src/rpn2solr.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2012 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 "solr.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 0;
  73                 case Z_ProximityOperator_Prox_lessThanOrEqual:
  74                     return 0;
  75                 case Z_ProximityOperator_Prox_equal:
  76                     return ":";
  77                 case Z_ProximityOperator_Prox_greaterThanOrEqual:
  78                     return 0;
  79                 case Z_ProximityOperator_Prox_greaterThan:
  80                     return 0;
  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 rpn2solr_attr(solr_transform_t ct,
 108                          Z_AttributeList *attributes, WRBUF w)
 109 {
 110     const char *relation = solr_lookup_reverse(ct, "relation.", attributes);
 111     const char *index = solr_lookup_reverse(ct, "index.", attributes);
 112     const char *structure = solr_lookup_reverse(ct, "structure.", attributes);
 113
 114     /* if transform (properties) do not match, we'll just use a USE string attribute (bug #2978) */
 115     if (!index)
 116         index = lookup_index_from_string_attr(attributes);
 117
 118     /* Attempt to fix bug #2978: Look for a relation attribute */
 119     if (!relation)
 120         relation = lookup_relation_index_from_attr(attributes);
 121
 122     if (!index)
 123     {
 124         solr_transform_set_error(ct,
 125                                  YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
 126         return -1;
 127     }
 128     /* for serverChoice we omit index+relation+structure */
 129     if (strcmp(index, "solr.serverChoice"))
 130     {
 131         wrbuf_puts(w, index);
 132         if (relation)
 133         {
 134             if (!strcmp(relation, "exact"))
 135                 /* TODO Verify if a exact  SOLR exists */
 136                 relation = ":";
 137             else if (!strcmp(relation, "eq"))
 138                 relation = ":";
 139             else if (!strcmp(relation, "le")) {
 140                 /* TODO Not support as such, but could perhaps be transformed into a range
 141                    relation = ":[ * to ";
 142                    close_range = "]"
 143                 */
 144             }
 145             else if (!strcmp(relation, "ge")) {
 146                 /* TODO Not support as such, but could perhaps be transformed into a range
 147                    relation = "[";
 148                    relation = ":[ * to ";
 149                    close_range = "]"
 150                 */
 151             }
 152             /* Missing mapping of not equal, phonetic, stem and relevance */
 153             wrbuf_puts(w, relation);
 154         }
 155         else
 156             wrbuf_puts(w, ":");
 157
 158         if (structure)
 159         {
 160             if (strcmp(structure, "*"))
 161             {
 162                 wrbuf_puts(w, "/");
 163                 wrbuf_puts(w, structure);
 164                 wrbuf_puts(w, " ");
 165             }
 166         }
 167     }
 168     return 0;
 169 }
 170
 171 static Odr_int get_truncation(Z_AttributesPlusTerm *apt)
 172 {
 173     int j;
 174     Z_AttributeList *attributes = apt->attributes;
 175     for (j = 0; j < attributes->num_attributes; j++)
 176     {
 177         Z_AttributeElement *ae = attributes->attributes[j];
 178         if (*ae->attributeType == 5) /* truncation attribute */
 179         {
 180             if (ae->which == Z_AttributeValue_numeric)
 181             {
 182                 return *(ae->value.numeric);
 183             }
 184             else if (ae->which == Z_AttributeValue_complex) {
 185                 ;
 186                 //yaz_log(YLOG_DEBUG, "Z_Attribute_complex");
 187                 /* Complex: Shouldn't happen */
 188             }
 189         }
 190     }
 191     /* No truncation given */
 192     return 0;
 193 }
 194
 195 #define SOLR_SPECIAL "+-&|!(){}[]^\"~*?:\\"
 196
 197 static int rpn2solr_simple(solr_transform_t ct,
 198                            void (*pr)(const char *buf, void *client_data),
 199                            void *client_data,
 200                            Z_Operand *q, WRBUF w)
 201 {
 202     int ret = 0;
 203     if (q->which != Z_Operand_APT)
 204     {
 205         ret = -1;
 206         solr_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 207     }
 208     else
 209     {
 210         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
 211         Z_Term *term = apt->term;
 212         const char *sterm = 0;
 213         size_t lterm = 0;
 214         Odr_int trunc = get_truncation(apt);
 215
 216         wrbuf_rewind(w);
 217         ret = rpn2solr_attr(ct, apt->attributes, w);
 218
 219         if (trunc == 0 || trunc == 1 || trunc == 100 || trunc == 104)
 220             ;
 221         else
 222         {
 223             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
 224             return -1;
 225         }
 226         switch (term->which)
 227         {
 228         case Z_Term_general:
 229             lterm = term->u.general->len;
 230             sterm = (const char *) term->u.general->buf;
 231             break;
 232         case Z_Term_numeric:
 233             wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 234             break;
 235         case Z_Term_characterString:
 236             sterm = term->u.characterString;
 237             lterm = strlen(sterm);
 238             break;
 239         default:
 240             ret = -1;
 241             solr_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
 242         }
 243
 244         if (sterm)
 245         {
 246             size_t i;
 247             int must_quote = 0;
 248
 249             for (i = 0 ; i < lterm; i++)
 250                 if (sterm[i] == ' ')
 251                     must_quote = 1;
 252             if (must_quote)
 253                 wrbuf_puts(w, "\"");
 254             for (i = 0 ; i < lterm; i++)
 255             {
 256                 if (sterm[i] == '\\' && i < lterm - 1)
 257                 {
 258                     i++;
 259                     if (strchr(SOLR_SPECIAL, sterm[i]))
 260                         wrbuf_putc(w, '\\');
 261                     wrbuf_putc(w, sterm[i]);
 262                 }
 263                 else if (sterm[i] == '?' && trunc == 104)
 264                 {
 265                     wrbuf_putc(w, '*');
 266                 }
 267                 else if (sterm[i] == '#' && trunc == 104)
 268                 {
 269                     wrbuf_putc(w, '?');
 270                 }
 271                 else if (strchr(SOLR_SPECIAL, sterm[i]))
 272                 {
 273                     wrbuf_putc(w, '\\');
 274                     wrbuf_putc(w, sterm[i]);
 275                 }
 276                 else
 277                     wrbuf_putc(w, sterm[i]);
 278             }
 279             if (trunc == 1)
 280                 wrbuf_puts(w, "*");
 281             if (must_quote)
 282                 wrbuf_puts(w, "\"");
 283         }
 284         if (ret == 0)
 285             pr(wrbuf_cstr(w), client_data);
 286     }
 287     return ret;
 288 }
 289
 290
 291 static int rpn2solr_structure(solr_transform_t ct,
 292                               void (*pr)(const char *buf, void *client_data),
 293                               void *client_data,
 294                               Z_RPNStructure *q, int nested,
 295                               WRBUF w)
 296 {
 297     if (q->which == Z_RPNStructure_simple)
 298         return rpn2solr_simple(ct, pr, client_data, q->u.simple, w);
 299     else
 300     {
 301         Z_Operator *op = q->u.complex->roperator;
 302         int r;
 303
 304         if (nested)
 305             pr("(", client_data);
 306
 307         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 308         if (r)
 309             return r;
 310         switch(op->which)
 311         {
 312         case  Z_Operator_and:
 313             pr(" AND ", client_data);
 314             break;
 315         case  Z_Operator_or:
 316             pr(" OR ", client_data);
 317             break;
 318         case  Z_Operator_and_not:
 319             pr(" AND NOT ", client_data);
 320             break;
 321         case  Z_Operator_prox:
 322             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
 323             return -1;
 324         }
 325         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 326         if (nested)
 327             pr(")", client_data);
 328         return r;
 329     }
 330 }
 331
 332 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 333                                    void (*pr)(const char *buf, void *client_data),
 334                                    void *client_data,
 335                                    Z_RPNQuery *q)
 336 {
 337     int r;
 338     WRBUF w = wrbuf_alloc();
 339     solr_transform_set_error(ct, 0, 0);
 340     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 341     wrbuf_destroy(w);
 342     return r;
 343 }
 344
 345
 346 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 347                                   WRBUF w,
 348                                   Z_RPNQuery *q)
 349 {
 350     return solr_transform_rpn2solr_stream(ct, wrbuf_vputs, w, q);
 351 }
 352
 353 /*
 354  * Local variables:
 355  * c-basic-offset: 4
 356  * c-file-style: "Stroustrup"
 357  * indent-tabs-mode: nil
 358  * End:
 359  * vim: shiftwidth=4 tabstop=8 expandtab
 360  */
 361