index/retrieve.c

   1 /* $Id: retrieve.c,v 1.57 2006-11-23 14:01:21 adam Exp $
   2    Copyright (C) 1995-2006
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 #include <stdio.h>
  24 #include <assert.h>
  25
  26 #include <fcntl.h>
  27 #ifdef WIN32
  28 #include <io.h>
  29 #include <process.h>
  30 #endif
  31 #if HAVE_UNISTD_H
  32 #include <unistd.h>
  33 #endif
  34
  35 #include "index.h"
  36 #include <yaz/diagbib1.h>
  37 #include <direntz.h>
  38
  39
  40
  41
  42 static int zebra_create_record_stream(ZebraHandle zh,
  43                                Record *rec,
  44                                struct ZebraRecStream *stream){
  45
  46     RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
  47
  48     if ((*rec)->size[recInfo_storeData] > 0){
  49         zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData],
  50                                 (*rec)->size[recInfo_storeData]);
  51     }
  52     else {
  53         char full_rep[1024];
  54         int fd;
  55
  56         if (zh->path_reg && !yaz_is_abspath((*rec)->info[recInfo_filename])){
  57             strcpy(full_rep, zh->path_reg);
  58             strcat(full_rep, "/");
  59             strcat(full_rep, (*rec)->info[recInfo_filename]);
  60         }
  61         else
  62             strcpy(full_rep, (*rec)->info[recInfo_filename]);
  63
  64         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1){
  65             yaz_log (YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s",
  66                      full_rep);
  67             rec_free(rec);
  68             return 14;
  69         }
  70         zebra_create_stream_fd(stream, fd, recordAttr->recordOffset);
  71     }
  72     return 0;
  73 }
  74
  75
  76
  77 static int parse_zebra_elem(const char *elem,
  78                              const char **index, size_t *index_len,
  79                              const char **type, size_t *type_len)
  80 {
  81     *index = 0;
  82     *index_len = 0;
  83
  84     *type = 0;
  85     *type_len = 0;
  86
  87     if (elem && *elem)
  88     {
  89         char *cp;
  90         /* verify that '::' is in the beginning of *elem
  91            and something more follows */
  92         if (':' != *elem
  93             || !(elem +1) || ':' != *(elem +1)
  94             || !(elem +2) || '\0' == *(elem +2))
  95             return 0;
  96
  97         /* pick out info from string after '::' */
  98         elem = elem + 2;
  99         cp = strchr(elem, ':');
 100
 101         if (!cp) /* index, no colon, no type */
 102         {
 103             *index = elem;
 104             *index_len = strlen(elem);
 105         }
 106         else if (cp[1] == '\0') /* colon, but no following type */
 107         {
 108             return 0;
 109         }
 110         else  /* index, colon and type */
 111         {
 112             *index = elem;
 113             *index_len = cp - elem;
 114             *type = cp+1;
 115             *type_len = strlen(cp+1);
 116         }
 117     }
 118     return 1;
 119 }
 120
 121
 122 int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
 123                           Record rec,
 124                           const char *elemsetname,
 125                           oid_value input_format,
 126                           oid_value *output_format,
 127                           char **rec_bufp, int *rec_lenp)
 128 {
 129     const char *retrieval_index;
 130     size_t retrieval_index_len;
 131     const char *retrieval_type;
 132     size_t retrieval_type_len;
 133     WRBUF wrbuf = 0;
 134     zebra_rec_keys_t keys;
 135
 136     /* set output variables before processing possible error states */
 137     /* *rec_lenp = 0; */
 138
 139     /* only accept XML and SUTRS requests */
 140     if (input_format != VAL_TEXT_XML
 141         && input_format != VAL_SUTRS){
 142         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 143                 elemsetname);
 144         *output_format = VAL_NONE;
 145         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 146     }
 147
 148     if (!parse_zebra_elem(elemsetname,
 149                      &retrieval_index, &retrieval_index_len,
 150                      &retrieval_type,  &retrieval_type_len))
 151         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 152
 153     if (retrieval_type_len != 0 && retrieval_type_len != 1)
 154     {
 155         return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 156     }
 157
 158     if (retrieval_index_len)
 159     {
 160         char retrieval_index_cstr[256];
 161
 162         if (retrieval_index_len  < sizeof(retrieval_index_cstr) -1)
 163         {
 164             memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
 165             retrieval_index_cstr[retrieval_index_len] = '\0';
 166
 167             if (zebraExplain_lookup_attr_str(zh->reg->zei,
 168                                              zinfo_index_category_index,
 169                                              (retrieval_type_len == 0 ? -1 :
 170                                               retrieval_type[0]),
 171                                              retrieval_index_cstr) == -1)
 172                 return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 173         }
 174     }
 175
 176     keys = zebra_rec_keys_open();
 177     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 178                            rec->size[recInfo_delKeys], 0);
 179
 180     wrbuf = wrbuf_alloc();
 181     if (zebra_rec_keys_rewind(keys)){
 182         size_t slen;
 183         const char *str;
 184         struct it_key key_in;
 185
 186         if (input_format == VAL_TEXT_XML){
 187             *output_format = VAL_TEXT_XML;
 188             /*wrbuf_printf(wrbuf,
 189               "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");*/
 190
 191             wrbuf_printf(wrbuf,
 192                          "<record xmlns="
 193                          "\"http://www.indexdata.com/zebra/\""
 194                          " sysno=\"" ZINT_FORMAT "\""
 195                          " set=\"zebra::index%s/\">\n",
 196                          sysno, elemsetname);
 197         }
 198         else if (input_format == VAL_SUTRS)
 199             *output_format = VAL_SUTRS;
 200
 201
 202         while(zebra_rec_keys_read(keys, &str, &slen, &key_in)){
 203             int i;
 204             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 205             int index_type;
 206             const char *db = 0;
 207             const char *string_index = 0;
 208             size_t string_index_len;
 209             char dst_buf[IT_MAX_WORD];
 210
 211             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 212                                     &string_index);
 213             string_index_len = strlen(string_index);
 214
 215             /* process only if index is not defined,
 216                or if defined and matching */
 217             if (retrieval_index == 0
 218                 || (string_index_len == retrieval_index_len
 219                     && !memcmp(string_index, retrieval_index,
 220                                string_index_len))){
 221
 222                 /* process only if type is not defined, or is matching */
 223                 if (retrieval_type == 0
 224                     || (retrieval_type_len == 1
 225                         && retrieval_type[0] == index_type)){
 226
 227
 228                     zebra_term_untrans(zh, index_type, dst_buf, str);
 229                     if (strlen(dst_buf)){
 230
 231                         if (input_format == VAL_TEXT_XML){
 232                             wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 233                                          string_index);
 234
 235                             wrbuf_printf(wrbuf, " type=\"%c\"", index_type);
 236
 237                             wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 238                                          key_in.mem[key_in.len -1]);
 239
 240                             wrbuf_xmlputs(wrbuf, dst_buf);
 241                             wrbuf_printf(wrbuf, "</index>\n");
 242                         }
 243                         else if (input_format == VAL_SUTRS){
 244                             wrbuf_printf(wrbuf, "%s ", string_index);
 245
 246                             wrbuf_printf(wrbuf, "%c", index_type);
 247
 248                             for (i = 1; i < key_in.len; i++)
 249                                 wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 250                                              key_in.mem[i]);
 251
 252                         /* zebra_term_untrans(zh, index_type, dst_buf, str); */
 253                             wrbuf_printf(wrbuf, " %s", dst_buf);
 254
 255                             wrbuf_printf(wrbuf, "\n");
 256                         }
 257                     }
 258
 259                 }
 260             }
 261         }
 262         if (input_format == VAL_TEXT_XML)
 263             wrbuf_printf(wrbuf, "</record>\n");
 264      }
 265     *rec_lenp = wrbuf_len(wrbuf);
 266     *rec_bufp = odr_malloc(odr, *rec_lenp);
 267     memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 268     wrbuf_free(wrbuf, 1);
 269     zebra_rec_keys_close(keys);
 270     return 0;
 271 }
 272
 273
 274
 275 int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr,
 276                            const char *elemsetname,
 277                            oid_value input_format,
 278                            oid_value *output_format,
 279                            char **rec_bufp, int *rec_lenp)
 280 {
 281     Record rec;
 282
 283     /* set output variables before processing possible error states */
 284     /* *rec_lenp = 0; */
 285
 286     /* only accept XML and SUTRS requests */
 287     if (input_format != VAL_TEXT_XML
 288         && input_format != VAL_SUTRS){
 289         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 290                 elemsetname);
 291         return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
 292     }
 293
 294     /* processing zebra::meta::sysno elemset without fetching binary data */
 295     if (elemsetname  && 0 == strcmp(elemsetname, "meta::sysno"))
 296     {
 297         char rec_str[128];
 298         if (input_format == VAL_SUTRS){
 299             sprintf(rec_str, ZINT_FORMAT, sysno);
 300             *output_format = VAL_SUTRS;
 301         }
 302         else if (input_format == VAL_TEXT_XML){
 303             sprintf(rec_str, "<record xmlns="
 304                     "\"http://www.indexdata.com/zebra/\""
 305                              " sysno=\"" ZINT_FORMAT "\""
 306                              " set=\"zebra::%s\"/>\n",
 307                     sysno, elemsetname);
 308             *output_format = VAL_TEXT_XML;
 309         }
 310         *rec_lenp = strlen(rec_str);
 311         if (*rec_lenp){
 312             *rec_bufp = odr_strdup(odr, rec_str);
 313             return 0;
 314         } else {
 315             return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 316         }
 317     }
 318
 319     /* fetching binary record up for all other display elementsets */
 320     rec = rec_get(zh->reg->records, sysno);
 321     if (!rec){
 322         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 323         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 324     }
 325
 326     /* processing special elementsetnames zebra::meta:: */
 327     if (elemsetname && 0 == strcmp(elemsetname, "meta")){
 328         int ret = 0;
 329         char rec_str[1024];
 330         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 331
 332         if (input_format == VAL_TEXT_XML){
 333             *output_format = VAL_TEXT_XML;
 334
 335              sprintf(rec_str,
 336                      "<record xmlns="
 337                      "\"http://www.indexdata.com/zebra/\""
 338                      " sysno=\"" ZINT_FORMAT "\""
 339                      " base=\"%.*s\""
 340                      " file=\"%.*s\""
 341                      " type=\"%.*s\""
 342                      " score=\"%i\""
 343                      " rank=\"" ZINT_FORMAT "\""
 344                      " size=\"%i\""
 345                      " set=\"zebra::%s/\">\n",
 346                      sysno,
 347                      rec->size[recInfo_databaseName], rec->info[recInfo_databaseName],
 348                      rec->size[recInfo_filename], rec->info[recInfo_filename],
 349                      rec->size[recInfo_fileType], rec->info[recInfo_fileType],
 350                      score,
 351                      recordAttr->staticrank,
 352                      recordAttr->recordSize,
 353                      elemsetname);
 354         }
 355         else if (input_format == VAL_SUTRS){
 356             *output_format = VAL_SUTRS;
 357              sprintf(rec_str,
 358                      "sysno " ZINT_FORMAT "\n"
 359                      "base %.*s\n"
 360                      "file %.*s\n"
 361                      "type %.*s\n"
 362                      "score %i\n"
 363                      "rank " ZINT_FORMAT "\n"
 364                      "size %i\n"
 365                      "set zebra::%s\n",
 366                      sysno,
 367                      rec->size[recInfo_databaseName], rec->info[recInfo_databaseName],
 368                      rec->size[recInfo_filename], rec->info[recInfo_filename],
 369                      rec->size[recInfo_fileType], rec->info[recInfo_fileType],
 370                      score,
 371                      recordAttr->staticrank,
 372                      recordAttr->recordSize,
 373                      elemsetname);
 374         }
 375
 376
 377         *rec_lenp = strlen(rec_str);
 378         if (*rec_lenp){
 379             *rec_bufp = odr_strdup(odr, rec_str);
 380             ret = 0;
 381         } else {
 382             ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 383         }
 384
 385         rec_free(&rec);
 386         return ret;
 387     }
 388
 389     /* processing special elementsetnames zebra::index:: */
 390     if (elemsetname && 0 == strncmp(elemsetname, "index", 5)){
 391
 392         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
 393                                             elemsetname + 5,
 394                                             input_format, output_format,
 395                                             rec_bufp, rec_lenp);
 396
 397         rec_free(&rec);
 398         return ret;
 399     }
 400
 401     /* processing special elementsetnames zebra::data */
 402     if (elemsetname && 0 == strcmp(elemsetname, "data")){
 403         struct ZebraRecStream stream;
 404         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 405         zebra_create_record_stream(zh, &rec, &stream);
 406         *output_format = input_format;
 407         *rec_lenp = recordAttr->recordSize;
 408         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
 409         stream.readf(&stream, *rec_bufp, *rec_lenp);
 410         stream.destroy(&stream);
 411         rec_free(&rec);
 412         return 0;
 413     }
 414
 415     if (rec)
 416         rec_free(&rec);
 417     return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 418 }
 419
 420
 421 int zebra_record_fetch(ZebraHandle zh, zint sysno, int score,
 422                        zebra_snippets *hit_snippet, ODR odr,
 423                        oid_value input_format, Z_RecordComposition *comp,
 424                        oid_value *output_format,
 425                        char **rec_bufp, int *rec_lenp, char **basenamep,
 426                        char **addinfo)
 427 {
 428     Record rec;
 429     char *fname, *file_type, *basename;
 430     const char *elemsetname;
 431     struct ZebraRecStream stream;
 432     RecordAttr *recordAttr;
 433     void *clientData;
 434     int return_code = 0;
 435
 436     *basenamep = 0;
 437     *addinfo = 0;
 438     elemsetname = yaz_get_esn(comp);
 439
 440     /* processing zebra special elementset names of form 'zebra:: */
 441     if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
 442         return  zebra_special_fetch(zh, sysno, score, odr,
 443                                     elemsetname + 7,
 444                                     input_format, output_format,
 445                                     rec_bufp, rec_lenp);
 446
 447
 448     /* processing all other element set names */
 449     rec = rec_get(zh->reg->records, sysno);
 450     if (!rec)
 451     {
 452         yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 453         *basenamep = 0;
 454         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 455     }
 456
 457
 458     recordAttr = rec_init_attr(zh->reg->zei, rec);
 459
 460     file_type = rec->info[recInfo_fileType];
 461     fname = rec->info[recInfo_filename];
 462     basename = rec->info[recInfo_databaseName];
 463     *basenamep = (char *) odr_malloc (odr, strlen(basename)+1);
 464     strcpy (*basenamep, basename);
 465
 466     yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
 467             sysno, score);
 468
 469     zebra_create_record_stream(zh, &rec, &stream);
 470
 471     {
 472         /* snippets code */
 473         zebra_snippets *snippet;
 474         zebra_rec_keys_t reckeys = zebra_rec_keys_open();
 475         RecType rt;
 476         struct recRetrieveCtrl retrieveCtrl;
 477
 478         retrieveCtrl.stream = &stream;
 479         retrieveCtrl.fname = fname;
 480         retrieveCtrl.localno = sysno;
 481         retrieveCtrl.staticrank = recordAttr->staticrank;
 482         retrieveCtrl.score = score;
 483         retrieveCtrl.recordSize = recordAttr->recordSize;
 484         retrieveCtrl.odr = odr;
 485         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
 486         retrieveCtrl.comp = comp;
 487         retrieveCtrl.encoding = zh->record_encoding;
 488         retrieveCtrl.diagnostic = 0;
 489         retrieveCtrl.addinfo = 0;
 490         retrieveCtrl.dh = zh->reg->dh;
 491         retrieveCtrl.res = zh->res;
 492         retrieveCtrl.rec_buf = 0;
 493         retrieveCtrl.rec_len = -1;
 494         retrieveCtrl.hit_snippet = hit_snippet;
 495         retrieveCtrl.doc_snippet = zebra_snippets_create();
 496
 497         zebra_rec_keys_set_buf(reckeys,
 498                                rec->info[recInfo_delKeys],
 499                                rec->size[recInfo_delKeys],
 500                                0);
 501         zebra_rec_keys_to_snippets(zh, reckeys, retrieveCtrl.doc_snippet);
 502         zebra_rec_keys_close(reckeys);
 503
 504 #if 0
 505         /* for debugging purposes */
 506         yaz_log(YLOG_LOG, "DOC SNIPPET:");
 507         zebra_snippets_log(retrieveCtrl.doc_snippet, YLOG_LOG);
 508         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 509         zebra_snippets_log(retrieveCtrl.hit_snippet, YLOG_LOG);
 510 #endif
 511         snippet = zebra_snippets_window(retrieveCtrl.doc_snippet,
 512                                         retrieveCtrl.hit_snippet,
 513                                         10);
 514 #if 0
 515         /* for debugging purposes */
 516         yaz_log(YLOG_LOG, "WINDOW SNIPPET:");
 517         zebra_snippets_log(snippet, YLOG_LOG);
 518 #endif
 519
 520         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
 521                                   file_type, &clientData)))
 522         {
 523             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 524         }
 525         else
 526         {
 527             (*rt->retrieve)(clientData, &retrieveCtrl);
 528             return_code = retrieveCtrl.diagnostic;
 529
 530             *output_format = retrieveCtrl.output_format;
 531             *rec_bufp = (char *) retrieveCtrl.rec_buf;
 532             *rec_lenp = retrieveCtrl.rec_len;
 533             *addinfo = retrieveCtrl.addinfo;
 534         }
 535
 536         zebra_snippets_destroy(snippet);
 537         zebra_snippets_destroy(retrieveCtrl.doc_snippet);
 538      }
 539
 540     stream.destroy(&stream);
 541     rec_free(&rec);
 542
 543     return return_code;
 544 }
 545
 546 /*
 547  * Local variables:
 548  * c-basic-offset: 4
 549  * indent-tabs-mode: nil
 550  * End:
 551  * vim: shiftwidth=4 tabstop=8 expandtab
 552  */
 553