1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2008 Index Data
3 * See the file LICENSE for details.
7 * \brief Implements PQF parsing
14 #include <yaz/proto.h>
15 #include <yaz/oid_db.h>
16 #include <yaz/pquery.h>
18 struct yaz_pqf_parser {
19 const char *query_buf;
20 const char *query_ptr;
32 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
33 int num_attr, int max_attr,
34 int *attr_list, char **attr_clist,
37 static Odr_oid *query_oid_getvalbyname(struct yaz_pqf_parser *li, ODR o)
41 if (li->lex_len >= sizeof(buf)-1)
43 memcpy (buf, li->lex_buf, li->lex_len);
44 buf[li->lex_len] = '\0';
45 return yaz_string_to_oid_odr(yaz_oid_std(), CLASS_ATTSET, buf, o);
48 static int compare_term(struct yaz_pqf_parser *li, const char *src,
51 size_t len=strlen(src);
53 if (li->lex_len == len+off && !memcmp (li->lex_buf+off, src, len-off))
58 static int query_token(struct yaz_pqf_parser *li)
61 const char *sep_match;
62 const char **qptr = &li->query_ptr;
69 if ((sep_match = strchr (li->left_sep, **qptr)))
71 sep_char = li->right_sep[sep_match - li->left_sep];
76 if (**qptr == li->escape_char && isdigit (((const unsigned char *) *qptr)[1]))
82 while (**qptr && **qptr != sep_char)
94 if (sep_char == ' ' &&
95 li->lex_len >= 1 && li->lex_buf[0] == li->escape_char)
97 if (compare_term (li, "and", 1))
99 if (compare_term (li, "or", 1))
101 if (compare_term (li, "not", 1))
103 if (compare_term (li, "attr", 1))
105 if (compare_term (li, "set", 1))
107 if (compare_term (li, "attrset", 1))
109 if (compare_term (li, "prox", 1))
111 if (compare_term (li, "term", 1))
117 static int lex(struct yaz_pqf_parser *li)
119 return li->query_look = query_token(li);
122 static int escape_string(char *out_buf, const char *in, int len)
127 if (*in == '\\' && len > 0)
153 sscanf (s, "%x", &n);
170 sscanf (s, "%o", &n);
182 return out - out_buf;
185 static int p_query_parse_attr(struct yaz_pqf_parser *li, ODR o,
186 int num_attr, int *attr_list,
187 char **attr_clist, Odr_oid **attr_set)
191 if (!(cp = strchr (li->lex_buf, '=')) ||
192 (size_t) (cp-li->lex_buf) > li->lex_len)
194 attr_set[num_attr] = query_oid_getvalbyname (li, o);
195 if (attr_set[num_attr] == 0)
197 li->error = YAZ_PQF_ERROR_ATTSET;
202 li->error = YAZ_PQF_ERROR_MISSING;
205 if (!(cp = strchr (li->lex_buf, '=')))
207 li->error = YAZ_PQF_ERROR_BADATTR;
214 attr_set[num_attr] = attr_set[num_attr-1];
216 attr_set[num_attr] = 0;
218 if (*li->lex_buf < '0' || *li->lex_buf > '9')
220 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
223 attr_list[2*num_attr] = atoi(li->lex_buf);
225 if (*cp >= '0' && *cp <= '9')
227 attr_list[2*num_attr+1] = atoi (cp);
228 attr_clist[num_attr] = 0;
232 int len = li->lex_len - (cp - li->lex_buf);
233 attr_list[2*num_attr+1] = 0;
234 attr_clist[num_attr] = (char *) odr_malloc (o, len+1);
235 len = escape_string(attr_clist[num_attr], cp, len);
236 attr_clist[num_attr][len] = '\0';
241 static Z_AttributesPlusTerm *rpn_term(struct yaz_pqf_parser *li, ODR o,
242 int num_attr, int *attr_list,
243 char **attr_clist, Odr_oid **attr_set)
245 Z_AttributesPlusTerm *zapt;
248 Z_AttributeElement **elements;
250 zapt = (Z_AttributesPlusTerm *)odr_malloc (o, sizeof(*zapt));
251 term_octet = (Odr_oct *)odr_malloc (o, sizeof(*term_octet));
252 term = (Z_Term *)odr_malloc (o, sizeof(*term));
255 elements = (Z_AttributeElement**)odr_nullval();
261 elements = (Z_AttributeElement**)
262 odr_malloc (o, num_attr * sizeof(*elements));
264 attr_tmp = (int *)odr_malloc (o, num_attr * 2 * sizeof(int));
265 memcpy (attr_tmp, attr_list, num_attr * 2 * sizeof(int));
266 for (i = num_attr; --i >= 0; )
269 for (j = i+1; j<num_attr; j++)
270 if (attr_tmp[2*j] == attr_tmp[2*i])
275 (Z_AttributeElement*)odr_malloc (o,sizeof(**elements));
276 elements[k]->attributeType = &attr_tmp[2*i];
277 elements[k]->attributeSet = attr_set[i];
281 elements[k]->which = Z_AttributeValue_complex;
282 elements[k]->value.complex = (Z_ComplexAttribute *)
283 odr_malloc (o, sizeof(Z_ComplexAttribute));
284 elements[k]->value.complex->num_list = 1;
285 elements[k]->value.complex->list =
286 (Z_StringOrNumeric **)
287 odr_malloc (o, 1 * sizeof(Z_StringOrNumeric *));
288 elements[k]->value.complex->list[0] =
289 (Z_StringOrNumeric *)
290 odr_malloc (o, sizeof(Z_StringOrNumeric));
291 elements[k]->value.complex->list[0]->which =
292 Z_StringOrNumeric_string;
293 elements[k]->value.complex->list[0]->u.string =
295 elements[k]->value.complex->semanticAction = (int **)
297 elements[k]->value.complex->num_semanticAction = 0;
301 elements[k]->which = Z_AttributeValue_numeric;
302 elements[k]->value.numeric = &attr_tmp[2*i+1];
308 zapt->attributes = (Z_AttributeList *)
309 odr_malloc (o, sizeof(*zapt->attributes));
310 zapt->attributes->num_attributes = num_attr;
311 zapt->attributes->attributes = elements;
315 term_octet->buf = (unsigned char *)odr_malloc (o, 1 + li->lex_len);
316 term_octet->size = term_octet->len =
317 escape_string ((char *) (term_octet->buf), li->lex_buf, li->lex_len);
318 term_octet->buf[term_octet->size] = 0; /* null terminate */
320 switch (li->term_type)
323 term->which = Z_Term_general;
324 term->u.general = term_octet;
326 case Z_Term_characterString:
327 term->which = Z_Term_characterString;
328 term->u.characterString = (char*) term_octet->buf;
329 /* null terminated above */
332 term->which = Z_Term_numeric;
333 term->u.numeric = odr_intdup (o, atoi((char*) (term_octet->buf)));
336 term->which = Z_Term_null;
337 term->u.null = odr_nullval();
339 case Z_Term_external:
340 term->which = Z_Term_external;
341 term->u.external = 0;
344 term->which = Z_Term_null;
345 term->u.null = odr_nullval();
351 static Z_Operand *rpn_simple(struct yaz_pqf_parser *li, ODR o,
352 int num_attr, int *attr_list, char **attr_clist,
357 zo = (Z_Operand *)odr_malloc (o, sizeof(*zo));
358 switch (li->query_look)
361 zo->which = Z_Operand_APT;
362 if (!(zo->u.attributesPlusTerm =
363 rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set)))
371 li->error = YAZ_PQF_ERROR_MISSING;
374 zo->which = Z_Operand_resultSetId;
375 zo->u.resultSetId = (char *)odr_malloc (o, li->lex_len+1);
376 memcpy (zo->u.resultSetId, li->lex_buf, li->lex_len);
377 zo->u.resultSetId[li->lex_len] = '\0';
381 /* we're only called if one of the above types are seens so
382 this shouldn't happen */
383 li->error = YAZ_PQF_ERROR_INTERNAL;
389 static Z_ProximityOperator *rpn_proximity (struct yaz_pqf_parser *li, ODR o)
391 Z_ProximityOperator *p = (Z_ProximityOperator *)odr_malloc (o, sizeof(*p));
395 li->error = YAZ_PQF_ERROR_MISSING;
398 if (*li->lex_buf == '1')
399 p->exclusion = odr_intdup (o, 1);
400 else if (*li->lex_buf == '0')
401 p->exclusion = odr_intdup (o, 0);
402 else if (*li->lex_buf == 'v' || *li->lex_buf == 'n')
406 li->error = YAZ_PQF_ERROR_PROXIMITY;
412 li->error = YAZ_PQF_ERROR_MISSING;
415 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
416 p->distance = odr_intdup (o, atoi (li->lex_buf));
419 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
425 li->error = YAZ_PQF_ERROR_MISSING;
428 if (*li->lex_buf == '1')
429 p->ordered = odr_intdup (o, 1);
430 else if (*li->lex_buf == '0')
431 p->ordered = odr_intdup (o, 0);
434 li->error = YAZ_PQF_ERROR_PROXIMITY;
440 li->error = YAZ_PQF_ERROR_MISSING;
443 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
444 p->relationType = odr_intdup (o, atoi (li->lex_buf));
447 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
453 li->error = YAZ_PQF_ERROR_MISSING;
456 if (*li->lex_buf == 'k')
457 p->which = Z_ProximityOperator_known;
458 else if (*li->lex_buf == 'p')
459 p->which = Z_ProximityOperator_private;
461 p->which = atoi (li->lex_buf);
463 if (p->which != Z_ProximityOperator_known
464 && p->which != Z_ProximityOperator_private)
466 li->error = YAZ_PQF_ERROR_PROXIMITY;
472 li->error = YAZ_PQF_ERROR_MISSING;
475 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
476 p->u.known = odr_intdup (o, atoi(li->lex_buf));
479 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
485 static Z_Complex *rpn_complex(struct yaz_pqf_parser *li, ODR o,
486 int num_attr, int max_attr,
487 int *attr_list, char **attr_clist,
493 zc = (Z_Complex *)odr_malloc (o, sizeof(*zc));
494 zo = (Z_Operator *)odr_malloc (o, sizeof(*zo));
496 switch (li->query_look)
499 zo->which = Z_Operator_and;
500 zo->u.op_and = odr_nullval();
503 zo->which = Z_Operator_or;
504 zo->u.op_or = odr_nullval();
507 zo->which = Z_Operator_and_not;
508 zo->u.and_not = odr_nullval();
511 zo->which = Z_Operator_prox;
512 zo->u.prox = rpn_proximity (li, o);
517 /* we're only called if one of the above types are seens so
518 this shouldn't happen */
519 li->error = YAZ_PQF_ERROR_INTERNAL;
524 rpn_structure(li, o, num_attr, max_attr, attr_list,
525 attr_clist, attr_set)))
528 rpn_structure(li, o, num_attr, max_attr, attr_list,
529 attr_clist, attr_set)))
534 static void rpn_term_type(struct yaz_pqf_parser *li, ODR o)
538 if (compare_term (li, "general", 0))
539 li->term_type = Z_Term_general;
540 else if (compare_term (li, "numeric", 0))
541 li->term_type = Z_Term_numeric;
542 else if (compare_term (li, "string", 0))
543 li->term_type = Z_Term_characterString;
544 else if (compare_term (li, "oid", 0))
545 li->term_type = Z_Term_oid;
546 else if (compare_term (li, "datetime", 0))
547 li->term_type = Z_Term_dateTime;
548 else if (compare_term (li, "null", 0))
549 li->term_type = Z_Term_null;
551 else if (compare_term(li, "range", 0))
553 /* prepare for external: range search .. */
554 li->term_type = Z_Term_external;
555 li->external_type = VAL_MULTISRCH2;
561 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
562 int num_attr, int max_attr,
569 sz = (Z_RPNStructure *)odr_malloc (o, sizeof(*sz));
570 switch (li->query_look)
576 sz->which = Z_RPNStructure_complex;
577 if (!(sz->u.complex =
578 rpn_complex (li, o, num_attr, max_attr, attr_list,
579 attr_clist, attr_set)))
584 sz->which = Z_RPNStructure_simple;
586 rpn_simple (li, o, num_attr, attr_list,
587 attr_clist, attr_set)))
594 li->error = YAZ_PQF_ERROR_MISSING;
597 if (num_attr >= max_attr)
599 li->error = YAZ_PQF_ERROR_TOOMANY;
602 if (!p_query_parse_attr(li, o, num_attr, attr_list,
603 attr_clist, attr_set))
608 rpn_structure (li, o, num_attr, max_attr, attr_list,
609 attr_clist, attr_set);
612 rpn_term_type (li, o);
614 rpn_structure (li, o, num_attr, max_attr, attr_list,
615 attr_clist, attr_set);
616 case 0: /* operator/operand expected! */
617 li->error = YAZ_PQF_ERROR_MISSING;
623 Z_RPNQuery *p_query_rpn_mk(ODR o, struct yaz_pqf_parser *li, const char *qbuf)
626 int attr_array[1024];
627 char *attr_clist[512];
628 Odr_oid *attr_set[512];
629 Odr_oid *top_set = 0;
631 zq = (Z_RPNQuery *)odr_malloc (o, sizeof(*zq));
633 if (li->query_look == 'r')
636 top_set = query_oid_getvalbyname(li, o);
639 li->error = YAZ_PQF_ERROR_ATTSET;
646 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
649 zq->attributeSetId = top_set;
651 if (!zq->attributeSetId)
653 li->error = YAZ_PQF_ERROR_ATTSET;
657 if (!(zq->RPNStructure = rpn_structure(li, o, 0, 512,
658 attr_array, attr_clist, attr_set)))
662 li->error = YAZ_PQF_ERROR_EXTRA;
668 Z_RPNQuery *p_query_rpn(ODR o, const char *qbuf)
670 struct yaz_pqf_parser li;
674 li.right_sep = "}\"";
675 li.escape_char = '@';
676 li.term_type = Z_Term_general;
677 li.query_buf = li.query_ptr = qbuf;
679 return p_query_rpn_mk(o, &li, qbuf);
683 Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
684 ODR o, oid_proto proto,
685 Odr_oid **attributeSetP,
689 char *attr_clist[512];
690 Odr_oid *attr_set[512];
693 Odr_oid *top_set = 0;
694 Z_AttributesPlusTerm *apt;
697 if (li->query_look == 'r')
700 top_set = query_oid_getvalbyname(li, o);
703 li->error = YAZ_PQF_ERROR_ATTSET;
710 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
712 *attributeSetP = top_set;
716 if (li->query_look == 'l')
721 li->error = YAZ_PQF_ERROR_MISSING;
724 if (num_attr >= max_attr)
726 li->error = YAZ_PQF_ERROR_TOOMANY;
729 if (!p_query_parse_attr(li, o, num_attr, attr_list,
730 attr_clist, attr_set))
735 else if (li->query_look == 'y')
738 rpn_term_type (li, o);
745 li->error = YAZ_PQF_ERROR_MISSING;
748 apt = rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set);
752 if (li->query_look != 0)
754 li->error = YAZ_PQF_ERROR_EXTRA;
760 Z_AttributesPlusTerm *p_query_scan (ODR o, oid_proto proto,
761 Odr_oid **attributeSetP,
764 struct yaz_pqf_parser li;
768 li.right_sep = "}\"";
769 li.escape_char = '@';
770 li.term_type = Z_Term_general;
771 li.query_buf = li.query_ptr = qbuf;
774 return p_query_scan_mk (&li, o, proto, attributeSetP, qbuf);
777 YAZ_PQF_Parser yaz_pqf_create (void)
779 YAZ_PQF_Parser p = (YAZ_PQF_Parser) xmalloc (sizeof(*p));
783 p->right_sep = "}\"";
784 p->escape_char = '@';
785 p->term_type = Z_Term_general;
790 void yaz_pqf_destroy(YAZ_PQF_Parser p)
795 Z_RPNQuery *yaz_pqf_parse(YAZ_PQF_Parser p, ODR o, const char *qbuf)
799 p->query_buf = p->query_ptr = qbuf;
801 return p_query_rpn_mk (o, p, qbuf);
804 Z_AttributesPlusTerm *yaz_pqf_scan(YAZ_PQF_Parser p, ODR o,
805 Odr_oid **attributeSetP,
810 p->query_buf = p->query_ptr = qbuf;
812 return p_query_scan_mk (p, o, PROTO_Z3950, attributeSetP, qbuf);
815 int yaz_pqf_error (YAZ_PQF_Parser p, const char **msg, size_t *off)
819 case YAZ_PQF_ERROR_NONE:
820 *msg = "no error"; break;
821 case YAZ_PQF_ERROR_EXTRA:
822 *msg = "extra token"; break;
823 case YAZ_PQF_ERROR_MISSING:
824 *msg = "missing token"; break;
825 case YAZ_PQF_ERROR_ATTSET:
826 *msg = "unknown attribute set"; break;
827 case YAZ_PQF_ERROR_TOOMANY:
828 *msg = "too many attributes"; break;
829 case YAZ_PQF_ERROR_BADATTR:
830 *msg = "bad attribute specification"; break;
831 case YAZ_PQF_ERROR_INTERNAL:
832 *msg = "internal error"; break;
833 case YAZ_PQF_ERROR_PROXIMITY:
834 *msg = "proximity error"; break;
835 case YAZ_PQF_ERROR_BAD_INTEGER:
836 *msg = "bad integer"; break;
838 *msg = "unknown error"; break;
840 *off = p->query_ptr - p->query_buf;
846 * indent-tabs-mode: nil
848 * vim: shiftwidth=4 tabstop=8 expandtab