index/extract.c

   1 /* $Id: extract.c,v 1.262 2007-08-31 07:02:24 adam Exp $
   2    Copyright (C) 1995-2007
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 #include <stdio.h>
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #ifdef WIN32
  27 #include <io.h>
  28 #endif
  29 #if HAVE_UNISTD_H
  30 #include <unistd.h>
  31 #endif
  32 #include <fcntl.h>
  33
  34 #include "index.h"
  35 #include "orddict.h"
  36 #include <direntz.h>
  37 #include <charmap.h>
  38
  39 static int log_level_extract = 0;
  40 static int log_level_details = 0;
  41 static int log_level_initialized = 0;
  42
  43 /* 1 if we use eliminitate identical delete/insert keys */
  44 /* eventually this the 0-case code will be removed */
  45 #define FLUSH2 1
  46
  47 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  48                                 zebra_rec_keys_t ins_keys,
  49                                 zint ins_rank,
  50                                 zebra_rec_keys_t del_keys,
  51                                 zint del_rank);
  52
  53 static void zebra_init_log_level(void)
  54 {
  55     if (!log_level_initialized)
  56     {
  57         log_level_initialized = 1;
  58
  59         log_level_extract = yaz_log_module_level("extract");
  60         log_level_details = yaz_log_module_level("indexdetails");
  61     }
  62 }
  63
  64 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
  65                                       int cmd, zebra_rec_keys_t reckeys,
  66                                       zint staticrank);
  67 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  68                                     int cmd, zebra_rec_keys_t skp);
  69 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
  70 static void extract_token_add (RecWord *p);
  71
  72 static void check_log_limit(ZebraHandle zh)
  73 {
  74     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  75     {
  76         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  77                 zh->m_file_verbose_limit);
  78     }
  79 }
  80
  81 static void logRecord (ZebraHandle zh)
  82 {
  83     check_log_limit(zh);
  84     ++zh->records_processed;
  85     if (!(zh->records_processed % 1000))
  86     {
  87         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  88                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  89                 zh->records_processed, zh->records_inserted,
  90                 zh->records_updated, zh->records_deleted);
  91     }
  92 }
  93
  94 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
  95 {
  96     int i;
  97     for (i = 0; i<256; i++)
  98     {
  99         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
 100             ctrl->seqno[i] = 1;
 101         else
 102             ctrl->seqno[i] = 0;
 103     }
 104     ctrl->flagShowRecords = !zh->m_flag_rw;
 105 }
 106
 107
 108 static void extract_add_index_string (RecWord *p,
 109                                       zinfo_index_category_t cat,
 110                                       const char *str, int length);
 111
 112 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
 113
 114 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 115 {
 116     w->seqno = 1;
 117     w->index_name = "any";
 118     w->index_type = 'w';
 119     w->extractCtrl = p;
 120     w->record_id = 0;
 121     w->section_id = 0;
 122     w->segment = 0;
 123 }
 124
 125 struct snip_rec_info {
 126     ZebraHandle zh;
 127     zebra_snippets *snippets;
 128 };
 129
 130
 131 static void snippet_add_complete_field(RecWord *p, int ord)
 132 {
 133     struct snip_rec_info *h = p->extractCtrl->handle;
 134     ZebraHandle zh = h->zh;
 135
 136     const char *b = p->term_buf;
 137     char buf[IT_MAX_WORD+1];
 138     const char **map = 0;
 139     int i = 0, remain = p->term_len;
 140     const char *start = b;
 141     const char *last = 0;
 142
 143     if (remain > 0)
 144         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
 145
 146     while (remain > 0 && i < IT_MAX_WORD)
 147     {
 148         while (map && *map && **map == *CHR_SPACE)
 149         {
 150             remain = p->term_len - (b - p->term_buf);
 151
 152             if (i == 0)
 153                 start = b;  /* set to first non-ws area */
 154             if (remain > 0)
 155             {
 156                 int first = i ? 0 : 1;  /* first position */
 157
 158                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type,
 159                                        &b, remain, first);
 160             }
 161             else
 162                 map = 0;
 163         }
 164         if (!map)
 165             break;
 166
 167         if (i && i < IT_MAX_WORD)
 168             buf[i++] = *CHR_SPACE;
 169         while (map && *map && **map != *CHR_SPACE)
 170         {
 171             const char *cp = *map;
 172
 173             if (**map == *CHR_CUT)
 174             {
 175                 i = 0;
 176             }
 177             else
 178             {
 179                 if (i >= IT_MAX_WORD)
 180                     break;
 181                 while (i < IT_MAX_WORD && *cp)
 182                     buf[i++] = *(cp++);
 183             }
 184             last = b;
 185             remain = p->term_len  - (b - p->term_buf);
 186             if (remain > 0)
 187             {
 188                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
 189                                         remain, 0);
 190             }
 191             else
 192                 map = 0;
 193         }
 194     }
 195     if (!i)
 196         return;
 197     if (last && start != last)
 198         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 199                                start, last - start);
 200 }
 201
 202 static void snippet_add_incomplete_field(RecWord *p, int ord)
 203 {
 204     struct snip_rec_info *h = p->extractCtrl->handle;
 205     ZebraHandle zh = h->zh;
 206     const char *b = p->term_buf;
 207     int remain = p->term_len;
 208     int first = 1;
 209     const char **map = 0;
 210     const char *start = b;
 211     const char *last = b;
 212
 213     if (remain > 0)
 214         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
 215
 216     while (map)
 217     {
 218         char buf[IT_MAX_WORD+1];
 219         int i, remain;
 220
 221         /* Skip spaces */
 222         while (map && *map && **map == *CHR_SPACE)
 223         {
 224             remain = p->term_len - (b - p->term_buf);
 225             last = b;
 226             if (remain > 0)
 227                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
 228                                        remain, 0);
 229             else
 230                 map = 0;
 231         }
 232         if (!map)
 233             break;
 234         if (start != last)
 235         {
 236             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
 237                                    start, last - start);
 238
 239         }
 240         start = last;
 241
 242         i = 0;
 243         while (map && *map && **map != *CHR_SPACE)
 244         {
 245             const char *cp = *map;
 246
 247             while (i < IT_MAX_WORD && *cp)
 248                 buf[i++] = *(cp++);
 249             remain = p->term_len - (b - p->term_buf);
 250             last = b;
 251             if (remain > 0)
 252                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
 253             else
 254                 map = 0;
 255         }
 256         if (!i)
 257             return;
 258
 259         if (first)
 260         {
 261             first = 0;
 262             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
 263             {
 264                 /* first in field marker */
 265                 p->seqno++;
 266             }
 267         }
 268         if (start != last)
 269             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 270                                    start, last - start);
 271         start = last;
 272         p->seqno++;
 273     }
 274
 275 }
 276
 277 static void snippet_token_add(RecWord *p)
 278 {
 279     struct snip_rec_info *h = p->extractCtrl->handle;
 280     ZebraHandle zh = h->zh;
 281
 282     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
 283     {
 284         ZebraExplainInfo zei = zh->reg->zei;
 285         int ch = zebraExplain_lookup_attr_str(
 286             zei, zinfo_index_category_index, p->index_type, p->index_name);
 287
 288         if (zebra_maps_is_complete (h->zh->reg->zebra_maps, p->index_type))
 289             snippet_add_complete_field (p, ch);
 290         else
 291             snippet_add_incomplete_field(p, ch);
 292     }
 293 }
 294
 295 static void snippet_schema_add(
 296     struct recExtractCtrl *p, Odr_oid *oid)
 297 {
 298
 299 }
 300
 301 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 302                      struct ZebraRecStream *stream,
 303                      RecType rt, void *recTypeClientData)
 304 {
 305     struct recExtractCtrl extractCtrl;
 306     struct snip_rec_info info;
 307     int r;
 308
 309     extractCtrl.stream = stream;
 310     extractCtrl.first_record = 1;
 311     extractCtrl.init = extract_init;
 312     extractCtrl.tokenAdd = snippet_token_add;
 313     extractCtrl.schemaAdd = snippet_schema_add;
 314     assert(zh->reg);
 315     assert(zh->reg->dh);
 316
 317     extractCtrl.dh = zh->reg->dh;
 318
 319     info.zh = zh;
 320     info.snippets = sn;
 321     extractCtrl.handle = &info;
 322     extractCtrl.match_criteria[0] = '\0';
 323     extractCtrl.staticrank = 0;
 324     extractCtrl.action = action_insert;
 325
 326     init_extractCtrl(zh, &extractCtrl);
 327
 328     extractCtrl.setStoreData = 0;
 329
 330     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 331
 332 }
 333
 334 static void searchRecordKey(ZebraHandle zh,
 335                             zebra_rec_keys_t reckeys,
 336                             const char *index_name,
 337                             const char **ws, int ws_length)
 338 {
 339     int i;
 340     int ch = -1;
 341     zinfo_index_category_t cat = zinfo_index_category_index;
 342
 343     for (i = 0; i<ws_length; i++)
 344         ws[i] = NULL;
 345
 346     if (ch < 0)
 347         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
 348     if (ch < 0)
 349         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
 350     if (ch < 0)
 351         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
 352
 353     if (ch < 0)
 354         return ;
 355
 356     if (zebra_rec_keys_rewind(reckeys))
 357     {
 358         zint startSeq = -1;
 359         const char *str;
 360         size_t slen;
 361         struct it_key key;
 362         zint seqno;
 363         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 364         {
 365             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 366
 367             seqno = key.mem[key.len-1];
 368
 369             if (key.mem[0] == ch)
 370             {
 371                 zint woff;
 372
 373                 if (startSeq == -1)
 374                     startSeq = seqno;
 375                 woff = seqno - startSeq;
 376                 if (woff >= 0 && woff < ws_length)
 377                     ws[woff] = str;
 378             }
 379         }
 380     }
 381 }
 382
 383 #define FILE_MATCH_BLANK "\t "
 384
 385 static char *get_match_from_spec(ZebraHandle zh,
 386                           zebra_rec_keys_t reckeys,
 387                           const char *fname, const char *spec)
 388 {
 389     static char dstBuf[2048];      /* static here ??? */
 390     char *dst = dstBuf;
 391     const char *s = spec;
 392
 393     while (1)
 394     {
 395         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 396             ;
 397         if (!*s)
 398             break;
 399         if (*s == '(')
 400         {
 401             const char *ws[32];
 402             char attset_str[64], attname_str[64];
 403             int i;
 404             int first = 1;
 405
 406             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 407                 ;
 408             for (i = 0; *s && *s != ',' && *s != ')' &&
 409                      !strchr(FILE_MATCH_BLANK, *s); s++)
 410                 if (i+1 < sizeof(attset_str))
 411                     attset_str[i++] = *s;
 412             attset_str[i] = '\0';
 413
 414             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 415                 ;
 416             if (*s != ',')
 417                 strcpy(attname_str, attset_str);
 418             else
 419             {
 420                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 421                     ;
 422                 for (i = 0; *s && *s != ')' &&
 423                          !strchr(FILE_MATCH_BLANK, *s); s++)
 424                     if (i+1 < sizeof(attname_str))
 425                         attname_str[i++] = *s;
 426                 attname_str[i] = '\0';
 427             }
 428
 429             searchRecordKey (zh, reckeys, attname_str, ws, 32);
 430
 431             if (*s != ')')
 432             {
 433                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
 434                       spec, zh->m_group ? zh->m_group : "none");
 435                 return NULL;
 436             }
 437             s++;
 438
 439             for (i = 0; i<32; i++)
 440                 if (ws[i])
 441                 {
 442                     if (first)
 443                     {
 444                         *dst++ = ' ';
 445                         first = 0;
 446                     }
 447                     strcpy (dst, ws[i]);
 448                     dst += strlen(ws[i]);
 449                 }
 450             if (first)
 451             {
 452                 yaz_log (YLOG_WARN, "Record didn't contain match"
 453                       " fields in (%s,%s)", attset_str, attname_str);
 454                 return NULL;
 455             }
 456         }
 457         else if (*s == '$')
 458         {
 459             int spec_len;
 460             char special[64];
 461             const char *spec_src = NULL;
 462             const char *s1 = ++s;
 463             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 464                 s1++;
 465
 466             spec_len = s1 - s;
 467             if (spec_len > sizeof(special)-1)
 468                 spec_len = sizeof(special)-1;
 469             memcpy (special, s, spec_len);
 470             special[spec_len] = '\0';
 471             s = s1;
 472
 473             if (!strcmp (special, "group"))
 474                 spec_src = zh->m_group;
 475             else if (!strcmp (special, "database"))
 476                 spec_src = zh->basenames[0];
 477             else if (!strcmp (special, "filename")) {
 478                 spec_src = fname;
 479             }
 480             else if (!strcmp (special, "type"))
 481                 spec_src = zh->m_record_type;
 482             else
 483                 spec_src = NULL;
 484             if (spec_src)
 485             {
 486                 strcpy (dst, spec_src);
 487                 dst += strlen (spec_src);
 488             }
 489         }
 490         else if (*s == '\"' || *s == '\'')
 491         {
 492             int stopMarker = *s++;
 493             char tmpString[64];
 494             int i = 0;
 495
 496             while (*s && *s != stopMarker)
 497             {
 498                 if (i+1 < sizeof(tmpString))
 499                     tmpString[i++] = *s++;
 500             }
 501             if (*s)
 502                 s++;
 503             tmpString[i] = '\0';
 504             strcpy (dst, tmpString);
 505             dst += strlen (tmpString);
 506         }
 507         else
 508         {
 509             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
 510                   spec, zh->m_group ? zh->m_group : "none");
 511             return NULL;
 512         }
 513         *dst++ = 1;
 514     }
 515     if (dst == dstBuf)
 516     {
 517         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
 518               fname, zh->m_group ? zh->m_group : "none");
 519         return NULL;
 520     }
 521     *dst = '\0';
 522     return dstBuf;
 523 }
 524
 525 struct recordLogInfo {
 526     const char *fname;
 527     int recordOffset;
 528     struct recordGroup *rGroup;
 529 };
 530
 531 static void all_matches_add(struct recExtractCtrl *ctrl)
 532 {
 533     RecWord word;
 534     extract_init(ctrl, &word);
 535     word.index_name = "_ALLRECORDS";
 536     word.index_type = 'w';
 537     word.seqno = 1;
 538     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
 539                               "", 0);
 540 }
 541
 542 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 543                                        struct ZebraRecStream *stream,
 544                                        enum zebra_recctrl_action_t action,
 545                                        int test_mode,
 546                                        const char *recordType,
 547                                        zint *sysno,
 548                                        const char *match_criteria,
 549                                        const char *fname,
 550                                        RecType recType,
 551                                        void *recTypeClientData);
 552
 553
 554 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 555                              int deleteFlag)
 556 {
 557     ZEBRA_RES r = ZEBRA_OK;
 558     int i, fd;
 559     char gprefix[128];
 560     char ext[128];
 561     char ext_res[128];
 562     struct file_read_info *fi = 0;
 563     const char *original_record_type = 0;
 564     RecType recType;
 565     void *recTypeClientData;
 566     struct ZebraRecStream stream, *streamp;
 567
 568     zebra_init_log_level();
 569
 570     if (!zh->m_group || !*zh->m_group)
 571         *gprefix = '\0';
 572     else
 573         sprintf (gprefix, "%s.", zh->m_group);
 574
 575     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 576
 577     /* determine file extension */
 578     *ext = '\0';
 579     for (i = strlen(fname); --i >= 0; )
 580         if (fname[i] == '/')
 581             break;
 582         else if (fname[i] == '.')
 583         {
 584             strcpy (ext, fname+i+1);
 585             break;
 586         }
 587     /* determine file type - depending on extension */
 588     original_record_type = zh->m_record_type;
 589     if (!zh->m_record_type)
 590     {
 591         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
 592         zh->m_record_type = res_get (zh->res, ext_res);
 593     }
 594     if (!zh->m_record_type)
 595     {
 596         check_log_limit(zh);
 597         if (zh->records_processed + zh->records_skipped
 598             < zh->m_file_verbose_limit)
 599             yaz_log (YLOG_LOG, "? %s", fname);
 600         zh->records_skipped++;
 601         return 0;
 602     }
 603     /* determine match criteria */
 604     if (!zh->m_record_id)
 605     {
 606         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
 607         zh->m_record_id = res_get (zh->res, ext_res);
 608     }
 609
 610     if (!(recType =
 611           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
 612                           &recTypeClientData)))
 613     {
 614         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 615         return ZEBRA_FAIL;
 616     }
 617
 618     switch(recType->version)
 619     {
 620     case 0:
 621         break;
 622     default:
 623         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 624     }
 625     if (sysno && deleteFlag)
 626     {
 627         streamp = 0;
 628         fi = 0;
 629     }
 630     else
 631     {
 632         char full_rep[1024];
 633
 634         if (zh->path_reg && !yaz_is_abspath (fname))
 635         {
 636             strcpy (full_rep, zh->path_reg);
 637             strcat (full_rep, "/");
 638             strcat (full_rep, fname);
 639         }
 640         else
 641             strcpy (full_rep, fname);
 642
 643         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
 644         {
 645             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 646             zh->m_record_type = original_record_type;
 647             return ZEBRA_FAIL;
 648         }
 649         streamp = &stream;
 650         zebra_create_stream_fd(streamp, fd, 0);
 651     }
 652     r = zebra_extract_records_stream(zh, streamp,
 653                                      deleteFlag ?
 654                                      action_delete : action_update,
 655                                      0, /* tst_mode */
 656                                      zh->m_record_type,
 657                                      sysno,
 658                                      0, /*match_criteria */
 659                                      fname,
 660                                      recType, recTypeClientData);
 661     if (streamp)
 662         stream.destroy(streamp);
 663     zh->m_record_type = original_record_type;
 664     return r;
 665 }
 666
 667 /*
 668   If sysno is provided, then it's used to identify the reocord.
 669   If not, and match_criteria is provided, then sysno is guessed
 670   If not, and a record is provided, then sysno is got from there
 671
 672  */
 673
 674 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 675                                       const char *buf, size_t buf_size,
 676                                       enum zebra_recctrl_action_t action,
 677                                       int test_mode,
 678                                       const char *recordType,
 679                                       zint *sysno,
 680                                       const char *match_criteria,
 681                                       const char *fname)
 682 {
 683     struct ZebraRecStream stream;
 684     ZEBRA_RES res;
 685     void *clientData;
 686     RecType recType = 0;
 687
 688     if (recordType && *recordType)
 689     {
 690         yaz_log(log_level_extract,
 691                 "Record type explicitly specified: %s", recordType);
 692         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
 693                                   &clientData);
 694     }
 695     else
 696     {
 697         if (!(zh->m_record_type))
 698         {
 699             yaz_log (YLOG_WARN, "No such record type defined");
 700             return ZEBRA_FAIL;
 701         }
 702         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 703                 zh->m_record_type);
 704         recType = recType_byName (zh->reg->recTypes, zh->res,
 705                                   zh->m_record_type, &clientData);
 706         recordType = zh->m_record_type;
 707     }
 708
 709     if (!recType)
 710     {
 711         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
 712         return ZEBRA_FAIL;
 713     }
 714
 715     zebra_create_stream_mem(&stream, buf, buf_size);
 716
 717     res = zebra_extract_records_stream(zh, &stream,
 718                                        action,
 719                                        test_mode,
 720                                        recordType,
 721                                        sysno,
 722                                        match_criteria,
 723                                        fname,
 724                                        recType, clientData);
 725     stream.destroy(&stream);
 726     return res;
 727 }
 728
 729 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 730                                        struct ZebraRecStream *stream,
 731                                        enum zebra_recctrl_action_t action,
 732                                        int test_mode,
 733                                        const char *recordType,
 734                                        zint *sysno,
 735                                        const char *match_criteria,
 736                                        const char *fname,
 737                                        RecType recType,
 738                                        void *recTypeClientData)
 739 {
 740     ZEBRA_RES res = ZEBRA_OK;
 741     while (1)
 742     {
 743         int more = 0;
 744         res = zebra_extract_record_stream(zh, stream,
 745                                           action,
 746                                           test_mode,
 747                                           recordType,
 748                                           sysno,
 749                                           match_criteria,
 750                                           fname,
 751                                           recType, recTypeClientData, &more);
 752         if (!more)
 753         {
 754             res = ZEBRA_OK;
 755             break;
 756         }
 757         if (res != ZEBRA_OK)
 758             break;
 759         if (sysno)
 760             break;
 761     }
 762     return res;
 763 }
 764
 765
 766 static WRBUF wrbuf_hex_str(const char *cstr)
 767 {
 768     size_t i;
 769     WRBUF w = wrbuf_alloc();
 770     for (i = 0; cstr[i]; i++)
 771     {
 772         if (cstr[i] < ' ' || cstr[i] > 126)
 773             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 774         else
 775             wrbuf_putc(w, cstr[i]);
 776     }
 777     return w;
 778 }
 779
 780 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 781                                       struct ZebraRecStream *stream,
 782                                       enum zebra_recctrl_action_t action,
 783                                       int test_mode,
 784                                       const char *recordType,
 785                                       zint *sysno,
 786                                       const char *match_criteria,
 787                                       const char *fname,
 788                                       RecType recType,
 789                                       void *recTypeClientData,
 790                                       int *more)
 791
 792 {
 793     zint sysno0 = 0;
 794     RecordAttr *recordAttr;
 795     struct recExtractCtrl extractCtrl;
 796     int r;
 797     const char *matchStr = 0;
 798     Record rec;
 799     off_t start_offset = 0, end_offset = 0;
 800     const char *pr_fname = fname;  /* filename to print .. */
 801     int show_progress = zh->records_processed + zh->records_skipped
 802         < zh->m_file_verbose_limit ? 1:0;
 803
 804     zebra_init_log_level();
 805
 806     if (!pr_fname)
 807         pr_fname = "<no file>";  /* make it printable if file is omitted */
 808
 809     zebra_rec_keys_reset(zh->reg->keys);
 810     zebra_rec_keys_reset(zh->reg->sortKeys);
 811
 812     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
 813     {
 814         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
 815                                       zh->m_explain_database))
 816             return ZEBRA_FAIL;
 817     }
 818
 819     if (stream)
 820     {
 821         off_t null_offset = 0;
 822         extractCtrl.stream = stream;
 823
 824         start_offset = stream->tellf(stream);
 825
 826         extractCtrl.first_record = start_offset ? 0 : 1;
 827
 828         stream->endf(stream, &null_offset);;
 829
 830         extractCtrl.init = extract_init;
 831         extractCtrl.tokenAdd = extract_token_add;
 832         extractCtrl.schemaAdd = extract_schema_add;
 833         extractCtrl.dh = zh->reg->dh;
 834         extractCtrl.handle = zh;
 835         extractCtrl.match_criteria[0] = '\0';
 836         extractCtrl.staticrank = 0;
 837         extractCtrl.action = action;
 838
 839         init_extractCtrl(zh, &extractCtrl);
 840
 841         extract_set_store_data_prepare(&extractCtrl);
 842
 843         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 844
 845         if (action == action_update)
 846         {
 847             action = extractCtrl.action;
 848         }
 849
 850         switch (r)
 851         {
 852         case RECCTRL_EXTRACT_EOF:
 853             return ZEBRA_FAIL;
 854         case RECCTRL_EXTRACT_ERROR_GENERIC:
 855             /* error occured during extraction ... */
 856             yaz_log (YLOG_WARN, "extract error: generic");
 857             return ZEBRA_FAIL;
 858         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 859             /* error occured during extraction ... */
 860             yaz_log (YLOG_WARN, "extract error: no such filter");
 861             return ZEBRA_FAIL;
 862         case RECCTRL_EXTRACT_SKIP:
 863             if (show_progress)
 864                 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 865                          recordType, pr_fname, (zint) start_offset);
 866             *more = 1;
 867
 868             end_offset = stream->endf(stream, 0);
 869             if (end_offset)
 870                 stream->seekf(stream, end_offset);
 871
 872             return ZEBRA_OK;
 873         case RECCTRL_EXTRACT_OK:
 874             break;
 875         default:
 876             yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
 877             return ZEBRA_FAIL;
 878         }
 879         end_offset = stream->endf(stream, 0);
 880         if (end_offset)
 881             stream->seekf(stream, end_offset);
 882         else
 883             end_offset = stream->tellf(stream);
 884
 885         all_matches_add(&extractCtrl);
 886
 887         if (extractCtrl.match_criteria[0])
 888             match_criteria = extractCtrl.match_criteria;
 889     }
 890
 891     *more = 1;
 892     if (!sysno)
 893     {
 894         sysno = &sysno0;
 895
 896         if (match_criteria && *match_criteria) {
 897             matchStr = match_criteria;
 898         } else {
 899             if (zh->m_record_id && *zh->m_record_id) {
 900                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 901                                                zh->m_record_id);
 902                 if (!matchStr)
 903                 {
 904                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 905                              pr_fname, (zint) start_offset);
 906                     return ZEBRA_FAIL;
 907                 }
 908             }
 909         }
 910         if (matchStr)
 911         {
 912             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 913             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 914                                           matchStr);
 915
 916
 917             if (log_level_extract)
 918             {
 919                 WRBUF w = wrbuf_hex_str(matchStr);
 920                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 921                 wrbuf_destroy(w);
 922             }
 923             if (rinfo)
 924             {
 925                 assert(*rinfo == sizeof(*sysno));
 926                 memcpy (sysno, rinfo+1, sizeof(*sysno));
 927             }
 928        }
 929     }
 930     if (zebra_rec_keys_empty(zh->reg->keys))
 931     {
 932         /* the extraction process returned no information - the record
 933            is probably empty - unless flagShowRecords is in use */
 934         if (test_mode)
 935             return ZEBRA_OK;
 936     }
 937
 938     if (! *sysno)
 939     {
 940         /* new record */
 941         if (action == action_delete)
 942         {
 943             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 944                          pr_fname, (zint) start_offset);
 945             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
 946             return ZEBRA_FAIL;
 947         }
 948         else if (action == action_replace)
 949         {
 950             yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 951                          pr_fname, (zint) start_offset);
 952             yaz_log (YLOG_WARN, "cannot update record above (seems new)");
 953             return ZEBRA_FAIL;
 954         }
 955         if (show_progress)
 956             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 957                      (zint) start_offset);
 958         rec = rec_new (zh->reg->records);
 959
 960         *sysno = rec->sysno;
 961
 962         recordAttr = rec_init_attr (zh->reg->zei, rec);
 963         if (extractCtrl.staticrank < 0)
 964         {
 965             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
 966             extractCtrl.staticrank = 0;
 967         }
 968
 969         if (matchStr)
 970         {
 971             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 972             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
 973                             sizeof(*sysno), sysno);
 974         }
 975
 976         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 977 #if FLUSH2
 978         extract_flush_record_keys2(zh, *sysno,
 979                                    zh->reg->keys, extractCtrl.staticrank,
 980                                    0, recordAttr->staticrank);
 981 #else
 982         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 983                                   extractCtrl.staticrank);
 984 #endif
 985         recordAttr->staticrank = extractCtrl.staticrank;
 986         zh->records_inserted++;
 987     }
 988     else
 989     {
 990         /* record already exists */
 991         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 992         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
 993         if (action == action_insert)
 994         {
 995             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
 996                          recordType, pr_fname, (zint) start_offset);
 997             logRecord(zh);
 998             return ZEBRA_FAIL;
 999         }
1000
1001         rec = rec_get (zh->reg->records, *sysno);
1002         assert (rec);
1003
1004         recordAttr = rec_init_attr (zh->reg->zei, rec);
1005
1006         /* decrease total size */
1007         zebraExplain_recordBytesIncrement (zh->reg->zei,
1008                                            - recordAttr->recordSize);
1009
1010         zebra_rec_keys_set_buf(delkeys,
1011                                rec->info[recInfo_delKeys],
1012                                rec->size[recInfo_delKeys],
1013                                0);
1014         zebra_rec_keys_set_buf(sortKeys,
1015                                rec->info[recInfo_sortKeys],
1016                                rec->size[recInfo_sortKeys],
1017                                0);
1018
1019         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1020 #if !FLUSH2
1021         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1022                                   recordAttr->staticrank);
1023 #endif
1024         if (action == action_delete)
1025         {
1026             /* record going to be deleted */
1027 #if FLUSH2
1028             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1029                                        delkeys, recordAttr->staticrank);
1030 #endif
1031             if (zebra_rec_keys_empty(delkeys))
1032             {
1033                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1034                         pr_fname, (zint) start_offset);
1035                 yaz_log(YLOG_WARN, "cannot delete file above, "
1036                         "storeKeys false (3)");
1037             }
1038             else
1039             {
1040                 if (show_progress)
1041                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1042                             pr_fname, (zint) start_offset);
1043                 zh->records_deleted++;
1044                 if (matchStr)
1045                 {
1046                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1047                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1048                 }
1049                 rec_del (zh->reg->records, &rec);
1050             }
1051             zebra_rec_keys_close(delkeys);
1052             zebra_rec_keys_close(sortKeys);
1053             rec_free(&rec);
1054             logRecord(zh);
1055             return ZEBRA_OK;
1056         }
1057         else
1058         {   /* update or special_update */
1059             if (show_progress)
1060                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1061                         pr_fname, (zint) start_offset);
1062             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1063
1064 #if FLUSH2
1065             extract_flush_record_keys2(zh, *sysno,
1066                                        zh->reg->keys, extractCtrl.staticrank,
1067                                        delkeys, recordAttr->staticrank);
1068 #else
1069             extract_flush_record_keys(zh, *sysno, 1,
1070                                       zh->reg->keys, extractCtrl.staticrank);
1071 #endif
1072             recordAttr->staticrank = extractCtrl.staticrank;
1073             zh->records_updated++;
1074         }
1075         zebra_rec_keys_close(delkeys);
1076         zebra_rec_keys_close(sortKeys);
1077     }
1078     /* update file type */
1079     xfree (rec->info[recInfo_fileType]);
1080     rec->info[recInfo_fileType] =
1081         rec_strdup (recordType, &rec->size[recInfo_fileType]);
1082
1083     /* update filename */
1084     xfree (rec->info[recInfo_filename]);
1085     rec->info[recInfo_filename] =
1086         rec_strdup (fname, &rec->size[recInfo_filename]);
1087
1088     /* update delete keys */
1089     xfree (rec->info[recInfo_delKeys]);
1090     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1091     {
1092         zebra_rec_keys_get_buf(zh->reg->keys,
1093                                &rec->info[recInfo_delKeys],
1094                                &rec->size[recInfo_delKeys]);
1095     }
1096     else
1097     {
1098         rec->info[recInfo_delKeys] = NULL;
1099         rec->size[recInfo_delKeys] = 0;
1100     }
1101     /* update sort keys */
1102     xfree (rec->info[recInfo_sortKeys]);
1103
1104     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1105                            &rec->info[recInfo_sortKeys],
1106                            &rec->size[recInfo_sortKeys]);
1107
1108     if (stream)
1109     {
1110         recordAttr->recordSize = end_offset - start_offset;
1111         zebraExplain_recordBytesIncrement(zh->reg->zei,
1112                                           recordAttr->recordSize);
1113     }
1114
1115     /* set run-number for this record */
1116     recordAttr->runNumber =
1117         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1118
1119     /* update store data */
1120     xfree (rec->info[recInfo_storeData]);
1121
1122     /* update store data */
1123     if (zh->store_data_buf)
1124     {
1125         rec->size[recInfo_storeData] = zh->store_data_size;
1126         rec->info[recInfo_storeData] = zh->store_data_buf;
1127         zh->store_data_buf = 0;
1128         recordAttr->recordSize = zh->store_data_size;
1129     }
1130     else if (zh->m_store_data)
1131     {
1132         off_t cur_offset = stream->tellf(stream);
1133
1134         rec->size[recInfo_storeData] = recordAttr->recordSize;
1135         rec->info[recInfo_storeData] = (char *)
1136             xmalloc (recordAttr->recordSize);
1137         stream->seekf(stream, start_offset);
1138         stream->readf(stream, rec->info[recInfo_storeData],
1139                       recordAttr->recordSize);
1140         stream->seekf(stream, cur_offset);
1141     }
1142     else
1143     {
1144         rec->info[recInfo_storeData] = NULL;
1145         rec->size[recInfo_storeData] = 0;
1146     }
1147     /* update database name */
1148     xfree (rec->info[recInfo_databaseName]);
1149     rec->info[recInfo_databaseName] =
1150         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1151
1152     /* update offset */
1153     recordAttr->recordOffset = start_offset;
1154
1155     /* commit this record */
1156     rec_put (zh->reg->records, &rec);
1157     logRecord(zh);
1158     return ZEBRA_OK;
1159 }
1160
1161 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1162 {
1163     ZebraHandle zh = (ZebraHandle) handle;
1164     struct recExtractCtrl extractCtrl;
1165
1166     if (zebraExplain_curDatabase (zh->reg->zei,
1167                                   rec->info[recInfo_databaseName]))
1168     {
1169         abort();
1170         if (zebraExplain_newDatabase (zh->reg->zei,
1171                                       rec->info[recInfo_databaseName], 0))
1172             abort ();
1173     }
1174
1175     zebra_rec_keys_reset(zh->reg->keys);
1176     zebra_rec_keys_reset(zh->reg->sortKeys);
1177
1178     extractCtrl.init = extract_init;
1179     extractCtrl.tokenAdd = extract_token_add;
1180     extractCtrl.schemaAdd = extract_schema_add;
1181     extractCtrl.dh = zh->reg->dh;
1182
1183     init_extractCtrl(zh, &extractCtrl);
1184
1185     extractCtrl.flagShowRecords = 0;
1186     extractCtrl.match_criteria[0] = '\0';
1187     extractCtrl.staticrank = 0;
1188     extractCtrl.action = action_update;
1189
1190     extractCtrl.handle = handle;
1191     extractCtrl.first_record = 1;
1192
1193     extract_set_store_data_prepare(&extractCtrl);
1194
1195     if (n)
1196         grs_extract_tree(&extractCtrl, n);
1197
1198     if (rec->size[recInfo_delKeys])
1199     {
1200         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1201
1202         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1203
1204         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1205                                rec->size[recInfo_delKeys],
1206                                0);
1207 #if FLUSH2
1208         extract_flush_record_keys2(zh, rec->sysno,
1209                                    zh->reg->keys, 0, delkeys, 0);
1210 #else
1211         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1212         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1213 #endif
1214         zebra_rec_keys_close(delkeys);
1215
1216         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1217                                rec->size[recInfo_sortKeys],
1218                                0);
1219
1220         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1221         zebra_rec_keys_close(sortkeys);
1222     }
1223     else
1224     {
1225 #if FLUSH2
1226         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1227 #else
1228         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1229 #endif
1230     }
1231     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1232
1233     xfree (rec->info[recInfo_delKeys]);
1234     zebra_rec_keys_get_buf(zh->reg->keys,
1235                            &rec->info[recInfo_delKeys],
1236                            &rec->size[recInfo_delKeys]);
1237
1238     xfree (rec->info[recInfo_sortKeys]);
1239     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1240                            &rec->info[recInfo_sortKeys],
1241                            &rec->size[recInfo_sortKeys]);
1242     return ZEBRA_OK;
1243 }
1244
1245 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1246                           zebra_rec_keys_t reckeys,
1247                           int level)
1248 {
1249     if (zebra_rec_keys_rewind(reckeys))
1250     {
1251         size_t slen;
1252         const char *str;
1253         struct it_key key;
1254         NMEM nmem = nmem_create();
1255
1256         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1257         {
1258             char keystr[200]; /* room for zints to print */
1259             char *dst_term = 0;
1260             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1261             int index_type, i;
1262             const char *string_index;
1263
1264             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1265                                     0/* db */, &string_index);
1266             assert(index_type);
1267             zebra_term_untrans_iconv(zh, nmem, index_type,
1268                                      &dst_term, str);
1269             *keystr = '\0';
1270             for (i = 0; i<key.len; i++)
1271             {
1272                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1273             }
1274
1275             if (*str < CHR_BASE_CHAR)
1276             {
1277                 int i;
1278                 char dst_buf[200]; /* room for special chars */
1279
1280                 strcpy(dst_buf , "?");
1281
1282                 if (!strcmp(str, ""))
1283                     strcpy(dst_buf, "alwaysmatches");
1284                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1285                     strcpy(dst_buf, "firstinfield");
1286                 else if (!strcmp(str, CHR_UNKNOWN))
1287                     strcpy(dst_buf, "unknown");
1288                 else if (!strcmp(str, CHR_SPACE))
1289                     strcpy(dst_buf, "space");
1290
1291                 for (i = 0; i<slen; i++)
1292                 {
1293                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1294                 }
1295                 yaz_log(level, "%s%c %s %s", keystr, index_type,
1296                         string_index, dst_buf);
1297
1298             }
1299             else
1300                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1301                         string_index, dst_term);
1302
1303             nmem_reset(nmem);
1304         }
1305         nmem_destroy(nmem);
1306     }
1307 }
1308
1309 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1310                              zebra_rec_keys_t reckeys)
1311 {
1312     ZebraExplainInfo zei = zh->reg->zei;
1313     struct ord_stat {
1314         int no;
1315         int ord;
1316         struct ord_stat *next;
1317     };
1318
1319     if (zebra_rec_keys_rewind(reckeys))
1320     {
1321         struct ord_stat *ord_list = 0;
1322         struct ord_stat *p;
1323         size_t slen;
1324         const char *str;
1325         struct it_key key_in;
1326         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1327         {
1328             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1329
1330             for (p = ord_list; p ; p = p->next)
1331                 if (p->ord == ord)
1332                 {
1333                     p->no++;
1334                     break;
1335                 }
1336             if (!p)
1337             {
1338                 p = xmalloc(sizeof(*p));
1339                 p->no = 1;
1340                 p->ord = ord;
1341                 p->next = ord_list;
1342                 ord_list = p;
1343             }
1344         }
1345
1346         p = ord_list;
1347         while (p)
1348         {
1349             struct ord_stat *p1 = p;
1350
1351             if (is_insert)
1352                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1353             else
1354                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1355             p = p->next;
1356             xfree(p1);
1357         }
1358     }
1359 }
1360
1361 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1362                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1363                                 zebra_rec_keys_t del_keys, zint del_rank)
1364 {
1365     ZebraExplainInfo zei = zh->reg->zei;
1366     int normal = 0;
1367     int optimized = 0;
1368
1369     if (!zh->reg->key_block)
1370     {
1371         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1372         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1373         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1374         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1375     }
1376
1377     if (ins_keys)
1378     {
1379         extract_rec_keys_adjust(zh, 1, ins_keys);
1380         if (!del_keys)
1381             zebraExplain_recordCountIncrement (zei, 1);
1382         zebra_rec_keys_rewind(ins_keys);
1383     }
1384     if (del_keys)
1385     {
1386         extract_rec_keys_adjust(zh, 0, del_keys);
1387         if (!ins_keys)
1388             zebraExplain_recordCountIncrement (zei, -1);
1389         zebra_rec_keys_rewind(del_keys);
1390     }
1391
1392     while (1)
1393     {
1394         size_t del_slen;
1395         const char *del_str;
1396         struct it_key del_key_in;
1397         int del = 0;
1398
1399         size_t ins_slen;
1400         const char *ins_str;
1401         struct it_key ins_key_in;
1402         int ins = 0;
1403
1404         if (del_keys)
1405             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1406                                       &del_key_in);
1407         if (ins_keys)
1408             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1409                                       &ins_key_in);
1410
1411         if (del && ins && ins_rank == del_rank
1412             && !key_compare(&del_key_in, &ins_key_in)
1413             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1414         {
1415             optimized++;
1416             continue;
1417         }
1418         if (!del && !ins)
1419             break;
1420
1421         normal++;
1422         if (del)
1423             key_block_write(zh->reg->key_block, sysno,
1424                             &del_key_in, 0, del_str, del_slen,
1425                             del_rank, zh->m_staticrank);
1426         if (ins)
1427             key_block_write(zh->reg->key_block, sysno,
1428                             &ins_key_in, 1, ins_str, ins_slen,
1429                             ins_rank, zh->m_staticrank);
1430     }
1431     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1432 }
1433
1434 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1435                                zebra_rec_keys_t reckeys,
1436                                zint staticrank)
1437 {
1438     ZebraExplainInfo zei = zh->reg->zei;
1439
1440     extract_rec_keys_adjust(zh, cmd, reckeys);
1441
1442     if (log_level_details)
1443     {
1444         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1445                 sysno, cmd ? "insert" : "delete");
1446         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1447     }
1448
1449     if (!zh->reg->key_block)
1450     {
1451         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1452         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1453         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1454         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1455     }
1456     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1457
1458 #if 0
1459     yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
1460     print_rec_keys(zh, reckeys);
1461 #endif
1462     if (zebra_rec_keys_rewind(reckeys))
1463     {
1464         size_t slen;
1465         const char *str;
1466         struct it_key key_in;
1467         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1468         {
1469             key_block_write(zh->reg->key_block, sysno,
1470                             &key_in, cmd, str, slen,
1471                             staticrank, zh->m_staticrank);
1472         }
1473     }
1474 }
1475
1476 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1477                                      zebra_rec_keys_t reckeys,
1478                                      zebra_snippets *snippets)
1479 {
1480     NMEM nmem = nmem_create();
1481     if (zebra_rec_keys_rewind(reckeys))
1482     {
1483         const char *str;
1484         size_t slen;
1485         struct it_key key;
1486         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1487         {
1488             char *dst_term = 0;
1489             int ord;
1490             zint seqno;
1491             int index_type;
1492
1493             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1494             seqno = key.mem[key.len-1];
1495             ord = CAST_ZINT_TO_INT(key.mem[0]);
1496
1497             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1498                                     0/* db */, 0 /* string_index */);
1499             assert(index_type);
1500             zebra_term_untrans_iconv(zh, nmem, index_type,
1501                                      &dst_term, str);
1502             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1503             nmem_reset(nmem);
1504         }
1505     }
1506     nmem_destroy(nmem);
1507     return ZEBRA_OK;
1508 }
1509
1510 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1511 {
1512     yaz_log(YLOG_LOG, "print_rec_keys");
1513     if (zebra_rec_keys_rewind(reckeys))
1514     {
1515         const char *str;
1516         size_t slen;
1517         struct it_key key;
1518         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1519         {
1520             char dst_buf[IT_MAX_WORD];
1521             zint seqno;
1522             int index_type;
1523             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1524             const char *db = 0;
1525             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1526
1527             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1528
1529             seqno = key.mem[key.len-1];
1530
1531             zebra_term_untrans(zh, index_type, dst_buf, str);
1532
1533             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1534                     " term=%s", ord, seqno, dst_buf);
1535         }
1536     }
1537 }
1538
1539 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1540                                      const char *str, int length)
1541 {
1542     struct it_key key;
1543     ZebraHandle zh = p->extractCtrl->handle;
1544     ZebraExplainInfo zei = zh->reg->zei;
1545     int ch, i;
1546
1547     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1548     if (ch < 0)
1549         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1550
1551     i = 0;
1552     key.mem[i++] = ch;
1553     key.mem[i++] = p->record_id;
1554     key.mem[i++] = p->section_id;
1555
1556     if (zh->m_segment_indexing)
1557         key.mem[i++] = p->segment;
1558     key.mem[i++] = p->seqno;
1559     key.len = i;
1560
1561     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1562 }
1563
1564 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1565 {
1566     struct it_key key;
1567     ZebraHandle zh = p->extractCtrl->handle;
1568     ZebraExplainInfo zei = zh->reg->zei;
1569     int ch;
1570     zinfo_index_category_t cat = zinfo_index_category_sort;
1571
1572     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1573     if (ch < 0)
1574         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1575     key.len = 2;
1576     key.mem[0] = ch;
1577     key.mem[1] = p->record_id;
1578
1579     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1580 }
1581
1582 static void extract_add_staticrank_string(RecWord *p,
1583                                           const char *str, int length)
1584 {
1585     char valz[40];
1586     struct recExtractCtrl *ctrl = p->extractCtrl;
1587
1588     if (length > sizeof(valz)-1)
1589         length = sizeof(valz)-1;
1590
1591     memcpy(valz, str, length);
1592     valz[length] = '\0';
1593     ctrl->staticrank = atozint(valz);
1594 }
1595
1596 static void extract_add_string(RecWord *p, const char *string, int length)
1597 {
1598     ZebraHandle zh = p->extractCtrl->handle;
1599     assert (length > 0);
1600
1601     if (!p->index_name)
1602         return;
1603
1604     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1605     {
1606         extract_add_index_string(p, zinfo_index_category_index,
1607                                  string, length);
1608         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1609         {
1610             RecWord word;
1611             memcpy(&word, p, sizeof(word));
1612
1613             word.seqno = 1;
1614             extract_add_index_string(
1615                 &word, zinfo_index_category_alwaysmatches, "", 0);
1616         }
1617     }
1618     else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1619     {
1620         extract_add_sort_string(p, string, length);
1621     }
1622     else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1623     {
1624         extract_add_staticrank_string(p, string, length);
1625     }
1626 }
1627
1628 static void extract_add_incomplete_field(RecWord *p)
1629 {
1630     ZebraHandle zh = p->extractCtrl->handle;
1631     const char *b = p->term_buf;
1632     int remain = p->term_len;
1633     int first = 1;
1634     const char **map = 0;
1635
1636     if (remain > 0)
1637         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1638
1639     while (map)
1640     {
1641         char buf[IT_MAX_WORD+1];
1642         int i, remain;
1643
1644         /* Skip spaces */
1645         while (map && *map && **map == *CHR_SPACE)
1646         {
1647             remain = p->term_len - (b - p->term_buf);
1648             if (remain > 0)
1649                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1650                                        remain, 0);
1651             else
1652                 map = 0;
1653         }
1654         if (!map)
1655             break;
1656         i = 0;
1657         while (map && *map && **map != *CHR_SPACE)
1658         {
1659             const char *cp = *map;
1660
1661             while (i < IT_MAX_WORD && *cp)
1662                 buf[i++] = *(cp++);
1663             remain = p->term_len - (b - p->term_buf);
1664             if (remain > 0)
1665                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1666             else
1667                 map = 0;
1668         }
1669         if (!i)
1670             return;
1671
1672         if (first)
1673         {
1674             first = 0;
1675             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1676             {
1677                 /* first in field marker */
1678                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1679                 p->seqno++;
1680             }
1681         }
1682         extract_add_string (p, buf, i);
1683         p->seqno++;
1684     }
1685 }
1686
1687 static void extract_add_complete_field (RecWord *p)
1688 {
1689     ZebraHandle zh = p->extractCtrl->handle;
1690     const char *b = p->term_buf;
1691     char buf[IT_MAX_WORD+1];
1692     const char **map = 0;
1693     int i = 0, remain = p->term_len;
1694
1695     if (remain > 0)
1696         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1697
1698     while (remain > 0 && i < IT_MAX_WORD)
1699     {
1700         while (map && *map && **map == *CHR_SPACE)
1701         {
1702             remain = p->term_len - (b - p->term_buf);
1703
1704             if (remain > 0)
1705             {
1706                 int first = i ? 0 : 1;  /* first position */
1707                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1708             }
1709             else
1710                 map = 0;
1711         }
1712         if (!map)
1713             break;
1714
1715         if (i && i < IT_MAX_WORD)
1716             buf[i++] = *CHR_SPACE;
1717         while (map && *map && **map != *CHR_SPACE)
1718         {
1719             const char *cp = *map;
1720
1721             if (**map == *CHR_CUT)
1722             {
1723                 i = 0;
1724             }
1725             else
1726             {
1727                 if (i >= IT_MAX_WORD)
1728                     break;
1729                 while (i < IT_MAX_WORD && *cp)
1730                     buf[i++] = *(cp++);
1731             }
1732             remain = p->term_len  - (b - p->term_buf);
1733             if (remain > 0)
1734             {
1735                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1736                                         remain, 0);
1737             }
1738             else
1739                 map = 0;
1740         }
1741     }
1742     if (!i)
1743         return;
1744     extract_add_string (p, buf, i);
1745 }
1746
1747 static void extract_token_add(RecWord *p)
1748 {
1749     ZebraHandle zh = p->extractCtrl->handle;
1750     WRBUF wrbuf;
1751
1752     if (log_level_details)
1753     {
1754         yaz_log(log_level_details, "extract_token_add "
1755                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1756                 p->index_type, p->index_name,
1757                 p->seqno, p->term_len, p->term_buf);
1758     }
1759     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1760                                p->term_buf, p->term_len)))
1761     {
1762         p->term_buf = wrbuf_buf(wrbuf);
1763         p->term_len = wrbuf_len(wrbuf);
1764     }
1765     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1766         extract_add_complete_field (p);
1767     else
1768         extract_add_incomplete_field(p);
1769 }
1770
1771 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1772                                       void *buf, size_t sz)
1773 {
1774     ZebraHandle zh = (ZebraHandle) p->handle;
1775
1776     xfree(zh->store_data_buf);
1777     zh->store_data_buf = 0;
1778     zh->store_data_size = 0;
1779     if (buf && sz)
1780     {
1781         zh->store_data_buf = xmalloc(sz);
1782         zh->store_data_size = sz;
1783         memcpy(zh->store_data_buf, buf, sz);
1784     }
1785 }
1786
1787 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1788 {
1789     ZebraHandle zh = (ZebraHandle) p->handle;
1790     xfree(zh->store_data_buf);
1791     zh->store_data_buf = 0;
1792     zh->store_data_size = 0;
1793     p->setStoreData = extract_set_store_data_cb;
1794 }
1795
1796 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1797 {
1798     ZebraHandle zh = (ZebraHandle) p->handle;
1799     zebraExplain_addSchema (zh->reg->zei, oid);
1800 }
1801
1802 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1803                              int cmd, zebra_rec_keys_t reckeys)
1804 {
1805 #if 0
1806     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1807             cmd, sysno);
1808     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1809 #endif
1810
1811     if (zebra_rec_keys_rewind(reckeys))
1812     {
1813         zebra_sort_index_t si = zh->reg->sort_index;
1814         size_t slen;
1815         const char *str;
1816         struct it_key key_in;
1817
1818         zebra_sort_sysno(si, sysno);
1819
1820         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1821         {
1822             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1823
1824             zebra_sort_type(si, ord);
1825             if (cmd == 1)
1826                 zebra_sort_add(si, str, slen);
1827             else
1828                 zebra_sort_delete(si);
1829         }
1830     }
1831 }
1832
1833 /*
1834  * Local variables:
1835  * c-basic-offset: 4
1836  * indent-tabs-mode: nil
1837  * End:
1838  * vim: shiftwidth=4 tabstop=8 expandtab
1839  */
1840