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 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, char **close_range)
 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, YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
 125         return -1;
 126     }
 127     /* for serverChoice we omit index+relation+structure */
 128     if (strcmp(index, "cql.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             else if (!strcmp(relation, "ge")) {
 144                 /* TODO Not support as such, but could perhaps be transformed into a range */
 145                 relation = ":[";
 146                 *close_range = " TO *]";
 147             }
 148             /* Missing mapping of not equal, phonetic, stem and relevance */
 149             wrbuf_puts(w, relation);
 150         }
 151         else
 152             wrbuf_puts(w, ":");
 153
 154         if (structure)
 155         {
 156             if (strcmp(structure, "*"))
 157             {
 158                 wrbuf_puts(w, "/");
 159                 wrbuf_puts(w, structure);
 160                 wrbuf_puts(w, " ");
 161             }
 162         }
 163 //        if (close_range)
 164 //            wrbuf_puts(w, close_range);
 165     }
 166     return 0;
 167 }
 168
 169 static Odr_int get_truncation(Z_AttributesPlusTerm *apt)
 170 {
 171     int j;
 172     Z_AttributeList *attributes = apt->attributes;
 173     for (j = 0; j < attributes->num_attributes; j++)
 174     {
 175         Z_AttributeElement *ae = attributes->attributes[j];
 176         if (*ae->attributeType == 5) /* truncation attribute */
 177         {
 178             if (ae->which == Z_AttributeValue_numeric)
 179             {
 180                 return *(ae->value.numeric);
 181             }
 182             else if (ae->which == Z_AttributeValue_complex) {
 183                 ;
 184                 //yaz_log(YLOG_DEBUG, "Z_Attribute_complex");
 185                 /* Complex: Shouldn't happen */
 186             }
 187         }
 188     }
 189     /* No truncation given */
 190     return 0;
 191 }
 192
 193 #define SOLR_SPECIAL "+-&|!(){}[]^\"~*?:\\"
 194
 195 static int rpn2solr_simple(solr_transform_t ct,
 196                            void (*pr)(const char *buf, void *client_data),
 197                            void *client_data,
 198                            Z_Operand *q, WRBUF w)
 199 {
 200     int ret = 0;
 201     if (q->which != Z_Operand_APT)
 202     {
 203         ret = -1;
 204         solr_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 205     }
 206     else
 207     {
 208         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
 209         Z_Term *term = apt->term;
 210         const char *sterm = 0;
 211         size_t lterm = 0;
 212         Odr_int trunc = get_truncation(apt);
 213         char *close_range = 0;
 214
 215         wrbuf_rewind(w);
 216         ret = rpn2solr_attr(ct, apt->attributes, w, &close_range);
 217
 218         if (trunc == 0 || trunc == 1 || trunc == 100 || trunc == 104)
 219             ;
 220         else
 221         {
 222             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
 223             return -1;
 224         }
 225         switch (term->which)
 226         {
 227         case Z_Term_general:
 228             lterm = term->u.general->len;
 229             sterm = (const char *) term->u.general->buf;
 230             break;
 231         case Z_Term_numeric:
 232             wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 233             break;
 234         case Z_Term_characterString:
 235             sterm = term->u.characterString;
 236             lterm = strlen(sterm);
 237             break;
 238         default:
 239             ret = -1;
 240             solr_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
 241         }
 242
 243         if (sterm)
 244         {
 245             size_t i;
 246             int must_quote = 0;
 247
 248             for (i = 0 ; i < lterm; i++)
 249                 if (sterm[i] == ' ')
 250                     must_quote = 1;
 251             if (must_quote)
 252                 wrbuf_puts(w, "\"");
 253             for (i = 0 ; i < lterm; i++)
 254             {
 255                 if (sterm[i] == '\\' && i < lterm - 1)
 256                 {
 257                     i++;
 258                     if (strchr(SOLR_SPECIAL, sterm[i]))
 259                         wrbuf_putc(w, '\\');
 260                     wrbuf_putc(w, sterm[i]);
 261                 }
 262                 else if (sterm[i] == '?' && trunc == 104)
 263                 {
 264                     wrbuf_putc(w, '*');
 265                 }
 266                 else if (sterm[i] == '#' && trunc == 104)
 267                 {
 268                     wrbuf_putc(w, '?');
 269                 }
 270                 else if (strchr(SOLR_SPECIAL, sterm[i]))
 271                 {
 272                     wrbuf_putc(w, '\\');
 273                     wrbuf_putc(w, sterm[i]);
 274                 }
 275                 else
 276                     wrbuf_putc(w, sterm[i]);
 277             }
 278             if (trunc == 1)
 279                 wrbuf_puts(w, "*");
 280             if (must_quote)
 281                 wrbuf_puts(w, "\"");
 282             if (close_range)
 283                 wrbuf_puts(w, close_range);
 284         }
 285         if (ret == 0)
 286             pr(wrbuf_cstr(w), client_data);
 287     }
 288     return ret;
 289 }
 290
 291
 292 static int rpn2solr_structure(solr_transform_t ct,
 293                               void (*pr)(const char *buf, void *client_data),
 294                               void *client_data,
 295                               Z_RPNStructure *q, int nested,
 296                               WRBUF w)
 297 {
 298     if (q->which == Z_RPNStructure_simple)
 299         return rpn2solr_simple(ct, pr, client_data, q->u.simple, w);
 300     else
 301     {
 302         Z_Operator *op = q->u.complex->roperator;
 303         int r;
 304
 305         if (nested)
 306             pr("(", client_data);
 307
 308         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 309         if (r)
 310             return r;
 311         switch(op->which)
 312         {
 313         case  Z_Operator_and:
 314             pr(" AND ", client_data);
 315             break;
 316         case  Z_Operator_or:
 317             pr(" OR ", client_data);
 318             break;
 319         case  Z_Operator_and_not:
 320             pr(" AND NOT ", client_data);
 321             break;
 322         case  Z_Operator_prox:
 323             solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
 324             return -1;
 325         }
 326         r = rpn2solr_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 327         if (nested)
 328             pr(")", client_data);
 329         return r;
 330     }
 331 }
 332
 333 int solr_transform_rpn2solr_stream(solr_transform_t ct,
 334                                    void (*pr)(const char *buf, void *client_data),
 335                                    void *client_data,
 336                                    Z_RPNQuery *q)
 337 {
 338     int r;
 339     WRBUF w = wrbuf_alloc();
 340     solr_transform_set_error(ct, 0, 0);
 341     r = rpn2solr_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 342     wrbuf_destroy(w);
 343     return r;
 344 }
 345
 346
 347 int solr_transform_rpn2solr_wrbuf(solr_transform_t ct,
 348                                   WRBUF w,
 349                                   Z_RPNQuery *q)
 350 {
 351     return solr_transform_rpn2solr_stream(ct, wrbuf_vputs, w, q);
 352 }
 353
 354 /*
 355  * Local variables:
 356  * c-basic-offset: 4
 357  * c-file-style: "Stroustrup"
 358  * indent-tabs-mode: nil
 359  * End:
 360  * vim: shiftwidth=4 tabstop=8 expandtab
 361  */
 362