index/extract.c

   1 /* $Id: extract.c,v 1.269 2007-11-08 21:21:58 adam Exp $
   2    Copyright (C) 1995-2007
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 /** \file
  24     \brief indexes records and extract tokens for indexing and sorting
  25 */
  26
  27 #include <stdio.h>
  28 #include <assert.h>
  29 #include <ctype.h>
  30 #ifdef WIN32
  31 #include <io.h>
  32 #endif
  33 #if HAVE_UNISTD_H
  34 #include <unistd.h>
  35 #endif
  36 #include <fcntl.h>
  37
  38
  39 #include "index.h"
  40 #include "orddict.h"
  41 #include <direntz.h>
  42 #include <charmap.h>
  43 #include <yaz/snprintf.h>
  44
  45 static int log_level_extract = 0;
  46 static int log_level_details = 0;
  47 static int log_level_initialized = 0;
  48
  49 /* 1 if we use eliminitate identical delete/insert keys */
  50 /* eventually this the 0-case code will be removed */
  51 #define FLUSH2 1
  52
  53 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  54                                 zebra_rec_keys_t ins_keys,
  55                                 zint ins_rank,
  56                                 zebra_rec_keys_t del_keys,
  57                                 zint del_rank);
  58
  59 static void zebra_init_log_level(void)
  60 {
  61     if (!log_level_initialized)
  62     {
  63         log_level_initialized = 1;
  64
  65         log_level_extract = yaz_log_module_level("extract");
  66         log_level_details = yaz_log_module_level("indexdetails");
  67     }
  68 }
  69
  70 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  71                                     int cmd, zebra_rec_keys_t skp);
  72 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid);
  73 static void extract_token_add(RecWord *p);
  74
  75 static void check_log_limit(ZebraHandle zh)
  76 {
  77     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  78     {
  79         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  80                 zh->m_file_verbose_limit);
  81     }
  82 }
  83
  84 static void logRecord(ZebraHandle zh)
  85 {
  86     check_log_limit(zh);
  87     ++zh->records_processed;
  88     if (!(zh->records_processed % 1000))
  89     {
  90         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  91                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  92                 zh->records_processed, zh->records_inserted,
  93                 zh->records_updated, zh->records_deleted);
  94     }
  95 }
  96
  97 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
  98 {
  99     ctrl->flagShowRecords = !zh->m_flag_rw;
 100 }
 101
 102
 103 static void extract_add_index_string(RecWord *p,
 104                                       zinfo_index_category_t cat,
 105                                       const char *str, int length);
 106
 107 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
 108
 109 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 110 {
 111     w->seqno = 1;
 112     w->index_name = "any";
 113     w->index_type = "w";
 114     w->extractCtrl = p;
 115     w->record_id = 0;
 116     w->section_id = 0;
 117     w->segment = 0;
 118 }
 119
 120 struct snip_rec_info {
 121     ZebraHandle zh;
 122     zebra_snippets *snippets;
 123 };
 124
 125
 126 static void snippet_add_complete_field(RecWord *p, int ord,
 127                                        zebra_map_t zm)
 128 {
 129     struct snip_rec_info *h = p->extractCtrl->handle;
 130
 131     const char *b = p->term_buf;
 132     char buf[IT_MAX_WORD+1];
 133     const char **map = 0;
 134     int i = 0, remain = p->term_len;
 135     const char *start = b;
 136     const char *last = 0;
 137
 138     if (remain > 0)
 139         map = zebra_maps_input(zm, &b, remain, 1);
 140
 141     while (remain > 0 && i < IT_MAX_WORD)
 142     {
 143         while (map && *map && **map == *CHR_SPACE)
 144         {
 145             remain = p->term_len - (b - p->term_buf);
 146
 147             if (i == 0)
 148                 start = b;  /* set to first non-ws area */
 149             if (remain > 0)
 150             {
 151                 int first = i ? 0 : 1;  /* first position */
 152
 153                 map = zebra_maps_input(zm, &b, remain, first);
 154             }
 155             else
 156                 map = 0;
 157         }
 158         if (!map)
 159             break;
 160
 161         if (i && i < IT_MAX_WORD)
 162             buf[i++] = *CHR_SPACE;
 163         while (map && *map && **map != *CHR_SPACE)
 164         {
 165             const char *cp = *map;
 166
 167             if (**map == *CHR_CUT)
 168             {
 169                 i = 0;
 170             }
 171             else
 172             {
 173                 if (i >= IT_MAX_WORD)
 174                     break;
 175                 while (i < IT_MAX_WORD && *cp)
 176                     buf[i++] = *(cp++);
 177             }
 178             last = b;
 179             remain = p->term_len  - (b - p->term_buf);
 180             if (remain > 0)
 181             {
 182                 map = zebra_maps_input(zm, &b, remain, 0);
 183             }
 184             else
 185                 map = 0;
 186         }
 187     }
 188     if (!i)
 189         return;
 190     if (last && start != last)
 191         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 192                                start, last - start);
 193 }
 194
 195 static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 196 {
 197     struct snip_rec_info *h = p->extractCtrl->handle;
 198     const char *b = p->term_buf;
 199     int remain = p->term_len;
 200     int first = 1;
 201     const char **map = 0;
 202     const char *start = b;
 203     const char *last = b;
 204
 205     if (remain > 0)
 206         map = zebra_maps_input(zm, &b, remain, 0);
 207
 208     while (map)
 209     {
 210         char buf[IT_MAX_WORD+1];
 211         int i, remain;
 212
 213         /* Skip spaces */
 214         while (map && *map && **map == *CHR_SPACE)
 215         {
 216             remain = p->term_len - (b - p->term_buf);
 217             last = b;
 218             if (remain > 0)
 219                 map = zebra_maps_input(zm, &b, remain, 0);
 220             else
 221                 map = 0;
 222         }
 223         if (!map)
 224             break;
 225         if (start != last)
 226         {
 227             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
 228                                    start, last - start);
 229
 230         }
 231         start = last;
 232
 233         i = 0;
 234         while (map && *map && **map != *CHR_SPACE)
 235         {
 236             const char *cp = *map;
 237
 238             while (i < IT_MAX_WORD && *cp)
 239                 buf[i++] = *(cp++);
 240             remain = p->term_len - (b - p->term_buf);
 241             last = b;
 242             if (remain > 0)
 243                 map = zebra_maps_input(zm, &b, remain, 0);
 244             else
 245                 map = 0;
 246         }
 247         if (!i)
 248             return;
 249
 250         if (first)
 251         {
 252             first = 0;
 253             if (zebra_maps_is_first_in_field(zm))
 254             {
 255                 /* first in field marker */
 256                 p->seqno++;
 257             }
 258         }
 259         if (start != last)
 260             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 261                                    start, last - start);
 262         start = last;
 263         p->seqno++;
 264     }
 265
 266 }
 267
 268 static void snippet_token_add(RecWord *p)
 269 {
 270     struct snip_rec_info *h = p->extractCtrl->handle;
 271     ZebraHandle zh = h->zh;
 272     zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
 273
 274     if (zm && zebra_maps_is_index(zm))
 275     {
 276         ZebraExplainInfo zei = zh->reg->zei;
 277         int ch = zebraExplain_lookup_attr_str(
 278             zei, zinfo_index_category_index, p->index_type, p->index_name);
 279
 280         if (zebra_maps_is_complete(zm))
 281             snippet_add_complete_field(p, ch, zm);
 282         else
 283             snippet_add_incomplete_field(p, ch, zm);
 284     }
 285 }
 286
 287 static void snippet_schema_add(
 288     struct recExtractCtrl *p, Odr_oid *oid)
 289 {
 290
 291 }
 292
 293 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 294                      struct ZebraRecStream *stream,
 295                      RecType rt, void *recTypeClientData)
 296 {
 297     struct recExtractCtrl extractCtrl;
 298     struct snip_rec_info info;
 299     int r;
 300
 301     extractCtrl.stream = stream;
 302     extractCtrl.first_record = 1;
 303     extractCtrl.init = extract_init;
 304     extractCtrl.tokenAdd = snippet_token_add;
 305     extractCtrl.schemaAdd = snippet_schema_add;
 306     assert(zh->reg);
 307     assert(zh->reg->dh);
 308
 309     extractCtrl.dh = zh->reg->dh;
 310
 311     info.zh = zh;
 312     info.snippets = sn;
 313     extractCtrl.handle = &info;
 314     extractCtrl.match_criteria[0] = '\0';
 315     extractCtrl.staticrank = 0;
 316     extractCtrl.action = action_insert;
 317
 318     init_extractCtrl(zh, &extractCtrl);
 319
 320     extractCtrl.setStoreData = 0;
 321
 322     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 323
 324 }
 325
 326 static void searchRecordKey(ZebraHandle zh,
 327                             zebra_rec_keys_t reckeys,
 328                             const char *index_name,
 329                             const char **ws, int ws_length)
 330 {
 331     int i;
 332     int ch = -1;
 333     zinfo_index_category_t cat = zinfo_index_category_index;
 334
 335     for (i = 0; i<ws_length; i++)
 336         ws[i] = NULL;
 337
 338     if (ch < 0)
 339         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
 340     if (ch < 0)
 341         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
 342     if (ch < 0)
 343         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 344
 345     if (ch < 0)
 346         return ;
 347
 348     if (zebra_rec_keys_rewind(reckeys))
 349     {
 350         zint startSeq = -1;
 351         const char *str;
 352         size_t slen;
 353         struct it_key key;
 354         zint seqno;
 355         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 356         {
 357             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 358
 359             seqno = key.mem[key.len-1];
 360
 361             if (key.mem[0] == ch)
 362             {
 363                 zint woff;
 364
 365                 if (startSeq == -1)
 366                     startSeq = seqno;
 367                 woff = seqno - startSeq;
 368                 if (woff >= 0 && woff < ws_length)
 369                     ws[woff] = str;
 370             }
 371         }
 372     }
 373 }
 374
 375 #define FILE_MATCH_BLANK "\t "
 376
 377 static char *get_match_from_spec(ZebraHandle zh,
 378                           zebra_rec_keys_t reckeys,
 379                           const char *fname, const char *spec)
 380 {
 381     static char dstBuf[2048];      /* static here ??? */
 382     char *dst = dstBuf;
 383     const char *s = spec;
 384
 385     while (1)
 386     {
 387         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 388             ;
 389         if (!*s)
 390             break;
 391         if (*s == '(')
 392         {
 393             const char *ws[32];
 394             char attset_str[64], attname_str[64];
 395             int i;
 396             int first = 1;
 397
 398             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 399                 ;
 400             for (i = 0; *s && *s != ',' && *s != ')' &&
 401                      !strchr(FILE_MATCH_BLANK, *s); s++)
 402                 if (i+1 < sizeof(attset_str))
 403                     attset_str[i++] = *s;
 404             attset_str[i] = '\0';
 405
 406             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 407                 ;
 408             if (*s != ',')
 409                 strcpy(attname_str, attset_str);
 410             else
 411             {
 412                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 413                     ;
 414                 for (i = 0; *s && *s != ')' &&
 415                          !strchr(FILE_MATCH_BLANK, *s); s++)
 416                     if (i+1 < sizeof(attname_str))
 417                         attname_str[i++] = *s;
 418                 attname_str[i] = '\0';
 419             }
 420
 421             searchRecordKey(zh, reckeys, attname_str, ws, 32);
 422
 423             if (*s != ')')
 424             {
 425                 yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
 426                       spec, zh->m_group ? zh->m_group : "none");
 427                 return NULL;
 428             }
 429             s++;
 430
 431             for (i = 0; i<32; i++)
 432                 if (ws[i])
 433                 {
 434                     if (first)
 435                     {
 436                         *dst++ = ' ';
 437                         first = 0;
 438                     }
 439                     strcpy(dst, ws[i]);
 440                     dst += strlen(ws[i]);
 441                 }
 442             if (first)
 443             {
 444                 yaz_log(YLOG_WARN, "Record didn't contain match"
 445                       " fields in (%s,%s)", attset_str, attname_str);
 446                 return NULL;
 447             }
 448         }
 449         else if (*s == '$')
 450         {
 451             int spec_len;
 452             char special[64];
 453             const char *spec_src = NULL;
 454             const char *s1 = ++s;
 455             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 456                 s1++;
 457
 458             spec_len = s1 - s;
 459             if (spec_len > sizeof(special)-1)
 460                 spec_len = sizeof(special)-1;
 461             memcpy(special, s, spec_len);
 462             special[spec_len] = '\0';
 463             s = s1;
 464
 465             if (!strcmp(special, "group"))
 466                 spec_src = zh->m_group;
 467             else if (!strcmp(special, "database"))
 468                 spec_src = zh->basenames[0];
 469             else if (!strcmp(special, "filename")) {
 470                 spec_src = fname;
 471             }
 472             else if (!strcmp(special, "type"))
 473                 spec_src = zh->m_record_type;
 474             else
 475                 spec_src = NULL;
 476             if (spec_src)
 477             {
 478                 strcpy(dst, spec_src);
 479                 dst += strlen(spec_src);
 480             }
 481         }
 482         else if (*s == '\"' || *s == '\'')
 483         {
 484             int stopMarker = *s++;
 485             char tmpString[64];
 486             int i = 0;
 487
 488             while (*s && *s != stopMarker)
 489             {
 490                 if (i+1 < sizeof(tmpString))
 491                     tmpString[i++] = *s++;
 492             }
 493             if (*s)
 494                 s++;
 495             tmpString[i] = '\0';
 496             strcpy(dst, tmpString);
 497             dst += strlen(tmpString);
 498         }
 499         else
 500         {
 501             yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
 502                   spec, zh->m_group ? zh->m_group : "none");
 503             return NULL;
 504         }
 505         *dst++ = 1;
 506     }
 507     if (dst == dstBuf)
 508     {
 509         yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
 510               fname, zh->m_group ? zh->m_group : "none");
 511         return NULL;
 512     }
 513     *dst = '\0';
 514     return dstBuf;
 515 }
 516
 517 struct recordLogInfo {
 518     const char *fname;
 519     int recordOffset;
 520     struct recordGroup *rGroup;
 521 };
 522
 523 static void all_matches_add(struct recExtractCtrl *ctrl)
 524 {
 525     RecWord word;
 526     extract_init(ctrl, &word);
 527     word.index_name = "_ALLRECORDS";
 528     word.index_type = "w";
 529     word.seqno = 1;
 530     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
 531                               "", 0);
 532 }
 533
 534 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 535                                        struct ZebraRecStream *stream,
 536                                        enum zebra_recctrl_action_t action,
 537                                        int test_mode,
 538                                        const char *recordType,
 539                                        zint *sysno,
 540                                        const char *match_criteria,
 541                                        const char *fname,
 542                                        RecType recType,
 543                                        void *recTypeClientData);
 544
 545
 546 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 547                              int deleteFlag)
 548 {
 549     ZEBRA_RES r = ZEBRA_OK;
 550     int i, fd;
 551     char gprefix[128];
 552     char ext[128];
 553     char ext_res[128];
 554     struct file_read_info *fi = 0;
 555     const char *original_record_type = 0;
 556     RecType recType;
 557     void *recTypeClientData;
 558     struct ZebraRecStream stream, *streamp;
 559
 560     zebra_init_log_level();
 561
 562     if (!zh->m_group || !*zh->m_group)
 563         *gprefix = '\0';
 564     else
 565         sprintf(gprefix, "%s.", zh->m_group);
 566
 567     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 568
 569     /* determine file extension */
 570     *ext = '\0';
 571     for (i = strlen(fname); --i >= 0; )
 572         if (fname[i] == '/')
 573             break;
 574         else if (fname[i] == '.')
 575         {
 576             strcpy(ext, fname+i+1);
 577             break;
 578         }
 579     /* determine file type - depending on extension */
 580     original_record_type = zh->m_record_type;
 581     if (!zh->m_record_type)
 582     {
 583         sprintf(ext_res, "%srecordType.%s", gprefix, ext);
 584         zh->m_record_type = res_get(zh->res, ext_res);
 585     }
 586     if (!zh->m_record_type)
 587     {
 588         check_log_limit(zh);
 589         if (zh->records_processed + zh->records_skipped
 590             < zh->m_file_verbose_limit)
 591             yaz_log(YLOG_LOG, "? %s", fname);
 592         zh->records_skipped++;
 593         return 0;
 594     }
 595     /* determine match criteria */
 596     if (!zh->m_record_id)
 597     {
 598         sprintf(ext_res, "%srecordId.%s", gprefix, ext);
 599         zh->m_record_id = res_get(zh->res, ext_res);
 600     }
 601
 602     if (!(recType =
 603           recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
 604                           &recTypeClientData)))
 605     {
 606         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 607         return ZEBRA_FAIL;
 608     }
 609
 610     switch(recType->version)
 611     {
 612     case 0:
 613         break;
 614     default:
 615         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 616     }
 617     if (sysno && deleteFlag)
 618     {
 619         streamp = 0;
 620         fi = 0;
 621     }
 622     else
 623     {
 624         char full_rep[1024];
 625
 626         if (zh->path_reg && !yaz_is_abspath(fname))
 627         {
 628             strcpy(full_rep, zh->path_reg);
 629             strcat(full_rep, "/");
 630             strcat(full_rep, fname);
 631         }
 632         else
 633             strcpy(full_rep, fname);
 634
 635         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
 636         {
 637             yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 638             zh->m_record_type = original_record_type;
 639             return ZEBRA_FAIL;
 640         }
 641         streamp = &stream;
 642         zebra_create_stream_fd(streamp, fd, 0);
 643     }
 644     r = zebra_extract_records_stream(zh, streamp,
 645                                      deleteFlag ?
 646                                      action_delete : action_update,
 647                                      0, /* tst_mode */
 648                                      zh->m_record_type,
 649                                      sysno,
 650                                      0, /*match_criteria */
 651                                      fname,
 652                                      recType, recTypeClientData);
 653     if (streamp)
 654         stream.destroy(streamp);
 655     zh->m_record_type = original_record_type;
 656     return r;
 657 }
 658
 659 /*
 660   If sysno is provided, then it's used to identify the reocord.
 661   If not, and match_criteria is provided, then sysno is guessed
 662   If not, and a record is provided, then sysno is got from there
 663
 664  */
 665
 666 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 667                                       const char *buf, size_t buf_size,
 668                                       enum zebra_recctrl_action_t action,
 669                                       int test_mode,
 670                                       const char *recordType,
 671                                       zint *sysno,
 672                                       const char *match_criteria,
 673                                       const char *fname)
 674 {
 675     struct ZebraRecStream stream;
 676     ZEBRA_RES res;
 677     void *clientData;
 678     RecType recType = 0;
 679
 680     if (recordType && *recordType)
 681     {
 682         yaz_log(log_level_extract,
 683                 "Record type explicitly specified: %s", recordType);
 684         recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
 685                                   &clientData);
 686     }
 687     else
 688     {
 689         if (!(zh->m_record_type))
 690         {
 691             yaz_log(YLOG_WARN, "No such record type defined");
 692             return ZEBRA_FAIL;
 693         }
 694         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 695                 zh->m_record_type);
 696         recType = recType_byName(zh->reg->recTypes, zh->res,
 697                                   zh->m_record_type, &clientData);
 698         recordType = zh->m_record_type;
 699     }
 700
 701     if (!recType)
 702     {
 703         yaz_log(YLOG_WARN, "No such record type: %s", recordType);
 704         return ZEBRA_FAIL;
 705     }
 706
 707     zebra_create_stream_mem(&stream, buf, buf_size);
 708
 709     res = zebra_extract_records_stream(zh, &stream,
 710                                        action,
 711                                        test_mode,
 712                                        recordType,
 713                                        sysno,
 714                                        match_criteria,
 715                                        fname,
 716                                        recType, clientData);
 717     stream.destroy(&stream);
 718     return res;
 719 }
 720
 721 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 722                                        struct ZebraRecStream *stream,
 723                                        enum zebra_recctrl_action_t action,
 724                                        int test_mode,
 725                                        const char *recordType,
 726                                        zint *sysno,
 727                                        const char *match_criteria,
 728                                        const char *fname,
 729                                        RecType recType,
 730                                        void *recTypeClientData)
 731 {
 732     ZEBRA_RES res = ZEBRA_OK;
 733     while (1)
 734     {
 735         int more = 0;
 736         res = zebra_extract_record_stream(zh, stream,
 737                                           action,
 738                                           test_mode,
 739                                           recordType,
 740                                           sysno,
 741                                           match_criteria,
 742                                           fname,
 743                                           recType, recTypeClientData, &more);
 744         if (!more)
 745         {
 746             res = ZEBRA_OK;
 747             break;
 748         }
 749         if (res != ZEBRA_OK)
 750             break;
 751         if (sysno)
 752             break;
 753     }
 754     return res;
 755 }
 756
 757
 758 static WRBUF wrbuf_hex_str(const char *cstr)
 759 {
 760     size_t i;
 761     WRBUF w = wrbuf_alloc();
 762     for (i = 0; cstr[i]; i++)
 763     {
 764         if (cstr[i] < ' ' || cstr[i] > 126)
 765             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 766         else
 767             wrbuf_putc(w, cstr[i]);
 768     }
 769     return w;
 770 }
 771
 772 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 773                                       struct ZebraRecStream *stream,
 774                                       enum zebra_recctrl_action_t action,
 775                                       int test_mode,
 776                                       const char *recordType,
 777                                       zint *sysno,
 778                                       const char *match_criteria,
 779                                       const char *fname,
 780                                       RecType recType,
 781                                       void *recTypeClientData,
 782                                       int *more)
 783
 784 {
 785     zint sysno0 = 0;
 786     RecordAttr *recordAttr;
 787     struct recExtractCtrl extractCtrl;
 788     int r;
 789     const char *matchStr = 0;
 790     Record rec;
 791     off_t start_offset = 0, end_offset = 0;
 792     const char *pr_fname = fname;  /* filename to print .. */
 793     int show_progress = zh->records_processed + zh->records_skipped
 794         < zh->m_file_verbose_limit ? 1:0;
 795
 796     zebra_init_log_level();
 797
 798     if (!pr_fname)
 799         pr_fname = "<no file>";  /* make it printable if file is omitted */
 800
 801     zebra_rec_keys_reset(zh->reg->keys);
 802     zebra_rec_keys_reset(zh->reg->sortKeys);
 803
 804     if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
 805     {
 806         if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
 807                                       zh->m_explain_database))
 808             return ZEBRA_FAIL;
 809     }
 810
 811     if (stream)
 812     {
 813         off_t null_offset = 0;
 814         extractCtrl.stream = stream;
 815
 816         start_offset = stream->tellf(stream);
 817
 818         extractCtrl.first_record = start_offset ? 0 : 1;
 819
 820         stream->endf(stream, &null_offset);;
 821
 822         extractCtrl.init = extract_init;
 823         extractCtrl.tokenAdd = extract_token_add;
 824         extractCtrl.schemaAdd = extract_schema_add;
 825         extractCtrl.dh = zh->reg->dh;
 826         extractCtrl.handle = zh;
 827         extractCtrl.match_criteria[0] = '\0';
 828         extractCtrl.staticrank = 0;
 829         extractCtrl.action = action;
 830
 831         init_extractCtrl(zh, &extractCtrl);
 832
 833         extract_set_store_data_prepare(&extractCtrl);
 834
 835         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 836
 837         if (action == action_update)
 838         {
 839             action = extractCtrl.action;
 840         }
 841
 842         switch (r)
 843         {
 844         case RECCTRL_EXTRACT_EOF:
 845             return ZEBRA_FAIL;
 846         case RECCTRL_EXTRACT_ERROR_GENERIC:
 847             /* error occured during extraction ... */
 848             yaz_log(YLOG_WARN, "extract error: generic");
 849             return ZEBRA_FAIL;
 850         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 851             /* error occured during extraction ... */
 852             yaz_log(YLOG_WARN, "extract error: no such filter");
 853             return ZEBRA_FAIL;
 854         case RECCTRL_EXTRACT_SKIP:
 855             if (show_progress)
 856                 yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 857                          recordType, pr_fname, (zint) start_offset);
 858             *more = 1;
 859
 860             end_offset = stream->endf(stream, 0);
 861             if (end_offset)
 862                 stream->seekf(stream, end_offset);
 863
 864             return ZEBRA_OK;
 865         case RECCTRL_EXTRACT_OK:
 866             break;
 867         default:
 868             yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
 869             return ZEBRA_FAIL;
 870         }
 871         end_offset = stream->endf(stream, 0);
 872         if (end_offset)
 873             stream->seekf(stream, end_offset);
 874         else
 875             end_offset = stream->tellf(stream);
 876
 877         all_matches_add(&extractCtrl);
 878
 879         if (extractCtrl.match_criteria[0])
 880             match_criteria = extractCtrl.match_criteria;
 881     }
 882
 883     *more = 1;
 884     if (!sysno)
 885     {
 886         sysno = &sysno0;
 887
 888         if (match_criteria && *match_criteria) {
 889             matchStr = match_criteria;
 890         } else {
 891             if (zh->m_record_id && *zh->m_record_id) {
 892                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 893                                                zh->m_record_id);
 894                 if (!matchStr)
 895                 {
 896                     yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 897                              pr_fname, (zint) start_offset);
 898                     return ZEBRA_FAIL;
 899                 }
 900             }
 901         }
 902         if (matchStr)
 903         {
 904             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 905             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 906                                           matchStr);
 907
 908
 909             if (log_level_extract)
 910             {
 911                 WRBUF w = wrbuf_hex_str(matchStr);
 912                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 913                 wrbuf_destroy(w);
 914             }
 915             if (rinfo)
 916             {
 917                 assert(*rinfo == sizeof(*sysno));
 918                 memcpy(sysno, rinfo+1, sizeof(*sysno));
 919             }
 920        }
 921     }
 922     if (zebra_rec_keys_empty(zh->reg->keys))
 923     {
 924         /* the extraction process returned no information - the record
 925            is probably empty - unless flagShowRecords is in use */
 926         if (test_mode)
 927             return ZEBRA_OK;
 928     }
 929
 930     if (! *sysno)
 931     {
 932         /* new record */
 933         if (action == action_delete)
 934         {
 935             yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 936                          pr_fname, (zint) start_offset);
 937             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
 938             return ZEBRA_FAIL;
 939         }
 940         else if (action == action_replace)
 941         {
 942             yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 943                          pr_fname, (zint) start_offset);
 944             yaz_log(YLOG_WARN, "cannot update record above (seems new)");
 945             return ZEBRA_FAIL;
 946         }
 947         if (show_progress)
 948             yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 949                      (zint) start_offset);
 950         rec = rec_new(zh->reg->records);
 951
 952         *sysno = rec->sysno;
 953
 954         recordAttr = rec_init_attr(zh->reg->zei, rec);
 955         if (extractCtrl.staticrank < 0)
 956         {
 957             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
 958             extractCtrl.staticrank = 0;
 959         }
 960
 961         if (matchStr)
 962         {
 963             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 964             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
 965                             sizeof(*sysno), sysno);
 966         }
 967
 968         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
 969 #if FLUSH2
 970         extract_flush_record_keys2(zh, *sysno,
 971                                    zh->reg->keys, extractCtrl.staticrank,
 972                                    0, recordAttr->staticrank);
 973 #else
 974         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
 975                                   extractCtrl.staticrank);
 976 #endif
 977         recordAttr->staticrank = extractCtrl.staticrank;
 978         zh->records_inserted++;
 979     }
 980     else
 981     {
 982         /* record already exists */
 983         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
 984         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
 985         if (action == action_insert)
 986         {
 987             yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
 988                          recordType, pr_fname, (zint) start_offset);
 989             logRecord(zh);
 990             return ZEBRA_FAIL;
 991         }
 992
 993         rec = rec_get(zh->reg->records, *sysno);
 994         assert(rec);
 995
 996         recordAttr = rec_init_attr(zh->reg->zei, rec);
 997
 998         /* decrease total size */
 999         zebraExplain_recordBytesIncrement(zh->reg->zei,
1000                                            - recordAttr->recordSize);
1001
1002         zebra_rec_keys_set_buf(delkeys,
1003                                rec->info[recInfo_delKeys],
1004                                rec->size[recInfo_delKeys],
1005                                0);
1006         zebra_rec_keys_set_buf(sortKeys,
1007                                rec->info[recInfo_sortKeys],
1008                                rec->size[recInfo_sortKeys],
1009                                0);
1010
1011         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1012 #if !FLUSH2
1013         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1014                                   recordAttr->staticrank);
1015 #endif
1016         if (action == action_delete)
1017         {
1018             /* record going to be deleted */
1019 #if FLUSH2
1020             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1021                                        delkeys, recordAttr->staticrank);
1022 #endif
1023             if (zebra_rec_keys_empty(delkeys))
1024             {
1025                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1026                         pr_fname, (zint) start_offset);
1027                 yaz_log(YLOG_WARN, "cannot delete file above, "
1028                         "storeKeys false (3)");
1029             }
1030             else
1031             {
1032                 if (show_progress)
1033                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1034                             pr_fname, (zint) start_offset);
1035                 zh->records_deleted++;
1036                 if (matchStr)
1037                 {
1038                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1039                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1040                 }
1041                 rec_del(zh->reg->records, &rec);
1042             }
1043             zebra_rec_keys_close(delkeys);
1044             zebra_rec_keys_close(sortKeys);
1045             rec_free(&rec);
1046             logRecord(zh);
1047             return ZEBRA_OK;
1048         }
1049         else
1050         {   /* update or special_update */
1051             if (show_progress)
1052                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1053                         pr_fname, (zint) start_offset);
1054             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1055
1056 #if FLUSH2
1057             extract_flush_record_keys2(zh, *sysno,
1058                                        zh->reg->keys, extractCtrl.staticrank,
1059                                        delkeys, recordAttr->staticrank);
1060 #else
1061             extract_flush_record_keys(zh, *sysno, 1,
1062                                       zh->reg->keys, extractCtrl.staticrank);
1063 #endif
1064             recordAttr->staticrank = extractCtrl.staticrank;
1065             zh->records_updated++;
1066         }
1067         zebra_rec_keys_close(delkeys);
1068         zebra_rec_keys_close(sortKeys);
1069     }
1070     /* update file type */
1071     xfree(rec->info[recInfo_fileType]);
1072     rec->info[recInfo_fileType] =
1073         rec_strdup(recordType, &rec->size[recInfo_fileType]);
1074
1075     /* update filename */
1076     xfree(rec->info[recInfo_filename]);
1077     rec->info[recInfo_filename] =
1078         rec_strdup(fname, &rec->size[recInfo_filename]);
1079
1080     /* update delete keys */
1081     xfree(rec->info[recInfo_delKeys]);
1082     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1083     {
1084         zebra_rec_keys_get_buf(zh->reg->keys,
1085                                &rec->info[recInfo_delKeys],
1086                                &rec->size[recInfo_delKeys]);
1087     }
1088     else
1089     {
1090         rec->info[recInfo_delKeys] = NULL;
1091         rec->size[recInfo_delKeys] = 0;
1092     }
1093     /* update sort keys */
1094     xfree(rec->info[recInfo_sortKeys]);
1095
1096     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1097                            &rec->info[recInfo_sortKeys],
1098                            &rec->size[recInfo_sortKeys]);
1099
1100     if (stream)
1101     {
1102         recordAttr->recordSize = end_offset - start_offset;
1103         zebraExplain_recordBytesIncrement(zh->reg->zei,
1104                                           recordAttr->recordSize);
1105     }
1106
1107     /* set run-number for this record */
1108     recordAttr->runNumber =
1109         zebraExplain_runNumberIncrement(zh->reg->zei, 0);
1110
1111     /* update store data */
1112     xfree(rec->info[recInfo_storeData]);
1113
1114     /* update store data */
1115     if (zh->store_data_buf)
1116     {
1117         rec->size[recInfo_storeData] = zh->store_data_size;
1118         rec->info[recInfo_storeData] = zh->store_data_buf;
1119         zh->store_data_buf = 0;
1120         recordAttr->recordSize = zh->store_data_size;
1121     }
1122     else if (zh->m_store_data)
1123     {
1124         off_t cur_offset = stream->tellf(stream);
1125
1126         rec->size[recInfo_storeData] = recordAttr->recordSize;
1127         rec->info[recInfo_storeData] = (char *)
1128             xmalloc(recordAttr->recordSize);
1129         stream->seekf(stream, start_offset);
1130         stream->readf(stream, rec->info[recInfo_storeData],
1131                       recordAttr->recordSize);
1132         stream->seekf(stream, cur_offset);
1133     }
1134     else
1135     {
1136         rec->info[recInfo_storeData] = NULL;
1137         rec->size[recInfo_storeData] = 0;
1138     }
1139     /* update database name */
1140     xfree(rec->info[recInfo_databaseName]);
1141     rec->info[recInfo_databaseName] =
1142         rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
1143
1144     /* update offset */
1145     recordAttr->recordOffset = start_offset;
1146
1147     /* commit this record */
1148     rec_put(zh->reg->records, &rec);
1149     logRecord(zh);
1150     return ZEBRA_OK;
1151 }
1152
1153 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1154 {
1155     ZebraHandle zh = (ZebraHandle) handle;
1156     struct recExtractCtrl extractCtrl;
1157
1158     if (zebraExplain_curDatabase(zh->reg->zei,
1159                                   rec->info[recInfo_databaseName]))
1160     {
1161         abort();
1162         if (zebraExplain_newDatabase(zh->reg->zei,
1163                                       rec->info[recInfo_databaseName], 0))
1164             abort();
1165     }
1166
1167     zebra_rec_keys_reset(zh->reg->keys);
1168     zebra_rec_keys_reset(zh->reg->sortKeys);
1169
1170     extractCtrl.init = extract_init;
1171     extractCtrl.tokenAdd = extract_token_add;
1172     extractCtrl.schemaAdd = extract_schema_add;
1173     extractCtrl.dh = zh->reg->dh;
1174
1175     init_extractCtrl(zh, &extractCtrl);
1176
1177     extractCtrl.flagShowRecords = 0;
1178     extractCtrl.match_criteria[0] = '\0';
1179     extractCtrl.staticrank = 0;
1180     extractCtrl.action = action_update;
1181
1182     extractCtrl.handle = handle;
1183     extractCtrl.first_record = 1;
1184
1185     extract_set_store_data_prepare(&extractCtrl);
1186
1187     if (n)
1188         grs_extract_tree(&extractCtrl, n);
1189
1190     if (rec->size[recInfo_delKeys])
1191     {
1192         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1193
1194         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1195
1196         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1197                                rec->size[recInfo_delKeys],
1198                                0);
1199 #if FLUSH2
1200         extract_flush_record_keys2(zh, rec->sysno,
1201                                    zh->reg->keys, 0, delkeys, 0);
1202 #else
1203         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1204         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1205 #endif
1206         zebra_rec_keys_close(delkeys);
1207
1208         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1209                                rec->size[recInfo_sortKeys],
1210                                0);
1211
1212         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1213         zebra_rec_keys_close(sortkeys);
1214     }
1215     else
1216     {
1217 #if FLUSH2
1218         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1219 #else
1220         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1221 #endif
1222     }
1223     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1224
1225     xfree(rec->info[recInfo_delKeys]);
1226     zebra_rec_keys_get_buf(zh->reg->keys,
1227                            &rec->info[recInfo_delKeys],
1228                            &rec->size[recInfo_delKeys]);
1229
1230     xfree(rec->info[recInfo_sortKeys]);
1231     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1232                            &rec->info[recInfo_sortKeys],
1233                            &rec->size[recInfo_sortKeys]);
1234     return ZEBRA_OK;
1235 }
1236
1237 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1238                           zebra_rec_keys_t reckeys,
1239                           int level)
1240 {
1241     if (zebra_rec_keys_rewind(reckeys))
1242     {
1243         size_t slen;
1244         const char *str;
1245         struct it_key key;
1246         NMEM nmem = nmem_create();
1247
1248         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1249         {
1250             char keystr[200]; /* room for zints to print */
1251             char *dst_term = 0;
1252             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1253             const char *index_type;
1254             int i;
1255             const char *string_index;
1256
1257             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1258                                     0/* db */, &string_index);
1259             assert(index_type);
1260             zebra_term_untrans_iconv(zh, nmem, index_type,
1261                                      &dst_term, str);
1262             *keystr = '\0';
1263             for (i = 0; i<key.len; i++)
1264             {
1265                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1266             }
1267
1268             if (*str < CHR_BASE_CHAR)
1269             {
1270                 int i;
1271                 char dst_buf[200]; /* room for special chars */
1272
1273                 strcpy(dst_buf , "?");
1274
1275                 if (!strcmp(str, ""))
1276                     strcpy(dst_buf, "alwaysmatches");
1277                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1278                     strcpy(dst_buf, "firstinfield");
1279                 else if (!strcmp(str, CHR_UNKNOWN))
1280                     strcpy(dst_buf, "unknown");
1281                 else if (!strcmp(str, CHR_SPACE))
1282                     strcpy(dst_buf, "space");
1283
1284                 for (i = 0; i<slen; i++)
1285                 {
1286                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1287                 }
1288                 yaz_log(level, "%s%s %s %s", keystr, index_type,
1289                         string_index, dst_buf);
1290
1291             }
1292             else
1293                 yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
1294                         string_index, dst_term);
1295
1296             nmem_reset(nmem);
1297         }
1298         nmem_destroy(nmem);
1299     }
1300 }
1301
1302 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1303                              zebra_rec_keys_t reckeys)
1304 {
1305     ZebraExplainInfo zei = zh->reg->zei;
1306     struct ord_stat {
1307         int no;
1308         int ord;
1309         struct ord_stat *next;
1310     };
1311
1312     if (zebra_rec_keys_rewind(reckeys))
1313     {
1314         struct ord_stat *ord_list = 0;
1315         struct ord_stat *p;
1316         size_t slen;
1317         const char *str;
1318         struct it_key key_in;
1319         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1320         {
1321             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1322
1323             for (p = ord_list; p ; p = p->next)
1324                 if (p->ord == ord)
1325                 {
1326                     p->no++;
1327                     break;
1328                 }
1329             if (!p)
1330             {
1331                 p = xmalloc(sizeof(*p));
1332                 p->no = 1;
1333                 p->ord = ord;
1334                 p->next = ord_list;
1335                 ord_list = p;
1336             }
1337         }
1338
1339         p = ord_list;
1340         while (p)
1341         {
1342             struct ord_stat *p1 = p;
1343
1344             if (is_insert)
1345                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1346             else
1347                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1348             p = p->next;
1349             xfree(p1);
1350         }
1351     }
1352 }
1353
1354 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1355                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1356                                 zebra_rec_keys_t del_keys, zint del_rank)
1357 {
1358     ZebraExplainInfo zei = zh->reg->zei;
1359     int normal = 0;
1360     int optimized = 0;
1361
1362     if (!zh->reg->key_block)
1363     {
1364         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1365         const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
1366         int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
1367         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1368     }
1369
1370     if (ins_keys)
1371     {
1372         extract_rec_keys_adjust(zh, 1, ins_keys);
1373         if (!del_keys)
1374             zebraExplain_recordCountIncrement(zei, 1);
1375         zebra_rec_keys_rewind(ins_keys);
1376     }
1377     if (del_keys)
1378     {
1379         extract_rec_keys_adjust(zh, 0, del_keys);
1380         if (!ins_keys)
1381             zebraExplain_recordCountIncrement(zei, -1);
1382         zebra_rec_keys_rewind(del_keys);
1383     }
1384
1385     while (1)
1386     {
1387         size_t del_slen;
1388         const char *del_str;
1389         struct it_key del_key_in;
1390         int del = 0;
1391
1392         size_t ins_slen;
1393         const char *ins_str;
1394         struct it_key ins_key_in;
1395         int ins = 0;
1396
1397         if (del_keys)
1398             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1399                                       &del_key_in);
1400         if (ins_keys)
1401             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1402                                       &ins_key_in);
1403
1404         if (del && ins && ins_rank == del_rank
1405             && !key_compare(&del_key_in, &ins_key_in)
1406             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1407         {
1408             optimized++;
1409             continue;
1410         }
1411         if (!del && !ins)
1412             break;
1413
1414         normal++;
1415         if (del)
1416             key_block_write(zh->reg->key_block, sysno,
1417                             &del_key_in, 0, del_str, del_slen,
1418                             del_rank, zh->m_staticrank);
1419         if (ins)
1420             key_block_write(zh->reg->key_block, sysno,
1421                             &ins_key_in, 1, ins_str, ins_slen,
1422                             ins_rank, zh->m_staticrank);
1423     }
1424     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1425 }
1426
1427
1428 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1429                                      zebra_rec_keys_t reckeys,
1430                                      zebra_snippets *snippets)
1431 {
1432     NMEM nmem = nmem_create();
1433     if (zebra_rec_keys_rewind(reckeys))
1434     {
1435         const char *str;
1436         size_t slen;
1437         struct it_key key;
1438         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1439         {
1440             char *dst_term = 0;
1441             int ord;
1442             zint seqno;
1443             const char *index_type;
1444
1445             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1446             seqno = key.mem[key.len-1];
1447             ord = CAST_ZINT_TO_INT(key.mem[0]);
1448
1449             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1450                                     0/* db */, 0 /* string_index */);
1451             assert(index_type);
1452             zebra_term_untrans_iconv(zh, nmem, index_type,
1453                                      &dst_term, str);
1454             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1455             nmem_reset(nmem);
1456         }
1457     }
1458     nmem_destroy(nmem);
1459     return ZEBRA_OK;
1460 }
1461
1462 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1463 {
1464     yaz_log(YLOG_LOG, "print_rec_keys");
1465     if (zebra_rec_keys_rewind(reckeys))
1466     {
1467         const char *str;
1468         size_t slen;
1469         struct it_key key;
1470         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1471         {
1472             char dst_buf[IT_MAX_WORD];
1473             zint seqno;
1474             const char *index_type;
1475             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1476             const char *db = 0;
1477             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1478
1479             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1480
1481             seqno = key.mem[key.len-1];
1482
1483             zebra_term_untrans(zh, index_type, dst_buf, str);
1484
1485             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1486                     " term=%s", ord, seqno, dst_buf);
1487         }
1488     }
1489 }
1490
1491 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1492                                      const char *str, int length)
1493 {
1494     struct it_key key;
1495     ZebraHandle zh = p->extractCtrl->handle;
1496     ZebraExplainInfo zei = zh->reg->zei;
1497     int ch, i;
1498
1499     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1500     if (ch < 0)
1501         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1502
1503     i = 0;
1504     key.mem[i++] = ch;
1505     key.mem[i++] = p->record_id;
1506     key.mem[i++] = p->section_id;
1507
1508     if (zh->m_segment_indexing)
1509         key.mem[i++] = p->segment;
1510     key.mem[i++] = p->seqno;
1511     key.len = i;
1512
1513     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1514 }
1515
1516 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1517 {
1518     struct it_key key;
1519     ZebraHandle zh = p->extractCtrl->handle;
1520     ZebraExplainInfo zei = zh->reg->zei;
1521     int ch;
1522     zinfo_index_category_t cat = zinfo_index_category_sort;
1523
1524     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1525     if (ch < 0)
1526         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1527     key.len = 2;
1528     key.mem[0] = ch;
1529     key.mem[1] = p->record_id;
1530
1531     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1532 }
1533
1534 static void extract_add_staticrank_string(RecWord *p,
1535                                           const char *str, int length)
1536 {
1537     char valz[40];
1538     struct recExtractCtrl *ctrl = p->extractCtrl;
1539
1540     if (length > sizeof(valz)-1)
1541         length = sizeof(valz)-1;
1542
1543     memcpy(valz, str, length);
1544     valz[length] = '\0';
1545     ctrl->staticrank = atozint(valz);
1546 }
1547
1548 static void extract_add_string(RecWord *p, zebra_map_t zm,
1549                                const char *string, int length)
1550 {
1551     assert(length > 0);
1552
1553     if (!p->index_name)
1554         return;
1555
1556     if (zebra_maps_is_index(zm))
1557     {
1558         extract_add_index_string(p, zinfo_index_category_index,
1559                                  string, length);
1560         if (zebra_maps_is_alwaysmatches(zm))
1561         {
1562             RecWord word;
1563             memcpy(&word, p, sizeof(word));
1564
1565             word.seqno = 1;
1566             extract_add_index_string(
1567                 &word, zinfo_index_category_alwaysmatches, "", 0);
1568         }
1569     }
1570     else if (zebra_maps_is_sort(zm))
1571     {
1572         extract_add_sort_string(p, string, length);
1573     }
1574     else if (zebra_maps_is_staticrank(zm))
1575     {
1576         extract_add_staticrank_string(p, string, length);
1577     }
1578 }
1579
1580 static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
1581 {
1582     const char *b = p->term_buf;
1583     int remain = p->term_len;
1584     int first = 1;
1585     const char **map = 0;
1586
1587     if (remain > 0)
1588         map = zebra_maps_input(zm, &b, remain, 0);
1589
1590     while (map)
1591     {
1592         char buf[IT_MAX_WORD+1];
1593         int i, remain;
1594
1595         /* Skip spaces */
1596         while (map && *map && **map == *CHR_SPACE)
1597         {
1598             remain = p->term_len - (b - p->term_buf);
1599             if (remain > 0)
1600                 map = zebra_maps_input(zm, &b, remain, 0);
1601             else
1602                 map = 0;
1603         }
1604         if (!map)
1605             break;
1606         i = 0;
1607         while (map && *map && **map != *CHR_SPACE)
1608         {
1609             const char *cp = *map;
1610
1611             while (i < IT_MAX_WORD && *cp)
1612                 buf[i++] = *(cp++);
1613             remain = p->term_len - (b - p->term_buf);
1614             if (remain > 0)
1615                 map = zebra_maps_input(zm, &b, remain, 0);
1616             else
1617                 map = 0;
1618         }
1619         if (!i)
1620             return;
1621
1622         if (first)
1623         {
1624             first = 0;
1625             if (zebra_maps_is_first_in_field(zm))
1626             {
1627                 /* first in field marker */
1628                 extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1629                 p->seqno++;
1630             }
1631         }
1632         extract_add_string(p, zm, buf, i);
1633         p->seqno++;
1634     }
1635 }
1636
1637 static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
1638 {
1639     const char *b = p->term_buf;
1640     char buf[IT_MAX_WORD+1];
1641     const char **map = 0;
1642     int i = 0, remain = p->term_len;
1643
1644     if (remain > 0)
1645         map = zebra_maps_input(zm, &b, remain, 1);
1646
1647     while (remain > 0 && i < IT_MAX_WORD)
1648     {
1649         while (map && *map && **map == *CHR_SPACE)
1650         {
1651             remain = p->term_len - (b - p->term_buf);
1652
1653             if (remain > 0)
1654             {
1655                 int first = i ? 0 : 1;  /* first position */
1656                 map = zebra_maps_input(zm, &b, remain, first);
1657             }
1658             else
1659                 map = 0;
1660         }
1661         if (!map)
1662             break;
1663
1664         if (i && i < IT_MAX_WORD)
1665             buf[i++] = *CHR_SPACE;
1666         while (map && *map && **map != *CHR_SPACE)
1667         {
1668             const char *cp = *map;
1669
1670             if (**map == *CHR_CUT)
1671             {
1672                 i = 0;
1673             }
1674             else
1675             {
1676                 if (i >= IT_MAX_WORD)
1677                     break;
1678                 while (i < IT_MAX_WORD && *cp)
1679                     buf[i++] = *(cp++);
1680             }
1681             remain = p->term_len  - (b - p->term_buf);
1682             if (remain > 0)
1683             {
1684                 map = zebra_maps_input(zm, &b, remain, 0);
1685             }
1686             else
1687                 map = 0;
1688         }
1689     }
1690     if (!i)
1691         return;
1692     extract_add_string(p, zm, buf, i);
1693 }
1694
1695 static void extract_add_icu(RecWord *p, zebra_map_t zm)
1696 {
1697     struct it_key key;
1698     const char *res_buf = 0;
1699     size_t res_len = 0;
1700     ZebraHandle zh = p->extractCtrl->handle;
1701     int r = zebra_map_tokenize(zm, p->term_buf, p->term_len,
1702                                &res_buf, &res_len);
1703     int cat = zinfo_index_category_index;
1704     int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
1705     if (ch < 0)
1706         ch = zebraExplain_add_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
1707     while (r)
1708     {
1709         int i = 0;
1710         key.mem[i++] = ch;
1711         key.mem[i++] = p->record_id;
1712         key.mem[i++] = p->section_id;
1713
1714         if (zh->m_segment_indexing)
1715             key.mem[i++] = p->segment;
1716         key.mem[i++] = p->seqno;
1717         key.len = i;
1718
1719         zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
1720
1721         p->seqno++;
1722         r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len);
1723     }
1724 }
1725
1726
1727 /** \brief top-level indexing handler for recctrl system
1728     \param p token data to be indexed
1729
1730     Call sequence:
1731     extract_token
1732     zebra_add_{in}_complete
1733     extract_add_string
1734
1735     extract_add_index_string
1736     or
1737     extract_add_sort_string
1738     or
1739     extract_add_staticrank_string
1740
1741 */
1742 static void extract_token_add(RecWord *p)
1743 {
1744     ZebraHandle zh = p->extractCtrl->handle;
1745     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
1746     WRBUF wrbuf;
1747
1748     if (log_level_details)
1749     {
1750         yaz_log(log_level_details, "extract_token_add "
1751                 "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
1752                 p->index_type, p->index_name,
1753                 p->seqno, p->term_len, p->term_buf);
1754     }
1755     if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
1756     {
1757         p->term_buf = wrbuf_buf(wrbuf);
1758         p->term_len = wrbuf_len(wrbuf);
1759     }
1760     if (zebra_maps_is_icu(zm))
1761     {
1762         extract_add_icu(p, zm);
1763     }
1764     else
1765     {
1766         if (zebra_maps_is_complete(zm))
1767             extract_add_complete_field(p, zm);
1768         else
1769             extract_add_incomplete_field(p, zm);
1770     }
1771 }
1772
1773 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1774                                       void *buf, size_t sz)
1775 {
1776     ZebraHandle zh = (ZebraHandle) p->handle;
1777
1778     xfree(zh->store_data_buf);
1779     zh->store_data_buf = 0;
1780     zh->store_data_size = 0;
1781     if (buf && sz)
1782     {
1783         zh->store_data_buf = xmalloc(sz);
1784         zh->store_data_size = sz;
1785         memcpy(zh->store_data_buf, buf, sz);
1786     }
1787 }
1788
1789 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1790 {
1791     ZebraHandle zh = (ZebraHandle) p->handle;
1792     xfree(zh->store_data_buf);
1793     zh->store_data_buf = 0;
1794     zh->store_data_size = 0;
1795     p->setStoreData = extract_set_store_data_cb;
1796 }
1797
1798 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1799 {
1800     ZebraHandle zh = (ZebraHandle) p->handle;
1801     zebraExplain_addSchema(zh->reg->zei, oid);
1802 }
1803
1804 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1805                              int cmd, zebra_rec_keys_t reckeys)
1806 {
1807 #if 0
1808     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1809             cmd, sysno);
1810     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1811 #endif
1812
1813     if (zebra_rec_keys_rewind(reckeys))
1814     {
1815         zebra_sort_index_t si = zh->reg->sort_index;
1816         size_t slen;
1817         const char *str;
1818         struct it_key key_in;
1819
1820         zebra_sort_sysno(si, sysno);
1821
1822         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1823         {
1824             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1825
1826             zebra_sort_type(si, ord);
1827             if (cmd == 1)
1828                 zebra_sort_add(si, str, slen);
1829             else
1830                 zebra_sort_delete(si);
1831         }
1832     }
1833 }
1834
1835 /*
1836  * Local variables:
1837  * c-basic-offset: 4
1838  * indent-tabs-mode: nil
1839  * End:
1840  * vim: shiftwidth=4 tabstop=8 expandtab
1841  */
1842