1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2011 Index Data
3 * See the file LICENSE for details.
7 * \brief Implements PQF parsing
17 #include <yaz/proto.h>
18 #include <yaz/oid_db.h>
19 #include <yaz/pquery.h>
21 struct yaz_pqf_parser {
22 const char *query_buf;
23 const char *query_ptr;
35 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
36 int num_attr, int max_attr,
37 Odr_int *attr_list, char **attr_clist,
40 static Odr_oid *query_oid_getvalbyname(struct yaz_pqf_parser *li, ODR o)
44 if (li->lex_len >= sizeof(buf)-1)
46 memcpy(buf, li->lex_buf, li->lex_len);
47 buf[li->lex_len] = '\0';
48 return yaz_string_to_oid_odr(yaz_oid_std(), CLASS_ATTSET, buf, o);
51 static int compare_term(struct yaz_pqf_parser *li, const char *src,
54 size_t len=strlen(src);
56 if (li->lex_len == len+off && !memcmp(li->lex_buf+off, src, len-off))
61 static int query_token(struct yaz_pqf_parser *li)
64 const char *sep_match;
65 const char **qptr = &li->query_ptr;
72 if ((sep_match = strchr(li->left_sep, **qptr)))
74 sep_char = li->right_sep[sep_match - li->left_sep];
79 if (**qptr == li->escape_char && yaz_isdigit((*qptr)[1]))
85 while (**qptr && **qptr != sep_char)
97 if (sep_char == ' ' &&
98 li->lex_len >= 1 && li->lex_buf[0] == li->escape_char)
100 if (compare_term(li, "and", 1))
102 if (compare_term(li, "or", 1))
104 if (compare_term(li, "not", 1))
106 if (compare_term(li, "attr", 1))
108 if (compare_term(li, "set", 1))
110 if (compare_term(li, "attrset", 1))
112 if (compare_term(li, "prox", 1))
114 if (compare_term(li, "term", 1))
120 static int lex(struct yaz_pqf_parser *li)
122 return li->query_look = query_token(li);
125 int escape_string(char *out_buf, const char *in, int len)
130 if (*in == '\\' && len > 0)
185 return out - out_buf;
188 int p_query_parse_attr(struct yaz_pqf_parser *li, ODR o,
189 int num_attr, Odr_int *attr_list,
190 char **attr_clist, Odr_oid **attr_set)
195 if (!(cp = strchr(li->lex_buf, '=')) ||
196 (size_t) (cp-li->lex_buf) > li->lex_len)
198 attr_set[num_attr] = query_oid_getvalbyname(li, o);
199 if (attr_set[num_attr] == 0)
201 li->error = YAZ_PQF_ERROR_ATTSET;
206 li->error = YAZ_PQF_ERROR_MISSING;
209 if (!(cp = strchr(li->lex_buf, '=')))
211 li->error = YAZ_PQF_ERROR_BADATTR;
218 attr_set[num_attr] = attr_set[num_attr-1];
220 attr_set[num_attr] = 0;
222 if (*li->lex_buf < '0' || *li->lex_buf > '9')
224 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
227 attr_list[2*num_attr] = odr_atoi(li->lex_buf);
230 /* inspect value .. and make it a integer if it appears to be */
231 for (i = cp - li->lex_buf; i < li->lex_len; i++)
232 if (li->lex_buf[i] < '0' || li->lex_buf[i] > '9')
234 int len = li->lex_len - (cp - li->lex_buf);
235 attr_list[2*num_attr+1] = 0;
236 attr_clist[num_attr] = (char *) odr_malloc(o, len+1);
237 len = escape_string(attr_clist[num_attr], cp, len);
238 attr_clist[num_attr][len] = '\0';
241 attr_list[2*num_attr+1] = odr_atoi(cp);
242 attr_clist[num_attr] = 0;
246 Z_AttributeList *get_attributeList(ODR o,
247 int num_attr, Odr_int *attr_list,
248 char **attr_clist, Odr_oid **attr_set)
252 Z_AttributeElement **elements;
253 Z_AttributeList *attributes= (Z_AttributeList *) odr_malloc(o, sizeof(*attributes));
254 attributes->num_attributes = num_attr;
256 attributes->attributes = (Z_AttributeElement**)odr_nullval();
259 elements = (Z_AttributeElement**) odr_malloc (o, num_attr * sizeof(*elements));
261 attr_tmp = (Odr_int *)odr_malloc(o, num_attr * 2 * sizeof(*attr_tmp));
262 memcpy(attr_tmp, attr_list, num_attr * 2 * sizeof(*attr_tmp));
263 for (i = num_attr; --i >= 0; )
266 for (j = i+1; j<num_attr; j++)
267 if (attr_tmp[2*j] == attr_tmp[2*i])
272 (Z_AttributeElement*)odr_malloc(o,sizeof(**elements));
273 elements[k]->attributeType = &attr_tmp[2*i];
274 elements[k]->attributeSet = attr_set[i];
278 elements[k]->which = Z_AttributeValue_complex;
279 elements[k]->value.complex = (Z_ComplexAttribute *)
280 odr_malloc(o, sizeof(Z_ComplexAttribute));
281 elements[k]->value.complex->num_list = 1;
282 elements[k]->value.complex->list =
283 (Z_StringOrNumeric **)
284 odr_malloc(o, 1 * sizeof(Z_StringOrNumeric *));
285 elements[k]->value.complex->list[0] =
286 (Z_StringOrNumeric *)
287 odr_malloc(o, sizeof(Z_StringOrNumeric));
288 elements[k]->value.complex->list[0]->which =
289 Z_StringOrNumeric_string;
290 elements[k]->value.complex->list[0]->u.string =
292 elements[k]->value.complex->semanticAction = 0;
293 elements[k]->value.complex->num_semanticAction = 0;
297 elements[k]->which = Z_AttributeValue_numeric;
298 elements[k]->value.numeric = &attr_tmp[2*i+1];
302 attributes->num_attributes = k;
303 attributes->attributes = elements;
307 Z_Term *z_Term_create(ODR o, int term_type, const char *buf, size_t len)
309 Z_Term *term = (Z_Term *)odr_malloc(o, sizeof(*term));
310 Odr_oct *term_octet = (Odr_oct *)odr_malloc(o, sizeof(*term_octet));
311 term_octet->buf = (unsigned char *)odr_malloc(o, 1 + len);
312 memcpy(term_octet->buf, buf, len);
313 term_octet->size = term_octet->len = len;
314 term_octet->buf[term_octet->size] = 0; /* null terminate */
319 term->which = Z_Term_general;
320 term->u.general = term_octet;
322 case Z_Term_characterString:
323 term->which = Z_Term_characterString;
324 term->u.characterString = (char*) term_octet->buf;
325 /* null terminated above */
328 term->which = Z_Term_numeric;
329 term->u.numeric = odr_intdup(o, odr_atoi((const char*) term_octet->buf));
332 term->which = Z_Term_null;
333 term->u.null = odr_nullval();
335 case Z_Term_external:
336 term->which = Z_Term_external;
337 term->u.external = 0;
340 term->which = Z_Term_null;
341 term->u.null = odr_nullval();
347 static Z_AttributesPlusTerm *rpn_term_attributes(
348 struct yaz_pqf_parser *li, ODR o, Z_AttributeList *attributes)
350 char *es_str = odr_malloc(o, li->lex_len+1);
351 int es_len = escape_string(es_str, li->lex_buf, li->lex_len);
352 Z_Term *term = z_Term_create(o, li->term_type, es_str, es_len);
353 Z_AttributesPlusTerm *zapt = (Z_AttributesPlusTerm *)
354 odr_malloc(o, sizeof(*zapt));
357 zapt->attributes = attributes;
361 static Z_AttributesPlusTerm *rpn_term(struct yaz_pqf_parser *li, ODR o,
362 int num_attr, Odr_int *attr_list,
363 char **attr_clist, Odr_oid **attr_set)
365 return rpn_term_attributes(li, o, get_attributeList(o, num_attr, attr_list, attr_clist, attr_set));
368 static Z_Operand *rpn_simple(struct yaz_pqf_parser *li, ODR o,
369 int num_attr, Odr_int *attr_list,
375 zo = (Z_Operand *)odr_malloc(o, sizeof(*zo));
376 switch (li->query_look)
379 zo->which = Z_Operand_APT;
380 if (!(zo->u.attributesPlusTerm =
381 rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set)))
389 li->error = YAZ_PQF_ERROR_MISSING;
392 zo->which = Z_Operand_resultSetId;
393 zo->u.resultSetId = (char *)odr_malloc(o, li->lex_len+1);
394 memcpy(zo->u.resultSetId, li->lex_buf, li->lex_len);
395 zo->u.resultSetId[li->lex_len] = '\0';
399 /* we're only called if one of the above types are seens so
400 this shouldn't happen */
401 li->error = YAZ_PQF_ERROR_INTERNAL;
407 static Z_ProximityOperator *rpn_proximity(struct yaz_pqf_parser *li, ODR o)
409 Z_ProximityOperator *p = (Z_ProximityOperator *)odr_malloc(o, sizeof(*p));
413 li->error = YAZ_PQF_ERROR_MISSING;
416 if (*li->lex_buf == '1')
417 p->exclusion = odr_booldup(o, 1);
418 else if (*li->lex_buf == '0')
419 p->exclusion = odr_booldup(o, 0);
420 else if (*li->lex_buf == 'v' || *li->lex_buf == 'n')
424 li->error = YAZ_PQF_ERROR_PROXIMITY;
430 li->error = YAZ_PQF_ERROR_MISSING;
433 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
434 p->distance = odr_intdup(o, odr_atoi(li->lex_buf));
437 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
443 li->error = YAZ_PQF_ERROR_MISSING;
446 if (*li->lex_buf == '1')
447 p->ordered = odr_booldup(o, 1);
448 else if (*li->lex_buf == '0')
449 p->ordered = odr_booldup(o, 0);
452 li->error = YAZ_PQF_ERROR_PROXIMITY;
458 li->error = YAZ_PQF_ERROR_MISSING;
461 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
462 p->relationType = odr_intdup(o, odr_atoi(li->lex_buf));
465 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
471 li->error = YAZ_PQF_ERROR_MISSING;
474 if (*li->lex_buf == 'k')
475 p->which = Z_ProximityOperator_known;
476 else if (*li->lex_buf == 'p')
477 p->which = Z_ProximityOperator_private;
479 p->which = atoi(li->lex_buf);
481 if (p->which != Z_ProximityOperator_known
482 && p->which != Z_ProximityOperator_private)
484 li->error = YAZ_PQF_ERROR_PROXIMITY;
490 li->error = YAZ_PQF_ERROR_MISSING;
493 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
494 p->u.known = odr_intdup(o, odr_atoi(li->lex_buf));
497 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
503 static Z_Complex *rpn_complex(struct yaz_pqf_parser *li, ODR o,
504 int num_attr, int max_attr,
505 Odr_int *attr_list, char **attr_clist,
511 zc = (Z_Complex *)odr_malloc(o, sizeof(*zc));
512 zo = (Z_Operator *)odr_malloc(o, sizeof(*zo));
514 switch (li->query_look)
517 zo->which = Z_Operator_and;
518 zo->u.op_and = odr_nullval();
521 zo->which = Z_Operator_or;
522 zo->u.op_or = odr_nullval();
525 zo->which = Z_Operator_and_not;
526 zo->u.and_not = odr_nullval();
529 zo->which = Z_Operator_prox;
530 zo->u.prox = rpn_proximity(li, o);
535 /* we're only called if one of the above types are seens so
536 this shouldn't happen */
537 li->error = YAZ_PQF_ERROR_INTERNAL;
542 rpn_structure(li, o, num_attr, max_attr, attr_list,
543 attr_clist, attr_set)))
546 rpn_structure(li, o, num_attr, max_attr, attr_list,
547 attr_clist, attr_set)))
552 static void rpn_term_type(struct yaz_pqf_parser *li)
556 if (compare_term(li, "general", 0))
557 li->term_type = Z_Term_general;
558 else if (compare_term(li, "numeric", 0))
559 li->term_type = Z_Term_numeric;
560 else if (compare_term(li, "string", 0))
561 li->term_type = Z_Term_characterString;
562 else if (compare_term(li, "oid", 0))
563 li->term_type = Z_Term_oid;
564 else if (compare_term(li, "datetime", 0))
565 li->term_type = Z_Term_dateTime;
566 else if (compare_term(li, "null", 0))
567 li->term_type = Z_Term_null;
569 else if (compare_term(li, "range", 0))
571 /* prepare for external: range search .. */
572 li->term_type = Z_Term_external;
573 li->external_type = VAL_MULTISRCH2;
579 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
580 int num_attr, int max_attr,
587 sz = (Z_RPNStructure *)odr_malloc(o, sizeof(*sz));
588 switch (li->query_look)
594 sz->which = Z_RPNStructure_complex;
595 if (!(sz->u.complex =
596 rpn_complex(li, o, num_attr, max_attr, attr_list,
597 attr_clist, attr_set)))
602 sz->which = Z_RPNStructure_simple;
604 rpn_simple(li, o, num_attr, attr_list,
605 attr_clist, attr_set)))
612 li->error = YAZ_PQF_ERROR_MISSING;
615 if (num_attr >= max_attr)
617 li->error = YAZ_PQF_ERROR_TOOMANY;
620 if (!p_query_parse_attr(li, o, num_attr, attr_list,
621 attr_clist, attr_set))
626 rpn_structure(li, o, num_attr, max_attr, attr_list,
627 attr_clist, attr_set);
632 rpn_structure(li, o, num_attr, max_attr, attr_list,
633 attr_clist, attr_set);
634 case 0: /* operator/operand expected! */
635 li->error = YAZ_PQF_ERROR_MISSING;
641 static Z_RPNQuery *p_query_rpn_mk(ODR o, struct yaz_pqf_parser *li)
644 Odr_int attr_array[1024];
645 char *attr_clist[512];
646 Odr_oid *attr_set[512];
647 Odr_oid *top_set = 0;
649 zq = (Z_RPNQuery *)odr_malloc(o, sizeof(*zq));
651 if (li->query_look == 'r')
654 top_set = query_oid_getvalbyname(li, o);
657 li->error = YAZ_PQF_ERROR_ATTSET;
664 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
667 zq->attributeSetId = top_set;
669 if (!zq->attributeSetId)
671 li->error = YAZ_PQF_ERROR_ATTSET;
675 if (!(zq->RPNStructure = rpn_structure(li, o, 0, 512,
676 attr_array, attr_clist, attr_set)))
680 li->error = YAZ_PQF_ERROR_EXTRA;
686 Z_RPNQuery *p_query_rpn(ODR o, const char *qbuf)
688 struct yaz_pqf_parser li;
692 li.right_sep = "}\"";
693 li.escape_char = '@';
694 li.term_type = Z_Term_general;
695 li.query_buf = li.query_ptr = qbuf;
697 return p_query_rpn_mk(o, &li);
701 static Z_AttributeList *p_query_scan_attributes_mk(struct yaz_pqf_parser *li,
703 Odr_oid **attributeSetP)
705 Odr_int attr_list[1024];
706 char *attr_clist[512];
707 Odr_oid *attr_set[512];
710 Odr_oid *top_set = 0;
713 if (li->query_look == 'r')
716 top_set = query_oid_getvalbyname(li, o);
719 li->error = YAZ_PQF_ERROR_ATTSET;
726 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
728 *attributeSetP = top_set;
732 if (li->query_look == 'l')
737 li->error = YAZ_PQF_ERROR_MISSING;
740 if (num_attr >= max_attr)
742 li->error = YAZ_PQF_ERROR_TOOMANY;
745 if (!p_query_parse_attr(li, o, num_attr, attr_list,
746 attr_clist, attr_set))
751 else if (li->query_look == 'y')
759 return get_attributeList(o, num_attr, attr_list, attr_clist, attr_set);
762 static Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
764 Odr_oid **attributeSetP)
766 Z_AttributeList *attr_list = p_query_scan_attributes_mk(li, o, attributeSetP);
767 Z_AttributesPlusTerm *apt;
771 li->error = YAZ_PQF_ERROR_MISSING;
774 apt = rpn_term_attributes(li, o, attr_list);
778 if (li->query_look != 0)
780 li->error = YAZ_PQF_ERROR_EXTRA;
786 YAZ_PQF_Parser yaz_pqf_create(void)
788 YAZ_PQF_Parser p = (YAZ_PQF_Parser) xmalloc(sizeof(*p));
792 p->right_sep = "}\"";
793 p->escape_char = '@';
794 p->term_type = Z_Term_general;
799 void yaz_pqf_destroy(YAZ_PQF_Parser p)
804 Z_RPNQuery *yaz_pqf_parse(YAZ_PQF_Parser p, ODR o, const char *qbuf)
808 p->query_buf = p->query_ptr = qbuf;
810 return p_query_rpn_mk(o, p);
813 Z_AttributesPlusTerm *yaz_pqf_scan(YAZ_PQF_Parser p, ODR o,
814 Odr_oid **attributeSetP,
819 p->query_buf = p->query_ptr = qbuf;
821 return p_query_scan_mk(p, o, attributeSetP);
824 Z_AttributeList *yaz_pqf_scan_attribute_list(YAZ_PQF_Parser p, ODR o,
825 Odr_oid **attributeSetP,
830 p->query_buf = p->query_ptr = qbuf;
832 return p_query_scan_attributes_mk(p, o, attributeSetP);
835 static Z_FacetField* parse_facet(ODR odr, const char *facet, int length)
837 YAZ_PQF_Parser pqf_parser = yaz_pqf_create();
838 char *buffer = odr_strdupn(odr, facet, length);
839 Odr_oid *attributeSetId;
840 Z_FacetField *facet_field = 0;
841 Z_AttributeList *attribute_list =
842 yaz_pqf_scan_attribute_list(pqf_parser, odr, &attributeSetId, buffer);
846 facet_field = odr_malloc(odr, sizeof(*facet_field));
847 facet_field->attributes = attribute_list;
848 facet_field->num_terms = 0;
849 facet_field->terms = 0;
851 yaz_pqf_destroy(pqf_parser);
855 #define FACET_DELIMITER ','
857 static int scan_facet_argument(const char *arg) {
859 int length = strlen(arg);
861 for (index = 0; index < length; index++) {
862 if (arg[index] == FACET_DELIMITER)
869 * yax_pdg_parse_facet_list: Parses a comma-separated list of AttributeList(s) into a FacetList.
870 * It does not handle the optional facet term(s).
873 Z_FacetList *yaz_pqf_parse_facet_list(ODR odr, const char *facet) {
874 Z_FacetList *facet_list = 0;
875 Z_FacetField **elements;
877 int num_elements = scan_facet_argument(facet);
878 if (num_elements == 0)
880 facet_list = odr_malloc(odr, sizeof(*facet_list));
881 facet_list->num = num_elements;
882 elements = odr_malloc(odr, num_elements * sizeof(*elements));
883 facet_list->elements = elements;
884 for (index = 0; index < num_elements;) {
885 const char *pos = strchr(facet, FACET_DELIMITER);
887 pos = facet + strlen(facet);
888 elements[index] = parse_facet(odr, (const char *) facet, (pos - facet));
889 if (elements[index]) {
894 facet_list->num = num_elements;
903 int yaz_pqf_error(YAZ_PQF_Parser p, const char **msg, size_t *off)
907 case YAZ_PQF_ERROR_NONE:
908 *msg = "no error"; break;
909 case YAZ_PQF_ERROR_EXTRA:
910 *msg = "extra token"; break;
911 case YAZ_PQF_ERROR_MISSING:
912 *msg = "missing token"; break;
913 case YAZ_PQF_ERROR_ATTSET:
914 *msg = "unknown attribute set"; break;
915 case YAZ_PQF_ERROR_TOOMANY:
916 *msg = "too many attributes"; break;
917 case YAZ_PQF_ERROR_BADATTR:
918 *msg = "bad attribute specification"; break;
919 case YAZ_PQF_ERROR_INTERNAL:
920 *msg = "internal error"; break;
921 case YAZ_PQF_ERROR_PROXIMITY:
922 *msg = "proximity error"; break;
923 case YAZ_PQF_ERROR_BAD_INTEGER:
924 *msg = "bad integer"; break;
926 *msg = "unknown error"; break;
928 *off = p->query_ptr - p->query_buf;
934 * c-file-style: "Stroustrup"
935 * indent-tabs-mode: nil
937 * vim: shiftwidth=4 tabstop=8 expandtab