src/rpn2solr.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2010 Index Data
   3  * See the file LICENSE for details.
   4  */
   5
   6 /**
   7  * \file
   8  * \brief Implements RPN to SOLR conversion
   9  *
  10  */
  11
  12 #include <assert.h>
  13 #include <stdlib.h>
  14 #include <string.h>
  15 #include <yaz/rpn2solr.h>
  16 #include <yaz/xmalloc.h>
  17 #include <yaz/diagbib1.h>
  18 #include <yaz/z-core.h>
  19 #include <yaz/wrbuf.h>
  20
  21 static void wrbuf_vputs(const char *buf, void *client_data)
  22 {
  23     wrbuf_write((WRBUF) client_data, buf, strlen(buf));
  24 }
  25
  26 static const char *lookup_index_from_string_attr(Z_AttributeList *attributes)
  27 {
  28     int j;
  29     int server_choice = 1;
  30     for (j = 0; j < attributes->num_attributes; j++)
  31     {
  32         Z_AttributeElement *ae = attributes->attributes[j];
  33         if (*ae->attributeType == 1) /* use attribute */
  34         {
  35             if (ae->which == Z_AttributeValue_complex)
  36             {
  37                 Z_ComplexAttribute *ca = ae->value.complex;
  38                 int i;
  39                 for (i = 0; i < ca->num_list; i++)
  40                 {
  41                     Z_StringOrNumeric *son = ca->list[i];
  42                     if (son->which == Z_StringOrNumeric_string)
  43                         return son->u.string;
  44                 }
  45             }
  46             server_choice = 0; /* not serverChoice because we have use attr */
  47         }
  48     }
  49     if (server_choice)
  50         return "solr.serverChoice";
  51     return 0;
  52 }
  53
  54 static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
  55 {
  56     int j;
  57     for (j = 0; j < attributes->num_attributes; j++)
  58     {
  59         Z_AttributeElement *ae = attributes->attributes[j];
  60         if (*ae->attributeType == 2) /* relation attribute */
  61         {
  62             if (ae->which == Z_AttributeValue_numeric)
  63             {
  64                 /* Only support for numeric relation */
  65                 Odr_int *relation = ae->value.numeric;
  66                 /* map this numeric to representation in SOLR */
  67                 switch (*relation)
  68                 {
  69                     /* Unsure on whether this is the relation attribute constants? */
  70                 case Z_ProximityOperator_Prox_lessThan:
  71                     return 0;
  72                 case Z_ProximityOperator_Prox_lessThanOrEqual:
  73                     return 0;
  74                 case Z_ProximityOperator_Prox_equal:
  75                     return ":";
  76                 case Z_ProximityOperator_Prox_greaterThanOrEqual:
  77                     return 0;
  78                 case Z_ProximityOperator_Prox_greaterThan:
  79                     return 0;
  80                 case Z_ProximityOperator_Prox_notEqual:
  81                     return 0;
  82                 case 100:
  83                     /* phonetic is not implemented*/
  84                     return 0;
  85                 case 101:
  86                     /* stem is not not implemented */
  87                     return 0;
  88                 case 102:
  89                     /* relevance is supported in SOLR, but not implemented yet */
  90                     return 0;
  91                 default:
  92                     /* Invalid relation */
  93                     return 0;
  94                 }
  95             }
  96             else {
  97                 /*  Can we have a complex relation value?
  98                     Should we implement something?
  99                 */
 100             }
 101         }
 102     }
 103     return ":";
 104 }
 105
 106 static int rpn2solr_attr(solr_transform_t ct,
 107                         Z_AttributeList *attributes, WRBUF w)
 108 {
 109     const char *relation = solr_lookup_reverse(ct, "relation.", attributes);
 110     const char *index = solr_lookup_reverse(ct, "index.", attributes);
 111     const char *structure = solr_lookup_reverse(ct, "structure.", attributes);
 112
 113     /* if transform (properties) do not match, we'll just use a USE string attribute (bug #2978) */
 114     if (!index)
 115         index = lookup_index_from_string_attr(attributes);
 116
 117     /* Attempt to fix bug #2978: Look for a relation attribute */
 118     if (!relation)
 119         relation = lookup_relation_index_from_attr(attributes);
 120
 121     if (!index)
 122     {
 123         solr_transform_set_error(ct,
 124                                 YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
 125         return -1;
 126     }
 127     /* for serverChoice we omit index+relation+structure */
 128     if (strcmp(index, "solr.serverChoice"))
 129     {
 130         wrbuf_puts(w, index);
 131         if (relation)
 132         {
 133             if (!strcmp(relation, "exact"))
 134                 /* TODO Verify if a exact  SOLR exists */
 135                 relation = ":";
 136             else if (!strcmp(relation, "eq"))
 137                 relation = ":";
 138             else if (!strcmp(relation, "le")) {
 139                 /* TODO Not support as such, but could perhaps be transformed into a range
 140                 relation = ":[ * to ";
 141                 close_range = "]"
 142                 */
 143             }
 144             else if (!strcmp(relation, "ge")) {
 145                 /* TODO Not support as such, but could perhaps be transformed into a range
 146                 relation = "[";
 147                 relation = ":[ * to ";
 148                 close_range = "]"
 149                 */
 150             }
 151             /* Missing mapping of not equal, phonetic, stem and relevance */
 152             wrbuf_puts(w, relation);
 153         }
 154         else
 155             wrbuf_puts(w, ":");
 156
 157         if (structure)
 158         {
 159             if (strcmp(structure, "*"))
 160             {
 161                 wrbuf_puts(w, "/");
 162                 wrbuf_puts(w, structure);
 163                 wrbuf_puts(w, " ");
 164             }
 165         }
 166     }
 167     return 0;
 168 }
 169
 170 /* Bug 2878: Currently only support left and right truncation. Specific check for this */
 171 static int checkForTruncation(int flag, Z_AttributeList *attributes)
 172 {
 173     int j;
 174     for (j = 0; j < attributes->num_attributes; j++)
 175     {
 176         Z_AttributeElement *ae = attributes->attributes[j];
 177         if (*ae->attributeType == 5) /* truncation attribute */
 178         {
 179             if (ae->which == Z_AttributeValue_numeric)
 180             {
 181                 Odr_int truncation = *(ae->value.numeric);
 182                 /* This logic only works for Left, right and both. eg. 1,2,3 */
 183                 if (truncation <= 3)
 184                     return ((int) truncation & flag);
 185             }
 186             /* Complex: Shouldn't happen */
 187         }
 188     }
 189     /* No truncation or unsupported */
 190     return 0;
 191 };
 192
 193 static int checkForLeftTruncation(Z_AttributeList *attributes) {
 194         return checkForTruncation(1, attributes);
 195 }
 196
 197 static int checkForRightTruncation(Z_AttributeList *attributes) {
 198         return checkForTruncation(2, attributes);
 199 };
 200
 201 static int rpn2solr_simple(solr_transform_t ct,
 202                           void (*pr)(const char *buf, void *client_data),
 203                           void *client_data,
 204                           Z_Operand *q, WRBUF w)
 205 {
 206     int ret = 0;
 207     if (q->which != Z_Operand_APT)
 208     {
 209         ret = -1;
 210         solr_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 211     }
 212     else
 213     {
 214         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
 215         Z_Term *term = apt->term;
 216         const char *sterm = 0;
 217         size_t lterm = 0;
 218
 219         wrbuf_rewind(w);
 220         ret = rpn2solr_attr(ct, apt->attributes, w);
 221
 222         switch(term->which)
 223         {
 224         case Z_Term_general:
 225             lterm = term->u.general->len;
 226             sterm = (const char *) term->u.general->buf;
 227             break;
 228         case Z_Term_numeric:
 229             wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 230             break;
 231         case Z_Term_characterString:
 232             sterm = term->u.characterString;
 233             lterm = strlen(sterm);
 234             break;
 235         default:
 236             ret = -1;
 237             solr_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
 238         }
 239
 240         if (term)
 241         {
 242             size_t i;
 243             int must_quote = 0;
 244             for (i = 0 ; i < lterm; i++)
 245                 if (sterm[i] == ' ')
 246                     must_quote = 1;
 247             if (must_quote)
 248                 wrbuf_puts(w, "\"");
 249             /* Bug 2878: Check and add Truncation */
 250                         if (checkForLeftTruncation(apt->attributes))
 251                 wrbuf_puts(w, "*");
 252             wrbuf_write(w, sterm, lterm);
 253             /* Bug 2878: Check and add Truncation */
 254                         if (checkForRightTruncation(apt->attributes))
 255                 wrbuf_puts(w, "*");
 256             if (must_quote)
 257                 wrbuf_puts(w, "\"");
 258         }
 259         if (ret == 0)
 260             pr(wrbuf_cstr(w), client_data);
 261     }
 262     return ret;
 263 }
 264
 265
 266 static int rpn2solr_structure(solr_transform_t ct,
 267                              void (*pr)(const char *buf, void *client_data),
 268                              void *client_data,
 269                              Z_RPNStructure *q, int nested,
 270                              WRBUF w)
 271 {
 272     if (q->which == Z_RPNStructure_simple)
 273         return rpn2solr_simple(ct, pr, client_data, q->u.simple, w);
 274     else
 275     {
 276         Z_Operator *op = q->u.complex->roperator;
 277         int r;
 278
 279         if (nested)
 280             pr("(", client_data);
 281
 282         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 283         if (r)
 284             return r;
 285         switch(op->which)
 286         {
 287         case  Z_Operator_and:
 288             pr(" AND ", client_data);
 289             break;
 290         case  Z_Operator_or:
 291             pr(" OR ", client_data);
 292             break;
 293         case  Z_Operator_and_not:
 294             pr(" AND NOT ", client_data);
 295             break;
 296         case  Z_Operator_prox:
 297             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
 298             return -1;
 299         }
 300         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 301         if (nested)
 302             pr(")", client_data);
 303         return r;
 304     }
 305 }
 306
 307 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 308                                  void (*pr)(const char *buf, void *client_data),
 309                                  void *client_data,
 310                                  Z_RPNQuery *q)
 311 {
 312     int r;
 313     WRBUF w = wrbuf_alloc();
 314     solr_transform_set_error(ct, 0, 0);
 315     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 316     wrbuf_destroy(w);
 317     return r;
 318 }
 319
 320
 321 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 322                                 WRBUF w,
 323                                 Z_RPNQuery *q)
 324 {
 325     return solr_transform_rpn2solr_stream(ct, wrbuf_vputs, w, q);
 326 }
 327
 328 /*
 329  * Local variables:
 330  * c-basic-offset: 4
 331  * c-file-style: "Stroustrup"
 332  * indent-tabs-mode: nil
 333  * End:
 334  * vim: shiftwidth=4 tabstop=8 expandtab
 335  */
 336