index/extract.c

   1 /* This file is part of the Zebra server.
   2    Copyright (C) 1995-2008 Index Data
   3
   4 Zebra is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file
  21     \brief indexes records and extract tokens for indexing and sorting
  22 */
  23
  24 #include <stdio.h>
  25 #include <assert.h>
  26 #include <ctype.h>
  27 #ifdef WIN32
  28 #include <io.h>
  29 #endif
  30 #if HAVE_UNISTD_H
  31 #include <unistd.h>
  32 #endif
  33 #include <fcntl.h>
  34
  35
  36 #include "index.h"
  37 #include "orddict.h"
  38 #include <direntz.h>
  39 #include <charmap.h>
  40 #include <yaz/snprintf.h>
  41
  42 static int log_level_extract = 0;
  43 static int log_level_details = 0;
  44 static int log_level_initialized = 0;
  45
  46 /* 1 if we use eliminitate identical delete/insert keys */
  47 /* eventually this the 0-case code will be removed */
  48 #define FLUSH2 1
  49
  50 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  51                                 zebra_rec_keys_t ins_keys,
  52                                 zint ins_rank,
  53                                 zebra_rec_keys_t del_keys,
  54                                 zint del_rank);
  55
  56 static void zebra_init_log_level(void)
  57 {
  58     if (!log_level_initialized)
  59     {
  60         log_level_initialized = 1;
  61
  62         log_level_extract = yaz_log_module_level("extract");
  63         log_level_details = yaz_log_module_level("indexdetails");
  64     }
  65 }
  66
  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     ctrl->flagShowRecords = !zh->m_flag_rw;
  97 }
  98
  99
 100 static void extract_add_index_string(RecWord *p,
 101                                       zinfo_index_category_t cat,
 102                                       const char *str, int length);
 103
 104 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
 105
 106 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 107 {
 108     w->seqno = 1;
 109     w->index_name = "any";
 110     w->index_type = "w";
 111     w->extractCtrl = p;
 112     w->record_id = 0;
 113     w->section_id = 0;
 114     w->segment = 0;
 115 }
 116
 117 struct snip_rec_info {
 118     ZebraHandle zh;
 119     zebra_snippets *snippets;
 120 };
 121
 122
 123 static void snippet_add_complete_field(RecWord *p, int ord,
 124                                        zebra_map_t zm)
 125 {
 126     struct snip_rec_info *h = p->extractCtrl->handle;
 127
 128     const char *b = p->term_buf;
 129     char buf[IT_MAX_WORD+1];
 130     const char **map = 0;
 131     int i = 0, remain = p->term_len;
 132     const char *start = b;
 133     const char *last = 0;
 134
 135     if (remain > 0)
 136         map = zebra_maps_input(zm, &b, remain, 1);
 137
 138     while (remain > 0 && i < IT_MAX_WORD)
 139     {
 140         while (map && *map && **map == *CHR_SPACE)
 141         {
 142             remain = p->term_len - (b - p->term_buf);
 143
 144             if (i == 0)
 145                 start = b;  /* set to first non-ws area */
 146             if (remain > 0)
 147             {
 148                 int first = i ? 0 : 1;  /* first position */
 149
 150                 map = zebra_maps_input(zm, &b, remain, first);
 151             }
 152             else
 153                 map = 0;
 154         }
 155         if (!map)
 156             break;
 157
 158         if (i && i < IT_MAX_WORD)
 159             buf[i++] = *CHR_SPACE;
 160         while (map && *map && **map != *CHR_SPACE)
 161         {
 162             const char *cp = *map;
 163
 164             if (**map == *CHR_CUT)
 165             {
 166                 i = 0;
 167             }
 168             else
 169             {
 170                 if (i >= IT_MAX_WORD)
 171                     break;
 172                 while (i < IT_MAX_WORD && *cp)
 173                     buf[i++] = *(cp++);
 174             }
 175             last = b;
 176             remain = p->term_len  - (b - p->term_buf);
 177             if (remain > 0)
 178             {
 179                 map = zebra_maps_input(zm, &b, remain, 0);
 180             }
 181             else
 182                 map = 0;
 183         }
 184     }
 185     if (!i)
 186         return;
 187     if (last && start != last && zebra_maps_is_index(zm))
 188         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 189                                start, last - start);
 190 }
 191
 192 static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 193 {
 194     struct snip_rec_info *h = p->extractCtrl->handle;
 195     const char *b = p->term_buf;
 196     int remain = p->term_len;
 197     int first = 1;
 198     const char **map = 0;
 199     const char *start = b;
 200     const char *last = b;
 201
 202     if (remain > 0)
 203         map = zebra_maps_input(zm, &b, remain, 0);
 204
 205     while (map)
 206     {
 207         char buf[IT_MAX_WORD+1];
 208         int i, remain;
 209
 210         /* Skip spaces */
 211         while (map && *map && **map == *CHR_SPACE)
 212         {
 213             remain = p->term_len - (b - p->term_buf);
 214             last = b;
 215             if (remain > 0)
 216                 map = zebra_maps_input(zm, &b, remain, 0);
 217             else
 218                 map = 0;
 219         }
 220         if (!map)
 221             break;
 222         if (start != last && zebra_maps_is_index(zm))
 223         {
 224             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
 225                                    start, last - start);
 226
 227         }
 228         start = last;
 229
 230         i = 0;
 231         while (map && *map && **map != *CHR_SPACE)
 232         {
 233             const char *cp = *map;
 234
 235             while (i < IT_MAX_WORD && *cp)
 236                 buf[i++] = *(cp++);
 237             remain = p->term_len - (b - p->term_buf);
 238             last = b;
 239             if (remain > 0)
 240                 map = zebra_maps_input(zm, &b, remain, 0);
 241             else
 242                 map = 0;
 243         }
 244         if (!i)
 245             return;
 246
 247         if (first)
 248         {
 249             first = 0;
 250             if (zebra_maps_is_first_in_field(zm))
 251             {
 252                 /* first in field marker */
 253                 p->seqno++;
 254             }
 255         }
 256         if (start != last && zebra_maps_is_index(zm))
 257             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 258                                    start, last - start);
 259         start = last;
 260         p->seqno++;
 261     }
 262
 263 }
 264
 265 static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
 266 {
 267     struct snip_rec_info *h = p->extractCtrl->handle;
 268
 269     const char *res_buf = 0;
 270     size_t res_len = 0;
 271
 272     const char *display_buf = 0;
 273     size_t display_len = 0;
 274
 275     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
 276     while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
 277                                    &display_buf, &display_len))
 278     {
 279         if (zebra_maps_is_index(zm))
 280             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 281                                    display_buf, display_len);
 282         p->seqno++;
 283     }
 284 }
 285
 286 static void snippet_token_add(RecWord *p)
 287 {
 288     struct snip_rec_info *h = p->extractCtrl->handle;
 289     ZebraHandle zh = h->zh;
 290     zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
 291
 292     if (zm)
 293     {
 294         ZebraExplainInfo zei = zh->reg->zei;
 295         int ch = zebraExplain_lookup_attr_str(
 296             zei, zinfo_index_category_index, p->index_type, p->index_name);
 297
 298         if (zebra_maps_is_icu(zm))
 299             snippet_add_icu(p, ch, zm);
 300         else
 301         {
 302             if (zebra_maps_is_complete(zm))
 303                 snippet_add_complete_field(p, ch, zm);
 304             else
 305                 snippet_add_incomplete_field(p, ch, zm);
 306         }
 307     }
 308 }
 309
 310 static void snippet_schema_add(
 311     struct recExtractCtrl *p, Odr_oid *oid)
 312 {
 313
 314 }
 315
 316 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 317                      struct ZebraRecStream *stream,
 318                      RecType rt, void *recTypeClientData)
 319 {
 320     struct recExtractCtrl extractCtrl;
 321     struct snip_rec_info info;
 322     int r;
 323
 324     extractCtrl.stream = stream;
 325     extractCtrl.first_record = 1;
 326     extractCtrl.init = extract_init;
 327     extractCtrl.tokenAdd = snippet_token_add;
 328     extractCtrl.schemaAdd = snippet_schema_add;
 329     assert(zh->reg);
 330     assert(zh->reg->dh);
 331
 332     extractCtrl.dh = zh->reg->dh;
 333
 334     info.zh = zh;
 335     info.snippets = sn;
 336     extractCtrl.handle = &info;
 337     extractCtrl.match_criteria[0] = '\0';
 338     extractCtrl.staticrank = 0;
 339     extractCtrl.action = action_insert;
 340
 341     init_extractCtrl(zh, &extractCtrl);
 342
 343     extractCtrl.setStoreData = 0;
 344
 345     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 346
 347 }
 348
 349 static void searchRecordKey(ZebraHandle zh,
 350                             zebra_rec_keys_t reckeys,
 351                             const char *index_name,
 352                             const char **ws, int ws_length)
 353 {
 354     int i;
 355     int ch = -1;
 356     zinfo_index_category_t cat = zinfo_index_category_index;
 357
 358     for (i = 0; i<ws_length; i++)
 359         ws[i] = NULL;
 360
 361     if (ch < 0)
 362         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
 363     if (ch < 0)
 364         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
 365     if (ch < 0)
 366         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 367
 368     if (ch < 0)
 369         return ;
 370
 371     if (zebra_rec_keys_rewind(reckeys))
 372     {
 373         zint startSeq = -1;
 374         const char *str;
 375         size_t slen;
 376         struct it_key key;
 377         zint seqno;
 378         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 379         {
 380             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 381
 382             seqno = key.mem[key.len-1];
 383
 384             if (key.mem[0] == ch)
 385             {
 386                 zint woff;
 387
 388                 if (startSeq == -1)
 389                     startSeq = seqno;
 390                 woff = seqno - startSeq;
 391                 if (woff >= 0 && woff < ws_length)
 392                     ws[woff] = str;
 393             }
 394         }
 395     }
 396 }
 397
 398 #define FILE_MATCH_BLANK "\t "
 399
 400 static char *get_match_from_spec(ZebraHandle zh,
 401                           zebra_rec_keys_t reckeys,
 402                           const char *fname, const char *spec)
 403 {
 404     static char dstBuf[2048];      /* static here ??? */
 405     char *dst = dstBuf;
 406     const char *s = spec;
 407
 408     while (1)
 409     {
 410         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 411             ;
 412         if (!*s)
 413             break;
 414         if (*s == '(')
 415         {
 416             const char *ws[32];
 417             char attset_str[64], attname_str[64];
 418             int i;
 419             int first = 1;
 420
 421             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 422                 ;
 423             for (i = 0; *s && *s != ',' && *s != ')' &&
 424                      !strchr(FILE_MATCH_BLANK, *s); s++)
 425                 if (i+1 < sizeof(attset_str))
 426                     attset_str[i++] = *s;
 427             attset_str[i] = '\0';
 428
 429             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 430                 ;
 431             if (*s != ',')
 432                 strcpy(attname_str, attset_str);
 433             else
 434             {
 435                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 436                     ;
 437                 for (i = 0; *s && *s != ')' &&
 438                          !strchr(FILE_MATCH_BLANK, *s); s++)
 439                     if (i+1 < sizeof(attname_str))
 440                         attname_str[i++] = *s;
 441                 attname_str[i] = '\0';
 442             }
 443
 444             searchRecordKey(zh, reckeys, attname_str, ws, 32);
 445
 446             if (*s != ')')
 447             {
 448                 yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
 449                       spec, zh->m_group ? zh->m_group : "none");
 450                 return NULL;
 451             }
 452             s++;
 453
 454             for (i = 0; i<32; i++)
 455                 if (ws[i])
 456                 {
 457                     if (first)
 458                     {
 459                         *dst++ = ' ';
 460                         first = 0;
 461                     }
 462                     strcpy(dst, ws[i]);
 463                     dst += strlen(ws[i]);
 464                 }
 465             if (first)
 466             {
 467                 yaz_log(YLOG_WARN, "Record didn't contain match"
 468                       " fields in (%s,%s)", attset_str, attname_str);
 469                 return NULL;
 470             }
 471         }
 472         else if (*s == '$')
 473         {
 474             int spec_len;
 475             char special[64];
 476             const char *spec_src = NULL;
 477             const char *s1 = ++s;
 478             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 479                 s1++;
 480
 481             spec_len = s1 - s;
 482             if (spec_len > sizeof(special)-1)
 483                 spec_len = sizeof(special)-1;
 484             memcpy(special, s, spec_len);
 485             special[spec_len] = '\0';
 486             s = s1;
 487
 488             if (!strcmp(special, "group"))
 489                 spec_src = zh->m_group;
 490             else if (!strcmp(special, "database"))
 491                 spec_src = zh->basenames[0];
 492             else if (!strcmp(special, "filename")) {
 493                 spec_src = fname;
 494             }
 495             else if (!strcmp(special, "type"))
 496                 spec_src = zh->m_record_type;
 497             else
 498                 spec_src = NULL;
 499             if (spec_src)
 500             {
 501                 strcpy(dst, spec_src);
 502                 dst += strlen(spec_src);
 503             }
 504         }
 505         else if (*s == '\"' || *s == '\'')
 506         {
 507             int stopMarker = *s++;
 508             char tmpString[64];
 509             int i = 0;
 510
 511             while (*s && *s != stopMarker)
 512             {
 513                 if (i+1 < sizeof(tmpString))
 514                     tmpString[i++] = *s++;
 515             }
 516             if (*s)
 517                 s++;
 518             tmpString[i] = '\0';
 519             strcpy(dst, tmpString);
 520             dst += strlen(tmpString);
 521         }
 522         else
 523         {
 524             yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
 525                   spec, zh->m_group ? zh->m_group : "none");
 526             return NULL;
 527         }
 528         *dst++ = 1;
 529     }
 530     if (dst == dstBuf)
 531     {
 532         yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
 533               fname, zh->m_group ? zh->m_group : "none");
 534         return NULL;
 535     }
 536     *dst = '\0';
 537     return dstBuf;
 538 }
 539
 540 struct recordLogInfo {
 541     const char *fname;
 542     int recordOffset;
 543     struct recordGroup *rGroup;
 544 };
 545
 546 /** \brief add the always-matches index entry and map to real record ID
 547     \param ctrl record control
 548     \param record_id custom record ID
 549     \param sysno system record ID
 550
 551     This function serves two purposes.. It adds the always matches
 552     entry and makes a pointer from the custom record ID (if defined)
 553     back to the system record ID (sysno)
 554     See zebra_recid_to_sysno .
 555   */
 556 static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
 557                             zint sysno)
 558 {
 559     RecWord word;
 560     extract_init(ctrl, &word);
 561     word.record_id = record_id;
 562     /* we use the seqno as placeholder for a way to get back to
 563        record database from _ALLRECORDS.. This is used if a custom
 564        RECORD was defined */
 565     word.seqno = sysno;
 566     word.index_name = "_ALLRECORDS";
 567     word.index_type = "w";
 568
 569     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
 570                               "", 0);
 571 }
 572
 573 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 574                                        struct ZebraRecStream *stream,
 575                                        enum zebra_recctrl_action_t action,
 576                                        int test_mode,
 577                                        const char *recordType,
 578                                        zint *sysno,
 579                                        const char *match_criteria,
 580                                        const char *fname,
 581                                        RecType recType,
 582                                        void *recTypeClientData);
 583
 584
 585 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 586                              enum zebra_recctrl_action_t action)
 587 {
 588     ZEBRA_RES r = ZEBRA_OK;
 589     int i, fd;
 590     char gprefix[128];
 591     char ext[128];
 592     char ext_res[128];
 593     struct file_read_info *fi = 0;
 594     const char *original_record_type = 0;
 595     RecType recType;
 596     void *recTypeClientData;
 597     struct ZebraRecStream stream, *streamp;
 598
 599     zebra_init_log_level();
 600
 601     if (!zh->m_group || !*zh->m_group)
 602         *gprefix = '\0';
 603     else
 604         sprintf(gprefix, "%s.", zh->m_group);
 605
 606     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 607
 608     /* determine file extension */
 609     *ext = '\0';
 610     for (i = strlen(fname); --i >= 0; )
 611         if (fname[i] == '/')
 612             break;
 613         else if (fname[i] == '.')
 614         {
 615             strcpy(ext, fname+i+1);
 616             break;
 617         }
 618     /* determine file type - depending on extension */
 619     original_record_type = zh->m_record_type;
 620     if (!zh->m_record_type)
 621     {
 622         sprintf(ext_res, "%srecordType.%s", gprefix, ext);
 623         zh->m_record_type = res_get(zh->res, ext_res);
 624     }
 625     if (!zh->m_record_type)
 626     {
 627         check_log_limit(zh);
 628         if (zh->records_processed + zh->records_skipped
 629             < zh->m_file_verbose_limit)
 630             yaz_log(YLOG_LOG, "? %s", fname);
 631         zh->records_skipped++;
 632         return 0;
 633     }
 634     /* determine match criteria */
 635     if (!zh->m_record_id)
 636     {
 637         sprintf(ext_res, "%srecordId.%s", gprefix, ext);
 638         zh->m_record_id = res_get(zh->res, ext_res);
 639     }
 640
 641     if (!(recType =
 642           recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
 643                           &recTypeClientData)))
 644     {
 645         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 646         return ZEBRA_FAIL;
 647     }
 648
 649     switch(recType->version)
 650     {
 651     case 0:
 652         break;
 653     default:
 654         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 655     }
 656     if (sysno && (action == action_delete || action == action_a_delete))
 657     {
 658         streamp = 0;
 659         fi = 0;
 660     }
 661     else
 662     {
 663         char full_rep[1024];
 664
 665         if (zh->path_reg && !yaz_is_abspath(fname))
 666         {
 667             strcpy(full_rep, zh->path_reg);
 668             strcat(full_rep, "/");
 669             strcat(full_rep, fname);
 670         }
 671         else
 672             strcpy(full_rep, fname);
 673
 674         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
 675         {
 676             yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 677             zh->m_record_type = original_record_type;
 678             return ZEBRA_FAIL;
 679         }
 680         streamp = &stream;
 681         zebra_create_stream_fd(streamp, fd, 0);
 682     }
 683     r = zebra_extract_records_stream(zh, streamp,
 684                                      action,
 685                                      0, /* tst_mode */
 686                                      zh->m_record_type,
 687                                      sysno,
 688                                      0, /*match_criteria */
 689                                      fname,
 690                                      recType, recTypeClientData);
 691     if (streamp)
 692         stream.destroy(streamp);
 693     zh->m_record_type = original_record_type;
 694     return r;
 695 }
 696
 697 /*
 698   If sysno is provided, then it's used to identify the reocord.
 699   If not, and match_criteria is provided, then sysno is guessed
 700   If not, and a record is provided, then sysno is got from there
 701
 702  */
 703
 704 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 705                                       const char *buf, size_t buf_size,
 706                                       enum zebra_recctrl_action_t action,
 707                                       int test_mode,
 708                                       const char *recordType,
 709                                       zint *sysno,
 710                                       const char *match_criteria,
 711                                       const char *fname)
 712 {
 713     struct ZebraRecStream stream;
 714     ZEBRA_RES res;
 715     void *clientData;
 716     RecType recType = 0;
 717
 718     if (recordType && *recordType)
 719     {
 720         yaz_log(log_level_extract,
 721                 "Record type explicitly specified: %s", recordType);
 722         recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
 723                                   &clientData);
 724     }
 725     else
 726     {
 727         if (!(zh->m_record_type))
 728         {
 729             yaz_log(YLOG_WARN, "No such record type defined");
 730             return ZEBRA_FAIL;
 731         }
 732         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 733                 zh->m_record_type);
 734         recType = recType_byName(zh->reg->recTypes, zh->res,
 735                                   zh->m_record_type, &clientData);
 736         recordType = zh->m_record_type;
 737     }
 738
 739     if (!recType)
 740     {
 741         yaz_log(YLOG_WARN, "No such record type: %s", recordType);
 742         return ZEBRA_FAIL;
 743     }
 744
 745     zebra_create_stream_mem(&stream, buf, buf_size);
 746
 747     res = zebra_extract_records_stream(zh, &stream,
 748                                        action,
 749                                        test_mode,
 750                                        recordType,
 751                                        sysno,
 752                                        match_criteria,
 753                                        fname,
 754                                        recType, clientData);
 755     stream.destroy(&stream);
 756     return res;
 757 }
 758
 759 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 760                                        struct ZebraRecStream *stream,
 761                                        enum zebra_recctrl_action_t action,
 762                                        int test_mode,
 763                                        const char *recordType,
 764                                        zint *sysno,
 765                                        const char *match_criteria,
 766                                        const char *fname,
 767                                        RecType recType,
 768                                        void *recTypeClientData)
 769 {
 770     ZEBRA_RES res = ZEBRA_OK;
 771     while (1)
 772     {
 773         int more = 0;
 774         res = zebra_extract_record_stream(zh, stream,
 775                                           action,
 776                                           test_mode,
 777                                           recordType,
 778                                           sysno,
 779                                           match_criteria,
 780                                           fname,
 781                                           recType, recTypeClientData, &more);
 782         if (!more)
 783         {
 784             res = ZEBRA_OK;
 785             break;
 786         }
 787         if (res != ZEBRA_OK)
 788             break;
 789         if (sysno)
 790             break;
 791     }
 792     return res;
 793 }
 794
 795
 796 static WRBUF wrbuf_hex_str(const char *cstr)
 797 {
 798     size_t i;
 799     WRBUF w = wrbuf_alloc();
 800     for (i = 0; cstr[i]; i++)
 801     {
 802         if (cstr[i] < ' ' || cstr[i] > 126)
 803             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 804         else
 805             wrbuf_putc(w, cstr[i]);
 806     }
 807     return w;
 808 }
 809
 810 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 811                                       struct ZebraRecStream *stream,
 812                                       enum zebra_recctrl_action_t action,
 813                                       int test_mode,
 814                                       const char *recordType,
 815                                       zint *sysno,
 816                                       const char *match_criteria,
 817                                       const char *fname,
 818                                       RecType recType,
 819                                       void *recTypeClientData,
 820                                       int *more)
 821
 822 {
 823     zint sysno0 = 0;
 824     RecordAttr *recordAttr;
 825     struct recExtractCtrl extractCtrl;
 826     int r;
 827     const char *matchStr = 0;
 828     Record rec;
 829     off_t start_offset = 0, end_offset = 0;
 830     const char *pr_fname = fname;  /* filename to print .. */
 831     int show_progress = zh->records_processed + zh->records_skipped
 832         < zh->m_file_verbose_limit ? 1:0;
 833
 834     zebra_init_log_level();
 835
 836     if (!pr_fname)
 837         pr_fname = "<no file>";  /* make it printable if file is omitted */
 838
 839     zebra_rec_keys_reset(zh->reg->keys);
 840     zebra_rec_keys_reset(zh->reg->sortKeys);
 841
 842     if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
 843     {
 844         if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
 845                                       zh->m_explain_database))
 846             return ZEBRA_FAIL;
 847     }
 848
 849     if (stream)
 850     {
 851         off_t null_offset = 0;
 852         extractCtrl.stream = stream;
 853
 854         start_offset = stream->tellf(stream);
 855
 856         extractCtrl.first_record = start_offset ? 0 : 1;
 857
 858         stream->endf(stream, &null_offset);;
 859
 860         extractCtrl.init = extract_init;
 861         extractCtrl.tokenAdd = extract_token_add;
 862         extractCtrl.schemaAdd = extract_schema_add;
 863         extractCtrl.dh = zh->reg->dh;
 864         extractCtrl.handle = zh;
 865         extractCtrl.match_criteria[0] = '\0';
 866         extractCtrl.staticrank = 0;
 867         extractCtrl.action = action;
 868
 869         init_extractCtrl(zh, &extractCtrl);
 870
 871         extract_set_store_data_prepare(&extractCtrl);
 872
 873         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 874
 875         if (action == action_update)
 876         {
 877             action = extractCtrl.action;
 878         }
 879
 880         switch (r)
 881         {
 882         case RECCTRL_EXTRACT_EOF:
 883             return ZEBRA_FAIL;
 884         case RECCTRL_EXTRACT_ERROR_GENERIC:
 885             /* error occured during extraction ... */
 886             yaz_log(YLOG_WARN, "extract error: generic");
 887             return ZEBRA_FAIL;
 888         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 889             /* error occured during extraction ... */
 890             yaz_log(YLOG_WARN, "extract error: no such filter");
 891             return ZEBRA_FAIL;
 892         case RECCTRL_EXTRACT_SKIP:
 893             if (show_progress)
 894                 yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 895                          recordType, pr_fname, (zint) start_offset);
 896             *more = 1;
 897
 898             end_offset = stream->endf(stream, 0);
 899             if (end_offset)
 900                 stream->seekf(stream, end_offset);
 901
 902             return ZEBRA_OK;
 903         case RECCTRL_EXTRACT_OK:
 904             break;
 905         default:
 906             yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
 907             return ZEBRA_FAIL;
 908         }
 909         end_offset = stream->endf(stream, 0);
 910         if (end_offset)
 911             stream->seekf(stream, end_offset);
 912         else
 913             end_offset = stream->tellf(stream);
 914
 915         if (extractCtrl.match_criteria[0])
 916             match_criteria = extractCtrl.match_criteria;
 917     }
 918
 919     *more = 1;
 920     if (!sysno)
 921     {
 922         sysno = &sysno0;
 923
 924         if (match_criteria && *match_criteria) {
 925             matchStr = match_criteria;
 926         } else {
 927             if (zh->m_record_id && *zh->m_record_id) {
 928                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 929                                                zh->m_record_id);
 930                 if (!matchStr)
 931                 {
 932                     yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 933                              pr_fname, (zint) start_offset);
 934                     return ZEBRA_FAIL;
 935                 }
 936             }
 937         }
 938         if (matchStr)
 939         {
 940             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 941             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 942                                           matchStr);
 943
 944
 945             if (log_level_extract)
 946             {
 947                 WRBUF w = wrbuf_hex_str(matchStr);
 948                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 949                 wrbuf_destroy(w);
 950             }
 951             if (rinfo)
 952             {
 953                 assert(*rinfo == sizeof(*sysno));
 954                 memcpy(sysno, rinfo+1, sizeof(*sysno));
 955             }
 956        }
 957     }
 958
 959     if (zebra_rec_keys_empty(zh->reg->keys))
 960     {
 961         /* the extraction process returned no information - the record
 962            is probably empty - unless flagShowRecords is in use */
 963         if (test_mode)
 964             return ZEBRA_OK;
 965     }
 966
 967     if (! *sysno)
 968     {
 969         /* new record AKA does not exist already */
 970         if (action == action_delete)
 971         {
 972                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 973                         pr_fname, (zint) start_offset);
 974             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
 975             return ZEBRA_FAIL;
 976         }
 977         else if (action == action_a_delete)
 978         {
 979             if (show_progress)
 980                 yaz_log(YLOG_LOG, "adelete %s %s " ZINT_FORMAT, recordType,
 981                         pr_fname, (zint) start_offset);
 982             return ZEBRA_OK;
 983         }
 984         else if (action == action_replace)
 985         {
 986             yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 987                          pr_fname, (zint) start_offset);
 988             yaz_log(YLOG_WARN, "cannot update record above (seems new)");
 989             return ZEBRA_FAIL;
 990         }
 991         if (show_progress)
 992             yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 993                      (zint) start_offset);
 994         rec = rec_new(zh->reg->records);
 995
 996         *sysno = rec->sysno;
 997
 998
 999         if (stream)
1000         {
1001             all_matches_add(&extractCtrl,
1002                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1003                             *sysno);
1004         }
1005
1006
1007         recordAttr = rec_init_attr(zh->reg->zei, rec);
1008         if (extractCtrl.staticrank < 0)
1009         {
1010             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
1011             extractCtrl.staticrank = 0;
1012         }
1013
1014         if (matchStr)
1015         {
1016             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1017             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
1018                             sizeof(*sysno), sysno);
1019         }
1020
1021         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1022 #if FLUSH2
1023         extract_flush_record_keys2(zh, *sysno,
1024                                    zh->reg->keys, extractCtrl.staticrank,
1025                                    0, recordAttr->staticrank);
1026 #else
1027         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
1028                                   extractCtrl.staticrank);
1029 #endif
1030         recordAttr->staticrank = extractCtrl.staticrank;
1031         zh->records_inserted++;
1032     }
1033     else
1034     {
1035         /* record already exists */
1036         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1037         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1038         if (action == action_insert)
1039         {
1040             yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
1041                          recordType, pr_fname, (zint) start_offset);
1042             logRecord(zh);
1043             return ZEBRA_FAIL;
1044         }
1045
1046         rec = rec_get(zh->reg->records, *sysno);
1047         assert(rec);
1048
1049         if (stream)
1050         {
1051             all_matches_add(&extractCtrl,
1052                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1053                             *sysno);
1054         }
1055
1056         recordAttr = rec_init_attr(zh->reg->zei, rec);
1057
1058         /* decrease total size */
1059         zebraExplain_recordBytesIncrement(zh->reg->zei,
1060                                            - recordAttr->recordSize);
1061
1062         zebra_rec_keys_set_buf(delkeys,
1063                                rec->info[recInfo_delKeys],
1064                                rec->size[recInfo_delKeys],
1065                                0);
1066         zebra_rec_keys_set_buf(sortKeys,
1067                                rec->info[recInfo_sortKeys],
1068                                rec->size[recInfo_sortKeys],
1069                                0);
1070
1071         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1072 #if !FLUSH2
1073         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1074                                   recordAttr->staticrank);
1075 #endif
1076         if (action == action_delete || action == action_a_delete)
1077         {
1078             /* record going to be deleted */
1079 #if FLUSH2
1080             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1081                                        delkeys, recordAttr->staticrank);
1082 #endif
1083             if (zebra_rec_keys_empty(delkeys))
1084             {
1085                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1086                         pr_fname, (zint) start_offset);
1087                 yaz_log(YLOG_WARN, "cannot delete file above, "
1088                         "storeKeys false (3)");
1089             }
1090             else
1091             {
1092                 if (show_progress)
1093                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1094                             pr_fname, (zint) start_offset);
1095                 zh->records_deleted++;
1096                 if (matchStr)
1097                 {
1098                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1099                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1100                 }
1101                 rec_del(zh->reg->records, &rec);
1102             }
1103             zebra_rec_keys_close(delkeys);
1104             zebra_rec_keys_close(sortKeys);
1105             rec_free(&rec);
1106             logRecord(zh);
1107             return ZEBRA_OK;
1108         }
1109         else
1110         {   /* update or special_update */
1111             if (show_progress)
1112                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1113                         pr_fname, (zint) start_offset);
1114             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1115
1116 #if FLUSH2
1117             extract_flush_record_keys2(zh, *sysno,
1118                                        zh->reg->keys, extractCtrl.staticrank,
1119                                        delkeys, recordAttr->staticrank);
1120 #else
1121             extract_flush_record_keys(zh, *sysno, 1,
1122                                       zh->reg->keys, extractCtrl.staticrank);
1123 #endif
1124             recordAttr->staticrank = extractCtrl.staticrank;
1125             zh->records_updated++;
1126         }
1127         zebra_rec_keys_close(delkeys);
1128         zebra_rec_keys_close(sortKeys);
1129     }
1130     /* update file type */
1131     xfree(rec->info[recInfo_fileType]);
1132     rec->info[recInfo_fileType] =
1133         rec_strdup(recordType, &rec->size[recInfo_fileType]);
1134
1135     /* update filename */
1136     xfree(rec->info[recInfo_filename]);
1137     rec->info[recInfo_filename] =
1138         rec_strdup(fname, &rec->size[recInfo_filename]);
1139
1140     /* update delete keys */
1141     xfree(rec->info[recInfo_delKeys]);
1142     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1143     {
1144         zebra_rec_keys_get_buf(zh->reg->keys,
1145                                &rec->info[recInfo_delKeys],
1146                                &rec->size[recInfo_delKeys]);
1147     }
1148     else
1149     {
1150         rec->info[recInfo_delKeys] = NULL;
1151         rec->size[recInfo_delKeys] = 0;
1152     }
1153     /* update sort keys */
1154     xfree(rec->info[recInfo_sortKeys]);
1155
1156     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1157                            &rec->info[recInfo_sortKeys],
1158                            &rec->size[recInfo_sortKeys]);
1159
1160     if (stream)
1161     {
1162         recordAttr->recordSize = end_offset - start_offset;
1163         zebraExplain_recordBytesIncrement(zh->reg->zei,
1164                                           recordAttr->recordSize);
1165     }
1166
1167     /* set run-number for this record */
1168     recordAttr->runNumber =
1169         zebraExplain_runNumberIncrement(zh->reg->zei, 0);
1170
1171     /* update store data */
1172     xfree(rec->info[recInfo_storeData]);
1173
1174     /* update store data */
1175     if (zh->store_data_buf)
1176     {
1177         rec->size[recInfo_storeData] = zh->store_data_size;
1178         rec->info[recInfo_storeData] = zh->store_data_buf;
1179         zh->store_data_buf = 0;
1180         recordAttr->recordSize = zh->store_data_size;
1181     }
1182     else if (zh->m_store_data)
1183     {
1184         off_t cur_offset = stream->tellf(stream);
1185
1186         rec->size[recInfo_storeData] = recordAttr->recordSize;
1187         rec->info[recInfo_storeData] = (char *)
1188             xmalloc(recordAttr->recordSize);
1189         stream->seekf(stream, start_offset);
1190         stream->readf(stream, rec->info[recInfo_storeData],
1191                       recordAttr->recordSize);
1192         stream->seekf(stream, cur_offset);
1193     }
1194     else
1195     {
1196         rec->info[recInfo_storeData] = NULL;
1197         rec->size[recInfo_storeData] = 0;
1198     }
1199     /* update database name */
1200     xfree(rec->info[recInfo_databaseName]);
1201     rec->info[recInfo_databaseName] =
1202         rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
1203
1204     /* update offset */
1205     recordAttr->recordOffset = start_offset;
1206
1207     /* commit this record */
1208     rec_put(zh->reg->records, &rec);
1209     logRecord(zh);
1210     return ZEBRA_OK;
1211 }
1212
1213 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1214 {
1215     ZebraHandle zh = (ZebraHandle) handle;
1216     struct recExtractCtrl extractCtrl;
1217
1218     if (zebraExplain_curDatabase(zh->reg->zei,
1219                                   rec->info[recInfo_databaseName]))
1220     {
1221         abort();
1222         if (zebraExplain_newDatabase(zh->reg->zei,
1223                                       rec->info[recInfo_databaseName], 0))
1224             abort();
1225     }
1226
1227     zebra_rec_keys_reset(zh->reg->keys);
1228     zebra_rec_keys_reset(zh->reg->sortKeys);
1229
1230     extractCtrl.init = extract_init;
1231     extractCtrl.tokenAdd = extract_token_add;
1232     extractCtrl.schemaAdd = extract_schema_add;
1233     extractCtrl.dh = zh->reg->dh;
1234
1235     init_extractCtrl(zh, &extractCtrl);
1236
1237     extractCtrl.flagShowRecords = 0;
1238     extractCtrl.match_criteria[0] = '\0';
1239     extractCtrl.staticrank = 0;
1240     extractCtrl.action = action_update;
1241
1242     extractCtrl.handle = handle;
1243     extractCtrl.first_record = 1;
1244
1245     extract_set_store_data_prepare(&extractCtrl);
1246
1247     if (n)
1248         grs_extract_tree(&extractCtrl, n);
1249
1250     if (rec->size[recInfo_delKeys])
1251     {
1252         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1253
1254         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1255
1256         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1257                                rec->size[recInfo_delKeys],
1258                                0);
1259 #if FLUSH2
1260         extract_flush_record_keys2(zh, rec->sysno,
1261                                    zh->reg->keys, 0, delkeys, 0);
1262 #else
1263         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1264         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1265 #endif
1266         zebra_rec_keys_close(delkeys);
1267
1268         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1269                                rec->size[recInfo_sortKeys],
1270                                0);
1271
1272         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1273         zebra_rec_keys_close(sortkeys);
1274     }
1275     else
1276     {
1277 #if FLUSH2
1278         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1279 #else
1280         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1281 #endif
1282     }
1283     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1284
1285     xfree(rec->info[recInfo_delKeys]);
1286     zebra_rec_keys_get_buf(zh->reg->keys,
1287                            &rec->info[recInfo_delKeys],
1288                            &rec->size[recInfo_delKeys]);
1289
1290     xfree(rec->info[recInfo_sortKeys]);
1291     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1292                            &rec->info[recInfo_sortKeys],
1293                            &rec->size[recInfo_sortKeys]);
1294     return ZEBRA_OK;
1295 }
1296
1297 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1298                           zebra_rec_keys_t reckeys,
1299                           int level)
1300 {
1301     if (zebra_rec_keys_rewind(reckeys))
1302     {
1303         size_t slen;
1304         const char *str;
1305         struct it_key key;
1306         NMEM nmem = nmem_create();
1307
1308         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1309         {
1310             char keystr[200]; /* room for zints to print */
1311             char *dst_term = 0;
1312             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1313             const char *index_type;
1314             int i;
1315             const char *string_index;
1316
1317             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1318                                     0/* db */, &string_index);
1319             assert(index_type);
1320             zebra_term_untrans_iconv(zh, nmem, index_type,
1321                                      &dst_term, str);
1322             *keystr = '\0';
1323             for (i = 0; i<key.len; i++)
1324             {
1325                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1326             }
1327
1328             if (*str < CHR_BASE_CHAR)
1329             {
1330                 int i;
1331                 char dst_buf[200]; /* room for special chars */
1332
1333                 strcpy(dst_buf , "?");
1334
1335                 if (!strcmp(str, ""))
1336                     strcpy(dst_buf, "alwaysmatches");
1337                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1338                     strcpy(dst_buf, "firstinfield");
1339                 else if (!strcmp(str, CHR_UNKNOWN))
1340                     strcpy(dst_buf, "unknown");
1341                 else if (!strcmp(str, CHR_SPACE))
1342                     strcpy(dst_buf, "space");
1343
1344                 for (i = 0; i<slen; i++)
1345                 {
1346                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1347                 }
1348                 yaz_log(level, "%s%s %s %s", keystr, index_type,
1349                         string_index, dst_buf);
1350
1351             }
1352             else
1353                 yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
1354                         string_index, dst_term);
1355
1356             nmem_reset(nmem);
1357         }
1358         nmem_destroy(nmem);
1359     }
1360 }
1361
1362 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1363                              zebra_rec_keys_t reckeys)
1364 {
1365     ZebraExplainInfo zei = zh->reg->zei;
1366     struct ord_stat {
1367         int no;
1368         int ord;
1369         struct ord_stat *next;
1370     };
1371
1372     if (zebra_rec_keys_rewind(reckeys))
1373     {
1374         struct ord_stat *ord_list = 0;
1375         struct ord_stat *p;
1376         size_t slen;
1377         const char *str;
1378         struct it_key key_in;
1379         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1380         {
1381             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1382
1383             for (p = ord_list; p ; p = p->next)
1384                 if (p->ord == ord)
1385                 {
1386                     p->no++;
1387                     break;
1388                 }
1389             if (!p)
1390             {
1391                 p = xmalloc(sizeof(*p));
1392                 p->no = 1;
1393                 p->ord = ord;
1394                 p->next = ord_list;
1395                 ord_list = p;
1396             }
1397         }
1398
1399         p = ord_list;
1400         while (p)
1401         {
1402             struct ord_stat *p1 = p;
1403
1404             if (is_insert)
1405                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1406             else
1407                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1408             p = p->next;
1409             xfree(p1);
1410         }
1411     }
1412 }
1413
1414 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1415                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1416                                 zebra_rec_keys_t del_keys, zint del_rank)
1417 {
1418     ZebraExplainInfo zei = zh->reg->zei;
1419     int normal = 0;
1420     int optimized = 0;
1421
1422     if (!zh->reg->key_block)
1423     {
1424         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1425         const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
1426         int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
1427         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1428     }
1429
1430     if (ins_keys)
1431     {
1432         extract_rec_keys_adjust(zh, 1, ins_keys);
1433         if (!del_keys)
1434             zebraExplain_recordCountIncrement(zei, 1);
1435         zebra_rec_keys_rewind(ins_keys);
1436     }
1437     if (del_keys)
1438     {
1439         extract_rec_keys_adjust(zh, 0, del_keys);
1440         if (!ins_keys)
1441             zebraExplain_recordCountIncrement(zei, -1);
1442         zebra_rec_keys_rewind(del_keys);
1443     }
1444
1445     while (1)
1446     {
1447         size_t del_slen;
1448         const char *del_str;
1449         struct it_key del_key_in;
1450         int del = 0;
1451
1452         size_t ins_slen;
1453         const char *ins_str;
1454         struct it_key ins_key_in;
1455         int ins = 0;
1456
1457         if (del_keys)
1458             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1459                                       &del_key_in);
1460         if (ins_keys)
1461             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1462                                       &ins_key_in);
1463
1464         if (del && ins && ins_rank == del_rank
1465             && !key_compare(&del_key_in, &ins_key_in)
1466             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1467         {
1468             optimized++;
1469             continue;
1470         }
1471         if (!del && !ins)
1472             break;
1473
1474         normal++;
1475         if (del)
1476             key_block_write(zh->reg->key_block, sysno,
1477                             &del_key_in, 0, del_str, del_slen,
1478                             del_rank, zh->m_staticrank);
1479         if (ins)
1480             key_block_write(zh->reg->key_block, sysno,
1481                             &ins_key_in, 1, ins_str, ins_slen,
1482                             ins_rank, zh->m_staticrank);
1483     }
1484     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1485 }
1486
1487
1488 ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
1489                                      zebra_rec_keys_t reckeys,
1490                                      zebra_snippets *snippets)
1491 {
1492     NMEM nmem = nmem_create();
1493     if (zebra_rec_keys_rewind(reckeys))
1494     {
1495         const char *str;
1496         size_t slen;
1497         struct it_key key;
1498         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1499         {
1500             char *dst_term = 0;
1501             int ord;
1502             zint seqno;
1503             const char *index_type;
1504
1505             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1506             seqno = key.mem[key.len-1];
1507             ord = CAST_ZINT_TO_INT(key.mem[0]);
1508
1509             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1510                                     0/* db */, 0 /* string_index */);
1511             assert(index_type);
1512             zebra_term_untrans_iconv(zh, nmem, index_type,
1513                                      &dst_term, str);
1514             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1515             nmem_reset(nmem);
1516         }
1517     }
1518     nmem_destroy(nmem);
1519     return ZEBRA_OK;
1520 }
1521
1522 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1523 {
1524     yaz_log(YLOG_LOG, "print_rec_keys");
1525     if (zebra_rec_keys_rewind(reckeys))
1526     {
1527         const char *str;
1528         size_t slen;
1529         struct it_key key;
1530         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1531         {
1532             char dst_buf[IT_MAX_WORD];
1533             zint seqno;
1534             const char *index_type;
1535             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1536             const char *db = 0;
1537             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1538
1539             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1540
1541             seqno = key.mem[key.len-1];
1542
1543             zebra_term_untrans(zh, index_type, dst_buf, str);
1544
1545             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1546                     " term=%s", ord, seqno, dst_buf);
1547         }
1548     }
1549 }
1550
1551 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1552                                      const char *str, int length)
1553 {
1554     struct it_key key;
1555     ZebraHandle zh = p->extractCtrl->handle;
1556     ZebraExplainInfo zei = zh->reg->zei;
1557     int ch, i;
1558
1559     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1560     if (ch < 0)
1561         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1562
1563     i = 0;
1564     key.mem[i++] = ch;
1565     key.mem[i++] = p->record_id;
1566     key.mem[i++] = p->section_id;
1567
1568     if (zh->m_segment_indexing)
1569         key.mem[i++] = p->segment;
1570     key.mem[i++] = p->seqno;
1571     key.len = i;
1572
1573     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1574 }
1575
1576 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1577 {
1578     struct it_key key;
1579     ZebraHandle zh = p->extractCtrl->handle;
1580     ZebraExplainInfo zei = zh->reg->zei;
1581     int ch;
1582     zinfo_index_category_t cat = zinfo_index_category_sort;
1583
1584     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1585     if (ch < 0)
1586         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1587     key.len = 2;
1588     key.mem[0] = ch;
1589     key.mem[1] = p->record_id;
1590
1591     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1592 }
1593
1594 static void extract_add_staticrank_string(RecWord *p,
1595                                           const char *str, int length)
1596 {
1597     char valz[40];
1598     struct recExtractCtrl *ctrl = p->extractCtrl;
1599
1600     if (length > sizeof(valz)-1)
1601         length = sizeof(valz)-1;
1602
1603     memcpy(valz, str, length);
1604     valz[length] = '\0';
1605     ctrl->staticrank = atozint(valz);
1606 }
1607
1608 static void extract_add_string(RecWord *p, zebra_map_t zm,
1609                                const char *string, int length)
1610 {
1611     assert(length > 0);
1612
1613     if (!p->index_name)
1614         return;
1615     if (log_level_details)
1616     {
1617
1618         WRBUF w = wrbuf_alloc();
1619
1620         wrbuf_write_escaped(w, string, length);
1621         yaz_log(log_level_details, "extract_add_string: %s", wrbuf_cstr(w));
1622         wrbuf_destroy(w);
1623     }
1624     if (zebra_maps_is_index(zm))
1625     {
1626         extract_add_index_string(p, zinfo_index_category_index,
1627                                  string, length);
1628         if (zebra_maps_is_alwaysmatches(zm))
1629         {
1630             RecWord word;
1631             memcpy(&word, p, sizeof(word));
1632
1633             word.seqno = 1;
1634             extract_add_index_string(
1635                 &word, zinfo_index_category_alwaysmatches, "", 0);
1636         }
1637     }
1638     else if (zebra_maps_is_sort(zm))
1639     {
1640         extract_add_sort_string(p, string, length);
1641     }
1642     else if (zebra_maps_is_staticrank(zm))
1643     {
1644         extract_add_staticrank_string(p, string, length);
1645     }
1646 }
1647
1648 static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
1649 {
1650     const char *b = p->term_buf;
1651     int remain = p->term_len;
1652     int first = 1;
1653     const char **map = 0;
1654
1655     if (remain > 0)
1656         map = zebra_maps_input(zm, &b, remain, 0);
1657
1658     while (map)
1659     {
1660         char buf[IT_MAX_WORD+1];
1661         int i, remain;
1662
1663         /* Skip spaces */
1664         while (map && *map && **map == *CHR_SPACE)
1665         {
1666             remain = p->term_len - (b - p->term_buf);
1667             if (remain > 0)
1668                 map = zebra_maps_input(zm, &b, remain, 0);
1669             else
1670                 map = 0;
1671         }
1672         if (!map)
1673             break;
1674         i = 0;
1675         while (map && *map && **map != *CHR_SPACE)
1676         {
1677             const char *cp = *map;
1678
1679             while (i < IT_MAX_WORD && *cp)
1680                 buf[i++] = *(cp++);
1681             remain = p->term_len - (b - p->term_buf);
1682             if (remain > 0)
1683                 map = zebra_maps_input(zm, &b, remain, 0);
1684             else
1685                 map = 0;
1686         }
1687         if (!i)
1688             return;
1689
1690         if (first)
1691         {
1692             first = 0;
1693             if (zebra_maps_is_first_in_field(zm))
1694             {
1695                 /* first in field marker */
1696                 extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1697                 p->seqno++;
1698             }
1699         }
1700         extract_add_string(p, zm, buf, i);
1701         p->seqno++;
1702     }
1703 }
1704
1705 static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
1706 {
1707     const char *b = p->term_buf;
1708     char buf[IT_MAX_WORD+1];
1709     const char **map = 0;
1710     int i = 0, remain = p->term_len;
1711
1712     if (remain > 0)
1713         map = zebra_maps_input(zm, &b, remain, 1);
1714
1715     while (remain > 0 && i < IT_MAX_WORD)
1716     {
1717         while (map && *map && **map == *CHR_SPACE)
1718         {
1719             remain = p->term_len - (b - p->term_buf);
1720
1721             if (remain > 0)
1722             {
1723                 int first = i ? 0 : 1;  /* first position */
1724                 map = zebra_maps_input(zm, &b, remain, first);
1725             }
1726             else
1727                 map = 0;
1728         }
1729         if (!map)
1730             break;
1731
1732         if (i && i < IT_MAX_WORD)
1733             buf[i++] = *CHR_SPACE;
1734         while (map && *map && **map != *CHR_SPACE)
1735         {
1736             const char *cp = *map;
1737
1738             if (**map == *CHR_CUT)
1739             {
1740                 i = 0;
1741             }
1742             else
1743             {
1744                 if (i >= IT_MAX_WORD)
1745                     break;
1746                 while (i < IT_MAX_WORD && *cp)
1747                     buf[i++] = *(cp++);
1748             }
1749             remain = p->term_len  - (b - p->term_buf);
1750             if (remain > 0)
1751             {
1752                 map = zebra_maps_input(zm, &b, remain, 0);
1753             }
1754             else
1755                 map = 0;
1756         }
1757     }
1758     if (!i)
1759         return;
1760     extract_add_string(p, zm, buf, i);
1761 }
1762
1763 static void extract_add_icu(RecWord *p, zebra_map_t zm)
1764 {
1765     const char *res_buf = 0;
1766     size_t res_len = 0;
1767
1768     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
1769     while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
1770     {
1771         extract_add_string(p, zm, res_buf, res_len);
1772         p->seqno++;
1773     }
1774 }
1775
1776
1777 /** \brief top-level indexing handler for recctrl system
1778     \param p token data to be indexed
1779
1780     Call sequence:
1781     extract_token_add
1782     extract_add_{in}_complete / extract_add_icu
1783     extract_add_string
1784
1785     extract_add_index_string
1786     or
1787     extract_add_sort_string
1788     or
1789     extract_add_staticrank_string
1790
1791 */
1792 static void extract_token_add(RecWord *p)
1793 {
1794     ZebraHandle zh = p->extractCtrl->handle;
1795     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
1796     WRBUF wrbuf;
1797
1798     if (log_level_details)
1799     {
1800         yaz_log(log_level_details, "extract_token_add "
1801                 "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
1802                 p->index_type, p->index_name,
1803                 p->seqno, p->term_len, p->term_buf);
1804     }
1805     if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
1806     {
1807         p->term_buf = wrbuf_buf(wrbuf);
1808         p->term_len = wrbuf_len(wrbuf);
1809     }
1810     if (zebra_maps_is_icu(zm))
1811     {
1812         extract_add_icu(p, zm);
1813     }
1814     else
1815     {
1816         if (zebra_maps_is_complete(zm))
1817             extract_add_complete_field(p, zm);
1818         else
1819             extract_add_incomplete_field(p, zm);
1820     }
1821 }
1822
1823 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1824                                       void *buf, size_t sz)
1825 {
1826     ZebraHandle zh = (ZebraHandle) p->handle;
1827
1828     xfree(zh->store_data_buf);
1829     zh->store_data_buf = 0;
1830     zh->store_data_size = 0;
1831     if (buf && sz)
1832     {
1833         zh->store_data_buf = xmalloc(sz);
1834         zh->store_data_size = sz;
1835         memcpy(zh->store_data_buf, buf, sz);
1836     }
1837 }
1838
1839 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1840 {
1841     ZebraHandle zh = (ZebraHandle) p->handle;
1842     xfree(zh->store_data_buf);
1843     zh->store_data_buf = 0;
1844     zh->store_data_size = 0;
1845     p->setStoreData = extract_set_store_data_cb;
1846 }
1847
1848 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1849 {
1850     ZebraHandle zh = (ZebraHandle) p->handle;
1851     zebraExplain_addSchema(zh->reg->zei, oid);
1852 }
1853
1854 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1855                              int cmd, zebra_rec_keys_t reckeys)
1856 {
1857 #if 0
1858     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1859             cmd, sysno);
1860     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1861 #endif
1862
1863     if (zebra_rec_keys_rewind(reckeys))
1864     {
1865         zebra_sort_index_t si = zh->reg->sort_index;
1866         size_t slen;
1867         const char *str;
1868         struct it_key key_in;
1869
1870 #define USE_SORT_ENT 1
1871 #if USE_SORT_ENT
1872         NMEM nmem = nmem_create();
1873         struct sort_add_ent {
1874             int ord;
1875             int cmd;
1876             struct sort_add_ent *next;
1877             struct zebra_sort_ent sort_ent;
1878         };
1879         struct sort_add_ent *sort_ent_list = 0;
1880 #endif
1881         zebra_sort_sysno(si, sysno);
1882
1883 #if USE_SORT_ENT
1884         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1885         {
1886             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1887
1888             struct sort_add_ent **e = &sort_ent_list;
1889             while (*e && (*e)->ord != ord)
1890                 e = &(*e)->next;
1891             if (!*e)
1892             {
1893                 *e = nmem_malloc(nmem, sizeof(**e));
1894                 (*e)->next = 0;
1895                 (*e)->sort_ent.wrbuf = wrbuf_alloc();
1896                 (*e)->sort_ent.num = 0;
1897                 (*e)->ord = ord;
1898                 (*e)->cmd = cmd;
1899             }
1900
1901             wrbuf_write((*e)->sort_ent.wrbuf, str, slen);
1902             wrbuf_putc((*e)->sort_ent.wrbuf, '\0');
1903             (*e)->sort_ent.num++;
1904         }
1905         if (sort_ent_list)
1906         {
1907             struct sort_add_ent *e = sort_ent_list;
1908             for (; e; e = e->next)
1909             {
1910                 zebra_sort_type(si, e->ord);
1911                 if (e->cmd == 1)
1912                     zebra_sort_add_ent(si, &e->sort_ent);
1913                 else
1914                     zebra_sort_delete(si);
1915                 wrbuf_destroy(e->sort_ent.wrbuf);
1916             }
1917         }
1918         nmem_destroy(nmem);
1919 #else
1920         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1921         {
1922             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1923
1924             zebra_sort_type(si, ord);
1925             if (cmd == 1)
1926                 zebra_sort_add(si, str, slen);
1927             else
1928                 zebra_sort_delete(si);
1929         }
1930 #endif
1931     }
1932 }
1933
1934 /*
1935  * Local variables:
1936  * c-basic-offset: 4
1937  * indent-tabs-mode: nil
1938  * End:
1939  * vim: shiftwidth=4 tabstop=8 expandtab
1940  */
1941