index/extract.c

   1 /* $Id: extract.c,v 1.270 2007-11-30 12:19:08 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_token_add(RecWord *p)
 269 {
 270     struct snip_rec_info *h = p->extractCtrl->handle;
 271     ZebraHandle zh = h->zh;
 272     zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
 273
 274     if (zm && zebra_maps_is_index(zm))
 275     {
 276         ZebraExplainInfo zei = zh->reg->zei;
 277         int ch = zebraExplain_lookup_attr_str(
 278             zei, zinfo_index_category_index, p->index_type, p->index_name);
 279
 280         if (zebra_maps_is_complete(zm))
 281             snippet_add_complete_field(p, ch, zm);
 282         else
 283             snippet_add_incomplete_field(p, ch, zm);
 284     }
 285 }
 286
 287 static void snippet_schema_add(
 288     struct recExtractCtrl *p, Odr_oid *oid)
 289 {
 290
 291 }
 292
 293 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 294                      struct ZebraRecStream *stream,
 295                      RecType rt, void *recTypeClientData)
 296 {
 297     struct recExtractCtrl extractCtrl;
 298     struct snip_rec_info info;
 299     int r;
 300
 301     extractCtrl.stream = stream;
 302     extractCtrl.first_record = 1;
 303     extractCtrl.init = extract_init;
 304     extractCtrl.tokenAdd = snippet_token_add;
 305     extractCtrl.schemaAdd = snippet_schema_add;
 306     assert(zh->reg);
 307     assert(zh->reg->dh);
 308
 309     extractCtrl.dh = zh->reg->dh;
 310
 311     info.zh = zh;
 312     info.snippets = sn;
 313     extractCtrl.handle = &info;
 314     extractCtrl.match_criteria[0] = '\0';
 315     extractCtrl.staticrank = 0;
 316     extractCtrl.action = action_insert;
 317
 318     init_extractCtrl(zh, &extractCtrl);
 319
 320     extractCtrl.setStoreData = 0;
 321
 322     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 323
 324 }
 325
 326 static void searchRecordKey(ZebraHandle zh,
 327                             zebra_rec_keys_t reckeys,
 328                             const char *index_name,
 329                             const char **ws, int ws_length)
 330 {
 331     int i;
 332     int ch = -1;
 333     zinfo_index_category_t cat = zinfo_index_category_index;
 334
 335     for (i = 0; i<ws_length; i++)
 336         ws[i] = NULL;
 337
 338     if (ch < 0)
 339         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
 340     if (ch < 0)
 341         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
 342     if (ch < 0)
 343         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 344
 345     if (ch < 0)
 346         return ;
 347
 348     if (zebra_rec_keys_rewind(reckeys))
 349     {
 350         zint startSeq = -1;
 351         const char *str;
 352         size_t slen;
 353         struct it_key key;
 354         zint seqno;
 355         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 356         {
 357             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 358
 359             seqno = key.mem[key.len-1];
 360
 361             if (key.mem[0] == ch)
 362             {
 363                 zint woff;
 364
 365                 if (startSeq == -1)
 366                     startSeq = seqno;
 367                 woff = seqno - startSeq;
 368                 if (woff >= 0 && woff < ws_length)
 369                     ws[woff] = str;
 370             }
 371         }
 372     }
 373 }
 374
 375 #define FILE_MATCH_BLANK "\t "
 376
 377 static char *get_match_from_spec(ZebraHandle zh,
 378                           zebra_rec_keys_t reckeys,
 379                           const char *fname, const char *spec)
 380 {
 381     static char dstBuf[2048];      /* static here ??? */
 382     char *dst = dstBuf;
 383     const char *s = spec;
 384
 385     while (1)
 386     {
 387         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 388             ;
 389         if (!*s)
 390             break;
 391         if (*s == '(')
 392         {
 393             const char *ws[32];
 394             char attset_str[64], attname_str[64];
 395             int i;
 396             int first = 1;
 397
 398             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 399                 ;
 400             for (i = 0; *s && *s != ',' && *s != ')' &&
 401                      !strchr(FILE_MATCH_BLANK, *s); s++)
 402                 if (i+1 < sizeof(attset_str))
 403                     attset_str[i++] = *s;
 404             attset_str[i] = '\0';
 405
 406             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 407                 ;
 408             if (*s != ',')
 409                 strcpy(attname_str, attset_str);
 410             else
 411             {
 412                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 413                     ;
 414                 for (i = 0; *s && *s != ')' &&
 415                          !strchr(FILE_MATCH_BLANK, *s); s++)
 416                     if (i+1 < sizeof(attname_str))
 417                         attname_str[i++] = *s;
 418                 attname_str[i] = '\0';
 419             }
 420
 421             searchRecordKey(zh, reckeys, attname_str, ws, 32);
 422
 423             if (*s != ')')
 424             {
 425                 yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
 426                       spec, zh->m_group ? zh->m_group : "none");
 427                 return NULL;
 428             }
 429             s++;
 430
 431             for (i = 0; i<32; i++)
 432                 if (ws[i])
 433                 {
 434                     if (first)
 435                     {
 436                         *dst++ = ' ';
 437                         first = 0;
 438                     }
 439                     strcpy(dst, ws[i]);
 440                     dst += strlen(ws[i]);
 441                 }
 442             if (first)
 443             {
 444                 yaz_log(YLOG_WARN, "Record didn't contain match"
 445                       " fields in (%s,%s)", attset_str, attname_str);
 446                 return NULL;
 447             }
 448         }
 449         else if (*s == '$')
 450         {
 451             int spec_len;
 452             char special[64];
 453             const char *spec_src = NULL;
 454             const char *s1 = ++s;
 455             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 456                 s1++;
 457
 458             spec_len = s1 - s;
 459             if (spec_len > sizeof(special)-1)
 460                 spec_len = sizeof(special)-1;
 461             memcpy(special, s, spec_len);
 462             special[spec_len] = '\0';
 463             s = s1;
 464
 465             if (!strcmp(special, "group"))
 466                 spec_src = zh->m_group;
 467             else if (!strcmp(special, "database"))
 468                 spec_src = zh->basenames[0];
 469             else if (!strcmp(special, "filename")) {
 470                 spec_src = fname;
 471             }
 472             else if (!strcmp(special, "type"))
 473                 spec_src = zh->m_record_type;
 474             else
 475                 spec_src = NULL;
 476             if (spec_src)
 477             {
 478                 strcpy(dst, spec_src);
 479                 dst += strlen(spec_src);
 480             }
 481         }
 482         else if (*s == '\"' || *s == '\'')
 483         {
 484             int stopMarker = *s++;
 485             char tmpString[64];
 486             int i = 0;
 487
 488             while (*s && *s != stopMarker)
 489             {
 490                 if (i+1 < sizeof(tmpString))
 491                     tmpString[i++] = *s++;
 492             }
 493             if (*s)
 494                 s++;
 495             tmpString[i] = '\0';
 496             strcpy(dst, tmpString);
 497             dst += strlen(tmpString);
 498         }
 499         else
 500         {
 501             yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
 502                   spec, zh->m_group ? zh->m_group : "none");
 503             return NULL;
 504         }
 505         *dst++ = 1;
 506     }
 507     if (dst == dstBuf)
 508     {
 509         yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
 510               fname, zh->m_group ? zh->m_group : "none");
 511         return NULL;
 512     }
 513     *dst = '\0';
 514     return dstBuf;
 515 }
 516
 517 struct recordLogInfo {
 518     const char *fname;
 519     int recordOffset;
 520     struct recordGroup *rGroup;
 521 };
 522
 523 /** \brief add the always-matches index entry and map to real record ID
 524     \param ctrl record control
 525     \param record_id custom record ID
 526     \param sysno system record ID
 527
 528     This function serves two purposes.. It adds the always matches
 529     entry and makes a pointer from the custom record ID (if defined)
 530     back to the system record ID (sysno)
 531     See zebra_recid_to_sysno .
 532   */
 533 static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
 534                             zint sysno)
 535 {
 536     RecWord word;
 537     extract_init(ctrl, &word);
 538     word.record_id = record_id;
 539     /* we use the seqno as placeholder for a way to get back to
 540        record database from _ALLRECORDS.. This is used if a custom
 541        RECORD was defined */
 542     word.seqno = sysno;
 543     word.index_name = "_ALLRECORDS";
 544     word.index_type = "w";
 545
 546     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
 547                               "", 0);
 548 }
 549
 550 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 551                                        struct ZebraRecStream *stream,
 552                                        enum zebra_recctrl_action_t action,
 553                                        int test_mode,
 554                                        const char *recordType,
 555                                        zint *sysno,
 556                                        const char *match_criteria,
 557                                        const char *fname,
 558                                        RecType recType,
 559                                        void *recTypeClientData);
 560
 561
 562 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 563                              int deleteFlag)
 564 {
 565     ZEBRA_RES r = ZEBRA_OK;
 566     int i, fd;
 567     char gprefix[128];
 568     char ext[128];
 569     char ext_res[128];
 570     struct file_read_info *fi = 0;
 571     const char *original_record_type = 0;
 572     RecType recType;
 573     void *recTypeClientData;
 574     struct ZebraRecStream stream, *streamp;
 575
 576     zebra_init_log_level();
 577
 578     if (!zh->m_group || !*zh->m_group)
 579         *gprefix = '\0';
 580     else
 581         sprintf(gprefix, "%s.", zh->m_group);
 582
 583     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 584
 585     /* determine file extension */
 586     *ext = '\0';
 587     for (i = strlen(fname); --i >= 0; )
 588         if (fname[i] == '/')
 589             break;
 590         else if (fname[i] == '.')
 591         {
 592             strcpy(ext, fname+i+1);
 593             break;
 594         }
 595     /* determine file type - depending on extension */
 596     original_record_type = zh->m_record_type;
 597     if (!zh->m_record_type)
 598     {
 599         sprintf(ext_res, "%srecordType.%s", gprefix, ext);
 600         zh->m_record_type = res_get(zh->res, ext_res);
 601     }
 602     if (!zh->m_record_type)
 603     {
 604         check_log_limit(zh);
 605         if (zh->records_processed + zh->records_skipped
 606             < zh->m_file_verbose_limit)
 607             yaz_log(YLOG_LOG, "? %s", fname);
 608         zh->records_skipped++;
 609         return 0;
 610     }
 611     /* determine match criteria */
 612     if (!zh->m_record_id)
 613     {
 614         sprintf(ext_res, "%srecordId.%s", gprefix, ext);
 615         zh->m_record_id = res_get(zh->res, ext_res);
 616     }
 617
 618     if (!(recType =
 619           recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
 620                           &recTypeClientData)))
 621     {
 622         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 623         return ZEBRA_FAIL;
 624     }
 625
 626     switch(recType->version)
 627     {
 628     case 0:
 629         break;
 630     default:
 631         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 632     }
 633     if (sysno && deleteFlag)
 634     {
 635         streamp = 0;
 636         fi = 0;
 637     }
 638     else
 639     {
 640         char full_rep[1024];
 641
 642         if (zh->path_reg && !yaz_is_abspath(fname))
 643         {
 644             strcpy(full_rep, zh->path_reg);
 645             strcat(full_rep, "/");
 646             strcat(full_rep, fname);
 647         }
 648         else
 649             strcpy(full_rep, fname);
 650
 651         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
 652         {
 653             yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 654             zh->m_record_type = original_record_type;
 655             return ZEBRA_FAIL;
 656         }
 657         streamp = &stream;
 658         zebra_create_stream_fd(streamp, fd, 0);
 659     }
 660     r = zebra_extract_records_stream(zh, streamp,
 661                                      deleteFlag ?
 662                                      action_delete : action_update,
 663                                      0, /* tst_mode */
 664                                      zh->m_record_type,
 665                                      sysno,
 666                                      0, /*match_criteria */
 667                                      fname,
 668                                      recType, recTypeClientData);
 669     if (streamp)
 670         stream.destroy(streamp);
 671     zh->m_record_type = original_record_type;
 672     return r;
 673 }
 674
 675 /*
 676   If sysno is provided, then it's used to identify the reocord.
 677   If not, and match_criteria is provided, then sysno is guessed
 678   If not, and a record is provided, then sysno is got from there
 679
 680  */
 681
 682 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 683                                       const char *buf, size_t buf_size,
 684                                       enum zebra_recctrl_action_t action,
 685                                       int test_mode,
 686                                       const char *recordType,
 687                                       zint *sysno,
 688                                       const char *match_criteria,
 689                                       const char *fname)
 690 {
 691     struct ZebraRecStream stream;
 692     ZEBRA_RES res;
 693     void *clientData;
 694     RecType recType = 0;
 695
 696     if (recordType && *recordType)
 697     {
 698         yaz_log(log_level_extract,
 699                 "Record type explicitly specified: %s", recordType);
 700         recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
 701                                   &clientData);
 702     }
 703     else
 704     {
 705         if (!(zh->m_record_type))
 706         {
 707             yaz_log(YLOG_WARN, "No such record type defined");
 708             return ZEBRA_FAIL;
 709         }
 710         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 711                 zh->m_record_type);
 712         recType = recType_byName(zh->reg->recTypes, zh->res,
 713                                   zh->m_record_type, &clientData);
 714         recordType = zh->m_record_type;
 715     }
 716
 717     if (!recType)
 718     {
 719         yaz_log(YLOG_WARN, "No such record type: %s", recordType);
 720         return ZEBRA_FAIL;
 721     }
 722
 723     zebra_create_stream_mem(&stream, buf, buf_size);
 724
 725     res = zebra_extract_records_stream(zh, &stream,
 726                                        action,
 727                                        test_mode,
 728                                        recordType,
 729                                        sysno,
 730                                        match_criteria,
 731                                        fname,
 732                                        recType, clientData);
 733     stream.destroy(&stream);
 734     return res;
 735 }
 736
 737 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 738                                        struct ZebraRecStream *stream,
 739                                        enum zebra_recctrl_action_t action,
 740                                        int test_mode,
 741                                        const char *recordType,
 742                                        zint *sysno,
 743                                        const char *match_criteria,
 744                                        const char *fname,
 745                                        RecType recType,
 746                                        void *recTypeClientData)
 747 {
 748     ZEBRA_RES res = ZEBRA_OK;
 749     while (1)
 750     {
 751         int more = 0;
 752         res = zebra_extract_record_stream(zh, stream,
 753                                           action,
 754                                           test_mode,
 755                                           recordType,
 756                                           sysno,
 757                                           match_criteria,
 758                                           fname,
 759                                           recType, recTypeClientData, &more);
 760         if (!more)
 761         {
 762             res = ZEBRA_OK;
 763             break;
 764         }
 765         if (res != ZEBRA_OK)
 766             break;
 767         if (sysno)
 768             break;
 769     }
 770     return res;
 771 }
 772
 773
 774 static WRBUF wrbuf_hex_str(const char *cstr)
 775 {
 776     size_t i;
 777     WRBUF w = wrbuf_alloc();
 778     for (i = 0; cstr[i]; i++)
 779     {
 780         if (cstr[i] < ' ' || cstr[i] > 126)
 781             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 782         else
 783             wrbuf_putc(w, cstr[i]);
 784     }
 785     return w;
 786 }
 787
 788 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 789                                       struct ZebraRecStream *stream,
 790                                       enum zebra_recctrl_action_t action,
 791                                       int test_mode,
 792                                       const char *recordType,
 793                                       zint *sysno,
 794                                       const char *match_criteria,
 795                                       const char *fname,
 796                                       RecType recType,
 797                                       void *recTypeClientData,
 798                                       int *more)
 799
 800 {
 801     zint sysno0 = 0;
 802     RecordAttr *recordAttr;
 803     struct recExtractCtrl extractCtrl;
 804     int r;
 805     const char *matchStr = 0;
 806     Record rec;
 807     off_t start_offset = 0, end_offset = 0;
 808     const char *pr_fname = fname;  /* filename to print .. */
 809     int show_progress = zh->records_processed + zh->records_skipped
 810         < zh->m_file_verbose_limit ? 1:0;
 811
 812     zebra_init_log_level();
 813
 814     if (!pr_fname)
 815         pr_fname = "<no file>";  /* make it printable if file is omitted */
 816
 817     zebra_rec_keys_reset(zh->reg->keys);
 818     zebra_rec_keys_reset(zh->reg->sortKeys);
 819
 820     if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
 821     {
 822         if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
 823                                       zh->m_explain_database))
 824             return ZEBRA_FAIL;
 825     }
 826
 827     if (stream)
 828     {
 829         off_t null_offset = 0;
 830         extractCtrl.stream = stream;
 831
 832         start_offset = stream->tellf(stream);
 833
 834         extractCtrl.first_record = start_offset ? 0 : 1;
 835
 836         stream->endf(stream, &null_offset);;
 837
 838         extractCtrl.init = extract_init;
 839         extractCtrl.tokenAdd = extract_token_add;
 840         extractCtrl.schemaAdd = extract_schema_add;
 841         extractCtrl.dh = zh->reg->dh;
 842         extractCtrl.handle = zh;
 843         extractCtrl.match_criteria[0] = '\0';
 844         extractCtrl.staticrank = 0;
 845         extractCtrl.action = action;
 846
 847         init_extractCtrl(zh, &extractCtrl);
 848
 849         extract_set_store_data_prepare(&extractCtrl);
 850
 851         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 852
 853         if (action == action_update)
 854         {
 855             action = extractCtrl.action;
 856         }
 857
 858         switch (r)
 859         {
 860         case RECCTRL_EXTRACT_EOF:
 861             return ZEBRA_FAIL;
 862         case RECCTRL_EXTRACT_ERROR_GENERIC:
 863             /* error occured during extraction ... */
 864             yaz_log(YLOG_WARN, "extract error: generic");
 865             return ZEBRA_FAIL;
 866         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 867             /* error occured during extraction ... */
 868             yaz_log(YLOG_WARN, "extract error: no such filter");
 869             return ZEBRA_FAIL;
 870         case RECCTRL_EXTRACT_SKIP:
 871             if (show_progress)
 872                 yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 873                          recordType, pr_fname, (zint) start_offset);
 874             *more = 1;
 875
 876             end_offset = stream->endf(stream, 0);
 877             if (end_offset)
 878                 stream->seekf(stream, end_offset);
 879
 880             return ZEBRA_OK;
 881         case RECCTRL_EXTRACT_OK:
 882             break;
 883         default:
 884             yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
 885             return ZEBRA_FAIL;
 886         }
 887         end_offset = stream->endf(stream, 0);
 888         if (end_offset)
 889             stream->seekf(stream, end_offset);
 890         else
 891             end_offset = stream->tellf(stream);
 892
 893         if (extractCtrl.match_criteria[0])
 894             match_criteria = extractCtrl.match_criteria;
 895     }
 896
 897     *more = 1;
 898     if (!sysno)
 899     {
 900         sysno = &sysno0;
 901
 902         if (match_criteria && *match_criteria) {
 903             matchStr = match_criteria;
 904         } else {
 905             if (zh->m_record_id && *zh->m_record_id) {
 906                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 907                                                zh->m_record_id);
 908                 if (!matchStr)
 909                 {
 910                     yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 911                              pr_fname, (zint) start_offset);
 912                     return ZEBRA_FAIL;
 913                 }
 914             }
 915         }
 916         if (matchStr)
 917         {
 918             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 919             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 920                                           matchStr);
 921
 922
 923             if (log_level_extract)
 924             {
 925                 WRBUF w = wrbuf_hex_str(matchStr);
 926                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 927                 wrbuf_destroy(w);
 928             }
 929             if (rinfo)
 930             {
 931                 assert(*rinfo == sizeof(*sysno));
 932                 memcpy(sysno, rinfo+1, sizeof(*sysno));
 933             }
 934        }
 935     }
 936
 937     if (zebra_rec_keys_empty(zh->reg->keys))
 938     {
 939         /* the extraction process returned no information - the record
 940            is probably empty - unless flagShowRecords is in use */
 941         if (test_mode)
 942             return ZEBRA_OK;
 943     }
 944
 945     if (! *sysno)
 946     {
 947         /* new record */
 948         if (action == action_delete)
 949         {
 950             yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 951                          pr_fname, (zint) start_offset);
 952             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
 953             return ZEBRA_FAIL;
 954         }
 955         else if (action == action_replace)
 956         {
 957             yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 958                          pr_fname, (zint) start_offset);
 959             yaz_log(YLOG_WARN, "cannot update record above (seems new)");
 960             return ZEBRA_FAIL;
 961         }
 962         if (show_progress)
 963             yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 964                      (zint) start_offset);
 965         rec = rec_new(zh->reg->records);
 966
 967         *sysno = rec->sysno;
 968
 969
 970         if (stream)
 971         {
 972             all_matches_add(&extractCtrl,
 973                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
 974                             *sysno);
 975         }
 976
 977
 978         recordAttr = rec_init_attr(zh->reg->zei, rec);
 979         if (extractCtrl.staticrank < 0)
 980         {
 981             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
 982             extractCtrl.staticrank = 0;
 983         }
 984
 985         if (matchStr)
 986         {
 987             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 988             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
 989                             sizeof(*sysno), sysno);
 990         }
 991
 992         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 993 #if FLUSH2
 994         extract_flush_record_keys2(zh, *sysno,
 995                                    zh->reg->keys, extractCtrl.staticrank,
 996                                    0, recordAttr->staticrank);
 997 #else
 998         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 999                                   extractCtrl.staticrank);
1000 #endif
1001         recordAttr->staticrank = extractCtrl.staticrank;
1002         zh->records_inserted++;
1003     }
1004     else
1005     {
1006         /* record already exists */
1007         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1008         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1009         if (action == action_insert)
1010         {
1011             yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
1012                          recordType, pr_fname, (zint) start_offset);
1013             logRecord(zh);
1014             return ZEBRA_FAIL;
1015         }
1016
1017         rec = rec_get(zh->reg->records, *sysno);
1018         assert(rec);
1019
1020         if (stream)
1021         {
1022             all_matches_add(&extractCtrl,
1023                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1024                             *sysno);
1025         }
1026
1027         recordAttr = rec_init_attr(zh->reg->zei, rec);
1028
1029         /* decrease total size */
1030         zebraExplain_recordBytesIncrement(zh->reg->zei,
1031                                            - recordAttr->recordSize);
1032
1033         zebra_rec_keys_set_buf(delkeys,
1034                                rec->info[recInfo_delKeys],
1035                                rec->size[recInfo_delKeys],
1036                                0);
1037         zebra_rec_keys_set_buf(sortKeys,
1038                                rec->info[recInfo_sortKeys],
1039                                rec->size[recInfo_sortKeys],
1040                                0);
1041
1042         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1043 #if !FLUSH2
1044         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1045                                   recordAttr->staticrank);
1046 #endif
1047         if (action == action_delete)
1048         {
1049             /* record going to be deleted */
1050 #if FLUSH2
1051             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1052                                        delkeys, recordAttr->staticrank);
1053 #endif
1054             if (zebra_rec_keys_empty(delkeys))
1055             {
1056                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1057                         pr_fname, (zint) start_offset);
1058                 yaz_log(YLOG_WARN, "cannot delete file above, "
1059                         "storeKeys false (3)");
1060             }
1061             else
1062             {
1063                 if (show_progress)
1064                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1065                             pr_fname, (zint) start_offset);
1066                 zh->records_deleted++;
1067                 if (matchStr)
1068                 {
1069                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1070                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1071                 }
1072                 rec_del(zh->reg->records, &rec);
1073             }
1074             zebra_rec_keys_close(delkeys);
1075             zebra_rec_keys_close(sortKeys);
1076             rec_free(&rec);
1077             logRecord(zh);
1078             return ZEBRA_OK;
1079         }
1080         else
1081         {   /* update or special_update */
1082             if (show_progress)
1083                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1084                         pr_fname, (zint) start_offset);
1085             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1086
1087 #if FLUSH2
1088             extract_flush_record_keys2(zh, *sysno,
1089                                        zh->reg->keys, extractCtrl.staticrank,
1090                                        delkeys, recordAttr->staticrank);
1091 #else
1092             extract_flush_record_keys(zh, *sysno, 1,
1093                                       zh->reg->keys, extractCtrl.staticrank);
1094 #endif
1095             recordAttr->staticrank = extractCtrl.staticrank;
1096             zh->records_updated++;
1097         }
1098         zebra_rec_keys_close(delkeys);
1099         zebra_rec_keys_close(sortKeys);
1100     }
1101     /* update file type */
1102     xfree(rec->info[recInfo_fileType]);
1103     rec->info[recInfo_fileType] =
1104         rec_strdup(recordType, &rec->size[recInfo_fileType]);
1105
1106     /* update filename */
1107     xfree(rec->info[recInfo_filename]);
1108     rec->info[recInfo_filename] =
1109         rec_strdup(fname, &rec->size[recInfo_filename]);
1110
1111     /* update delete keys */
1112     xfree(rec->info[recInfo_delKeys]);
1113     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1114     {
1115         zebra_rec_keys_get_buf(zh->reg->keys,
1116                                &rec->info[recInfo_delKeys],
1117                                &rec->size[recInfo_delKeys]);
1118     }
1119     else
1120     {
1121         rec->info[recInfo_delKeys] = NULL;
1122         rec->size[recInfo_delKeys] = 0;
1123     }
1124     /* update sort keys */
1125     xfree(rec->info[recInfo_sortKeys]);
1126
1127     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1128                            &rec->info[recInfo_sortKeys],
1129                            &rec->size[recInfo_sortKeys]);
1130
1131     if (stream)
1132     {
1133         recordAttr->recordSize = end_offset - start_offset;
1134         zebraExplain_recordBytesIncrement(zh->reg->zei,
1135                                           recordAttr->recordSize);
1136     }
1137
1138     /* set run-number for this record */
1139     recordAttr->runNumber =
1140         zebraExplain_runNumberIncrement(zh->reg->zei, 0);
1141
1142     /* update store data */
1143     xfree(rec->info[recInfo_storeData]);
1144
1145     /* update store data */
1146     if (zh->store_data_buf)
1147     {
1148         rec->size[recInfo_storeData] = zh->store_data_size;
1149         rec->info[recInfo_storeData] = zh->store_data_buf;
1150         zh->store_data_buf = 0;
1151         recordAttr->recordSize = zh->store_data_size;
1152     }
1153     else if (zh->m_store_data)
1154     {
1155         off_t cur_offset = stream->tellf(stream);
1156
1157         rec->size[recInfo_storeData] = recordAttr->recordSize;
1158         rec->info[recInfo_storeData] = (char *)
1159             xmalloc(recordAttr->recordSize);
1160         stream->seekf(stream, start_offset);
1161         stream->readf(stream, rec->info[recInfo_storeData],
1162                       recordAttr->recordSize);
1163         stream->seekf(stream, cur_offset);
1164     }
1165     else
1166     {
1167         rec->info[recInfo_storeData] = NULL;
1168         rec->size[recInfo_storeData] = 0;
1169     }
1170     /* update database name */
1171     xfree(rec->info[recInfo_databaseName]);
1172     rec->info[recInfo_databaseName] =
1173         rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
1174
1175     /* update offset */
1176     recordAttr->recordOffset = start_offset;
1177
1178     /* commit this record */
1179     rec_put(zh->reg->records, &rec);
1180     logRecord(zh);
1181     return ZEBRA_OK;
1182 }
1183
1184 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1185 {
1186     ZebraHandle zh = (ZebraHandle) handle;
1187     struct recExtractCtrl extractCtrl;
1188
1189     if (zebraExplain_curDatabase(zh->reg->zei,
1190                                   rec->info[recInfo_databaseName]))
1191     {
1192         abort();
1193         if (zebraExplain_newDatabase(zh->reg->zei,
1194                                       rec->info[recInfo_databaseName], 0))
1195             abort();
1196     }
1197
1198     zebra_rec_keys_reset(zh->reg->keys);
1199     zebra_rec_keys_reset(zh->reg->sortKeys);
1200
1201     extractCtrl.init = extract_init;
1202     extractCtrl.tokenAdd = extract_token_add;
1203     extractCtrl.schemaAdd = extract_schema_add;
1204     extractCtrl.dh = zh->reg->dh;
1205
1206     init_extractCtrl(zh, &extractCtrl);
1207
1208     extractCtrl.flagShowRecords = 0;
1209     extractCtrl.match_criteria[0] = '\0';
1210     extractCtrl.staticrank = 0;
1211     extractCtrl.action = action_update;
1212
1213     extractCtrl.handle = handle;
1214     extractCtrl.first_record = 1;
1215
1216     extract_set_store_data_prepare(&extractCtrl);
1217
1218     if (n)
1219         grs_extract_tree(&extractCtrl, n);
1220
1221     if (rec->size[recInfo_delKeys])
1222     {
1223         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1224
1225         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1226
1227         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1228                                rec->size[recInfo_delKeys],
1229                                0);
1230 #if FLUSH2
1231         extract_flush_record_keys2(zh, rec->sysno,
1232                                    zh->reg->keys, 0, delkeys, 0);
1233 #else
1234         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1235         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1236 #endif
1237         zebra_rec_keys_close(delkeys);
1238
1239         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1240                                rec->size[recInfo_sortKeys],
1241                                0);
1242
1243         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1244         zebra_rec_keys_close(sortkeys);
1245     }
1246     else
1247     {
1248 #if FLUSH2
1249         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1250 #else
1251         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1252 #endif
1253     }
1254     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1255
1256     xfree(rec->info[recInfo_delKeys]);
1257     zebra_rec_keys_get_buf(zh->reg->keys,
1258                            &rec->info[recInfo_delKeys],
1259                            &rec->size[recInfo_delKeys]);
1260
1261     xfree(rec->info[recInfo_sortKeys]);
1262     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1263                            &rec->info[recInfo_sortKeys],
1264                            &rec->size[recInfo_sortKeys]);
1265     return ZEBRA_OK;
1266 }
1267
1268 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1269                           zebra_rec_keys_t reckeys,
1270                           int level)
1271 {
1272     if (zebra_rec_keys_rewind(reckeys))
1273     {
1274         size_t slen;
1275         const char *str;
1276         struct it_key key;
1277         NMEM nmem = nmem_create();
1278
1279         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1280         {
1281             char keystr[200]; /* room for zints to print */
1282             char *dst_term = 0;
1283             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1284             const char *index_type;
1285             int i;
1286             const char *string_index;
1287
1288             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1289                                     0/* db */, &string_index);
1290             assert(index_type);
1291             zebra_term_untrans_iconv(zh, nmem, index_type,
1292                                      &dst_term, str);
1293             *keystr = '\0';
1294             for (i = 0; i<key.len; i++)
1295             {
1296                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1297             }
1298
1299             if (*str < CHR_BASE_CHAR)
1300             {
1301                 int i;
1302                 char dst_buf[200]; /* room for special chars */
1303
1304                 strcpy(dst_buf , "?");
1305
1306                 if (!strcmp(str, ""))
1307                     strcpy(dst_buf, "alwaysmatches");
1308                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1309                     strcpy(dst_buf, "firstinfield");
1310                 else if (!strcmp(str, CHR_UNKNOWN))
1311                     strcpy(dst_buf, "unknown");
1312                 else if (!strcmp(str, CHR_SPACE))
1313                     strcpy(dst_buf, "space");
1314
1315                 for (i = 0; i<slen; i++)
1316                 {
1317                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1318                 }
1319                 yaz_log(level, "%s%s %s %s", keystr, index_type,
1320                         string_index, dst_buf);
1321
1322             }
1323             else
1324                 yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
1325                         string_index, dst_term);
1326
1327             nmem_reset(nmem);
1328         }
1329         nmem_destroy(nmem);
1330     }
1331 }
1332
1333 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1334                              zebra_rec_keys_t reckeys)
1335 {
1336     ZebraExplainInfo zei = zh->reg->zei;
1337     struct ord_stat {
1338         int no;
1339         int ord;
1340         struct ord_stat *next;
1341     };
1342
1343     if (zebra_rec_keys_rewind(reckeys))
1344     {
1345         struct ord_stat *ord_list = 0;
1346         struct ord_stat *p;
1347         size_t slen;
1348         const char *str;
1349         struct it_key key_in;
1350         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1351         {
1352             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1353
1354             for (p = ord_list; p ; p = p->next)
1355                 if (p->ord == ord)
1356                 {
1357                     p->no++;
1358                     break;
1359                 }
1360             if (!p)
1361             {
1362                 p = xmalloc(sizeof(*p));
1363                 p->no = 1;
1364                 p->ord = ord;
1365                 p->next = ord_list;
1366                 ord_list = p;
1367             }
1368         }
1369
1370         p = ord_list;
1371         while (p)
1372         {
1373             struct ord_stat *p1 = p;
1374
1375             if (is_insert)
1376                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1377             else
1378                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1379             p = p->next;
1380             xfree(p1);
1381         }
1382     }
1383 }
1384
1385 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1386                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1387                                 zebra_rec_keys_t del_keys, zint del_rank)
1388 {
1389     ZebraExplainInfo zei = zh->reg->zei;
1390     int normal = 0;
1391     int optimized = 0;
1392
1393     if (!zh->reg->key_block)
1394     {
1395         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1396         const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
1397         int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
1398         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1399     }
1400
1401     if (ins_keys)
1402     {
1403         extract_rec_keys_adjust(zh, 1, ins_keys);
1404         if (!del_keys)
1405             zebraExplain_recordCountIncrement(zei, 1);
1406         zebra_rec_keys_rewind(ins_keys);
1407     }
1408     if (del_keys)
1409     {
1410         extract_rec_keys_adjust(zh, 0, del_keys);
1411         if (!ins_keys)
1412             zebraExplain_recordCountIncrement(zei, -1);
1413         zebra_rec_keys_rewind(del_keys);
1414     }
1415
1416     while (1)
1417     {
1418         size_t del_slen;
1419         const char *del_str;
1420         struct it_key del_key_in;
1421         int del = 0;
1422
1423         size_t ins_slen;
1424         const char *ins_str;
1425         struct it_key ins_key_in;
1426         int ins = 0;
1427
1428         if (del_keys)
1429             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1430                                       &del_key_in);
1431         if (ins_keys)
1432             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1433                                       &ins_key_in);
1434
1435         if (del && ins && ins_rank == del_rank
1436             && !key_compare(&del_key_in, &ins_key_in)
1437             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1438         {
1439             optimized++;
1440             continue;
1441         }
1442         if (!del && !ins)
1443             break;
1444
1445         normal++;
1446         if (del)
1447             key_block_write(zh->reg->key_block, sysno,
1448                             &del_key_in, 0, del_str, del_slen,
1449                             del_rank, zh->m_staticrank);
1450         if (ins)
1451             key_block_write(zh->reg->key_block, sysno,
1452                             &ins_key_in, 1, ins_str, ins_slen,
1453                             ins_rank, zh->m_staticrank);
1454     }
1455     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1456 }
1457
1458
1459 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1460                                      zebra_rec_keys_t reckeys,
1461                                      zebra_snippets *snippets)
1462 {
1463     NMEM nmem = nmem_create();
1464     if (zebra_rec_keys_rewind(reckeys))
1465     {
1466         const char *str;
1467         size_t slen;
1468         struct it_key key;
1469         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1470         {
1471             char *dst_term = 0;
1472             int ord;
1473             zint seqno;
1474             const char *index_type;
1475
1476             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1477             seqno = key.mem[key.len-1];
1478             ord = CAST_ZINT_TO_INT(key.mem[0]);
1479
1480             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1481                                     0/* db */, 0 /* string_index */);
1482             assert(index_type);
1483             zebra_term_untrans_iconv(zh, nmem, index_type,
1484                                      &dst_term, str);
1485             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1486             nmem_reset(nmem);
1487         }
1488     }
1489     nmem_destroy(nmem);
1490     return ZEBRA_OK;
1491 }
1492
1493 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1494 {
1495     yaz_log(YLOG_LOG, "print_rec_keys");
1496     if (zebra_rec_keys_rewind(reckeys))
1497     {
1498         const char *str;
1499         size_t slen;
1500         struct it_key key;
1501         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1502         {
1503             char dst_buf[IT_MAX_WORD];
1504             zint seqno;
1505             const char *index_type;
1506             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1507             const char *db = 0;
1508             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1509
1510             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1511
1512             seqno = key.mem[key.len-1];
1513
1514             zebra_term_untrans(zh, index_type, dst_buf, str);
1515
1516             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1517                     " term=%s", ord, seqno, dst_buf);
1518         }
1519     }
1520 }
1521
1522 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1523                                      const char *str, int length)
1524 {
1525     struct it_key key;
1526     ZebraHandle zh = p->extractCtrl->handle;
1527     ZebraExplainInfo zei = zh->reg->zei;
1528     int ch, i;
1529
1530     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1531     if (ch < 0)
1532         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1533
1534     i = 0;
1535     key.mem[i++] = ch;
1536     key.mem[i++] = p->record_id;
1537     key.mem[i++] = p->section_id;
1538
1539     if (zh->m_segment_indexing)
1540         key.mem[i++] = p->segment;
1541     key.mem[i++] = p->seqno;
1542     key.len = i;
1543
1544     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1545 }
1546
1547 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1548 {
1549     struct it_key key;
1550     ZebraHandle zh = p->extractCtrl->handle;
1551     ZebraExplainInfo zei = zh->reg->zei;
1552     int ch;
1553     zinfo_index_category_t cat = zinfo_index_category_sort;
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     key.len = 2;
1559     key.mem[0] = ch;
1560     key.mem[1] = p->record_id;
1561
1562     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1563 }
1564
1565 static void extract_add_staticrank_string(RecWord *p,
1566                                           const char *str, int length)
1567 {
1568     char valz[40];
1569     struct recExtractCtrl *ctrl = p->extractCtrl;
1570
1571     if (length > sizeof(valz)-1)
1572         length = sizeof(valz)-1;
1573
1574     memcpy(valz, str, length);
1575     valz[length] = '\0';
1576     ctrl->staticrank = atozint(valz);
1577 }
1578
1579 static void extract_add_string(RecWord *p, zebra_map_t zm,
1580                                const char *string, int length)
1581 {
1582     assert(length > 0);
1583
1584     if (!p->index_name)
1585         return;
1586
1587     if (zebra_maps_is_index(zm))
1588     {
1589         extract_add_index_string(p, zinfo_index_category_index,
1590                                  string, length);
1591         if (zebra_maps_is_alwaysmatches(zm))
1592         {
1593             RecWord word;
1594             memcpy(&word, p, sizeof(word));
1595
1596             word.seqno = 1;
1597             extract_add_index_string(
1598                 &word, zinfo_index_category_alwaysmatches, "", 0);
1599         }
1600     }
1601     else if (zebra_maps_is_sort(zm))
1602     {
1603         extract_add_sort_string(p, string, length);
1604     }
1605     else if (zebra_maps_is_staticrank(zm))
1606     {
1607         extract_add_staticrank_string(p, string, length);
1608     }
1609 }
1610
1611 static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
1612 {
1613     const char *b = p->term_buf;
1614     int remain = p->term_len;
1615     int first = 1;
1616     const char **map = 0;
1617
1618     if (remain > 0)
1619         map = zebra_maps_input(zm, &b, remain, 0);
1620
1621     while (map)
1622     {
1623         char buf[IT_MAX_WORD+1];
1624         int i, remain;
1625
1626         /* Skip spaces */
1627         while (map && *map && **map == *CHR_SPACE)
1628         {
1629             remain = p->term_len - (b - p->term_buf);
1630             if (remain > 0)
1631                 map = zebra_maps_input(zm, &b, remain, 0);
1632             else
1633                 map = 0;
1634         }
1635         if (!map)
1636             break;
1637         i = 0;
1638         while (map && *map && **map != *CHR_SPACE)
1639         {
1640             const char *cp = *map;
1641
1642             while (i < IT_MAX_WORD && *cp)
1643                 buf[i++] = *(cp++);
1644             remain = p->term_len - (b - p->term_buf);
1645             if (remain > 0)
1646                 map = zebra_maps_input(zm, &b, remain, 0);
1647             else
1648                 map = 0;
1649         }
1650         if (!i)
1651             return;
1652
1653         if (first)
1654         {
1655             first = 0;
1656             if (zebra_maps_is_first_in_field(zm))
1657             {
1658                 /* first in field marker */
1659                 extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1660                 p->seqno++;
1661             }
1662         }
1663         extract_add_string(p, zm, buf, i);
1664         p->seqno++;
1665     }
1666 }
1667
1668 static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
1669 {
1670     const char *b = p->term_buf;
1671     char buf[IT_MAX_WORD+1];
1672     const char **map = 0;
1673     int i = 0, remain = p->term_len;
1674
1675     if (remain > 0)
1676         map = zebra_maps_input(zm, &b, remain, 1);
1677
1678     while (remain > 0 && i < IT_MAX_WORD)
1679     {
1680         while (map && *map && **map == *CHR_SPACE)
1681         {
1682             remain = p->term_len - (b - p->term_buf);
1683
1684             if (remain > 0)
1685             {
1686                 int first = i ? 0 : 1;  /* first position */
1687                 map = zebra_maps_input(zm, &b, remain, first);
1688             }
1689             else
1690                 map = 0;
1691         }
1692         if (!map)
1693             break;
1694
1695         if (i && i < IT_MAX_WORD)
1696             buf[i++] = *CHR_SPACE;
1697         while (map && *map && **map != *CHR_SPACE)
1698         {
1699             const char *cp = *map;
1700
1701             if (**map == *CHR_CUT)
1702             {
1703                 i = 0;
1704             }
1705             else
1706             {
1707                 if (i >= IT_MAX_WORD)
1708                     break;
1709                 while (i < IT_MAX_WORD && *cp)
1710                     buf[i++] = *(cp++);
1711             }
1712             remain = p->term_len  - (b - p->term_buf);
1713             if (remain > 0)
1714             {
1715                 map = zebra_maps_input(zm, &b, remain, 0);
1716             }
1717             else
1718                 map = 0;
1719         }
1720     }
1721     if (!i)
1722         return;
1723     extract_add_string(p, zm, buf, i);
1724 }
1725
1726 static void extract_add_icu(RecWord *p, zebra_map_t zm)
1727 {
1728     struct it_key key;
1729     const char *res_buf = 0;
1730     size_t res_len = 0;
1731     ZebraHandle zh = p->extractCtrl->handle;
1732     int r = zebra_map_tokenize(zm, p->term_buf, p->term_len,
1733                                &res_buf, &res_len);
1734     int cat = zinfo_index_category_index;
1735     int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
1736     if (ch < 0)
1737         ch = zebraExplain_add_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
1738     while (r)
1739     {
1740         int i = 0;
1741         key.mem[i++] = ch;
1742         key.mem[i++] = p->record_id;
1743         key.mem[i++] = p->section_id;
1744
1745         if (zh->m_segment_indexing)
1746             key.mem[i++] = p->segment;
1747         key.mem[i++] = p->seqno;
1748         key.len = i;
1749
1750         zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
1751
1752         p->seqno++;
1753         r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len);
1754     }
1755 }
1756
1757
1758 /** \brief top-level indexing handler for recctrl system
1759     \param p token data to be indexed
1760
1761     Call sequence:
1762     extract_token
1763     zebra_add_{in}_complete
1764     extract_add_string
1765
1766     extract_add_index_string
1767     or
1768     extract_add_sort_string
1769     or
1770     extract_add_staticrank_string
1771
1772 */
1773 static void extract_token_add(RecWord *p)
1774 {
1775     ZebraHandle zh = p->extractCtrl->handle;
1776     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
1777     WRBUF wrbuf;
1778
1779     if (log_level_details)
1780     {
1781         yaz_log(log_level_details, "extract_token_add "
1782                 "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
1783                 p->index_type, p->index_name,
1784                 p->seqno, p->term_len, p->term_buf);
1785     }
1786     if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
1787     {
1788         p->term_buf = wrbuf_buf(wrbuf);
1789         p->term_len = wrbuf_len(wrbuf);
1790     }
1791     if (zebra_maps_is_icu(zm))
1792     {
1793         extract_add_icu(p, zm);
1794     }
1795     else
1796     {
1797         if (zebra_maps_is_complete(zm))
1798             extract_add_complete_field(p, zm);
1799         else
1800             extract_add_incomplete_field(p, zm);
1801     }
1802 }
1803
1804 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1805                                       void *buf, size_t sz)
1806 {
1807     ZebraHandle zh = (ZebraHandle) p->handle;
1808
1809     xfree(zh->store_data_buf);
1810     zh->store_data_buf = 0;
1811     zh->store_data_size = 0;
1812     if (buf && sz)
1813     {
1814         zh->store_data_buf = xmalloc(sz);
1815         zh->store_data_size = sz;
1816         memcpy(zh->store_data_buf, buf, sz);
1817     }
1818 }
1819
1820 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1821 {
1822     ZebraHandle zh = (ZebraHandle) p->handle;
1823     xfree(zh->store_data_buf);
1824     zh->store_data_buf = 0;
1825     zh->store_data_size = 0;
1826     p->setStoreData = extract_set_store_data_cb;
1827 }
1828
1829 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1830 {
1831     ZebraHandle zh = (ZebraHandle) p->handle;
1832     zebraExplain_addSchema(zh->reg->zei, oid);
1833 }
1834
1835 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1836                              int cmd, zebra_rec_keys_t reckeys)
1837 {
1838 #if 0
1839     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1840             cmd, sysno);
1841     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1842 #endif
1843
1844     if (zebra_rec_keys_rewind(reckeys))
1845     {
1846         zebra_sort_index_t si = zh->reg->sort_index;
1847         size_t slen;
1848         const char *str;
1849         struct it_key key_in;
1850
1851         zebra_sort_sysno(si, sysno);
1852
1853         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1854         {
1855             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1856
1857             zebra_sort_type(si, ord);
1858             if (cmd == 1)
1859                 zebra_sort_add(si, str, slen);
1860             else
1861                 zebra_sort_delete(si);
1862         }
1863     }
1864 }
1865
1866 /*
1867  * Local variables:
1868  * c-basic-offset: 4
1869  * indent-tabs-mode: nil
1870  * End:
1871  * vim: shiftwidth=4 tabstop=8 expandtab
1872  */
1873