src/rpn2cql.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2011 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 void wrbuf_vputs(const char *buf, void *client_data)
  24 {
  25     wrbuf_write((WRBUF) client_data, buf, strlen(buf));
  26 }
  27
  28 static const char *lookup_index_from_string_attr(Z_AttributeList *attributes)
  29 {
  30     int j;
  31     int server_choice = 1;
  32     for (j = 0; j < attributes->num_attributes; j++)
  33     {
  34         Z_AttributeElement *ae = attributes->attributes[j];
  35         if (*ae->attributeType == 1) /* use attribute */
  36         {
  37             if (ae->which == Z_AttributeValue_complex)
  38             {
  39                 Z_ComplexAttribute *ca = ae->value.complex;
  40                 int i;
  41                 for (i = 0; i < ca->num_list; i++)
  42                 {
  43                     Z_StringOrNumeric *son = ca->list[i];
  44                     if (son->which == Z_StringOrNumeric_string)
  45                         return son->u.string;
  46                 }
  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 just use a USE string attribute (bug #2978) */
 116     if (!index)
 117         index = lookup_index_from_string_attr(attributes);
 118
 119     /* Attempt to fix bug #2978: Look for a relation attribute */
 120     if (!relation)
 121         relation = lookup_relation_index_from_attr(attributes);
 122
 123     if (!index)
 124     {
 125         cql_transform_set_error(ct,
 126                                 YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
 127         return -1;
 128     }
 129     /* for serverChoice we omit index+relation+structure */
 130     if (strcmp(index, "cql.serverChoice"))
 131     {
 132         wrbuf_puts(w, index);
 133         if (relation)
 134         {
 135             if (!strcmp(relation, "exact"))
 136                 relation = "==";
 137             else if (!strcmp(relation, "eq"))
 138                 relation = "=";
 139             else if (!strcmp(relation, "le"))
 140                 relation = "<=";
 141             else if (!strcmp(relation, "ge"))
 142                 relation = ">=";
 143             /* Missing mapping of not equal, phonetic, stem and relevance */
 144             wrbuf_puts(w, relation);
 145         }
 146         else
 147             wrbuf_puts(w, "=");
 148
 149         if (structure)
 150         {
 151             if (strcmp(structure, "*"))
 152             {
 153                 wrbuf_puts(w, "/");
 154                 wrbuf_puts(w, structure);
 155                 wrbuf_puts(w, " ");
 156             }
 157         }
 158     }
 159     return 0;
 160 }
 161
 162 static Odr_int lookup_truncation(Z_AttributeList *attributes)
 163 {
 164     int j;
 165     for (j = 0; j < attributes->num_attributes; j++)
 166     {
 167         Z_AttributeElement *ae = attributes->attributes[j];
 168         if (*ae->attributeType == 5) /* truncation attribute */
 169         {
 170             if (ae->which == Z_AttributeValue_numeric)
 171                 return *(ae->value.numeric);
 172         }
 173     }
 174     /* No truncation specified */
 175     return 0;
 176 };
 177
 178 static int rpn2cql_simple(cql_transform_t ct,
 179                           void (*pr)(const char *buf, void *client_data),
 180                           void *client_data,
 181                           Z_Operand *q, WRBUF w)
 182 {
 183     int ret = 0;
 184     if (q->which != Z_Operand_APT)
 185     {
 186         ret = -1;
 187         cql_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
 188     }
 189     else
 190     {
 191         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
 192         Z_Term *term = apt->term;
 193         const char *sterm = 0;
 194         size_t lterm = 0;
 195         Odr_int trunc = lookup_truncation(apt->attributes);
 196         size_t i;
 197         int must_quote = 0;
 198
 199         wrbuf_rewind(w);
 200         ret = rpn2cql_attr(ct, apt->attributes, w);
 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             cql_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
 217             return -1;
 218         }
 219
 220         if (trunc <= 3 || trunc == 100 || trunc == 102 || trunc == 104)
 221         {
 222             for (i = 0 ; i < lterm; i++)
 223                 if (strchr(" ()=></", sterm[i]))
 224                     break;
 225             if (i < lterm || lterm == 0)
 226                 must_quote = 1;
 227             if (must_quote)
 228                 wrbuf_puts(w, "\"");
 229             if (trunc == 2 || trunc == 3)
 230                 wrbuf_puts(w, "*");
 231             for (i = 0; i < lterm; i++)
 232             {
 233                 if (sterm[i] == '\\' && i < lterm - 1)
 234                 {
 235                     i++;
 236                     if (strchr("*?\"\\", sterm[i]))
 237                         wrbuf_putc(w, '\\');
 238                     wrbuf_putc(w, sterm[i]);
 239                 }
 240                 else if (trunc == 102 && sterm[i] == '.' && sterm[i+1] == '*')
 241                 {
 242                     wrbuf_putc(w, '*');
 243                     i++;
 244                 }
 245                 else if (trunc == 102 && sterm[i] == '.')
 246                     wrbuf_putc(w, '?');
 247                 else if (trunc == 104 && sterm[i] == '?')
 248                     wrbuf_putc(w, '*');
 249                 else if (trunc == 104 && sterm[i] == '#')
 250                     wrbuf_putc(w, '?');
 251                 else if (strchr("*?\"", sterm[i]))
 252                 {
 253                     wrbuf_putc(w, '\\');
 254                     wrbuf_putc(w, sterm[i]);
 255                 }
 256                 else
 257                     wrbuf_putc(w, sterm[i]);
 258             }
 259             if (trunc == 1 || trunc == 3)
 260                 wrbuf_puts(w, "*");
 261             if (must_quote)
 262                 wrbuf_puts(w, "\"");
 263         }
 264         else
 265         {
 266             cql_transform_set_error(
 267                 ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
 268             ret = -1;
 269         }
 270         if (ret == 0)
 271             pr(wrbuf_cstr(w), client_data);
 272     }
 273     return ret;
 274 }
 275
 276
 277 static int rpn2cql_structure(cql_transform_t ct,
 278                              void (*pr)(const char *buf, void *client_data),
 279                              void *client_data,
 280                              Z_RPNStructure *q, int nested,
 281                              WRBUF w)
 282 {
 283     if (q->which == Z_RPNStructure_simple)
 284         return rpn2cql_simple(ct, pr, client_data, q->u.simple, w);
 285     else
 286     {
 287         Z_Operator *op = q->u.complex->roperator;
 288         Z_ProximityOperator *prox;
 289         int r;
 290
 291         if (nested)
 292             pr("(", client_data);
 293
 294         r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
 295         if (r)
 296             return r;
 297         switch(op->which)
 298         {
 299         case  Z_Operator_and:
 300             pr(" and ", client_data);
 301             break;
 302         case  Z_Operator_or:
 303             pr(" or ", client_data);
 304             break;
 305         case  Z_Operator_and_not:
 306             pr(" not ", client_data);
 307             break;
 308         case  Z_Operator_prox: {
 309             pr(" prox", client_data);
 310             prox = op->u.prox;
 311             /* No way to express Odr_bool *exclusion -- ignore it */
 312             if (prox->distance) {
 313                 char buf[21]; /* Enough for any 64-bit int */
 314                 char *op2name[6] = { "<", "<=", "=", ">=", ">","<>" };
 315                 pr("/distance", client_data);
 316                 if (!prox->relationType ||
 317                     *prox->relationType < Z_ProximityOperator_Prox_lessThan ||
 318                     *prox->relationType > Z_ProximityOperator_Prox_notEqual) {
 319                     cql_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH,
 320                         "unrecognised proximity relationType");
 321                     return -1;
 322                 }
 323                 pr(op2name[*prox->relationType-1], client_data);
 324                 sprintf(buf, "%ld", (long) *prox->distance);
 325                 pr(buf, client_data);
 326             }
 327             if (prox->ordered) {
 328                 if (*prox->ordered) {
 329                     pr("/ordered", client_data);
 330                 } else {
 331                     pr("/unordered", client_data);
 332                 }
 333             }
 334             if (prox->which != Z_ProximityOperator_known ||
 335                 *prox->u.known != Z_ProxUnit_word) {
 336                     pr("/unit=", client_data);
 337                     pr(yaz_prox_unit_name(prox), client_data);
 338             }
 339             pr(" ", client_data);
 340             break;
 341         }
 342         }
 343         r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
 344         if (nested)
 345             pr(")", client_data);
 346         return r;
 347     }
 348 }
 349
 350 int cql_transform_rpn2cql_stream(cql_transform_t ct,
 351                                  void (*pr)(const char *buf, void *client_data),
 352                                  void *client_data,
 353                                  Z_RPNQuery *q)
 354 {
 355     int r;
 356     WRBUF w = wrbuf_alloc();
 357     cql_transform_set_error(ct, 0, 0);
 358     r = rpn2cql_structure(ct, pr, client_data, q->RPNStructure, 0, w);
 359     wrbuf_destroy(w);
 360     return r;
 361 }
 362
 363
 364 int cql_transform_rpn2cql_wrbuf(cql_transform_t ct,
 365                                 WRBUF w,
 366                                 Z_RPNQuery *q)
 367 {
 368     return cql_transform_rpn2cql_stream(ct, wrbuf_vputs, w, q);
 369 }
 370
 371 /*
 372  * Local variables:
 373  * c-basic-offset: 4
 374  * c-file-style: "Stroustrup"
 375  * indent-tabs-mode: nil
 376  * End:
 377  * vim: shiftwidth=4 tabstop=8 expandtab
 378  */
 379