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 no real match, try string attribute */
 157     if (!index)
 158         index = lookup_index_from_string_attr(attributes);
 159     if (!index)
 160         return YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
 161     /* for serverChoice we omit index+relation+structure */
 162     if (strcmp(index, "cql.serverChoice"))
 163     {
 164         wrbuf_puts(w, index);
 165         wrbuf_puts(w, ":");
 166         if (structure)
 167         {
 168             if (strcmp(structure, "*"))
 169             {
 170                 wrbuf_puts(w, "/");
 171                 wrbuf_puts(w, structure);
 172                 wrbuf_puts(w, " ");
 173             }
 174         }
 175     }
 176     return 0;
 177 }
 178
 179 static Odr_int get_truncation(Z_AttributesPlusTerm *apt)
 180 {
 181     int j;
 182     Z_AttributeList *attributes = apt->attributes;
 183     for (j = 0; j < attributes->num_attributes; j++)
 184     {
 185         Z_AttributeElement *ae = attributes->attributes[j];
 186         if (*ae->attributeType == 5) /* truncation attribute */
 187         {
 188             if (ae->which == Z_AttributeValue_numeric)
 189             {
 190                 return *(ae->value.numeric);
 191             }
 192             else if (ae->which == Z_AttributeValue_complex) {
 193                 ;
 194                 //yaz_log(YLOG_DEBUG, "Z_Attribute_complex");
 195                 /* Complex: Shouldn't happen */
 196             }
 197         }
 198     }
 199     /* No truncation given */
 200     return 0;
 201 }
 202
 203 #define SOLR_SPECIAL "+-&|!(){}[]^\"~*?:\\"
 204
 205 static int emit_term(solr_transform_t ct, WRBUF w, Z_Term *term, Odr_int trunc)
 206 {
 207     size_t lterm = 0;
 208     const char *sterm = 0;
 209     switch (term->which)
 210     {
 211     case Z_Term_general:
 212         lterm = term->u.general->len;
 213         sterm = (const char *) term->u.general->buf;
 214         break;
 215     case Z_Term_numeric:
 216         wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 217         break;
 218     case Z_Term_characterString:
 219         sterm = term->u.characterString;
 220         lterm = strlen(sterm);
 221         break;
 222     default:
 223         return YAZ_BIB1_TERM_TYPE_UNSUPP;
 224     }
 225
 226     if (sterm)
 227     {
 228         size_t i;
 229         int must_quote = 0;
 230
 231         for (i = 0 ; i < lterm; i++)
 232             if (sterm[i] == ' ')
 233                 must_quote = 1;
 234         if (must_quote)
 235             wrbuf_puts(w, "\"");
 236         if (trunc == 2 || trunc == 3)
 237             wrbuf_puts(w, "*");
 238         for (i = 0 ; i < lterm; i++)
 239         {
 240             if (sterm[i] == '\\' && i < lterm - 1)
 241             {
 242                 i++;
 243                 if (strchr(SOLR_SPECIAL, sterm[i]))
 244                     wrbuf_putc(w, '\\');
 245                 wrbuf_putc(w, sterm[i]);
 246             }
 247             else if (sterm[i] == '?' && trunc == 104)
 248             {
 249                 wrbuf_putc(w, '*');
 250             }
 251             else if (sterm[i] == '#' && trunc == 104)
 252             {
 253                 wrbuf_putc(w, '?');
 254             }
 255             else if (strchr(SOLR_SPECIAL, sterm[i]))
 256             {
 257                 wrbuf_putc(w, '\\');
 258                 wrbuf_putc(w, sterm[i]);
 259             }
 260             else
 261                 wrbuf_putc(w, sterm[i]);
 262         }
 263         if (trunc == 1 || trunc == 3)
 264             wrbuf_puts(w, "*");
 265         if (must_quote)
 266             wrbuf_puts(w, "\"");
 267     }
 268     return 0;
 269 }
 270
 271 static int rpn2solr_simple(solr_transform_t ct,
 272                            void (*pr)(const char *buf, void *client_data),
 273                            void *client_data,
 274                            Z_AttributesPlusTerm *apt, WRBUF w,
 275                            Z_AttributesPlusTerm *apt2)
 276  {
 277      int ret = 0;
 278      Z_Term *term = apt->term;
 279      Odr_int trunc = get_truncation(apt);
 280      const char *relation2 = 0;
 281      const char *relation1 = solr_lookup_reverse(ct, "relation.",
 282                                                  apt->attributes);
 283      /* Attempt to fix bug #2978: Look for a relation attribute */
 284      if (!relation1)
 285          relation1 = lookup_relation_index_from_attr(apt->attributes);
 286      if (!relation1)
 287      {
 288          return YAZ_BIB1_UNSUPP_RELATION_ATTRIBUTE;
 289      }
 290      if (apt2)
 291      {
 292          relation2 = solr_lookup_reverse(ct, "relation.",
 293                                          apt2->attributes);
 294          if (!relation2)
 295              relation2 = lookup_relation_index_from_attr(apt2->attributes);
 296      }
 297      wrbuf_rewind(w);
 298      ret = rpn2solr_attr(ct, apt->attributes, w);
 299      if (ret)
 300          return ret;
 301      if ((trunc >= 0 && trunc <= 3) || trunc == 100 || trunc == 104)
 302              ;
 303      else
 304      {
 305          return YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE;
 306      }
 307
 308      if (!relation1)
 309          ret = emit_term(ct, w, term, trunc);
 310      else if (relation1[0] == '<' || relation1[0] == 'l')
 311      {
 312          wrbuf_puts(w, "[* TO ");
 313          ret = emit_term(ct, w, term, trunc);
 314          if (!strcmp(relation1, "le") || !strcmp(relation1, "<="))
 315              wrbuf_puts(w, "]");
 316          else
 317              wrbuf_puts(w, "}");
 318      }
 319      else if (relation1[0] == '>' || relation1[0] == 'g')
 320      {
 321          if (!strcmp(relation1, ">=") || !strcmp(relation1, "ge"))
 322              wrbuf_puts(w, "[");
 323          else
 324              wrbuf_puts(w, "{");
 325          ret = emit_term(ct, w, term, trunc);
 326          wrbuf_puts(w, " TO ");
 327          if (apt2)
 328          {
 329              emit_term(ct, w, apt2->term, 0);
 330              if (!relation2 || !strcmp(relation2, "<=") ||
 331                  !strcmp(relation2, "le"))
 332                  wrbuf_puts(w, "]");
 333              else
 334                  wrbuf_puts(w, "}");
 335          }
 336          else
 337              wrbuf_puts(w, "*]");
 338      }
 339      else
 340          ret = emit_term(ct, w, term, trunc);
 341      if (ret == 0)
 342          pr(wrbuf_cstr(w), client_data);
 343      return ret;
 344  }
 345
 346
 347 static int rpn2solr_structure(solr_transform_t ct,
 348                               void (*pr)(const char *buf, void *client_data),
 349                               void *client_data,
 350                               Z_RPNStructure *q, int nested,
 351                               WRBUF w)
 352 {
 353     if (q->which == Z_RPNStructure_simple)
 354     {
 355         if (q->u.simple->which != Z_Operand_APT)
 356             return YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM;
 357         else
 358             return rpn2solr_simple(ct, pr, client_data,
 359                                    q->u.simple->u.attributesPlusTerm, w, 0);
 360     }
 361     else
 362     {
 363         Z_Operator *op = q->u.complex->roperator;
 364         Z_AttributesPlusTerm *apt1, *apt2;
 365         int r;
 366
 367         if (check_range(ct, q->u.complex, &apt1, &apt2))
 368             return rpn2solr_simple(ct, pr, client_data, apt1, w, apt2);
 369         if (nested)
 370             pr("(", client_data);
 371
 372         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 373         if (r)
 374             return r;
 375         switch (op->which)
 376         {
 377         case Z_Operator_and:
 378             pr(" AND ", client_data);
 379             break;
 380         case Z_Operator_or:
 381             pr(" OR ", client_data);
 382             break;
 383         case Z_Operator_and_not:
 384             pr(" AND NOT ", client_data);
 385             break;
 386         case Z_Operator_prox:
 387             return YAZ_BIB1_UNSUPP_SEARCH;
 388         }
 389         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 390         if (nested)
 391             pr(")", client_data);
 392         return r;
 393     }
 394 }
 395
 396 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 397                                    void (*pr)(const char *buf, void *client_data),
 398                                    void *client_data,
 399                                    Z_RPNQuery *q)
 400 {
 401     int r;
 402     WRBUF w = wrbuf_alloc();
 403     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 404     if (r)
 405         solr_transform_set_error(ct, r, 0);
 406     wrbuf_destroy(w);
 407     return r;
 408 }
 409
 410
 411 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 412                                   WRBUF w,
 413                                   Z_RPNQuery *q)
 414 {
 415     return solr_transform_rpn2solr_stream(ct, wrbuf_vp_puts, w, q);
 416 }
 417
 418 /*
 419  * Local variables:
 420  * c-basic-offset: 4
 421  * c-file-style: "Stroustrup"
 422  * indent-tabs-mode: nil
 423  * End:
 424  * vim: shiftwidth=4 tabstop=8 expandtab
 425  */
 426