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 = 0;
296 elements[k]->value.complex->num_semanticAction = 0;
300 elements[k]->which = Z_AttributeValue_numeric;
301 elements[k]->value.numeric = &attr_tmp[2*i+1];
307 zapt->attributes = (Z_AttributeList *)
308 odr_malloc (o, sizeof(*zapt->attributes));
309 zapt->attributes->num_attributes = num_attr;
310 zapt->attributes->attributes = elements;
314 term_octet->buf = (unsigned char *)odr_malloc (o, 1 + li->lex_len);
315 term_octet->size = term_octet->len =
316 escape_string ((char *) (term_octet->buf), li->lex_buf, li->lex_len);
317 term_octet->buf[term_octet->size] = 0; /* null terminate */
319 switch (li->term_type)
322 term->which = Z_Term_general;
323 term->u.general = term_octet;
325 case Z_Term_characterString:
326 term->which = Z_Term_characterString;
327 term->u.characterString = (char*) term_octet->buf;
328 /* null terminated above */
331 term->which = Z_Term_numeric;
332 term->u.numeric = odr_intdup (o, atoi((char*) (term_octet->buf)));
335 term->which = Z_Term_null;
336 term->u.null = odr_nullval();
338 case Z_Term_external:
339 term->which = Z_Term_external;
340 term->u.external = 0;
343 term->which = Z_Term_null;
344 term->u.null = odr_nullval();
350 static Z_Operand *rpn_simple(struct yaz_pqf_parser *li, ODR o,
351 int num_attr, int *attr_list, char **attr_clist,
356 zo = (Z_Operand *)odr_malloc (o, sizeof(*zo));
357 switch (li->query_look)
360 zo->which = Z_Operand_APT;
361 if (!(zo->u.attributesPlusTerm =
362 rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set)))
370 li->error = YAZ_PQF_ERROR_MISSING;
373 zo->which = Z_Operand_resultSetId;
374 zo->u.resultSetId = (char *)odr_malloc (o, li->lex_len+1);
375 memcpy (zo->u.resultSetId, li->lex_buf, li->lex_len);
376 zo->u.resultSetId[li->lex_len] = '\0';
380 /* we're only called if one of the above types are seens so
381 this shouldn't happen */
382 li->error = YAZ_PQF_ERROR_INTERNAL;
388 static Z_ProximityOperator *rpn_proximity (struct yaz_pqf_parser *li, ODR o)
390 Z_ProximityOperator *p = (Z_ProximityOperator *)odr_malloc (o, sizeof(*p));
394 li->error = YAZ_PQF_ERROR_MISSING;
397 if (*li->lex_buf == '1')
398 p->exclusion = odr_intdup (o, 1);
399 else if (*li->lex_buf == '0')
400 p->exclusion = odr_intdup (o, 0);
401 else if (*li->lex_buf == 'v' || *li->lex_buf == 'n')
405 li->error = YAZ_PQF_ERROR_PROXIMITY;
411 li->error = YAZ_PQF_ERROR_MISSING;
414 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
415 p->distance = odr_intdup (o, atoi (li->lex_buf));
418 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
424 li->error = YAZ_PQF_ERROR_MISSING;
427 if (*li->lex_buf == '1')
428 p->ordered = odr_intdup (o, 1);
429 else if (*li->lex_buf == '0')
430 p->ordered = odr_intdup (o, 0);
433 li->error = YAZ_PQF_ERROR_PROXIMITY;
439 li->error = YAZ_PQF_ERROR_MISSING;
442 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
443 p->relationType = odr_intdup (o, atoi (li->lex_buf));
446 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
452 li->error = YAZ_PQF_ERROR_MISSING;
455 if (*li->lex_buf == 'k')
456 p->which = Z_ProximityOperator_known;
457 else if (*li->lex_buf == 'p')
458 p->which = Z_ProximityOperator_private;
460 p->which = atoi (li->lex_buf);
462 if (p->which != Z_ProximityOperator_known
463 && p->which != Z_ProximityOperator_private)
465 li->error = YAZ_PQF_ERROR_PROXIMITY;
471 li->error = YAZ_PQF_ERROR_MISSING;
474 if (*li->lex_buf >= '0' && *li->lex_buf <= '9')
475 p->u.known = odr_intdup (o, atoi(li->lex_buf));
478 li->error = YAZ_PQF_ERROR_BAD_INTEGER;
484 static Z_Complex *rpn_complex(struct yaz_pqf_parser *li, ODR o,
485 int num_attr, int max_attr,
486 int *attr_list, char **attr_clist,
492 zc = (Z_Complex *)odr_malloc (o, sizeof(*zc));
493 zo = (Z_Operator *)odr_malloc (o, sizeof(*zo));
495 switch (li->query_look)
498 zo->which = Z_Operator_and;
499 zo->u.op_and = odr_nullval();
502 zo->which = Z_Operator_or;
503 zo->u.op_or = odr_nullval();
506 zo->which = Z_Operator_and_not;
507 zo->u.and_not = odr_nullval();
510 zo->which = Z_Operator_prox;
511 zo->u.prox = rpn_proximity (li, o);
516 /* we're only called if one of the above types are seens so
517 this shouldn't happen */
518 li->error = YAZ_PQF_ERROR_INTERNAL;
523 rpn_structure(li, o, num_attr, max_attr, attr_list,
524 attr_clist, attr_set)))
527 rpn_structure(li, o, num_attr, max_attr, attr_list,
528 attr_clist, attr_set)))
533 static void rpn_term_type(struct yaz_pqf_parser *li, ODR o)
537 if (compare_term (li, "general", 0))
538 li->term_type = Z_Term_general;
539 else if (compare_term (li, "numeric", 0))
540 li->term_type = Z_Term_numeric;
541 else if (compare_term (li, "string", 0))
542 li->term_type = Z_Term_characterString;
543 else if (compare_term (li, "oid", 0))
544 li->term_type = Z_Term_oid;
545 else if (compare_term (li, "datetime", 0))
546 li->term_type = Z_Term_dateTime;
547 else if (compare_term (li, "null", 0))
548 li->term_type = Z_Term_null;
550 else if (compare_term(li, "range", 0))
552 /* prepare for external: range search .. */
553 li->term_type = Z_Term_external;
554 li->external_type = VAL_MULTISRCH2;
560 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
561 int num_attr, int max_attr,
568 sz = (Z_RPNStructure *)odr_malloc (o, sizeof(*sz));
569 switch (li->query_look)
575 sz->which = Z_RPNStructure_complex;
576 if (!(sz->u.complex =
577 rpn_complex (li, o, num_attr, max_attr, attr_list,
578 attr_clist, attr_set)))
583 sz->which = Z_RPNStructure_simple;
585 rpn_simple (li, o, num_attr, attr_list,
586 attr_clist, attr_set)))
593 li->error = YAZ_PQF_ERROR_MISSING;
596 if (num_attr >= max_attr)
598 li->error = YAZ_PQF_ERROR_TOOMANY;
601 if (!p_query_parse_attr(li, o, num_attr, attr_list,
602 attr_clist, attr_set))
607 rpn_structure (li, o, num_attr, max_attr, attr_list,
608 attr_clist, attr_set);
611 rpn_term_type (li, o);
613 rpn_structure (li, o, num_attr, max_attr, attr_list,
614 attr_clist, attr_set);
615 case 0: /* operator/operand expected! */
616 li->error = YAZ_PQF_ERROR_MISSING;
622 Z_RPNQuery *p_query_rpn_mk(ODR o, struct yaz_pqf_parser *li, const char *qbuf)
625 int attr_array[1024];
626 char *attr_clist[512];
627 Odr_oid *attr_set[512];
628 Odr_oid *top_set = 0;
630 zq = (Z_RPNQuery *)odr_malloc (o, sizeof(*zq));
632 if (li->query_look == 'r')
635 top_set = query_oid_getvalbyname(li, o);
638 li->error = YAZ_PQF_ERROR_ATTSET;
645 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
648 zq->attributeSetId = top_set;
650 if (!zq->attributeSetId)
652 li->error = YAZ_PQF_ERROR_ATTSET;
656 if (!(zq->RPNStructure = rpn_structure(li, o, 0, 512,
657 attr_array, attr_clist, attr_set)))
661 li->error = YAZ_PQF_ERROR_EXTRA;
667 Z_RPNQuery *p_query_rpn(ODR o, const char *qbuf)
669 struct yaz_pqf_parser li;
673 li.right_sep = "}\"";
674 li.escape_char = '@';
675 li.term_type = Z_Term_general;
676 li.query_buf = li.query_ptr = qbuf;
678 return p_query_rpn_mk(o, &li, qbuf);
682 Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
683 ODR o, oid_proto proto,
684 Odr_oid **attributeSetP,
688 char *attr_clist[512];
689 Odr_oid *attr_set[512];
692 Odr_oid *top_set = 0;
693 Z_AttributesPlusTerm *apt;
696 if (li->query_look == 'r')
699 top_set = query_oid_getvalbyname(li, o);
702 li->error = YAZ_PQF_ERROR_ATTSET;
709 top_set = odr_oiddup(o, yaz_oid_attset_bib_1);
711 *attributeSetP = top_set;
715 if (li->query_look == 'l')
720 li->error = YAZ_PQF_ERROR_MISSING;
723 if (num_attr >= max_attr)
725 li->error = YAZ_PQF_ERROR_TOOMANY;
728 if (!p_query_parse_attr(li, o, num_attr, attr_list,
729 attr_clist, attr_set))
734 else if (li->query_look == 'y')
737 rpn_term_type (li, o);
744 li->error = YAZ_PQF_ERROR_MISSING;
747 apt = rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set);
751 if (li->query_look != 0)
753 li->error = YAZ_PQF_ERROR_EXTRA;
759 Z_AttributesPlusTerm *p_query_scan (ODR o, oid_proto proto,
760 Odr_oid **attributeSetP,
763 struct yaz_pqf_parser li;
767 li.right_sep = "}\"";
768 li.escape_char = '@';
769 li.term_type = Z_Term_general;
770 li.query_buf = li.query_ptr = qbuf;
773 return p_query_scan_mk (&li, o, proto, attributeSetP, qbuf);
776 YAZ_PQF_Parser yaz_pqf_create (void)
778 YAZ_PQF_Parser p = (YAZ_PQF_Parser) xmalloc (sizeof(*p));
782 p->right_sep = "}\"";
783 p->escape_char = '@';
784 p->term_type = Z_Term_general;
789 void yaz_pqf_destroy(YAZ_PQF_Parser p)
794 Z_RPNQuery *yaz_pqf_parse(YAZ_PQF_Parser p, ODR o, const char *qbuf)
798 p->query_buf = p->query_ptr = qbuf;
800 return p_query_rpn_mk (o, p, qbuf);
803 Z_AttributesPlusTerm *yaz_pqf_scan(YAZ_PQF_Parser p, ODR o,
804 Odr_oid **attributeSetP,
809 p->query_buf = p->query_ptr = qbuf;
811 return p_query_scan_mk (p, o, PROTO_Z3950, attributeSetP, qbuf);
814 int yaz_pqf_error (YAZ_PQF_Parser p, const char **msg, size_t *off)
818 case YAZ_PQF_ERROR_NONE:
819 *msg = "no error"; break;
820 case YAZ_PQF_ERROR_EXTRA:
821 *msg = "extra token"; break;
822 case YAZ_PQF_ERROR_MISSING:
823 *msg = "missing token"; break;
824 case YAZ_PQF_ERROR_ATTSET:
825 *msg = "unknown attribute set"; break;
826 case YAZ_PQF_ERROR_TOOMANY:
827 *msg = "too many attributes"; break;
828 case YAZ_PQF_ERROR_BADATTR:
829 *msg = "bad attribute specification"; break;
830 case YAZ_PQF_ERROR_INTERNAL:
831 *msg = "internal error"; break;
832 case YAZ_PQF_ERROR_PROXIMITY:
833 *msg = "proximity error"; break;
834 case YAZ_PQF_ERROR_BAD_INTEGER:
835 *msg = "bad integer"; break;
837 *msg = "unknown error"; break;
839 *off = p->query_ptr - p->query_buf;
845 * indent-tabs-mode: nil
847 * vim: shiftwidth=4 tabstop=8 expandtab