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