh->reg->records, &rec);
 840             }
 841             zebra_rec_keys_close(delkeys);
 842             zebra_rec_keys_close(sortKeys);
 843             rec_free(&rec);
 844             logRecord(zh);
 845             return ZEBRA_OK;
 846         }
 847         else
 848         {   /* update or special_update */
 849             if (show_progress)
 850                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 851                         pr_fname, (zint) start_offset);
 852             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 853
 854 #if FLUSH2
 855             extract_flush_record_keys2(zh, *sysno,
 856                                        zh->reg->keys, extractCtrl.staticrank,
 857                                        delkeys, recordAttr->staticrank);
 858 #else
 859             extract_flush_record_keys(zh, *sysno, 1,
 860                                       zh->reg->keys, extractCtrl.staticrank);
 861 #endif
 862             recordAttr->staticrank = extractCtrl.staticrank;
 863             zh->records_updated++;
 864         }
 865         zebra_rec_keys_close(delkeys);
 866         zebra_rec_keys_close(sortKeys);
 867     }
 868     /* update file type */
 869     xfree (rec->info[recInfo_fileType]);
 870     rec->info[recInfo_fileType] =
 871         rec_strdup (recordType, &rec->size[recInfo_fileType]);
 872
 873     /* update filename */
 874     xfree (rec->info[recInfo_filename]);
 875     rec->info[recInfo_filename] =
 876         rec_strdup (fname, &rec->size[recInfo_filename]);
 877
 878     /* update delete keys */
 879     xfree (rec->info[recInfo_delKeys]);
 880     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
 881     {
 882         zebra_rec_keys_get_buf(zh->reg->keys,
 883                                &rec->info[recInfo_delKeys],
 884                                &rec->size[recInfo_delKeys]);
 885     }
 886     else
 887     {
 888         rec->info[recInfo_delKeys] = NULL;
 889         rec->size[recInfo_delKeys] = 0;
 890     }
 891     /* update sort keys */
 892     xfree (rec->info[recInfo_sortKeys]);
 893
 894     zebra_rec_keys_get_buf(zh->reg->sortKeys,
 895                            &rec->info[recInfo_sortKeys],
 896                            &rec->size[recInfo_sortKeys]);
 897
 898     if (stream)
 899     {
 900         recordAttr->recordSize = end_offset - start_offset;
 901         zebraExplain_recordBytesIncrement(zh->reg->zei,
 902                                           recordAttr->recordSize);
 903     }
 904
 905     /* set run-number for this record */
 906     recordAttr->runNumber =
 907         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
 908
 909     /* update store data */
 910     xfree (rec->info[recInfo_storeData]);
 911
 912     /* update store data */
 913     if (zh->store_data_buf)
 914     {
 915         rec->size[recInfo_storeData] = zh->store_data_size;
 916         rec->info[recInfo_storeData] = zh->store_data_buf;
 917         zh->store_data_buf = 0;
 918         recordAttr->recordSize = zh->store_data_size;
 919     }
 920     else if (zh->m_store_data)
 921     {
 922         off_t cur_offset = stream->tellf(stream);
 923
 924         rec->size[recInfo_storeData] = recordAttr->recordSize;
 925         rec->info[recInfo_storeData] = (char *)
 926             xmalloc (recordAttr->recordSize);
 927         stream->seekf(stream, start_offset);
 928         stream->readf(stream, rec->info[recInfo_storeData],
 929                       recordAttr->recordSize);
 930         stream->seekf(stream, cur_offset);
 931     }
 932     else
 933     {
 934         rec->info[recInfo_storeData] = NULL;
 935         rec->size[recInfo_storeData] = 0;
 936     }
 937     /* update database name */
 938     xfree (rec->info[recInfo_databaseName]);
 939     rec->info[recInfo_databaseName] =
 940         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
 941
 942     /* update offset */
 943     recordAttr->recordOffset = start_offset;
 944
 945     /* commit this record */
 946     rec_put (zh->reg->records, &rec);
 947     logRecord(zh);
 948     return ZEBRA_OK;
 949 }
 950
 951 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
 952 {
 953     ZebraHandle zh = (ZebraHandle) handle;
 954     struct recExtractCtrl extractCtrl;
 955
 956     if (zebraExplain_curDatabase (zh->reg->zei,
 957                                   rec->info[recInfo_databaseName]))
 958     {
 959         abort();
 960         if (zebraExplain_newDatabase (zh->reg->zei,
 961                                       rec->info[recInfo_databaseName], 0))
 962             abort ();
 963     }
 964
 965     zebra_rec_keys_reset(zh->reg->keys);
 966     zebra_rec_keys_reset(zh->reg->sortKeys);
 967
 968     extractCtrl.init = extract_init;
 969     extractCtrl.tokenAdd = extract_token_add;
 970     extractCtrl.schemaAdd = extract_schema_add;
 971     extractCtrl.dh = zh->reg->dh;
 972
 973     init_extractCtrl(zh, &extractCtrl);
 974
 975     extractCtrl.flagShowRecords = 0;
 976     extractCtrl.match_criteria[0] = '\0';
 977     extractCtrl.staticrank = 0;
 978     extractCtrl.action = action_update;
 979
 980     extractCtrl.handle = handle;
 981     extractCtrl.first_record = 1;
 982
 983     extract_set_store_data_prepare(&extractCtrl);
 984
 985     if (n)
 986         grs_extract_tree(&extractCtrl, n);
 987
 988     if (rec->size[recInfo_delKeys])
 989     {
 990         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 991
 992         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
 993
 994         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
 995                                rec->size[recInfo_delKeys],
 996                                0);
 997 #if FLUSH2
 998         extract_flush_record_keys2(zh, rec->sysno,
 999                                    zh->reg->keys, 0, delkeys, 0);
1000 #else
1001         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1002         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1003 #endif
1004         zebra_rec_keys_close(delkeys);
1005
1006         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1007                                rec->size[recInfo_sortKeys],
1008                                0);
1009
1010         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1011         zebra_rec_keys_close(sortkeys);
1012     }
1013     else
1014     {
1015 #if FLUSH2
1016         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1017 #else
1018         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1019 #endif
1020     }
1021     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1022
1023     xfree (rec->info[recInfo_delKeys]);
1024     zebra_rec_keys_get_buf(zh->reg->keys,
1025                            &rec->info[recInfo_delKeys],
1026                            &rec->size[recInfo_delKeys]);
1027
1028     xfree (rec->info[recInfo_sortKeys]);
1029     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1030                            &rec->info[recInfo_sortKeys],
1031                            &rec->size[recInfo_sortKeys]);
1032     return ZEBRA_OK;
1033 }
1034
1035 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1036                           zebra_rec_keys_t reckeys,
1037                           int level)
1038 {
1039     if (zebra_rec_keys_rewind(reckeys))
1040     {
1041         size_t slen;
1042         const char *str;
1043         struct it_key key;
1044         NMEM nmem = nmem_create();
1045
1046         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1047         {
1048             char keystr[200]; /* room for zints to print */
1049             char *dst_term = 0;
1050             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1051             int index_type, i;
1052             const char *string_index;
1053
1054             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1055                                     0/* db */, &string_index);
1056             assert(index_type);
1057             zebra_term_untrans_iconv(zh, nmem, index_type,
1058                                      &dst_term, str);
1059             *keystr = '\0';
1060             for (i = 0; i<key.len; i++)
1061             {
1062                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1063             }
1064
1065             if (*str < CHR_BASE_CHAR)
1066             {
1067                 int i;
1068                 char dst_buf[200]; /* room for special chars */
1069
1070                 strcpy(dst_buf , "?");
1071
1072                 if (!strcmp(str, ""))
1073                     strcpy(dst_buf, "alwaysmatches");
1074                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1075                     strcpy(dst_buf, "firstinfield");
1076                 else if (!strcmp(str, CHR_UNKNOWN))
1077                     strcpy(dst_buf, "unknown");
1078                 else if (!strcmp(str, CHR_SPACE))
1079                     strcpy(dst_buf, "space");
1080
1081                 for (i = 0; i<slen; i++)
1082                 {
1083                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1084                 }
1085                 yaz_log(level, "%s%c %s %s", keystr, index_type,
1086                         string_index, dst_buf);
1087
1088             }
1089             else
1090                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1091                         string_index, dst_term);
1092
1093             nmem_reset(nmem);
1094         }
1095         nmem_destroy(nmem);
1096     }
1097 }
1098
1099 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1100                              zebra_rec_keys_t reckeys)
1101 {
1102     ZebraExplainInfo zei = zh->reg->zei;
1103     struct ord_stat {
1104         int no;
1105         int ord;
1106         struct ord_stat *next;
1107     };
1108
1109     if (zebra_rec_keys_rewind(reckeys))
1110     {
1111         struct ord_stat *ord_list = 0;
1112         struct ord_stat *p;
1113         size_t slen;
1114         const char *str;
1115         struct it_key key_in;
1116         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1117         {
1118             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1119
1120             for (p = ord_list; p ; p = p->next)
1121                 if (p->ord == ord)
1122                 {
1123                     p->no++;
1124                     break;
1125                 }
1126             if (!p)
1127             {
1128                 p = xmalloc(sizeof(*p));
1129                 p->no = 1;
1130                 p->ord = ord;
1131                 p->next = ord_list;
1132                 ord_list = p;
1133             }
1134         }
1135
1136         p = ord_list;
1137         while (p)
1138         {
1139             struct ord_stat *p1 = p;
1140
1141             if (is_insert)
1142                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1143             else
1144                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1145             p = p->next;
1146             xfree(p1);
1147         }
1148     }
1149 }
1150
1151 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1152                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1153                                 zebra_rec_keys_t del_keys, zint del_rank)
1154 {
1155     ZebraExplainInfo zei = zh->reg->zei;
1156     int normal = 0;
1157     int optimized = 0;
1158
1159     if (!zh->reg->key_block)
1160     {
1161         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1162         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1163         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1164         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1165     }
1166
1167     if (ins_keys)
1168     {
1169         extract_rec_keys_adjust(zh, 1, ins_keys);
1170         if (!del_keys)
1171             zebraExplain_recordCountIncrement (zei, 1);
1172         zebra_rec_keys_rewind(ins_keys);
1173     }
1174     if (del_keys)
1175     {
1176         extract_rec_keys_adjust(zh, 0, del_keys);
1177         if (!ins_keys)
1178             zebraExplain_recordCountIncrement (zei, -1);
1179         zebra_rec_keys_rewind(del_keys);
1180     }
1181
1182     while (1)
1183     {
1184         size_t del_slen;
1185         const char *del_str;
1186         struct it_key del_key_in;
1187         int del = 0;
1188
1189         size_t ins_slen;
1190         const char *ins_str;
1191         struct it_key ins_key_in;
1192         int ins = 0;
1193
1194         if (del_keys)
1195             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1196                                       &del_key_in);
1197         if (ins_keys)
1198             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1199                                       &ins_key_in);
1200
1201         if (del && ins && ins_rank == del_rank
1202             && !key_compare(&del_key_in, &ins_key_in)
1203             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1204         {
1205             optimized++;
1206             continue;
1207         }
1208         if (!del && !ins)
1209             break;
1210
1211         normal++;
1212         if (del)
1213             key_block_write(zh->reg->key_block, sysno,
1214                             &del_key_in, 0, del_str, del_slen,
1215                             del_rank, zh->m_staticrank);
1216         if (ins)
1217             key_block_write(zh->reg->key_block, sysno,
1218                             &ins_key_in, 1, ins_str, ins_slen,
1219                             ins_rank, zh->m_staticrank);
1220     }
1221     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1222 }
1223
1224 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1225                                zebra_rec_keys_t reckeys,
1226                                zint staticrank)
1227 {
1228     ZebraExplainInfo zei = zh->reg->zei;
1229
1230     extract_rec_keys_adjust(zh, cmd, reckeys);
1231
1232     if (log_level_details)
1233     {
1234         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1235                 sysno, cmd ? "insert" : "delete");
1236         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1237     }
1238
1239     if (!zh->reg->key_block)
1240     {
1241         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1242         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1243         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1244         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1245     }
1246     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1247
1248 #if 0
1249     yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
1250     print_rec_keys(zh, reckeys);
1251 #endif
1252     if (zebra_rec_keys_rewind(reckeys))
1253     {
1254         size_t slen;
1255         const char *str;
1256         struct it_key key_in;
1257         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1258         {
1259             key_block_write(zh->reg->key_block, sysno,
1260                             &key_in, cmd, str, slen,
1261                             staticrank, zh->m_staticrank);
1262         }
1263     }
1264 }
1265
1266 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1267                                      zebra_rec_keys_t reckeys,
1268                                      zebra_snippets *snippets)
1269 {
1270     NMEM nmem = nmem_create();
1271     if (zebra_rec_keys_rewind(reckeys))
1272     {
1273         const char *str;
1274         size_t slen;
1275         struct it_key key;
1276         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1277         {
1278             char *dst_term = 0;
1279             int ord;
1280             zint seqno;
1281             int index_type;
1282
1283             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1284             seqno = key.mem[key.len-1];
1285             ord = CAST_ZINT_TO_INT(key.mem[0]);
1286
1287             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1288                                     0/* db */, 0 /* string_index */);
1289             assert(index_type);
1290             zebra_term_untrans_iconv(zh, nmem, index_type,
1291                                      &dst_term, str);
1292             zebra_snippets_append(snippets, seqno, ord, dst_term);
1293             nmem_reset(nmem);
1294         }
1295     }
1296     nmem_destroy(nmem);
1297     return ZEBRA_OK;
1298 }
1299
1300 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1301 {
1302     yaz_log(YLOG_LOG, "print_rec_keys");
1303     if (zebra_rec_keys_rewind(reckeys))
1304     {
1305         const char *str;
1306         size_t slen;
1307         struct it_key key;
1308         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1309         {
1310             char dst_buf[IT_MAX_WORD];
1311             zint seqno;
1312             int index_type;
1313             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1314             const char *db = 0;
1315             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1316
1317             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1318
1319             seqno = key.mem[key.len-1];
1320
1321             zebra_term_untrans(zh, index_type, dst_buf, str);
1322
1323             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1324                     " term=%s", ord, seqno, dst_buf);
1325         }
1326     }
1327 }
1328
1329 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1330                                      const char *str, int length)
1331 {
1332     struct it_key key;
1333     ZebraHandle zh = p->extractCtrl->handle;
1334     ZebraExplainInfo zei = zh->reg->zei;
1335     int ch, i;
1336
1337     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1338     if (ch < 0)
1339         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1340
1341     i = 0;
1342     key.mem[i++] = ch;
1343     key.mem[i++] = p->record_id;
1344     key.mem[i++] = p->section_id;
1345
1346     if (zh->m_segment_indexing)
1347         key.mem[i++] = p->segment;
1348     key.mem[i++] = p->seqno;
1349     key.len = i;
1350
1351     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1352 }
1353
1354 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1355 {
1356     struct it_key key;
1357     ZebraHandle zh = p->extractCtrl->handle;
1358     ZebraExplainInfo zei = zh->reg->zei;
1359     int ch;
1360     zinfo_index_category_t cat = zinfo_index_category_sort;
1361
1362     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1363     if (ch < 0)
1364         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1365     key.len = 2;
1366     key.mem[0] = ch;
1367     key.mem[1] = p->record_id;
1368
1369     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1370 }
1371
1372 static void extract_add_staticrank_string(RecWord *p,
1373                                           const char *str, int length)
1374 {
1375     char valz[40];
1376     struct recExtractCtrl *ctrl = p->extractCtrl;
1377
1378     if (length > sizeof(valz)-1)
1379         length = sizeof(valz)-1;
1380
1381     memcpy(valz, str, length);
1382     valz[length] = '\0';
1383     ctrl->staticrank = atozint(valz);
1384 }
1385
1386 static void extract_add_string(RecWord *p, const char *string, int length)
1387 {
1388     ZebraHandle zh = p->extractCtrl->handle;
1389     assert (length > 0);
1390
1391     if (!p->index_name)
1392         return;
1393
1394     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1395     {
1396         extract_add_index_string(p, zinfo_index_category_index,
1397                                  string, length);
1398         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1399         {
1400             RecWord word;
1401             memcpy(&word, p, sizeof(word));
1402
1403             word.seqno = 1;
1404             extract_add_index_string(
1405                 &word, zinfo_index_category_alwaysmatches, "", 0);
1406         }
1407     }
1408     else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1409     {
1410         extract_add_sort_string(p, string, length);
1411     }
1412     else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1413     {
1414         extract_add_staticrank_string(p, string, length);
1415     }
1416 }
1417
1418 static void extract_add_incomplete_field(RecWord *p)
1419 {
1420     ZebraHandle zh = p->extractCtrl->handle;
1421     const char *b = p->term_buf;
1422     int remain = p->term_len;
1423     int first = 1;
1424     const char **map = 0;
1425
1426     if (remain > 0)
1427         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1428
1429     while (map)
1430     {
1431         char buf[IT_MAX_WORD+1];
1432         int i, remain;
1433
1434         /* Skip spaces */
1435         while (map && *map && **map == *CHR_SPACE)
1436         {
1437             remain = p->term_len - (b - p->term_buf);
1438             if (remain > 0)
1439                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1440                                        remain, 0);
1441             else
1442                 map = 0;
1443         }
1444         if (!map)
1445             break;
1446         i = 0;
1447         while (map && *map && **map != *CHR_SPACE)
1448         {
1449             const char *cp = *map;
1450
1451             while (i < IT_MAX_WORD && *cp)
1452                 buf[i++] = *(cp++);
1453             remain = p->term_len - (b - p->term_buf);
1454             if (remain > 0)
1455                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1456             else
1457                 map = 0;
1458         }
1459         if (!i)
1460             return;
1461
1462         if (first)
1463         {
1464             first = 0;
1465             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1466             {
1467                 /* first in field marker */
1468                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1469                 p->seqno++;
1470             }
1471         }
1472         extract_add_string (p, buf, i);
1473         p->seqno++;
1474     }
1475 }
1476
1477 static void extract_add_complete_field (RecWord *p)
1478 {
1479     ZebraHandle zh = p->extractCtrl->handle;
1480     const char *b = p->term_buf;
1481     char buf[IT_MAX_WORD+1];
1482     const char **map = 0;
1483     int i = 0, remain = p->term_len;
1484
1485     if (remain > 0)
1486         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1487
1488     while (remain > 0 && i < IT_MAX_WORD)
1489     {
1490         while (map && *map && **map == *CHR_SPACE)
1491         {
1492             remain = p->term_len - (b - p->term_buf);
1493
1494             if (remain > 0)
1495             {
1496                 int first = i ? 0 : 1;  /* first position */
1497                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1498             }
1499             else
1500                 map = 0;
1501         }
1502         if (!map)
1503             break;
1504
1505         if (i && i < IT_MAX_WORD)
1506             buf[i++] = *CHR_SPACE;
1507         while (map && *map && **map != *CHR_SPACE)
1508         {
1509             const char *cp = *map;
1510
1511             if (**map == *CHR_CUT)
1512             {
1513                 i = 0;
1514             }
1515             else
1516             {
1517                 if (i >= IT_MAX_WORD)
1518                     break;
1519                 while (i < IT_MAX_WORD && *cp)
1520                     buf[i++] = *(cp++);
1521             }
1522             remain = p->term_len  - (b - p->term_buf);
1523             if (remain > 0)
1524             {
1525                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1526                                         remain, 0);
1527             }
1528             else
1529                 map = 0;
1530         }
1531     }
1532     if (!i)
1533         return;
1534     extract_add_string (p, buf, i);
1535 }
1536
1537 static void extract_token_add(RecWord *p)
1538 {
1539     ZebraHandle zh = p->extractCtrl->handle;
1540     WRBUF wrbuf;
1541
1542     if (log_level_details)
1543     {
1544         yaz_log(log_level_details, "extract_token_add "
1545                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1546                 p->index_type, p->index_name,
1547                 p->seqno, p->term_len, p->term_buf);
1548     }
1549     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1550                                p->term_buf, p->term_len)))
1551     {
1552         p->term_buf = wrbuf_buf(wrbuf);
1553         p->term_len = wrbuf_len(wrbuf);
1554     }
1555     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1556         extract_add_complete_field (p);
1557     else
1558         extract_add_incomplete_field(p);
1559 }
1560
1561 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1562                                       void *buf, size_t sz)
1563 {
1564     ZebraHandle zh = (ZebraHandle) p->handle;
1565
1566     xfree(zh->store_data_buf);
1567     zh->store_data_buf = 0;
1568     zh->store_data_size = 0;
1569     if (buf && sz)
1570     {
1571         zh->store_data_buf = xmalloc(sz);
1572         zh->store_data_size = sz;
1573         memcpy(zh->store_data_buf, buf, sz);
1574     }
1575 }
1576
1577 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1578 {
1579     ZebraHandle zh = (ZebraHandle) p->handle;
1580     xfree(zh->store_data_buf);
1581     zh->store_data_buf = 0;
1582     zh->store_data_size = 0;
1583     p->setStoreData = extract_set_store_data_cb;
1584 }
1585
1586 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1587 {
1588     ZebraHandle zh = (ZebraHandle) p->handle;
1589     zebraExplain_addSchema (zh->reg->zei, oid);
1590 }
1591
1592 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1593                              int cmd, zebra_rec_keys_t reckeys)
1594 {
1595 #if 0
1596     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1597             cmd, sysno);
1598     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1599 #endif
1600
1601     if (zebra_rec_keys_rewind(reckeys))
1602     {
1603         zebra_sort_index_t si = zh->reg->sort_index;
1604         size_t slen;
1605         const char *str;
1606         struct it_key key_in;
1607
1608         zebra_sort_sysno(si, sysno);
1609
1610         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1611         {
1612             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1613
1614             zebra_sort_type(si, ord);
1615             if (cmd == 1)
1616                 zebra_sort_add(si, str, slen);
1617             else
1618                 zebra_sort_delete(si);
1619         }
1620     }
1621 }
1622
1623 /*
1624  * Local variables:
1625  * c-basic-offset: 4
1626  * indent-tabs-mode: nil
1627  * End:
1628  * vim: shiftwidth=4 tabstop=8 expandtab
1629  */
1630