index/extract.c

   1 /* $Id: extract.c,v 1.273 2007-12-13 11:09:20 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 /** \file
  24     \brief indexes records and extract tokens for indexing and sorting
  25 */
  26
  27 #include <stdio.h>
  28 #include <assert.h>
  29 #include <ctype.h>
  30 #ifdef WIN32
  31 #include <io.h>
  32 #endif
  33 #if HAVE_UNISTD_H
  34 #include <unistd.h>
  35 #endif
  36 #include <fcntl.h>
  37
  38
  39 #include "index.h"
  40 #include "orddict.h"
  41 #include <direntz.h>
  42 #include <charmap.h>
  43 #include <yaz/snprintf.h>
  44
  45 static int log_level_extract = 0;
  46 static int log_level_details = 0;
  47 static int log_level_initialized = 0;
  48
  49 /* 1 if we use eliminitate identical delete/insert keys */
  50 /* eventually this the 0-case code will be removed */
  51 #define FLUSH2 1
  52
  53 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  54                                 zebra_rec_keys_t ins_keys,
  55                                 zint ins_rank,
  56                                 zebra_rec_keys_t del_keys,
  57                                 zint del_rank);
  58
  59 static void zebra_init_log_level(void)
  60 {
  61     if (!log_level_initialized)
  62     {
  63         log_level_initialized = 1;
  64
  65         log_level_extract = yaz_log_module_level("extract");
  66         log_level_details = yaz_log_module_level("indexdetails");
  67     }
  68 }
  69
  70 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  71                                     int cmd, zebra_rec_keys_t skp);
  72 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid);
  73 static void extract_token_add(RecWord *p);
  74
  75 static void check_log_limit(ZebraHandle zh)
  76 {
  77     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  78     {
  79         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  80                 zh->m_file_verbose_limit);
  81     }
  82 }
  83
  84 static void logRecord(ZebraHandle zh)
  85 {
  86     check_log_limit(zh);
  87     ++zh->records_processed;
  88     if (!(zh->records_processed % 1000))
  89     {
  90         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  91                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  92                 zh->records_processed, zh->records_inserted,
  93                 zh->records_updated, zh->records_deleted);
  94     }
  95 }
  96
  97 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
  98 {
  99     ctrl->flagShowRecords = !zh->m_flag_rw;
 100 }
 101
 102
 103 static void extract_add_index_string(RecWord *p,
 104                                       zinfo_index_category_t cat,
 105                                       const char *str, int length);
 106
 107 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
 108
 109 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 110 {
 111     w->seqno = 1;
 112     w->index_name = "any";
 113     w->index_type = "w";
 114     w->extractCtrl = p;
 115     w->record_id = 0;
 116     w->section_id = 0;
 117     w->segment = 0;
 118 }
 119
 120 struct snip_rec_info {
 121     ZebraHandle zh;
 122     zebra_snippets *snippets;
 123 };
 124
 125
 126 static void snippet_add_complete_field(RecWord *p, int ord,
 127                                        zebra_map_t zm)
 128 {
 129     struct snip_rec_info *h = p->extractCtrl->handle;
 130
 131     const char *b = p->term_buf;
 132     char buf[IT_MAX_WORD+1];
 133     const char **map = 0;
 134     int i = 0, remain = p->term_len;
 135     const char *start = b;
 136     const char *last = 0;
 137
 138     if (remain > 0)
 139         map = zebra_maps_input(zm, &b, remain, 1);
 140
 141     while (remain > 0 && i < IT_MAX_WORD)
 142     {
 143         while (map && *map && **map == *CHR_SPACE)
 144         {
 145             remain = p->term_len - (b - p->term_buf);
 146
 147             if (i == 0)
 148                 start = b;  /* set to first non-ws area */
 149             if (remain > 0)
 150             {
 151                 int first = i ? 0 : 1;  /* first position */
 152
 153                 map = zebra_maps_input(zm, &b, remain, first);
 154             }
 155             else
 156                 map = 0;
 157         }
 158         if (!map)
 159             break;
 160
 161         if (i && i < IT_MAX_WORD)
 162             buf[i++] = *CHR_SPACE;
 163         while (map && *map && **map != *CHR_SPACE)
 164         {
 165             const char *cp = *map;
 166
 167             if (**map == *CHR_CUT)
 168             {
 169                 i = 0;
 170             }
 171             else
 172             {
 173                 if (i >= IT_MAX_WORD)
 174                     break;
 175                 while (i < IT_MAX_WORD && *cp)
 176                     buf[i++] = *(cp++);
 177             }
 178             last = b;
 179             remain = p->term_len  - (b - p->term_buf);
 180             if (remain > 0)
 181             {
 182                 map = zebra_maps_input(zm, &b, remain, 0);
 183             }
 184             else
 185                 map = 0;
 186         }
 187     }
 188     if (!i)
 189         return;
 190     if (last && start != last)
 191         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 192                                start, last - start);
 193 }
 194
 195 static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 196 {
 197     struct snip_rec_info *h = p->extractCtrl->handle;
 198     const char *b = p->term_buf;
 199     int remain = p->term_len;
 200     int first = 1;
 201     const char **map = 0;
 202     const char *start = b;
 203     const char *last = b;
 204
 205     if (remain > 0)
 206         map = zebra_maps_input(zm, &b, remain, 0);
 207
 208     while (map)
 209     {
 210         char buf[IT_MAX_WORD+1];
 211         int i, remain;
 212
 213         /* Skip spaces */
 214         while (map && *map && **map == *CHR_SPACE)
 215         {
 216             remain = p->term_len - (b - p->term_buf);
 217             last = b;
 218             if (remain > 0)
 219                 map = zebra_maps_input(zm, &b, remain, 0);
 220             else
 221                 map = 0;
 222         }
 223         if (!map)
 224             break;
 225         if (start != last)
 226         {
 227             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
 228                                    start, last - start);
 229
 230         }
 231         start = last;
 232
 233         i = 0;
 234         while (map && *map && **map != *CHR_SPACE)
 235         {
 236             const char *cp = *map;
 237
 238             while (i < IT_MAX_WORD && *cp)
 239                 buf[i++] = *(cp++);
 240             remain = p->term_len - (b - p->term_buf);
 241             last = b;
 242             if (remain > 0)
 243                 map = zebra_maps_input(zm, &b, remain, 0);
 244             else
 245                 map = 0;
 246         }
 247         if (!i)
 248             return;
 249
 250         if (first)
 251         {
 252             first = 0;
 253             if (zebra_maps_is_first_in_field(zm))
 254             {
 255                 /* first in field marker */
 256                 p->seqno++;
 257             }
 258         }
 259         if (start != last)
 260             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 261                                    start, last - start);
 262         start = last;
 263         p->seqno++;
 264     }
 265
 266 }
 267
 268 static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
 269 {
 270     struct snip_rec_info *h = p->extractCtrl->handle;
 271
 272     const char *res_buf = 0;
 273     size_t res_len = 0;
 274
 275     const char *display_buf = 0;
 276     size_t display_len = 0;
 277
 278     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
 279     while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
 280                                    &display_buf, &display_len))
 281     {
 282         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 283                                display_buf, display_len);
 284         p->seqno++;
 285     }
 286 }
 287
 288 static void snippet_token_add(RecWord *p)
 289 {
 290     struct snip_rec_info *h = p->extractCtrl->handle;
 291     ZebraHandle zh = h->zh;
 292     zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
 293
 294     if (zm && zebra_maps_is_index(zm))
 295     {
 296         ZebraExplainInfo zei = zh->reg->zei;
 297         int ch = zebraExplain_lookup_attr_str(
 298             zei, zinfo_index_category_index, p->index_type, p->index_name);
 299
 300         if (zebra_maps_is_icu(zm))
 301             snippet_add_icu(p, ch, zm);
 302         else
 303         {
 304             if (zebra_maps_is_complete(zm))
 305                 snippet_add_complete_field(p, ch, zm);
 306             else
 307                 snippet_add_incomplete_field(p, ch, zm);
 308         }
 309     }
 310 }
 311
 312 static void snippet_schema_add(
 313     struct recExtractCtrl *p, Odr_oid *oid)
 314 {
 315
 316 }
 317
 318 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 319                      struct ZebraRecStream *stream,
 320                      RecType rt, void *recTypeClientData)
 321 {
 322     struct recExtractCtrl extractCtrl;
 323     struct snip_rec_info info;
 324     int r;
 325
 326     extractCtrl.stream = stream;
 327     extractCtrl.first_record = 1;
 328     extractCtrl.init = extract_init;
 329     extractCtrl.tokenAdd = snippet_token_add;
 330     extractCtrl.schemaAdd = snippet_schema_add;
 331     assert(zh->reg);
 332     assert(zh->reg->dh);
 333
 334     extractCtrl.dh = zh->reg->dh;
 335
 336     info.zh = zh;
 337     info.snippets = sn;
 338     extractCtrl.handle = &info;
 339     extractCtrl.match_criteria[0] = '\0';
 340     extractCtrl.staticrank = 0;
 341     extractCtrl.action = action_insert;
 342
 343     init_extractCtrl(zh, &extractCtrl);
 344
 345     extractCtrl.setStoreData = 0;
 346
 347     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 348
 349 }
 350
 351 static void searchRecordKey(ZebraHandle zh,
 352                             zebra_rec_keys_t reckeys,
 353                             const char *index_name,
 354                             const char **ws, int ws_length)
 355 {
 356     int i;
 357     int ch = -1;
 358     zinfo_index_category_t cat = zinfo_index_category_index;
 359
 360     for (i = 0; i<ws_length; i++)
 361         ws[i] = NULL;
 362
 363     if (ch < 0)
 364         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
 365     if (ch < 0)
 366         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
 367     if (ch < 0)
 368         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 369
 370     if (ch < 0)
 371         return ;
 372
 373     if (zebra_rec_keys_rewind(reckeys))
 374     {
 375         zint startSeq = -1;
 376         const char *str;
 377         size_t slen;
 378         struct it_key key;
 379         zint seqno;
 380         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 381         {
 382             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 383
 384             seqno = key.mem[key.len-1];
 385
 386             if (key.mem[0] == ch)
 387             {
 388                 zint woff;
 389
 390                 if (startSeq == -1)
 391                     startSeq = seqno;
 392                 woff = seqno - startSeq;
 393                 if (woff >= 0 && woff < ws_length)
 394                     ws[woff] = str;
 395             }
 396         }
 397     }
 398 }
 399
 400 #define FILE_MATCH_BLANK "\t "
 401
 402 static char *get_match_from_spec(ZebraHandle zh,
 403                           zebra_rec_keys_t reckeys,
 404                           const char *fname, const char *spec)
 405 {
 406     static char dstBuf[2048];      /* static here ??? */
 407     char *dst = dstBuf;
 408     const char *s = spec;
 409
 410     while (1)
 411     {
 412         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 413             ;
 414         if (!*s)
 415             break;
 416         if (*s == '(')
 417         {
 418             const char *ws[32];
 419             char attset_str[64], attname_str[64];
 420             int i;
 421             int first = 1;
 422
 423             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 424                 ;
 425             for (i = 0; *s && *s != ',' && *s != ')' &&
 426                      !strchr(FILE_MATCH_BLANK, *s); s++)
 427                 if (i+1 < sizeof(attset_str))
 428                     attset_str[i++] = *s;
 429             attset_str[i] = '\0';
 430
 431             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 432                 ;
 433             if (*s != ',')
 434                 strcpy(attname_str, attset_str);
 435             else
 436             {
 437                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 438                     ;
 439                 for (i = 0; *s && *s != ')' &&
 440                          !strchr(FILE_MATCH_BLANK, *s); s++)
 441                     if (i+1 < sizeof(attname_str))
 442                         attname_str[i++] = *s;
 443                 attname_str[i] = '\0';
 444             }
 445
 446             searchRecordKey(zh, reckeys, attname_str, ws, 32);
 447
 448             if (*s != ')')
 449             {
 450                 yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
 451                       spec, zh->m_group ? zh->m_group : "none");
 452                 return NULL;
 453             }
 454             s++;
 455
 456             for (i = 0; i<32; i++)
 457                 if (ws[i])
 458                 {
 459                     if (first)
 460                     {
 461                         *dst++ = ' ';
 462                         first = 0;
 463                     }
 464                     strcpy(dst, ws[i]);
 465                     dst += strlen(ws[i]);
 466                 }
 467             if (first)
 468             {
 469                 yaz_log(YLOG_WARN, "Record didn't contain match"
 470                       " fields in (%s,%s)", attset_str, attname_str);
 471                 return NULL;
 472             }
 473         }
 474         else if (*s == '$')
 475         {
 476             int spec_len;
 477             char special[64];
 478             const char *spec_src = NULL;
 479             const char *s1 = ++s;
 480             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 481                 s1++;
 482
 483             spec_len = s1 - s;
 484             if (spec_len > sizeof(special)-1)
 485                 spec_len = sizeof(special)-1;
 486             memcpy(special, s, spec_len);
 487             special[spec_len] = '\0';
 488             s = s1;
 489
 490             if (!strcmp(special, "group"))
 491                 spec_src = zh->m_group;
 492             else if (!strcmp(special, "database"))
 493                 spec_src = zh->basenames[0];
 494             else if (!strcmp(special, "filename")) {
 495                 spec_src = fname;
 496             }
 497             else if (!strcmp(special, "type"))
 498                 spec_src = zh->m_record_type;
 499             else
 500                 spec_src = NULL;
 501             if (spec_src)
 502             {
 503                 strcpy(dst, spec_src);
 504                 dst += strlen(spec_src);
 505             }
 506         }
 507         else if (*s == '\"' || *s == '\'')
 508         {
 509             int stopMarker = *s++;
 510             char tmpString[64];
 511             int i = 0;
 512
 513             while (*s && *s != stopMarker)
 514             {
 515                 if (i+1 < sizeof(tmpString))
 516                     tmpString[i++] = *s++;
 517             }
 518             if (*s)
 519                 s++;
 520             tmpString[i] = '\0';
 521             strcpy(dst, tmpString);
 522             dst += strlen(tmpString);
 523         }
 524         else
 525         {
 526             yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
 527                   spec, zh->m_group ? zh->m_group : "none");
 528             return NULL;
 529         }
 530         *dst++ = 1;
 531     }
 532     if (dst == dstBuf)
 533     {
 534         yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
 535               fname, zh->m_group ? zh->m_group : "none");
 536         return NULL;
 537     }
 538     *dst = '\0';
 539     return dstBuf;
 540 }
 541
 542 struct recordLogInfo {
 543     const char *fname;
 544     int recordOffset;
 545     struct recordGroup *rGroup;
 546 };
 547
 548 /** \brief add the always-matches index entry and map to real record ID
 549     \param ctrl record control
 550     \param record_id custom record ID
 551     \param sysno system record ID
 552
 553     This function serves two purposes.. It adds the always matches
 554     entry and makes a pointer from the custom record ID (if defined)
 555     back to the system record ID (sysno)
 556     See zebra_recid_to_sysno .
 557   */
 558 static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
 559                             zint sysno)
 560 {
 561     RecWord word;
 562     extract_init(ctrl, &word);
 563     word.record_id = record_id;
 564     /* we use the seqno as placeholder for a way to get back to
 565        record database from _ALLRECORDS.. This is used if a custom
 566        RECORD was defined */
 567     word.seqno = sysno;
 568     word.index_name = "_ALLRECORDS";
 569     word.index_type = "w";
 570
 571     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
 572                               "", 0);
 573 }
 574
 575 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 576                                        struct ZebraRecStream *stream,
 577                                        enum zebra_recctrl_action_t action,
 578                                        int test_mode,
 579                                        const char *recordType,
 580                                        zint *sysno,
 581                                        const char *match_criteria,
 582                                        const char *fname,
 583                                        RecType recType,
 584                                        void *recTypeClientData);
 585
 586
 587 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 588                              int deleteFlag)
 589 {
 590     ZEBRA_RES r = ZEBRA_OK;
 591     int i, fd;
 592     char gprefix[128];
 593     char ext[128];
 594     char ext_res[128];
 595     struct file_read_info *fi = 0;
 596     const char *original_record_type = 0;
 597     RecType recType;
 598     void *recTypeClientData;
 599     struct ZebraRecStream stream, *streamp;
 600
 601     zebra_init_log_level();
 602
 603     if (!zh->m_group || !*zh->m_group)
 604         *gprefix = '\0';
 605     else
 606         sprintf(gprefix, "%s.", zh->m_group);
 607
 608     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 609
 610     /* determine file extension */
 611     *ext = '\0';
 612     for (i = strlen(fname); --i >= 0; )
 613         if (fname[i] == '/')
 614             break;
 615         else if (fname[i] == '.')
 616         {
 617             strcpy(ext, fname+i+1);
 618             break;
 619         }
 620     /* determine file type - depending on extension */
 621     original_record_type = zh->m_record_type;
 622     if (!zh->m_record_type)
 623     {
 624         sprintf(ext_res, "%srecordType.%s", gprefix, ext);
 625         zh->m_record_type = res_get(zh->res, ext_res);
 626     }
 627     if (!zh->m_record_type)
 628     {
 629         check_log_limit(zh);
 630         if (zh->records_processed + zh->records_skipped
 631             < zh->m_file_verbose_limit)
 632             yaz_log(YLOG_LOG, "? %s", fname);
 633         zh->records_skipped++;
 634         return 0;
 635     }
 636     /* determine match criteria */
 637     if (!zh->m_record_id)
 638     {
 639         sprintf(ext_res, "%srecordId.%s", gprefix, ext);
 640         zh->m_record_id = res_get(zh->res, ext_res);
 641     }
 642
 643     if (!(recType =
 644           recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
 645                           &recTypeClientData)))
 646     {
 647         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 648         return ZEBRA_FAIL;
 649     }
 650
 651     switch(recType->version)
 652     {
 653     case 0:
 654         break;
 655     default:
 656         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 657     }
 658     if (sysno && deleteFlag)
 659     {
 660         streamp = 0;
 661         fi = 0;
 662     }
 663     else
 664     {
 665         char full_rep[1024];
 666
 667         if (zh->path_reg && !yaz_is_abspath(fname))
 668         {
 669             strcpy(full_rep, zh->path_reg);
 670             strcat(full_rep, "/");
 671             strcat(full_rep, fname);
 672         }
 673         else
 674             strcpy(full_rep, fname);
 675
 676         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
 677         {
 678             yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 679             zh->m_record_type = original_record_type;
 680             return ZEBRA_FAIL;
 681         }
 682         streamp = &stream;
 683         zebra_create_stream_fd(streamp, fd, 0);
 684     }
 685     r = zebra_extract_records_stream(zh, streamp,
 686                                      deleteFlag ?
 687                                      action_delete : action_update,
 688                                      0, /* tst_mode */
 689                                      zh->m_record_type,
 690                                      sysno,
 691                                      0, /*match_criteria */
 692                                      fname,
 693                                      recType, recTypeClientData);
 694     if (streamp)
 695         stream.destroy(streamp);
 696     zh->m_record_type = original_record_type;
 697     return r;
 698 }
 699
 700 /*
 701   If sysno is provided, then it's used to identify the reocord.
 702   If not, and match_criteria is provided, then sysno is guessed
 703   If not, and a record is provided, then sysno is got from there
 704
 705  */
 706
 707 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 708                                       const char *buf, size_t buf_size,
 709                                       enum zebra_recctrl_action_t action,
 710                                       int test_mode,
 711                                       const char *recordType,
 712                                       zint *sysno,
 713                                       const char *match_criteria,
 714                                       const char *fname)
 715 {
 716     struct ZebraRecStream stream;
 717     ZEBRA_RES res;
 718     void *clientData;
 719     RecType recType = 0;
 720
 721     if (recordType && *recordType)
 722     {
 723         yaz_log(log_level_extract,
 724                 "Record type explicitly specified: %s", recordType);
 725         recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
 726                                   &clientData);
 727     }
 728     else
 729     {
 730         if (!(zh->m_record_type))
 731         {
 732             yaz_log(YLOG_WARN, "No such record type defined");
 733             return ZEBRA_FAIL;
 734         }
 735         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 736                 zh->m_record_type);
 737         recType = recType_byName(zh->reg->recTypes, zh->res,
 738                                   zh->m_record_type, &clientData);
 739         recordType = zh->m_record_type;
 740     }
 741
 742     if (!recType)
 743     {
 744         yaz_log(YLOG_WARN, "No such record type: %s", recordType);
 745         return ZEBRA_FAIL;
 746     }
 747
 748     zebra_create_stream_mem(&stream, buf, buf_size);
 749
 750     res = zebra_extract_records_stream(zh, &stream,
 751                                        action,
 752                                        test_mode,
 753                                        recordType,
 754                                        sysno,
 755                                        match_criteria,
 756                                        fname,
 757                                        recType, clientData);
 758     stream.destroy(&stream);
 759     return res;
 760 }
 761
 762 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 763                                        struct ZebraRecStream *stream,
 764                                        enum zebra_recctrl_action_t action,
 765                                        int test_mode,
 766                                        const char *recordType,
 767                                        zint *sysno,
 768                                        const char *match_criteria,
 769                                        const char *fname,
 770                                        RecType recType,
 771                                        void *recTypeClientData)
 772 {
 773     ZEBRA_RES res = ZEBRA_OK;
 774     while (1)
 775     {
 776         int more = 0;
 777         res = zebra_extract_record_stream(zh, stream,
 778                                           action,
 779                                           test_mode,
 780                                           recordType,
 781                                           sysno,
 782                                           match_criteria,
 783                                           fname,
 784                                           recType, recTypeClientData, &more);
 785         if (!more)
 786         {
 787             res = ZEBRA_OK;
 788             break;
 789         }
 790         if (res != ZEBRA_OK)
 791             break;
 792         if (sysno)
 793             break;
 794     }
 795     return res;
 796 }
 797
 798
 799 static WRBUF wrbuf_hex_str(const char *cstr)
 800 {
 801     size_t i;
 802     WRBUF w = wrbuf_alloc();
 803     for (i = 0; cstr[i]; i++)
 804     {
 805         if (cstr[i] < ' ' || cstr[i] > 126)
 806             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 807         else
 808             wrbuf_putc(w, cstr[i]);
 809     }
 810     return w;
 811 }
 812
 813 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 814                                       struct ZebraRecStream *stream,
 815                                       enum zebra_recctrl_action_t action,
 816                                       int test_mode,
 817                                       const char *recordType,
 818                                       zint *sysno,
 819                                       const char *match_criteria,
 820                                       const char *fname,
 821                                       RecType recType,
 822                                       void *recTypeClientData,
 823                                       int *more)
 824
 825 {
 826     zint sysno0 = 0;
 827     RecordAttr *recordAttr;
 828     struct recExtractCtrl extractCtrl;
 829     int r;
 830     const char *matchStr = 0;
 831     Record rec;
 832     off_t start_offset = 0, end_offset = 0;
 833     const char *pr_fname = fname;  /* filename to print .. */
 834     int show_progress = zh->records_processed + zh->records_skipped
 835         < zh->m_file_verbose_limit ? 1:0;
 836
 837     zebra_init_log_level();
 838
 839     if (!pr_fname)
 840         pr_fname = "<no file>";  /* make it printable if file is omitted */
 841
 842     zebra_rec_keys_reset(zh->reg->keys);
 843     zebra_rec_keys_reset(zh->reg->sortKeys);
 844
 845     if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
 846     {
 847         if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
 848                                       zh->m_explain_database))
 849             return ZEBRA_FAIL;
 850     }
 851
 852     if (stream)
 853     {
 854         off_t null_offset = 0;
 855         extractCtrl.stream = stream;
 856
 857         start_offset = stream->tellf(stream);
 858
 859         extractCtrl.first_record = start_offset ? 0 : 1;
 860
 861         stream->endf(stream, &null_offset);;
 862
 863         extractCtrl.init = extract_init;
 864         extractCtrl.tokenAdd = extract_token_add;
 865         extractCtrl.schemaAdd = extract_schema_add;
 866         extractCtrl.dh = zh->reg->dh;
 867         extractCtrl.handle = zh;
 868         extractCtrl.match_criteria[0] = '\0';
 869         extractCtrl.staticrank = 0;
 870         extractCtrl.action = action;
 871
 872         init_extractCtrl(zh, &extractCtrl);
 873
 874         extract_set_store_data_prepare(&extractCtrl);
 875
 876         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 877
 878         if (action == action_update)
 879         {
 880             action = extractCtrl.action;
 881         }
 882
 883         switch (r)
 884         {
 885         case RECCTRL_EXTRACT_EOF:
 886             return ZEBRA_FAIL;
 887         case RECCTRL_EXTRACT_ERROR_GENERIC:
 888             /* error occured during extraction ... */
 889             yaz_log(YLOG_WARN, "extract error: generic");
 890             return ZEBRA_FAIL;
 891         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 892             /* error occured during extraction ... */
 893             yaz_log(YLOG_WARN, "extract error: no such filter");
 894             return ZEBRA_FAIL;
 895         case RECCTRL_EXTRACT_SKIP:
 896             if (show_progress)
 897                 yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 898                          recordType, pr_fname, (zint) start_offset);
 899             *more = 1;
 900
 901             end_offset = stream->endf(stream, 0);
 902             if (end_offset)
 903                 stream->seekf(stream, end_offset);
 904
 905             return ZEBRA_OK;
 906         case RECCTRL_EXTRACT_OK:
 907             break;
 908         default:
 909             yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
 910             return ZEBRA_FAIL;
 911         }
 912         end_offset = stream->endf(stream, 0);
 913         if (end_offset)
 914             stream->seekf(stream, end_offset);
 915         else
 916             end_offset = stream->tellf(stream);
 917
 918         if (extractCtrl.match_criteria[0])
 919             match_criteria = extractCtrl.match_criteria;
 920     }
 921
 922     *more = 1;
 923     if (!sysno)
 924     {
 925         sysno = &sysno0;
 926
 927         if (match_criteria && *match_criteria) {
 928             matchStr = match_criteria;
 929         } else {
 930             if (zh->m_record_id && *zh->m_record_id) {
 931                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 932                                                zh->m_record_id);
 933                 if (!matchStr)
 934                 {
 935                     yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 936                              pr_fname, (zint) start_offset);
 937                     return ZEBRA_FAIL;
 938                 }
 939             }
 940         }
 941         if (matchStr)
 942         {
 943             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 944             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 945                                           matchStr);
 946
 947
 948             if (log_level_extract)
 949             {
 950                 WRBUF w = wrbuf_hex_str(matchStr);
 951                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 952                 wrbuf_destroy(w);
 953             }
 954             if (rinfo)
 955             {
 956                 assert(*rinfo == sizeof(*sysno));
 957                 memcpy(sysno, rinfo+1, sizeof(*sysno));
 958             }
 959        }
 960     }
 961
 962     if (zebra_rec_keys_empty(zh->reg->keys))
 963     {
 964         /* the extraction process returned no information - the record
 965            is probably empty - unless flagShowRecords is in use */
 966         if (test_mode)
 967             return ZEBRA_OK;
 968     }
 969
 970     if (! *sysno)
 971     {
 972         /* new record */
 973         if (action == action_delete)
 974         {
 975             yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 976                          pr_fname, (zint) start_offset);
 977             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
 978             return ZEBRA_FAIL;
 979         }
 980         else if (action == action_replace)
 981         {
 982             yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 983                          pr_fname, (zint) start_offset);
 984             yaz_log(YLOG_WARN, "cannot update record above (seems new)");
 985             return ZEBRA_FAIL;
 986         }
 987         if (show_progress)
 988             yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 989                      (zint) start_offset);
 990         rec = rec_new(zh->reg->records);
 991
 992         *sysno = rec->sysno;
 993
 994
 995         if (stream)
 996         {
 997             all_matches_add(&extractCtrl,
 998                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
 999                             *sysno);
1000         }
1001
1002
1003         recordAttr = rec_init_attr(zh->reg->zei, rec);
1004         if (extractCtrl.staticrank < 0)
1005         {
1006             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
1007             extractCtrl.staticrank = 0;
1008         }
1009
1010         if (matchStr)
1011         {
1012             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1013             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
1014                             sizeof(*sysno), sysno);
1015         }
1016
1017         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1018 #if FLUSH2
1019         extract_flush_record_keys2(zh, *sysno,
1020                                    zh->reg->keys, extractCtrl.staticrank,
1021                                    0, recordAttr->staticrank);
1022 #else
1023         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
1024                                   extractCtrl.staticrank);
1025 #endif
1026         recordAttr->staticrank = extractCtrl.staticrank;
1027         zh->records_inserted++;
1028     }
1029     else
1030     {
1031         /* record already exists */
1032         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1033         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1034         if (action == action_insert)
1035         {
1036             yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
1037                          recordType, pr_fname, (zint) start_offset);
1038             logRecord(zh);
1039             return ZEBRA_FAIL;
1040         }
1041
1042         rec = rec_get(zh->reg->records, *sysno);
1043         assert(rec);
1044
1045         if (stream)
1046         {
1047             all_matches_add(&extractCtrl,
1048                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1049                             *sysno);
1050         }
1051
1052         recordAttr = rec_init_attr(zh->reg->zei, rec);
1053
1054         /* decrease total size */
1055         zebraExplain_recordBytesIncrement(zh->reg->zei,
1056                                            - recordAttr->recordSize);
1057
1058         zebra_rec_keys_set_buf(delkeys,
1059                                rec->info[recInfo_delKeys],
1060                                rec->size[recInfo_delKeys],
1061                                0);
1062         zebra_rec_keys_set_buf(sortKeys,
1063                                rec->info[recInfo_sortKeys],
1064                                rec->size[recInfo_sortKeys],
1065                                0);
1066
1067         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1068 #if !FLUSH2
1069         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1070                                   recordAttr->staticrank);
1071 #endif
1072         if (action == action_delete)
1073         {
1074             /* record going to be deleted */
1075 #if FLUSH2
1076             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1077                                        delkeys, recordAttr->staticrank);
1078 #endif
1079             if (zebra_rec_keys_empty(delkeys))
1080             {
1081                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1082                         pr_fname, (zint) start_offset);
1083                 yaz_log(YLOG_WARN, "cannot delete file above, "
1084                         "storeKeys false (3)");
1085             }
1086             else
1087             {
1088                 if (show_progress)
1089                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1090                             pr_fname, (zint) start_offset);
1091                 zh->records_deleted++;
1092                 if (matchStr)
1093                 {
1094                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1095                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1096                 }
1097                 rec_del(zh->reg->records, &rec);
1098             }
1099             zebra_rec_keys_close(delkeys);
1100             zebra_rec_keys_close(sortKeys);
1101             rec_free(&rec);
1102             logRecord(zh);
1103             return ZEBRA_OK;
1104         }
1105         else
1106         {   /* update or special_update */
1107             if (show_progress)
1108                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1109                         pr_fname, (zint) start_offset);
1110             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1111
1112 #if FLUSH2
1113             extract_flush_record_keys2(zh, *sysno,
1114                                        zh->reg->keys, extractCtrl.staticrank,
1115                                        delkeys, recordAttr->staticrank);
1116 #else
1117             extract_flush_record_keys(zh, *sysno, 1,
1118                                       zh->reg->keys, extractCtrl.staticrank);
1119 #endif
1120             recordAttr->staticrank = extractCtrl.staticrank;
1121             zh->records_updated++;
1122         }
1123         zebra_rec_keys_close(delkeys);
1124         zebra_rec_keys_close(sortKeys);
1125     }
1126     /* update file type */
1127     xfree(rec->info[recInfo_fileType]);
1128     rec->info[recInfo_fileType] =
1129         rec_strdup(recordType, &rec->size[recInfo_fileType]);
1130
1131     /* update filename */
1132     xfree(rec->info[recInfo_filename]);
1133     rec->info[recInfo_filename] =
1134         rec_strdup(fname, &rec->size[recInfo_filename]);
1135
1136     /* update delete keys */
1137     xfree(rec->info[recInfo_delKeys]);
1138     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1139     {
1140         zebra_rec_keys_get_buf(zh->reg->keys,
1141                                &rec->info[recInfo_delKeys],
1142                                &rec->size[recInfo_delKeys]);
1143     }
1144     else
1145     {
1146         rec->info[recInfo_delKeys] = NULL;
1147         rec->size[recInfo_delKeys] = 0;
1148     }
1149     /* update sort keys */
1150     xfree(rec->info[recInfo_sortKeys]);
1151
1152     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1153                            &rec->info[recInfo_sortKeys],
1154                            &rec->size[recInfo_sortKeys]);
1155
1156     if (stream)
1157     {
1158         recordAttr->recordSize = end_offset - start_offset;
1159         zebraExplain_recordBytesIncrement(zh->reg->zei,
1160                                           recordAttr->recordSize);
1161     }
1162
1163     /* set run-number for this record */
1164     recordAttr->runNumber =
1165         zebraExplain_runNumberIncrement(zh->reg->zei, 0);
1166
1167     /* update store data */
1168     xfree(rec->info[recInfo_storeData]);
1169
1170     /* update store data */
1171     if (zh->store_data_buf)
1172     {
1173         rec->size[recInfo_storeData] = zh->store_data_size;
1174         rec->info[recInfo_storeData] = zh->store_data_buf;
1175         zh->store_data_buf = 0;
1176         recordAttr->recordSize = zh->store_data_size;
1177     }
1178     else if (zh->m_store_data)
1179     {
1180         off_t cur_offset = stream->tellf(stream);
1181
1182         rec->size[recInfo_storeData] = recordAttr->recordSize;
1183         rec->info[recInfo_storeData] = (char *)
1184             xmalloc(recordAttr->recordSize);
1185         stream->seekf(stream, start_offset);
1186         stream->readf(stream, rec->info[recInfo_storeData],
1187                       recordAttr->recordSize);
1188         stream->seekf(stream, cur_offset);
1189     }
1190     else
1191     {
1192         rec->info[recInfo_storeData] = NULL;
1193         rec->size[recInfo_storeData] = 0;
1194     }
1195     /* update database name */
1196     xfree(rec->info[recInfo_databaseName]);
1197     rec->info[recInfo_databaseName] =
1198         rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
1199
1200     /* update offset */
1201     recordAttr->recordOffset = start_offset;
1202
1203     /* commit this record */
1204     rec_put(zh->reg->records, &rec);
1205     logRecord(zh);
1206     return ZEBRA_OK;
1207 }
1208
1209 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1210 {
1211     ZebraHandle zh = (ZebraHandle) handle;
1212     struct recExtractCtrl extractCtrl;
1213
1214     if (zebraExplain_curDatabase(zh->reg->zei,
1215                                   rec->info[recInfo_databaseName]))
1216     {
1217         abort();
1218         if (zebraExplain_newDatabase(zh->reg->zei,
1219                                       rec->info[recInfo_databaseName], 0))
1220             abort();
1221     }
1222
1223     zebra_rec_keys_reset(zh->reg->keys);
1224     zebra_rec_keys_reset(zh->reg->sortKeys);
1225
1226     extractCtrl.init = extract_init;
1227     extractCtrl.tokenAdd = extract_token_add;
1228     extractCtrl.schemaAdd = extract_schema_add;
1229     extractCtrl.dh = zh->reg->dh;
1230
1231     init_extractCtrl(zh, &extractCtrl);
1232
1233     extractCtrl.flagShowRecords = 0;
1234     extractCtrl.match_criteria[0] = '\0';
1235     extractCtrl.staticrank = 0;
1236     extractCtrl.action = action_update;
1237
1238     extractCtrl.handle = handle;
1239     extractCtrl.first_record = 1;
1240
1241     extract_set_store_data_prepare(&extractCtrl);
1242
1243     if (n)
1244         grs_extract_tree(&extractCtrl, n);
1245
1246     if (rec->size[recInfo_delKeys])
1247     {
1248         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1249
1250         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1251
1252         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1253                                rec->size[recInfo_delKeys],
1254                                0);
1255 #if FLUSH2
1256         extract_flush_record_keys2(zh, rec->sysno,
1257                                    zh->reg->keys, 0, delkeys, 0);
1258 #else
1259         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1260         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1261 #endif
1262         zebra_rec_keys_close(delkeys);
1263
1264         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1265                                rec->size[recInfo_sortKeys],
1266                                0);
1267
1268         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1269         zebra_rec_keys_close(sortkeys);
1270     }
1271     else
1272     {
1273 #if FLUSH2
1274         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1275 #else
1276         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1277 #endif
1278     }
1279     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1280
1281     xfree(rec->info[recInfo_delKeys]);
1282     zebra_rec_keys_get_buf(zh->reg->keys,
1283                            &rec->info[recInfo_delKeys],
1284                            &rec->size[recInfo_delKeys]);
1285
1286     xfree(rec->info[recInfo_sortKeys]);
1287     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1288                            &rec->info[recInfo_sortKeys],
1289                            &rec->size[recInfo_sortKeys]);
1290     return ZEBRA_OK;
1291 }
1292
1293 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1294                           zebra_rec_keys_t reckeys,
1295                           int level)
1296 {
1297     if (zebra_rec_keys_rewind(reckeys))
1298     {
1299         size_t slen;
1300         const char *str;
1301         struct it_key key;
1302         NMEM nmem = nmem_create();
1303
1304         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1305         {
1306             char keystr[200]; /* room for zints to print */
1307             char *dst_term = 0;
1308             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1309             const char *index_type;
1310             int i;
1311             const char *string_index;
1312
1313             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1314                                     0/* db */, &string_index);
1315             assert(index_type);
1316             zebra_term_untrans_iconv(zh, nmem, index_type,
1317                                      &dst_term, str);
1318             *keystr = '\0';
1319             for (i = 0; i<key.len; i++)
1320             {
1321                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1322             }
1323
1324             if (*str < CHR_BASE_CHAR)
1325             {
1326                 int i;
1327                 char dst_buf[200]; /* room for special chars */
1328
1329                 strcpy(dst_buf , "?");
1330
1331                 if (!strcmp(str, ""))
1332                     strcpy(dst_buf, "alwaysmatches");
1333                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1334                     strcpy(dst_buf, "firstinfield");
1335                 else if (!strcmp(str, CHR_UNKNOWN))
1336                     strcpy(dst_buf, "unknown");
1337                 else if (!strcmp(str, CHR_SPACE))
1338                     strcpy(dst_buf, "space");
1339
1340                 for (i = 0; i<slen; i++)
1341                 {
1342                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1343                 }
1344                 yaz_log(level, "%s%s %s %s", keystr, index_type,
1345                         string_index, dst_buf);
1346
1347             }
1348             else
1349                 yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
1350                         string_index, dst_term);
1351
1352             nmem_reset(nmem);
1353         }
1354         nmem_destroy(nmem);
1355     }
1356 }
1357
1358 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1359                              zebra_rec_keys_t reckeys)
1360 {
1361     ZebraExplainInfo zei = zh->reg->zei;
1362     struct ord_stat {
1363         int no;
1364         int ord;
1365         struct ord_stat *next;
1366     };
1367
1368     if (zebra_rec_keys_rewind(reckeys))
1369     {
1370         struct ord_stat *ord_list = 0;
1371         struct ord_stat *p;
1372         size_t slen;
1373         const char *str;
1374         struct it_key key_in;
1375         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1376         {
1377             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1378
1379             for (p = ord_list; p ; p = p->next)
1380                 if (p->ord == ord)
1381                 {
1382                     p->no++;
1383                     break;
1384                 }
1385             if (!p)
1386             {
1387                 p = xmalloc(sizeof(*p));
1388                 p->no = 1;
1389                 p->ord = ord;
1390                 p->next = ord_list;
1391                 ord_list = p;
1392             }
1393         }
1394
1395         p = ord_list;
1396         while (p)
1397         {
1398             struct ord_stat *p1 = p;
1399
1400             if (is_insert)
1401                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1402             else
1403                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1404             p = p->next;
1405             xfree(p1);
1406         }
1407     }
1408 }
1409
1410 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1411                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1412                                 zebra_rec_keys_t del_keys, zint del_rank)
1413 {
1414     ZebraExplainInfo zei = zh->reg->zei;
1415     int normal = 0;
1416     int optimized = 0;
1417
1418     if (!zh->reg->key_block)
1419     {
1420         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1421         const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
1422         int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
1423         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1424     }
1425
1426     if (ins_keys)
1427     {
1428         extract_rec_keys_adjust(zh, 1, ins_keys);
1429         if (!del_keys)
1430             zebraExplain_recordCountIncrement(zei, 1);
1431         zebra_rec_keys_rewind(ins_keys);
1432     }
1433     if (del_keys)
1434     {
1435         extract_rec_keys_adjust(zh, 0, del_keys);
1436         if (!ins_keys)
1437             zebraExplain_recordCountIncrement(zei, -1);
1438         zebra_rec_keys_rewind(del_keys);
1439     }
1440
1441     while (1)
1442     {
1443         size_t del_slen;
1444         const char *del_str;
1445         struct it_key del_key_in;
1446         int del = 0;
1447
1448         size_t ins_slen;
1449         const char *ins_str;
1450         struct it_key ins_key_in;
1451         int ins = 0;
1452
1453         if (del_keys)
1454             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1455                                       &del_key_in);
1456         if (ins_keys)
1457             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1458                                       &ins_key_in);
1459
1460         if (del && ins && ins_rank == del_rank
1461             && !key_compare(&del_key_in, &ins_key_in)
1462             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1463         {
1464             optimized++;
1465             continue;
1466         }
1467         if (!del && !ins)
1468             break;
1469
1470         normal++;
1471         if (del)
1472             key_block_write(zh->reg->key_block, sysno,
1473                             &del_key_in, 0, del_str, del_slen,
1474                             del_rank, zh->m_staticrank);
1475         if (ins)
1476             key_block_write(zh->reg->key_block, sysno,
1477                             &ins_key_in, 1, ins_str, ins_slen,
1478                             ins_rank, zh->m_staticrank);
1479     }
1480     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1481 }
1482
1483
1484 ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
1485                                      zebra_rec_keys_t reckeys,
1486                                      zebra_snippets *snippets)
1487 {
1488     NMEM nmem = nmem_create();
1489     if (zebra_rec_keys_rewind(reckeys))
1490     {
1491         const char *str;
1492         size_t slen;
1493         struct it_key key;
1494         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1495         {
1496             char *dst_term = 0;
1497             int ord;
1498             zint seqno;
1499             const char *index_type;
1500
1501             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1502             seqno = key.mem[key.len-1];
1503             ord = CAST_ZINT_TO_INT(key.mem[0]);
1504
1505             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1506                                     0/* db */, 0 /* string_index */);
1507             assert(index_type);
1508             zebra_term_untrans_iconv(zh, nmem, index_type,
1509                                      &dst_term, str);
1510             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1511             nmem_reset(nmem);
1512         }
1513     }
1514     nmem_destroy(nmem);
1515     return ZEBRA_OK;
1516 }
1517
1518 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1519 {
1520     yaz_log(YLOG_LOG, "print_rec_keys");
1521     if (zebra_rec_keys_rewind(reckeys))
1522     {
1523         const char *str;
1524         size_t slen;
1525         struct it_key key;
1526         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1527         {
1528             char dst_buf[IT_MAX_WORD];
1529             zint seqno;
1530             const char *index_type;
1531             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1532             const char *db = 0;
1533             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1534
1535             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1536
1537             seqno = key.mem[key.len-1];
1538
1539             zebra_term_untrans(zh, index_type, dst_buf, str);
1540
1541             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1542                     " term=%s", ord, seqno, dst_buf);
1543         }
1544     }
1545 }
1546
1547 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1548                                      const char *str, int length)
1549 {
1550     struct it_key key;
1551     ZebraHandle zh = p->extractCtrl->handle;
1552     ZebraExplainInfo zei = zh->reg->zei;
1553     int ch, i;
1554
1555     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1556     if (ch < 0)
1557         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1558
1559     i = 0;
1560     key.mem[i++] = ch;
1561     key.mem[i++] = p->record_id;
1562     key.mem[i++] = p->section_id;
1563
1564     if (zh->m_segment_indexing)
1565         key.mem[i++] = p->segment;
1566     key.mem[i++] = p->seqno;
1567     key.len = i;
1568
1569     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1570 }
1571
1572 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1573 {
1574     struct it_key key;
1575     ZebraHandle zh = p->extractCtrl->handle;
1576     ZebraExplainInfo zei = zh->reg->zei;
1577     int ch;
1578     zinfo_index_category_t cat = zinfo_index_category_sort;
1579
1580     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1581     if (ch < 0)
1582         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1583     key.len = 2;
1584     key.mem[0] = ch;
1585     key.mem[1] = p->record_id;
1586
1587     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1588 }
1589
1590 static void extract_add_staticrank_string(RecWord *p,
1591                                           const char *str, int length)
1592 {
1593     char valz[40];
1594     struct recExtractCtrl *ctrl = p->extractCtrl;
1595
1596     if (length > sizeof(valz)-1)
1597         length = sizeof(valz)-1;
1598
1599     memcpy(valz, str, length);
1600     valz[length] = '\0';
1601     ctrl->staticrank = atozint(valz);
1602 }
1603
1604 static void extract_add_string(RecWord *p, zebra_map_t zm,
1605                                const char *string, int length)
1606 {
1607     assert(length > 0);
1608
1609     if (!p->index_name)
1610         return;
1611
1612     if (zebra_maps_is_index(zm))
1613     {
1614         extract_add_index_string(p, zinfo_index_category_index,
1615                                  string, length);
1616         if (zebra_maps_is_alwaysmatches(zm))
1617         {
1618             RecWord word;
1619             memcpy(&word, p, sizeof(word));
1620
1621             word.seqno = 1;
1622             extract_add_index_string(
1623                 &word, zinfo_index_category_alwaysmatches, "", 0);
1624         }
1625     }
1626     else if (zebra_maps_is_sort(zm))
1627     {
1628         extract_add_sort_string(p, string, length);
1629     }
1630     else if (zebra_maps_is_staticrank(zm))
1631     {
1632         extract_add_staticrank_string(p, string, length);
1633     }
1634 }
1635
1636 static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
1637 {
1638     const char *b = p->term_buf;
1639     int remain = p->term_len;
1640     int first = 1;
1641     const char **map = 0;
1642
1643     if (remain > 0)
1644         map = zebra_maps_input(zm, &b, remain, 0);
1645
1646     while (map)
1647     {
1648         char buf[IT_MAX_WORD+1];
1649         int i, remain;
1650
1651         /* Skip spaces */
1652         while (map && *map && **map == *CHR_SPACE)
1653         {
1654             remain = p->term_len - (b - p->term_buf);
1655             if (remain > 0)
1656                 map = zebra_maps_input(zm, &b, remain, 0);
1657             else
1658                 map = 0;
1659         }
1660         if (!map)
1661             break;
1662         i = 0;
1663         while (map && *map && **map != *CHR_SPACE)
1664         {
1665             const char *cp = *map;
1666
1667             while (i < IT_MAX_WORD && *cp)
1668                 buf[i++] = *(cp++);
1669             remain = p->term_len - (b - p->term_buf);
1670             if (remain > 0)
1671                 map = zebra_maps_input(zm, &b, remain, 0);
1672             else
1673                 map = 0;
1674         }
1675         if (!i)
1676             return;
1677
1678         if (first)
1679         {
1680             first = 0;
1681             if (zebra_maps_is_first_in_field(zm))
1682             {
1683                 /* first in field marker */
1684                 extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1685                 p->seqno++;
1686             }
1687         }
1688         extract_add_string(p, zm, buf, i);
1689         p->seqno++;
1690     }
1691 }
1692
1693 static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
1694 {
1695     const char *b = p->term_buf;
1696     char buf[IT_MAX_WORD+1];
1697     const char **map = 0;
1698     int i = 0, remain = p->term_len;
1699
1700     if (remain > 0)
1701         map = zebra_maps_input(zm, &b, remain, 1);
1702
1703     while (remain > 0 && i < IT_MAX_WORD)
1704     {
1705         while (map && *map && **map == *CHR_SPACE)
1706         {
1707             remain = p->term_len - (b - p->term_buf);
1708
1709             if (remain > 0)
1710             {
1711                 int first = i ? 0 : 1;  /* first position */
1712                 map = zebra_maps_input(zm, &b, remain, first);
1713             }
1714             else
1715                 map = 0;
1716         }
1717         if (!map)
1718             break;
1719
1720         if (i && i < IT_MAX_WORD)
1721             buf[i++] = *CHR_SPACE;
1722         while (map && *map && **map != *CHR_SPACE)
1723         {
1724             const char *cp = *map;
1725
1726             if (**map == *CHR_CUT)
1727             {
1728                 i = 0;
1729             }
1730             else
1731             {
1732                 if (i >= IT_MAX_WORD)
1733                     break;
1734                 while (i < IT_MAX_WORD && *cp)
1735                     buf[i++] = *(cp++);
1736             }
1737             remain = p->term_len  - (b - p->term_buf);
1738             if (remain > 0)
1739             {
1740                 map = zebra_maps_input(zm, &b, remain, 0);
1741             }
1742             else
1743                 map = 0;
1744         }
1745     }
1746     if (!i)
1747         return;
1748     extract_add_string(p, zm, buf, i);
1749 }
1750
1751 static void extract_add_icu(RecWord *p, zebra_map_t zm)
1752 {
1753     struct it_key key;
1754     const char *res_buf = 0;
1755     size_t res_len = 0;
1756     ZebraHandle zh = p->extractCtrl->handle;
1757
1758     int cat = zinfo_index_category_index;
1759     int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat,
1760                                           p->index_type, p->index_name);
1761     if (ch < 0)
1762         ch = zebraExplain_add_attr_str(zh->reg->zei, cat,
1763                                        p->index_type, p->index_name);
1764     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
1765     while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
1766     {
1767         int i = 0;
1768         key.mem[i++] = ch;
1769         key.mem[i++] = p->record_id;
1770         key.mem[i++] = p->section_id;
1771
1772         if (zh->m_segment_indexing)
1773             key.mem[i++] = p->segment;
1774         key.mem[i++] = p->seqno;
1775         key.len = i;
1776
1777         zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
1778
1779         p->seqno++;
1780     }
1781 }
1782
1783
1784 /** \brief top-level indexing handler for recctrl system
1785     \param p token data to be indexed
1786
1787     Call sequence:
1788     extract_token_add
1789     extract_add_{in}_complete
1790     extract_add_string
1791
1792     extract_add_index_string
1793     or
1794     extract_add_sort_string
1795     or
1796     extract_add_staticrank_string
1797
1798 */
1799 static void extract_token_add(RecWord *p)
1800 {
1801     ZebraHandle zh = p->extractCtrl->handle;
1802     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
1803     WRBUF wrbuf;
1804
1805     if (log_level_details)
1806     {
1807         yaz_log(log_level_details, "extract_token_add "
1808                 "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
1809                 p->index_type, p->index_name,
1810                 p->seqno, p->term_len, p->term_buf);
1811     }
1812     if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
1813     {
1814         p->term_buf = wrbuf_buf(wrbuf);
1815         p->term_len = wrbuf_len(wrbuf);
1816     }
1817     if (zebra_maps_is_icu(zm))
1818     {
1819         extract_add_icu(p, zm);
1820     }
1821     else
1822     {
1823         if (zebra_maps_is_complete(zm))
1824             extract_add_complete_field(p, zm);
1825         else
1826             extract_add_incomplete_field(p, zm);
1827     }
1828 }
1829
1830 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1831                                       void *buf, size_t sz)
1832 {
1833     ZebraHandle zh = (ZebraHandle) p->handle;
1834
1835     xfree(zh->store_data_buf);
1836     zh->store_data_buf = 0;
1837     zh->store_data_size = 0;
1838     if (buf && sz)
1839     {
1840         zh->store_data_buf = xmalloc(sz);
1841         zh->store_data_size = sz;
1842         memcpy(zh->store_data_buf, buf, sz);
1843     }
1844 }
1845
1846 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1847 {
1848     ZebraHandle zh = (ZebraHandle) p->handle;
1849     xfree(zh->store_data_buf);
1850     zh->store_data_buf = 0;
1851     zh->store_data_size = 0;
1852     p->setStoreData = extract_set_store_data_cb;
1853 }
1854
1855 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1856 {
1857     ZebraHandle zh = (ZebraHandle) p->handle;
1858     zebraExplain_addSchema(zh->reg->zei, oid);
1859 }
1860
1861 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1862                              int cmd, zebra_rec_keys_t reckeys)
1863 {
1864 #if 0
1865     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1866             cmd, sysno);
1867     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1868 #endif
1869
1870     if (zebra_rec_keys_rewind(reckeys))
1871     {
1872         zebra_sort_index_t si = zh->reg->sort_index;
1873         size_t slen;
1874         const char *str;
1875         struct it_key key_in;
1876
1877         zebra_sort_sysno(si, sysno);
1878
1879         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1880         {
1881             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1882
1883             zebra_sort_type(si, ord);
1884             if (cmd == 1)
1885                 zebra_sort_add(si, str, slen);
1886             else
1887                 zebra_sort_delete(si);
1888         }
1889     }
1890 }
1891
1892 /*
1893  * Local variables:
1894  * c-basic-offset: 4
1895  * indent-tabs-mode: nil
1896  * End:
1897  * vim: shiftwidth=4 tabstop=8 expandtab
1898  */
1899