index/retrieve.c

   1 /* $Id: retrieve.c,v 1.50 2006-11-15 14:09:43 marc 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 void parse_zebra_elem(const char *elem,
  78                              const char **index, size_t *index_len,
  79                              const char **type, size_t *type_len)
  80 {
  81     *type = 0;
  82     *type_len = 0;
  83
  84     *index = 0;
  85     *index_len = 0;
  86
  87     if (elem && *elem)
  88     {
  89         const char *cp = strchr(elem, ':');
  90
  91         if (!cp) /* no colon */
  92         {
  93             *index = elem;
  94             *index_len = strlen(elem);
  95         }
  96         else if (cp[1] == '\0') /* 'index:' */
  97         {
  98             *index = elem;
  99             *index_len = cp - elem;
 100         }
 101         else
 102         {
 103             *index = elem;
 104             *index_len = cp - elem;
 105             *type = cp+1;
 106             *type_len = strlen(cp+1);
 107         }
 108     }
 109 }
 110
 111
 112 int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
 113                           Record rec,
 114                           const char *elemsetname,
 115                           oid_value input_format,
 116                           oid_value *output_format,
 117                           char **rec_bufp, int *rec_lenp)
 118 {
 119     const char *retrieval_index;
 120     size_t retrieval_index_len;
 121     const char *retrieval_type;
 122     size_t retrieval_type_len;
 123
 124     /* int return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; */
 125
 126     WRBUF wrbuf = wrbuf_alloc();
 127     zebra_rec_keys_t keys;
 128
 129     /* only accept XML and SUTRS requests */
 130     if (input_format != VAL_TEXT_XML
 131         && input_format != VAL_SUTRS){
 132         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 133                 elemsetname);
 134         *output_format = VAL_NONE;
 135         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 136     }
 137
 138     keys = zebra_rec_keys_open();
 139     zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
 140                            rec->size[recInfo_delKeys], 0);
 141
 142     parse_zebra_elem(elemsetname,
 143                      &retrieval_index, &retrieval_index_len,
 144                      &retrieval_type,  &retrieval_type_len);
 145
 146     if (zebra_rec_keys_rewind(keys)){
 147         size_t slen;
 148         const char *str;
 149         struct it_key key_in;
 150
 151         if (input_format == VAL_TEXT_XML){
 152             *output_format = VAL_TEXT_XML;
 153             /*wrbuf_printf(wrbuf,
 154               "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");*/
 155
 156             wrbuf_printf(wrbuf,
 157                          "<record xmlns="
 158                          "\"http://www.indexdata.com/zebra/\""
 159                          " sysno=\"" ZINT_FORMAT "\""
 160                          " set=\"zebra::index::%s\">\n",
 161                          sysno, elemsetname);
 162         }
 163         else if (input_format == VAL_SUTRS)
 164             *output_format = VAL_SUTRS;
 165
 166
 167         while(zebra_rec_keys_read(keys, &str, &slen, &key_in)){
 168             int i;
 169             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 170             int index_type;
 171             const char *db = 0;
 172             const char *string_index = 0;
 173             size_t string_index_len;
 174             char dst_buf[IT_MAX_WORD];
 175
 176             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
 177                                     &string_index);
 178             string_index_len = strlen(string_index);
 179             if (retrieval_index == 0
 180                 || (string_index_len == retrieval_index_len
 181                     && !memcmp(string_index, retrieval_index,
 182                                string_index_len))){
 183
 184                 if (retrieval_type == 0
 185                     || (retrieval_type_len == 1
 186                         && retrieval_type[0] == index_type)){
 187
 188                     if (input_format == VAL_TEXT_XML){
 189                         wrbuf_printf(wrbuf, "  <index name=\"%s\"",
 190                                      string_index);
 191
 192                         wrbuf_printf(wrbuf, " type=\"%c\"", index_type);
 193
 194                         wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">",
 195                                      key_in.mem[key_in.len -1]);
 196
 197                         zebra_term_untrans(zh, index_type, dst_buf, str);
 198                         wrbuf_xmlputs(wrbuf, dst_buf);
 199                         wrbuf_printf(wrbuf, "</index>\n");
 200                     }
 201                     else if (input_format == VAL_SUTRS){
 202                         wrbuf_printf(wrbuf, "%s ", string_index);
 203
 204                         wrbuf_printf(wrbuf, "%c", index_type);
 205
 206                         for (i = 1; i < key_in.len; i++)
 207                             wrbuf_printf(wrbuf, " " ZINT_FORMAT,
 208                                          key_in.mem[i]);
 209
 210                         zebra_term_untrans(zh, index_type, dst_buf, str);
 211                         wrbuf_printf(wrbuf, " %s", dst_buf);
 212
 213                         wrbuf_printf(wrbuf, "\n");
 214                     }
 215                 }
 216             }
 217         }
 218         if (input_format == VAL_TEXT_XML)
 219             wrbuf_printf(wrbuf, "</record>\n");
 220      }
 221
 222     *rec_lenp = wrbuf_len(wrbuf);
 223     *rec_bufp = odr_malloc(odr, *rec_lenp);
 224     memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
 225     wrbuf_free(wrbuf, 1);
 226     zebra_rec_keys_close(keys);
 227     return 0;
 228 }
 229
 230
 231 int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
 232                            const char *elemsetname,
 233                            oid_value input_format,
 234                            oid_value *output_format,
 235                            char **rec_bufp, int *rec_lenp)
 236 {
 237     Record rec;
 238
 239     /* only accept XML and SUTRS requests */
 240     if (input_format != VAL_TEXT_XML
 241         && input_format != VAL_SUTRS){
 242         yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
 243                 elemsetname);
 244         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 245     }
 246
 247     /* processing zebra::meta::sysno elemset without fetching binary data */
 248     if (elemsetname  && 0 == strcmp(elemsetname, "meta::sysno"))
 249     {
 250         char rec_str[128];
 251         if (input_format == VAL_SUTRS){
 252             sprintf(rec_str, ZINT_FORMAT, sysno);
 253             *output_format = VAL_SUTRS;
 254         }
 255         else if (input_format == VAL_TEXT_XML){
 256             sprintf(rec_str, "<record xmlns="
 257                              "\"http://www.indexdata.com/zebra/\""
 258                              " sysno=\"" ZINT_FORMAT "\""
 259                              " set=\"zebra::%s\">\n",
 260                              sysno, elemsetname);
 261             *output_format = VAL_TEXT_XML;
 262         }
 263         *rec_lenp = strlen(rec_str);
 264         if (*rec_lenp){
 265             *rec_bufp = odr_strdup(odr, rec_str);
 266             return 0;
 267         } else {
 268             return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 269         }
 270     }
 271
 272     /* fetching binary record up for all other display elementsets */
 273     rec = rec_get(zh->reg->records, sysno);
 274     if (!rec){
 275         yaz_log(YLOG_DEBUG, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 276         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 277     }
 278
 279     /* processing special elementsetnames zebra::index:: */
 280     if (elemsetname && 0 == strncmp(elemsetname, "index::", 7)){
 281
 282         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
 283                                             elemsetname + 7,
 284                                             input_format, output_format,
 285                                             rec_bufp, rec_lenp);
 286
 287         rec_free(&rec);
 288         return ret;
 289     }
 290
 291     /* processing special elementsetnames zebra::data */
 292     if (elemsetname && 0 == strcmp(elemsetname, "data")){
 293         struct ZebraRecStream stream;
 294         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
 295         zebra_create_record_stream(zh, &rec, &stream);
 296         *output_format = VAL_SUTRS;
 297         *rec_lenp = recordAttr->recordSize;
 298         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
 299         stream.readf(&stream, *rec_bufp, *rec_lenp);
 300         stream.destroy(&stream);
 301         rec_free(&rec);
 302         return 0;
 303     }
 304
 305     if (rec)
 306         rec_free(&rec);
 307     return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 308 }
 309
 310
 311 int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
 312                        zebra_snippets *hit_snippet, ODR odr,
 313                        oid_value input_format, Z_RecordComposition *comp,
 314                        oid_value *output_format,
 315                        char **rec_bufp, int *rec_lenp, char **basenamep,
 316                        char **addinfo)
 317 {
 318     Record rec;
 319     char *fname, *file_type, *basename;
 320     const char *elemsetname;
 321     struct ZebraRecStream stream;
 322     RecordAttr *recordAttr;
 323     void *clientData;
 324     int return_code = 0;
 325
 326     *basenamep = 0;
 327     *addinfo = 0;
 328     elemsetname = yaz_get_esn(comp);
 329
 330     /*
 331     yaz_log(YLOG_LOG, "ELEMENTSET: '%s'", elemsetname);
 332     if (comp && comp->which)
 333         {
 334             yaz_log(YLOG_LOG, "%i %i", comp, comp->which);
 335         }
 336         If SRU, comp->which ==2 , and yaz_get_esn(comp) does not work correctly
 337         IF Z3055, comp->which ==1 , and everything is fine - except reord
 338         encodning
 339     */
 340
 341     /* processing zebra special elementset names of form 'zebra:: */
 342     /* SUGGESTION: do not check elemset nema here, buuuut ... */
 343     if (elemsetname  && 0 == strncmp(elemsetname, "zebra::", 7))
 344         return  zebra_special_fetch(zh, sysno, odr,
 345                                     elemsetname + 7,
 346                                     input_format, output_format,
 347                                     rec_bufp, rec_lenp);
 348
 349
 350     /* processing all other element set names */
 351     rec = rec_get(zh->reg->records, sysno);
 352     if (!rec)
 353     {
 354         yaz_log(YLOG_DEBUG, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
 355         *basenamep = 0;
 356         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 357     }
 358
 359
 360     recordAttr = rec_init_attr(zh->reg->zei, rec);
 361
 362     file_type = rec->info[recInfo_fileType];
 363     fname = rec->info[recInfo_filename];
 364     basename = rec->info[recInfo_databaseName];
 365     *basenamep = (char *) odr_malloc (odr, strlen(basename)+1);
 366     strcpy (*basenamep, basename);
 367
 368
 369     yaz_log (YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
 370              sysno, score);
 371
 372     zebra_create_record_stream(zh, &rec, &stream);
 373
 374     {
 375         /* snippets code */
 376         zebra_snippets *snippet;
 377         zebra_rec_keys_t reckeys = zebra_rec_keys_open();
 378         RecType rt;
 379         struct recRetrieveCtrl retrieveCtrl;
 380
 381         retrieveCtrl.stream = &stream;
 382         retrieveCtrl.fname = fname;
 383         retrieveCtrl.localno = sysno;
 384         retrieveCtrl.staticrank = recordAttr->staticrank;
 385         retrieveCtrl.score = score;
 386         retrieveCtrl.recordSize = recordAttr->recordSize;
 387         retrieveCtrl.odr = odr;
 388         retrieveCtrl.input_format = retrieveCtrl.output_format = input_format;
 389         retrieveCtrl.comp = comp;
 390         retrieveCtrl.encoding = zh->record_encoding;
 391         retrieveCtrl.diagnostic = 0;
 392         retrieveCtrl.addinfo = 0;
 393         retrieveCtrl.dh = zh->reg->dh;
 394         retrieveCtrl.res = zh->res;
 395         retrieveCtrl.rec_buf = 0;
 396         retrieveCtrl.rec_len = -1;
 397         retrieveCtrl.hit_snippet = hit_snippet;
 398         retrieveCtrl.doc_snippet = zebra_snippets_create();
 399
 400         zebra_rec_keys_set_buf(reckeys,
 401                                rec->info[recInfo_delKeys],
 402                                rec->size[recInfo_delKeys],
 403                                0);
 404         zebra_rec_keys_to_snippets(zh, reckeys, retrieveCtrl.doc_snippet);
 405         zebra_rec_keys_close(reckeys);
 406
 407 #if 0
 408         /* for debugging purposes */
 409         yaz_log(YLOG_LOG, "DOC SNIPPET:");
 410         zebra_snippets_log(retrieveCtrl.doc_snippet, YLOG_LOG);
 411         yaz_log(YLOG_LOG, "HIT SNIPPET:");
 412         zebra_snippets_log(retrieveCtrl.hit_snippet, YLOG_LOG);
 413 #endif
 414         snippet = zebra_snippets_window(retrieveCtrl.doc_snippet,
 415                                         retrieveCtrl.hit_snippet,
 416                                         10);
 417 #if 0
 418         /* for debugging purposes */
 419         yaz_log(YLOG_LOG, "WINDOW SNIPPET:");
 420         zebra_snippets_log(snippet, YLOG_LOG);
 421 #endif
 422
 423         /* SUGGESTION: do not check elemset name here, buuuut ...
 424            add another recType Struct with zebra internal stuff here,
 425            which overrides the Alvis/GRS-1/Safari filters ....
 426         */
 427         if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
 428                                   file_type, &clientData)))
 429         {
 430             return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 431         }
 432         else
 433         {
 434         /* SUGGESTION: do not check elemset name here, buuuut ...
 435            add another recType Struct with zebra internal stuff here,
 436            which overrides the Alvis/GRS-1/Safari filters ....
 437            (*rt->retrieve) method to make the correct encoded, etc, retrieval,
 438            where all needed info already is found in the   &retrieveCtr
 439            parameter. This way, we do not need to re-code/dublicate a lot of
 440            logic.
 441         */
 442             (*rt->retrieve)(clientData, &retrieveCtrl);
 443             return_code = retrieveCtrl.diagnostic;
 444
 445             *output_format = retrieveCtrl.output_format;
 446             *rec_bufp = (char *) retrieveCtrl.rec_buf;
 447             *rec_lenp = retrieveCtrl.rec_len;
 448             *addinfo = retrieveCtrl.addinfo;
 449         }
 450         /* another SUGGESTION: throw out all this snippet stuff in this
 451            file, and do it correctly - either inside filters, or in another
 452            new   (*rt->retrieve)(clientData, &retrieveCtrl); type 'snippet'
 453         */
 454
 455         zebra_snippets_destroy(snippet);
 456         zebra_snippets_destroy(retrieveCtrl.doc_snippet);
 457      }
 458
 459     stream.destroy(&stream);
 460     rec_free(&rec);
 461
 462     return return_code;
 463 }
 464
 465 /*
 466  * Local variables:
 467  * c-basic-offset: 4
 468  * indent-tabs-mode: nil
 469  * End:
 470  * vim: shiftwidth=4 tabstop=8 expandtab
 471  */
 472