src/rpn2solr.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2013 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     if (sterm)
 236     {
 237         size_t i;
 238         int must_quote = 0;
 239
 240         for (i = 0 ; i < lterm; i++)
 241             if (sterm[i] == ' ')
 242                 must_quote = 1;
 243         if (must_quote)
 244             wrbuf_puts(w, "\"");
 245         for (i = 0 ; i < lterm; i++)
 246         {
 247             if (sterm[i] == '\\' && i < lterm - 1)
 248             {
 249                 i++;
 250                 if (strchr(SOLR_SPECIAL, sterm[i]))
 251                     wrbuf_putc(w, '\\');
 252                 wrbuf_putc(w, sterm[i]);
 253             }
 254             else if (sterm[i] == '?' && trunc == 104)
 255             {
 256                 wrbuf_putc(w, '*');
 257             }
 258             else if (sterm[i] == '#' && trunc == 104)
 259             {
 260                 wrbuf_putc(w, '?');
 261             }
 262             else if (strchr(SOLR_SPECIAL, sterm[i]))
 263             {
 264                 wrbuf_putc(w, '\\');
 265                 wrbuf_putc(w, sterm[i]);
 266             }
 267             else
 268                 wrbuf_putc(w, sterm[i]);
 269         }
 270         if (trunc == 1)
 271             wrbuf_puts(w, "*");
 272         if (must_quote)
 273             wrbuf_puts(w, "\"");
 274     }
 275     return 0;
 276 }
 277
 278 static int rpn2solr_simple(solr_transform_t ct,
 279                            void (*pr)(const char *buf, void *client_data),
 280                            void *client_data,
 281                            Z_AttributesPlusTerm *apt, WRBUF w,
 282                            Z_AttributesPlusTerm *apt2)
 283  {
 284      int ret = 0;
 285      Z_Term *term = apt->term;
 286      Odr_int trunc = get_truncation(apt);
 287      const char *relation2 = 0;
 288      const char *relation1 = solr_lookup_reverse(ct, "relation.",
 289                                                  apt->attributes);
 290      /* Attempt to fix bug #2978: Look for a relation attribute */
 291      if (!relation1)
 292          relation1 = lookup_relation_index_from_attr(apt->attributes);
 293      if (!relation1)
 294      {
 295          solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE, 0);
 296          return -1;
 297      }
 298      if (apt2)
 299      {
 300          relation2 = solr_lookup_reverse(ct, "relation.",
 301                                          apt2->attributes);
 302          if (!relation2)
 303              relation2 = lookup_relation_index_from_attr(apt2->attributes);
 304      }
 305      wrbuf_rewind(w);
 306      ret = rpn2solr_attr(ct, apt->attributes, w);
 307      if (ret)
 308          return ret;
 309      if (trunc == 0 || trunc == 1 || trunc == 100 || trunc == 104)
 310              ;
 311      else
 312      {
 313          solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
 314          return -1;
 315      }
 316
 317      if (!relation1)
 318          ret = emit_term(ct, w, term, trunc);
 319      else if (relation1[0] == '<' || relation1[0] == 'l')
 320      {
 321          wrbuf_puts(w, "[* TO ");
 322          ret = emit_term(ct, w, term, trunc);
 323          if (!strcmp(relation1, "le") || !strcmp(relation1, "<="))
 324              wrbuf_puts(w, "]");
 325          else
 326              wrbuf_puts(w, "}");
 327      }
 328      else if (relation1[0] == '>' || relation1[0] == 'g')
 329      {
 330          if (!strcmp(relation1, ">=") || !strcmp(relation1, "ge"))
 331              wrbuf_puts(w, "[");
 332          else
 333              wrbuf_puts(w, "{");
 334          ret = emit_term(ct, w, term, trunc);
 335          wrbuf_puts(w, " TO ");
 336          if (apt2)
 337          {
 338              emit_term(ct, w, apt2->term, 0);
 339              if (!relation2 || !strcmp(relation2, "<=") ||
 340                  !strcmp(relation2, "le"))
 341                  wrbuf_puts(w, "]");
 342              else
 343                  wrbuf_puts(w, "}");
 344          }
 345          else
 346              wrbuf_puts(w, "*]");
 347      }
 348      else
 349          ret = emit_term(ct, w, term, trunc);
 350      if (ret == 0)
 351          pr(wrbuf_cstr(w), client_data);
 352      return ret;
 353  }
 354
 355
 356 static int rpn2solr_structure(solr_transform_t ct,
 357                               void (*pr)(const char *buf, void *client_data),
 358                                void *client_data,
 359                               Z_RPNStructure *q, int nested,
 360                               WRBUF w)
 361 {
 362     if (q->which == Z_RPNStructure_simple)
 363     {
 364         if (q->u.simple->which != Z_Operand_APT)
 365         {
 366             solr_transform_set_error(
 367                 ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 368             return -1;
 369         }
 370         else
 371             return rpn2solr_simple(ct, pr, client_data,
 372                                    q->u.simple->u.attributesPlusTerm, w, 0);
 373     }
 374     else
 375     {
 376         Z_Operator *op = q->u.complex->roperator;
 377         Z_AttributesPlusTerm *apt1, *apt2;
 378         int r;
 379
 380         if (check_range(ct, q->u.complex, &apt1, &apt2))
 381             return rpn2solr_simple(ct, pr, client_data, apt1, w, apt2);
 382         if (nested)
 383             pr("(", client_data);
 384
 385         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 386         if (r)
 387             return r;
 388         switch(op->which)
 389         {
 390         case  Z_Operator_and:
 391             pr(" AND ", client_data);
 392             break;
 393         case  Z_Operator_or:
 394             pr(" OR ", client_data);
 395             break;
 396         case  Z_Operator_and_not:
 397             pr(" AND NOT ", client_data);
 398             break;
 399         case  Z_Operator_prox:
 400             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
 401             return -1;
 402         }
 403         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 404         if (nested)
 405             pr(")", client_data);
 406         return r;
 407     }
 408 }
 409
 410 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 411                                    void (*pr)(const char *buf, void *client_data),
 412                                    void *client_data,
 413                                    Z_RPNQuery *q)
 414 {
 415     int r;
 416     WRBUF w = wrbuf_alloc();
 417     solr_transform_set_error(ct, 0, 0);
 418     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 419     wrbuf_destroy(w);
 420     return r;
 421 }
 422
 423
 424 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 425                                   WRBUF w,
 426                                   Z_RPNQuery *q)
 427 {
 428     return solr_transform_rpn2solr_stream(ct, wrbuf_vputs, w, q);
 429 }
 430
 431 /*
 432  * Local variables:
 433  * c-basic-offset: 4
 434  * c-file-style: "Stroustrup"
 435  * indent-tabs-mode: nil
 436  * End:
 437  * vim: shiftwidth=4 tabstop=8 expandtab
 438  */
 439