src/rpn2cql.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 CQL 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/rpn2cql.h>
  17 #include <yaz/xmalloc.h>
  18 #include <yaz/diagbib1.h>
  19 #include <yaz/z-core.h>
  20 #include <yaz/wrbuf.h>
  21 #include <yaz/logrpn.h> /* For yaz_prox_unit_name() */
  22
  23 static const char *lookup_index_from_string_attr(Z_AttributeList *attributes,
  24                                                  Odr_int *numeric_value)
  25 {
  26     int j;
  27     int server_choice = 1;
  28     for (j = 0; j < attributes->num_attributes; j++)
  29     {
  30         Z_AttributeElement *ae = attributes->attributes[j];
  31         if (*ae->attributeType == 1) /* use attribute */
  32         {
  33             if (ae->which == Z_AttributeValue_complex)
  34             {
  35                 Z_ComplexAttribute *ca = ae->value.complex;
  36                 int i;
  37                 for (i = 0; i < ca->num_list; i++)
  38                 {
  39                     Z_StringOrNumeric *son = ca->list[i];
  40                     if (son->which == Z_StringOrNumeric_string)
  41                         return son->u.string;
  42                 }
  43             }
  44             else if (ae->which == Z_AttributeValue_numeric)
  45             {
  46                 *numeric_value = *ae->value.numeric;
  47             }
  48             server_choice = 0; /* not serverChoice because we have use attr */
  49         }
  50     }
  51     if (server_choice)
  52         return "cql.serverChoice";
  53     return 0;
  54 }
  55
  56 static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
  57 {
  58     int j;
  59     for (j = 0; j < attributes->num_attributes; j++)
  60     {
  61         Z_AttributeElement *ae = attributes->attributes[j];
  62         if (*ae->attributeType == 2) /* relation attribute */
  63         {
  64             if (ae->which == Z_AttributeValue_numeric)
  65             {
  66                 /* Only support for numeric relation */
  67                 Odr_int *relation = ae->value.numeric;
  68                 /* map this numeric to representation in CQL */
  69                 switch (*relation)
  70                 {
  71                     /* Unsure on whether this is the relation attribute constants? */
  72                 case Z_ProximityOperator_Prox_lessThan:
  73                     return "<";
  74                 case Z_ProximityOperator_Prox_lessThanOrEqual:
  75                     return "<=";
  76                 case Z_ProximityOperator_Prox_equal:
  77                     return "=";
  78                 case Z_ProximityOperator_Prox_greaterThanOrEqual:
  79                     return ">=";
  80                 case Z_ProximityOperator_Prox_greaterThan:
  81                     return ">";
  82                 case Z_ProximityOperator_Prox_notEqual:
  83                     return "<>";
  84                 case 100:
  85                     /* phonetic is not supported in CQL */
  86                     return 0;
  87                 case 101:
  88                     /* stem is not supported in CQL */
  89                     return 0;
  90                 case 102:
  91                     /* relevance is supported in CQL, but not implemented yet */
  92                     return 0;
  93                 default:
  94                     /* Invalid relation */
  95                     return 0;
  96                 }
  97             }
  98             else {
  99                 /*  Can we have a complex relation value?
 100                     Should we implement something?
 101                 */
 102             }
 103         }
 104     }
 105     return "=";
 106 }
 107
 108 static int rpn2cql_attr(cql_transform_t ct,
 109                         Z_AttributeList *attributes, WRBUF w)
 110 {
 111     const char *relation = cql_lookup_reverse(ct, "relation.", attributes);
 112     const char *index = cql_lookup_reverse(ct, "index.", attributes);
 113     const char *structure = cql_lookup_reverse(ct, "structure.", attributes);
 114
 115     /* if transform (properties) do not match, we'll fall back
 116        to string or report numeric attribute error */
 117     if (!index)
 118     {
 119         Odr_int use_attribute = -1;
 120         index = lookup_index_from_string_attr(attributes, &use_attribute);
 121         if (!index)
 122         {
 123             wrbuf_rewind(w);
 124             if (use_attribute != -1)
 125                 wrbuf_printf(w, ODR_INT_PRINTF, use_attribute);
 126             return YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
 127         }
 128     }
 129     if (!relation)
 130         relation = lookup_relation_index_from_attr(attributes);
 131
 132     if (!relation)
 133         relation = "=";
 134     else if (!strcmp(relation, "exact"))
 135         relation = "==";
 136     else if (!strcmp(relation, "eq"))
 137         relation = "=";
 138     else if (!strcmp(relation, "le"))
 139         relation = "<=";
 140     else if (!strcmp(relation, "ge"))
 141         relation = ">=";
 142
 143     if (strcmp(index, "cql.serverChoice") || strcmp(relation, "=")
 144         || (structure && strcmp(structure, "*")))
 145     {
 146         wrbuf_puts(w, index);
 147         wrbuf_puts(w, " ");
 148         wrbuf_puts(w, relation);
 149         wrbuf_puts(w, " ");
 150
 151         if (structure && strcmp(structure, "*"))
 152         {
 153             wrbuf_puts(w, "/");
 154             wrbuf_puts(w, structure);
 155             wrbuf_puts(w, " ");
 156         }
 157     }
 158     return 0;
 159 }
 160
 161 static Odr_int lookup_truncation(Z_AttributeList *attributes)
 162 {
 163     int j;
 164     for (j = 0; j < attributes->num_attributes; j++)
 165     {
 166         Z_AttributeElement *ae = attributes->attributes[j];
 167         if (*ae->attributeType == 5) /* truncation attribute */
 168         {
 169             if (ae->which == Z_AttributeValue_numeric)
 170                 return *(ae->value.numeric);
 171         }
 172     }
 173     /* No truncation specified */
 174     return 0;
 175 }
 176
 177 static int rpn2cql_simple(cql_transform_t ct,
 178                           void (*pr)(const char *buf, void *client_data),
 179                           void *client_data,
 180                           Z_Operand *q, WRBUF w)
 181 {
 182     if (q->which != Z_Operand_APT)
 183     {
 184         wrbuf_rewind(w);
 185         return YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM;
 186     }
 187     else
 188     {
 189         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
 190         Z_Term *term = apt->term;
 191         const char *sterm = 0;
 192         size_t lterm = 0;
 193         Odr_int trunc = lookup_truncation(apt->attributes);
 194         size_t i;
 195         int r;
 196
 197         wrbuf_rewind(w);
 198         r = rpn2cql_attr(ct, apt->attributes, w);
 199         if (r)
 200             return r;
 201
 202         switch (term->which)
 203         {
 204         case Z_Term_general:
 205             lterm = term->u.general->len;
 206             sterm = (const char *) term->u.general->buf;
 207             break;
 208         case Z_Term_numeric:
 209             wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
 210             break;
 211         case Z_Term_characterString:
 212             sterm = term->u.characterString;
 213             lterm = strlen(sterm);
 214             break;
 215         default:
 216             wrbuf_rewind(w);
 217             wrbuf_printf(w, "%d", term->which);
 218             return YAZ_BIB1_TERM_TYPE_UNSUPP;
 219         }
 220
 221         if (trunc <= 3 || trunc == 100 || trunc == 102 || trunc == 104)
 222         {
 223             int quote_it = 0;
 224             for (i = 0 ; i < lterm; i++)
 225                 if (strchr(" ()=></", sterm[i]))
 226                 {
 227                     quote_it = 1;
 228                     break;
 229                 }
 230             if (lterm == 0)
 231                 quote_it = 1;
 232             if (quote_it)
 233                 wrbuf_puts(w, "\"");
 234             if (trunc == 2 || trunc == 3)
 235                 wrbuf_puts(w, "*");
 236             for (i = 0; i < lterm; i++)
 237             {
 238                 if (sterm[i] == '\\' && i < lterm - 1)
 239                 {
 240                     i++;
 241                     if (strchr("*?\"\\", sterm[i]))
 242                         wrbuf_putc(w, '\\');
 243                     wrbuf_putc(w, sterm[i]);
 244                 }
 245                 else if (trunc == 102 && sterm[i] == '.' && sterm[i+1] == '*')
 246                 {
 247                     wrbuf_putc(w, '*');
 248                     i++;
 249                 }
 250                 else if (trunc == 102 && sterm[i] == '.')
 251                     wrbuf_putc(w, '?');
 252                 else if (trunc == 104 && sterm[i] == '?')
 253                     wrbuf_putc(w, '*');
 254                 else if (trunc == 104 && sterm[i] == '#')
 255                     wrbuf_putc(w, '?');
 256                 else if (strchr("*?\"", sterm[i]))
 257                 {
 258                     wrbuf_putc(w, '\\');
 259                     wrbuf_putc(w, sterm[i]);
 260                 }
 261                 else
 262                     wrbuf_putc(w, sterm[i]);
 263             }
 264             if (trunc == 1 || trunc == 3)
 265                 wrbuf_puts(w, "*");
 266             if (quote_it)
 267                 wrbuf_puts(w, "\"");
 268         }
 269         else
 270         {
 271             wrbuf_rewind(w);
 272             wrbuf_printf(w, ODR_INT_PRINTF, trunc);
 273             return YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE;
 274         }
 275         pr(wrbuf_cstr(w), client_data);
 276     }
 277     return 0;
 278 }
 279
 280
 281 static int rpn2cql_structure(cql_transform_t ct,
 282                              void (*pr)(const char *buf, void *client_data),
 283                              void *client_data,
 284                              Z_RPNStructure *q, int nested,
 285                              WRBUF w)
 286 {
 287     if (q->which == Z_RPNStructure_simple)
 288         return rpn2cql_simple(ct, pr, client_data, q->u.simple, w);
 289     else
 290     {
 291         Z_Operator *op = q->u.complex->roperator;
 292         Z_ProximityOperator *prox;
 293         int r;
 294
 295         if (nested)
 296             pr("(", client_data);
 297
 298         r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 299         if (r)
 300             return r;
 301         switch (op->which)
 302         {
 303         case  Z_Operator_and:
 304             pr(" and ", client_data);
 305             break;
 306         case  Z_Operator_or:
 307             pr(" or ", client_data);
 308             break;
 309         case  Z_Operator_and_not:
 310             pr(" not ", client_data);
 311             break;
 312         case  Z_Operator_prox: {
 313             pr(" prox", client_data);
 314             prox = op->u.prox;
 315             /* No way to express Odr_bool *exclusion -- ignore it */
 316             if (prox->distance) {
 317                 char buf[21]; /* Enough for any 64-bit int */
 318                 char *op2name[6] = { "<", "<=", "=", ">=", ">","<>" };
 319                 pr("/distance", client_data);
 320                 if (!prox->relationType ||
 321                     *prox->relationType < Z_ProximityOperator_Prox_lessThan ||
 322                     *prox->relationType > Z_ProximityOperator_Prox_notEqual)
 323                 {
 324                     wrbuf_rewind(w);
 325                     return YAZ_BIB1_UNSUPP_SEARCH;
 326                 }
 327                 pr(op2name[*prox->relationType-1], client_data);
 328                 sprintf(buf, "%ld", (long) *prox->distance);
 329                 pr(buf, client_data);
 330             }
 331             if (prox->ordered) {
 332                 if (*prox->ordered) {
 333                     pr("/ordered", client_data);
 334                 } else {
 335                     pr("/unordered", client_data);
 336                 }
 337             }
 338             if (prox->which != Z_ProximityOperator_known ||
 339                 *prox->u.known != Z_ProxUnit_word) {
 340                     pr("/unit=", client_data);
 341                     pr(yaz_prox_unit_name(prox), client_data);
 342             }
 343             pr(" ", client_data);
 344             break;
 345         }
 346         }
 347         r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 348         if (nested)
 349             pr(")", client_data);
 350         return r;
 351     }
 352 }
 353
 354 int cql_transform_rpn2cql_stream_r(cql_transform_t ct,
 355                                    WRBUF addinfo,
 356                                    void (*pr)(const char *buf, void *client_data),
 357                                    void *client_data,
 358                                    Z_RPNQuery *q)
 359 {
 360     /* addinfo (w) is used for both addinfo and house-keeping ! */
 361     int r = rpn2cql_structure(ct, pr, client_data, q->RPNStructure, 0, addinfo);
 362     if (!r)
 363         wrbuf_rewind(addinfo); /* no additional info if no error */
 364     return r;
 365 }
 366
 367
 368 int cql_transform_rpn2cql_stream(cql_transform_t ct,
 369                                  void (*pr)(const char *buf, void *client_data),
 370                                  void *client_data,
 371                                  Z_RPNQuery *q)
 372 {
 373     WRBUF w = wrbuf_alloc();
 374     int r = cql_transform_rpn2cql_stream_r(ct, w, pr, client_data, q);
 375     if (r)
 376         cql_transform_set_error(ct, r, wrbuf_len(w) ? wrbuf_cstr(w) : 0);
 377     wrbuf_destroy(w);
 378     return r;
 379 }
 380
 381
 382 int cql_transform_rpn2cql_wrbuf(cql_transform_t ct,
 383                                 WRBUF w,
 384                                 Z_RPNQuery *q)
 385 {
 386     return cql_transform_rpn2cql_stream(ct, wrbuf_vp_puts, w, q);
 387 }
 388
 389 /*
 390  * Local variables:
 391  * c-basic-offset: 4
 392  * c-file-style: "Stroustrup"
 393  * indent-tabs-mode: nil
 394  * End:
 395  * vim: shiftwidth=4 tabstop=8 expandtab
 396  */
 397