src/rpn2solr.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) Index Data
   3  * See the file LICENSE for details.
   4  */
   5 /**
   6  * \file
   7  * \brief Implements RPN to SOLR conversion
   8  */
   9 #if HAVE_CONFIG_H
  10 #include <config.h>
  11 #endif
  12
  13 #include <assert.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16 #include <yaz/rpn2solr.h>
  17 #include <yaz/xmalloc.h>
  18 #include <yaz/diagbib1.h>
  19 #include <yaz/z-core.h>
  20 #include <yaz/wrbuf.h>
  21
  22 static void wrbuf_vputs(const char *buf, void *client_data)
  23 {
  24     wrbuf_write((WRBUF) client_data, buf, strlen(buf));
  25 }
  26
  27 static const char *lookup_index_from_string_attr(Z_AttributeList *attributes)
  28 {
  29     int j;
  30     int server_choice = 1;
  31     for (j = 0; j < attributes->num_attributes; j++)
  32     {
  33         Z_AttributeElement *ae = attributes->attributes[j];
  34         if (*ae->attributeType == 1) /* use attribute */
  35         {
  36             if (ae->which == Z_AttributeValue_complex)
  37             {
  38                 Z_ComplexAttribute *ca = ae->value.complex;
  39                 int i;
  40                 for (i = 0; i < ca->num_list; i++)
  41                 {
  42                     Z_StringOrNumeric *son = ca->list[i];
  43                     if (son->which == Z_StringOrNumeric_string)
  44                         return son->u.string;
  45                 }
  46             }
  47             server_choice = 0; /* not serverChoice because we have use attr */
  48         }
  49     }
  50     if (server_choice)
  51         return "cql.serverChoice";
  52     return 0;
  53 }
  54
  55 static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
  56 {
  57     int j;
  58     for (j = 0; j < attributes->num_attributes; j++)
  59     {
  60         Z_AttributeElement *ae = attributes->attributes[j];
  61         if (*ae->attributeType == 2) /* relation attribute */
  62         {
  63             if (ae->which == Z_AttributeValue_numeric)
  64             {
  65                 /* Only support for numeric relation */
  66                 Odr_int *relation = ae->value.numeric;
  67                 /* map this numeric to representation in SOLR */
  68                 switch (*relation)
  69                 {
  70                     /* Unsure on whether this is the relation attribute constants? */
  71                 case Z_ProximityOperator_Prox_lessThan:
  72                     return "<";
  73                 case Z_ProximityOperator_Prox_lessThanOrEqual:
  74                     return "le";
  75                 case Z_ProximityOperator_Prox_equal:
  76                     return ":";
  77                 case Z_ProximityOperator_Prox_greaterThanOrEqual:
  78                     return "ge";
  79                 case Z_ProximityOperator_Prox_greaterThan:
  80                     return ">";
  81                 case Z_ProximityOperator_Prox_notEqual:
  82                     return 0;
  83                 case 100:
  84                     /* phonetic is not implemented */
  85                     return 0;
  86                 case 101:
  87                     /* stem is not not implemented */
  88                     return 0;
  89                 case 102:
  90                     /* relevance is supported in SOLR, but not implemented yet */
  91                     return 0;
  92                 default:
  93                     /* Invalid relation */
  94                     return 0;
  95                 }
  96             }
  97             else {
  98                 /*  Can we have a complex relation value?
  99                     Should we implement something?
 100                 */
 101             }
 102         }
 103     }
 104     return ":";
 105 }
 106
 107 static int check_range(solr_transform_t ct, Z_Complex *q,
 108                        Z_AttributesPlusTerm **p_apt1,
 109                        Z_AttributesPlusTerm **p_apt2)
 110 {
 111     Z_Operator *op = q->roperator;
 112     if (op->which == Z_Operator_and &&
 113         q->s1->which == Z_RPNStructure_simple &&
 114         q->s2->which == Z_RPNStructure_simple &&
 115         q->s1->u.simple->which == Z_Operand_APT &&
 116         q->s2->u.simple->which == Z_Operand_APT)
 117     {
 118         Z_AttributesPlusTerm *apt1 = q->s1->u.simple->u.attributesPlusTerm;
 119         Z_AttributesPlusTerm *apt2 = q->s2->u.simple->u.attributesPlusTerm;
 120         const char *i1 = solr_lookup_reverse(ct, "index.", apt1->attributes);
 121         const char *i2 = solr_lookup_reverse(ct, "index.", apt2->attributes);
 122         const char *rel1 = solr_lookup_reverse(ct, "relation.",
 123                                                apt1->attributes);
 124         const char *rel2 = solr_lookup_reverse(ct, "relation.",
 125                                                apt2->attributes);
 126         if (!rel1)
 127             rel1 = lookup_relation_index_from_attr(apt1->attributes);
 128         if (!rel2)
 129             rel2 = lookup_relation_index_from_attr(apt2->attributes);
 130         if (!i1)
 131             i1 = lookup_index_from_string_attr(apt1->attributes);
 132         if (!i2)
 133             i2 = lookup_index_from_string_attr(apt2->attributes);
 134         if (i1 && i2 && !strcmp(i1, i2) && rel1 && rel2)
 135         {
 136             if ((rel1[0] == '>' || rel1[0] == 'g') &&
 137                 (rel2[0] == '<' || rel2[0] == 'l'))
 138             {
 139                 *p_apt1 = apt1;
 140                 *p_apt2 = apt2;
 141                 return 1;
 142             }
 143             if ((rel2[0] == '>' || rel2[0] == 'g') &&
 144                 (rel1[0] == '<' || rel1[0] == 'l'))
 145             {
 146                 *p_apt1 = apt2;
 147                 *p_apt2 = apt1;
 148                 return 1;
 149             }
 150         }
 151     }
 152     return 0;
 153 }
 154
 155 static int rpn2solr_attr(solr_transform_t ct,
 156                          Z_AttributeList *attributes, WRBUF w)
 157 {
 158     const char *index = solr_lookup_reverse(ct, "index.", attributes);
 159     const char *structure = solr_lookup_reverse(ct, "structure.", attributes);
 160
 161     /* if transform (properties) do not match, we'll just use a USE string attribute (bug #2978) */
 162     if (!index)
 163         index = lookup_index_from_string_attr(attributes);
 164     if (!index)
 165     {
 166         solr_transform_set_error(ct,
 167                                  YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
 168         return -1;
 169     }
 170     /* for serverChoice we omit index+relation+structure */
 171     if (strcmp(index, "cql.serverChoice"))
 172     {
 173         wrbuf_puts(w, index);
 174         wrbuf_puts(w, ":");
 175         if (structure)
 176         {
 177             if (strcmp(structure, "*"))
 178             {
 179                 wrbuf_puts(w, "/");
 180                 wrbuf_puts(w, structure);
 181                 wrbuf_puts(w, " ");
 182             }
 183         }
 184     }
 185     return 0;
 186 }
 187
 188 static Odr_int get_truncation(Z_AttributesPlusTerm *apt)
 189 {
 190     int j;
 191     Z_AttributeList *attributes = apt->attributes;
 192     for (j = 0; j < attributes->num_attributes; j++)
 193     {
 194         Z_AttributeElement *ae = attributes->attributes[j];
 195         if (*ae->attributeType == 5) /* truncation attribute */
 196         {
 197             if (ae->which == Z_AttributeValue_numeric)
 198             {
 199                 return *(ae->value.numeric);
 200             }
 201             else if (ae->which == Z_AttributeValue_complex) {
 202                 ;
 203                 //yaz_log(YLOG_DEBUG, "Z_Attribute_complex");
 204                 /* Complex: Shouldn't happen */
 205             }
 206         }
 207     }
 208     /* No truncation given */
 209     return 0;
 210 }
 211
 212 #define SOLR_SPECIAL "+-&|!(){}[]^\"~*?:\\"
 213
 214 static int emit_term(solr_transform_t ct, WRBUF w, Z_Term *term, Odr_int trunc)
 215 {
 216     size_t lterm = 0;
 217     const char *sterm = 0;
 218     switch (term->which)
 219     {
 220     case Z_Term_general:
 221         lterm = term->u.general->len;
 222         sterm = (const char *) term->u.general->buf;
 223         break;
 224     case Z_Term_numeric:
 225         wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 226         break;
 227     case Z_Term_characterString:
 228         sterm = term->u.characterString;
 229         lterm = strlen(sterm);
 230         break;
 231     default:
 232         solr_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
 233         return -1;
 234     }
 235
 236     if (sterm)
 237     {
 238         size_t i;
 239         int must_quote = 0;
 240
 241         for (i = 0 ; i < lterm; i++)
 242             if (sterm[i] == ' ')
 243                 must_quote = 1;
 244         if (must_quote)
 245             wrbuf_puts(w, "\"");
 246         if (trunc == 2 || trunc == 3)
 247             wrbuf_puts(w, "*");
 248         for (i = 0 ; i < lterm; i++)
 249         {
 250             if (sterm[i] == '\\' && i < lterm - 1)
 251             {
 252                 i++;
 253                 if (strchr(SOLR_SPECIAL, sterm[i]))
 254                     wrbuf_putc(w, '\\');
 255                 wrbuf_putc(w, sterm[i]);
 256             }
 257             else if (sterm[i] == '?' && trunc == 104)
 258             {
 259                 wrbuf_putc(w, '*');
 260             }
 261             else if (sterm[i] == '#' && trunc == 104)
 262             {
 263                 wrbuf_putc(w, '?');
 264             }
 265             else if (strchr(SOLR_SPECIAL, sterm[i]))
 266             {
 267                 wrbuf_putc(w, '\\');
 268                 wrbuf_putc(w, sterm[i]);
 269             }
 270             else
 271                 wrbuf_putc(w, sterm[i]);
 272         }
 273         if (trunc == 1 || trunc == 3)
 274             wrbuf_puts(w, "*");
 275         if (must_quote)
 276             wrbuf_puts(w, "\"");
 277     }
 278     return 0;
 279 }
 280
 281 static int rpn2solr_simple(solr_transform_t ct,
 282                            void (*pr)(const char *buf, void *client_data),
 283                            void *client_data,
 284                            Z_AttributesPlusTerm *apt, WRBUF w,
 285                            Z_AttributesPlusTerm *apt2)
 286  {
 287      int ret = 0;
 288      Z_Term *term = apt->term;
 289      Odr_int trunc = get_truncation(apt);
 290      const char *relation2 = 0;
 291      const char *relation1 = solr_lookup_reverse(ct, "relation.",
 292                                                  apt->attributes);
 293      /* Attempt to fix bug #2978: Look for a relation attribute */
 294      if (!relation1)
 295          relation1 = lookup_relation_index_from_attr(apt->attributes);
 296      if (!relation1)
 297      {
 298          solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE, 0);
 299          return -1;
 300      }
 301      if (apt2)
 302      {
 303          relation2 = solr_lookup_reverse(ct, "relation.",
 304                                          apt2->attributes);
 305          if (!relation2)
 306              relation2 = lookup_relation_index_from_attr(apt2->attributes);
 307      }
 308      wrbuf_rewind(w);
 309      ret = rpn2solr_attr(ct, apt->attributes, w);
 310      if (ret)
 311          return ret;
 312      if ((trunc >= 0 && trunc <= 3) || trunc == 100 || trunc == 104)
 313              ;
 314      else
 315      {
 316          solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
 317          return -1;
 318      }
 319
 320      if (!relation1)
 321          ret = emit_term(ct, w, term, trunc);
 322      else if (relation1[0] == '<' || relation1[0] == 'l')
 323      {
 324          wrbuf_puts(w, "[* TO ");
 325          ret = emit_term(ct, w, term, trunc);
 326          if (!strcmp(relation1, "le") || !strcmp(relation1, "<="))
 327              wrbuf_puts(w, "]");
 328          else
 329              wrbuf_puts(w, "}");
 330      }
 331      else if (relation1[0] == '>' || relation1[0] == 'g')
 332      {
 333          if (!strcmp(relation1, ">=") || !strcmp(relation1, "ge"))
 334              wrbuf_puts(w, "[");
 335          else
 336              wrbuf_puts(w, "{");
 337          ret = emit_term(ct, w, term, trunc);
 338          wrbuf_puts(w, " TO ");
 339          if (apt2)
 340          {
 341              emit_term(ct, w, apt2->term, 0);
 342              if (!relation2 || !strcmp(relation2, "<=") ||
 343                  !strcmp(relation2, "le"))
 344                  wrbuf_puts(w, "]");
 345              else
 346                  wrbuf_puts(w, "}");
 347          }
 348          else
 349              wrbuf_puts(w, "*]");
 350      }
 351      else
 352          ret = emit_term(ct, w, term, trunc);
 353      if (ret == 0)
 354          pr(wrbuf_cstr(w), client_data);
 355      return ret;
 356  }
 357
 358
 359 static int rpn2solr_structure(solr_transform_t ct,
 360                               void (*pr)(const char *buf, void *client_data),
 361                                void *client_data,
 362                               Z_RPNStructure *q, int nested,
 363                               WRBUF w)
 364 {
 365     if (q->which == Z_RPNStructure_simple)
 366     {
 367         if (q->u.simple->which != Z_Operand_APT)
 368         {
 369             solr_transform_set_error(
 370                 ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 371             return -1;
 372         }
 373         else
 374             return rpn2solr_simple(ct, pr, client_data,
 375                                    q->u.simple->u.attributesPlusTerm, w, 0);
 376     }
 377     else
 378     {
 379         Z_Operator *op = q->u.complex->roperator;
 380         Z_AttributesPlusTerm *apt1, *apt2;
 381         int r;
 382
 383         if (check_range(ct, q->u.complex, &apt1, &apt2))
 384             return rpn2solr_simple(ct, pr, client_data, apt1, w, apt2);
 385         if (nested)
 386             pr("(", client_data);
 387
 388         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 389         if (r)
 390             return r;
 391         switch(op->which)
 392         {
 393         case  Z_Operator_and:
 394             pr(" AND ", client_data);
 395             break;
 396         case  Z_Operator_or:
 397             pr(" OR ", client_data);
 398             break;
 399         case  Z_Operator_and_not:
 400             pr(" AND NOT ", client_data);
 401             break;
 402         case  Z_Operator_prox:
 403             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
 404             return -1;
 405         }
 406         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 407         if (nested)
 408             pr(")", client_data);
 409         return r;
 410     }
 411 }
 412
 413 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 414                                    void (*pr)(const char *buf, void *client_data),
 415                                    void *client_data,
 416                                    Z_RPNQuery *q)
 417 {
 418     int r;
 419     WRBUF w = wrbuf_alloc();
 420     solr_transform_set_error(ct, 0, 0);
 421     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 422     wrbuf_destroy(w);
 423     return r;
 424 }
 425
 426
 427 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 428                                   WRBUF w,
 429                                   Z_RPNQuery *q)
 430 {
 431     return solr_transform_rpn2solr_stream(ct, wrbuf_vputs, w, q);
 432 }
 433
 434 /*
 435  * Local variables:
 436  * c-basic-offset: 4
 437  * c-file-style: "Stroustrup"
 438  * indent-tabs-mode: nil
 439  * End:
 440  * vim: shiftwidth=4 tabstop=8 expandtab
 441  */
 442