/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2009 Index Data
+ * Copyright (C) 1995-2013 Index Data
* See the file LICENSE for details.
*/
-/**
+/**
* \file cclfind.c
* \brief Implements parsing of a CCL FIND query.
*
* of lookahead in the handling of relational operations.. So
* it's not really pure.
*/
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
#include <stdlib.h>
#include <string.h>
+#include <assert.h>
#include "cclp.h"
/**
* strxcat: concatenate strings.
- * n: Null-terminated Destination string
+ * n: Null-terminated Destination string
* src: Source string to be appended (not null-terminated)
* len: Length of source string.
*/
const char *set, int type)
{
struct ccl_rpn_attr *n;
-
+
n = (struct ccl_rpn_attr *)xmalloc(sizeof(*n));
ccl_assert(n);
if (set)
n->type = type;
n->next = p->u.t.attr_list;
p->u.t.attr_list = n;
-
+
return n;
}
n->value.str = xstrdup(value);
}
+static size_t cmp_operator(const char **aliases, const char *input)
+{
+ for (; *aliases; aliases++)
+ {
+ const char *cp = *aliases;
+ size_t i;
+ for (i = 0; *cp && *cp == input[i]; i++, cp++)
+ ;
+ if (*cp == '\0')
+ return i;
+ }
+ return 0;
+}
+
+
+#define REGEX_CHARS "^[]{}()|.*+?!$"
+#define CCL_CHARS "#?\\"
+
+static int has_ccl_masking(const char *src_str,
+ int src_len,
+ const char **truncation_aliases,
+ const char **mask_aliases)
+{
+ size_t j;
+ int quote_mode = 0;
+
+ for (j = 0; j < src_len; j++)
+ {
+ size_t op_size;
+ if (j > 0 && src_str[j-1] == '\\')
+ ;
+ else if (src_str[j] == '"')
+ quote_mode = !quote_mode;
+ else if (!quote_mode &&
+ (op_size = cmp_operator(truncation_aliases,
+ src_str + j)))
+ return 1;
+ else if (!quote_mode &&
+ (op_size = cmp_operator(mask_aliases,
+ src_str + j)))
+ return 1;
+ }
+ return 0;
+}
+
+static int append_term(CCL_parser cclp, const char *src_str, size_t src_len,
+ char *dst_term, int regex_trunc, int z3958_trunc,
+ const char **truncation_aliases,
+ const char **mask_aliases,
+ int is_first, int is_last,
+ int *left_trunc, int *right_trunc)
+{
+ size_t j;
+ int quote_mode = 0;
+
+ for (j = 0; j < src_len; j++)
+ {
+ size_t op_size;
+ if (j > 0 && src_str[j-1] == '\\')
+ {
+ if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
+ strcat(dst_term, "\\");
+ else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
+ strcat(dst_term, "\\");
+ strxcat(dst_term, src_str + j, 1);
+ }
+ else if (src_str[j] == '"')
+ quote_mode = !quote_mode;
+ else if (!quote_mode &&
+ (op_size = cmp_operator(truncation_aliases,
+ src_str + j))
+ )
+ {
+ j += (op_size - 1); /* j++ in for loop */
+ if (regex_trunc)
+ strcat(dst_term, ".*");
+ else if (z3958_trunc)
+ strcat(dst_term, "?");
+ else if (is_first && j == 0)
+ *left_trunc = 1;
+ else if (is_last && j == src_len - 1)
+ *right_trunc = 1;
+ else
+ {
+ cclp->error_code = CCL_ERR_TRUNC_NOT_EMBED;
+ return -1;
+ }
+ }
+ else if (!quote_mode &&
+ (op_size = cmp_operator(mask_aliases, src_str + j)))
+ {
+ j += (op_size - 1); /* j++ in for loop */
+ if (regex_trunc)
+ strcat(dst_term, ".");
+ else if (z3958_trunc)
+ strcat(dst_term, "#");
+ else
+ {
+ cclp->error_code = CCL_ERR_TRUNC_NOT_SINGLE;
+ return -1;
+ }
+ }
+ else if (src_str[j] != '\\')
+ {
+ if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
+ strcat(dst_term, "\\");
+ else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
+ strcat(dst_term, "\\");
+ strxcat(dst_term, src_str + j, 1);
+ }
+ }
+ return 0;
+}
/**
- * search_term: Parse CCL search term.
+ * search_term: Parse CCL search term.
* cclp: CCL Parser
* qa: Qualifier attributes already applied.
* term_list: tokens we accept as terms in context
struct ccl_rpn_node *p_top = 0;
struct ccl_token *lookahead = cclp->look_token;
int and_list = 0;
+ int auto_group = 0;
int or_list = 0;
char *attset;
const char **truncation_aliases;
const char *t_default[2];
+ const char **mask_aliases;
+ const char *m_default[2];
truncation_aliases =
ccl_qual_search_special(cclp->bibset, "truncation");
t_default[1] = 0;
}
+ mask_aliases =
+ ccl_qual_search_special(cclp->bibset, "mask");
+ if (!mask_aliases)
+ {
+ mask_aliases = m_default;
+ m_default[0] = "#";
+ m_default[1] = 0;
+ }
+
+
if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
and_list = 1;
+ if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AUTO_GROUP, 0))
+ auto_group = 1;
if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
or_list = 1;
while (1)
{
struct ccl_rpn_node *p;
size_t no, i;
- int no_spaces = 0;
- int left_trunc = 0;
- int right_trunc = 0;
- int mid_trunc = 0;
+ int is_phrase = 0;
+ int is_ccl_masked = 0;
int relation_value = -1;
int position_value = -1;
int structure_value = -1;
int truncation_value = -1;
int completeness_value = -1;
int len = 0;
+ int left_trunc = 0;
+ int right_trunc = 0;
+ int regex_trunc = 0;
+ int z3958_trunc = 0;
size_t max = 200;
if (and_list || or_list || !multi)
max = 1;
-
+
/* ignore commas when dealing with and-lists .. */
if (and_list && lookahead && lookahead->kind == CCL_TOK_COMMA)
{
ADVANCE;
continue;
}
- /* go through each TERM token. If no truncation attribute is yet
- met, then look for left/right truncation markers (?) and
- set left_trunc/right_trunc/mid_trunc accordingly */
for (no = 0; no < max && is_term_ok(lookahead->kind, term_list); no++)
{
+ int this_is_phrase = 0;
for (i = 0; i<lookahead->len; i++)
if (lookahead->name[i] == ' ')
- no_spaces++;
- else if (strchr(truncation_aliases[0], lookahead->name[i]))
- {
- if (no == 0 && i == 0 && lookahead->len >= 1)
- left_trunc = 1;
- else if (!is_term_ok(lookahead->next->kind, term_list) &&
- i == lookahead->len-1 && i >= 1)
- right_trunc = 1;
- else
- mid_trunc = 1;
- }
+ this_is_phrase = 1;
+
+ if (has_ccl_masking(lookahead->name, lookahead->len,
+ truncation_aliases,
+ mask_aliases))
+ is_ccl_masked = 1;
+
+ if (auto_group)
+ {
+ if (no > 0 && (is_phrase || is_phrase != this_is_phrase))
+ break;
+ is_phrase = this_is_phrase;
+ }
+ else if (this_is_phrase || no > 0)
+ is_phrase = 1;
len += 1+lookahead->len+lookahead->ws_prefix_len;
lookahead = lookahead->next;
}
if (len == 0)
break; /* no more terms . stop . */
-
+
/* create the term node, but wait a moment before adding the term */
p = ccl_rpn_node_create(CCL_RPN_TERM);
p->u.t.attr_list = NULL;
for (i=0; qa && qa[i]; i++)
{
struct ccl_rpn_attr *attr;
-
+
for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
switch(attr->kind)
{
if (truncation_value != -1)
continue;
truncation_value = attr->value.numeric;
- left_trunc = right_trunc = mid_trunc = 0;
break;
case CCL_BIB1_COM:
if (completeness_value != -1)
}
}
}
- /* len now holds the number of characters in the RPN term */
- /* no holds the number of CCL tokens (1 or more) */
-
- if (structure_value == -1 &&
- qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
- { /* no structure attribute met. Apply either structure attribute
- WORD or PHRASE depending on number of CCL tokens */
- if (no == 1 && no_spaces == 0)
+ attset = 0;
+ if (structure_value == -1 && (
+ auto_group ||
+ qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
+ )
+ {
+ if (!is_phrase)
ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 2);
else
ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
}
+ if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_REGEX,
+ &attset))
+ {
+ if (is_ccl_masked)
+ regex_trunc = 1; /* regex trunc (102) allowed */
+ }
+ else if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_Z3958,
+ &attset))
+ {
+ if (is_ccl_masked)
+ z3958_trunc = 1; /* Z39.58 trunc (CCL) trunc allowed */
+ }
+
/* make the RPN token */
- p->u.t.term = (char *)xmalloc(len);
+ p->u.t.term = (char *)xmalloc(len * 2 + 2);
ccl_assert(p->u.t.term);
p->u.t.term[0] = '\0';
for (i = 0; i<no; i++)
{
const char *src_str = cclp->look_token->name;
size_t src_len = cclp->look_token->len;
-
- if (i == 0 && left_trunc)
+
+ if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
{
- src_len--;
- src_str++;
+ strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
+ cclp->look_token->ws_prefix_len);
}
- if (i == no-1 && right_trunc)
- src_len--;
- if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
+ if (append_term(cclp, src_str, src_len, p->u.t.term, regex_trunc,
+ z3958_trunc, truncation_aliases, mask_aliases,
+ i == 0, i == no - 1,
+ &left_trunc, &right_trunc))
{
- size_t len = strlen(p->u.t.term);
- memcpy(p->u.t.term + len, cclp->look_token->ws_prefix_buf,
- cclp->look_token->ws_prefix_len);
- p->u.t.term[len + cclp->look_token->ws_prefix_len] = '\0';
+ ccl_rpn_delete(p);
+ return NULL;
}
- strxcat(p->u.t.term, src_str, src_len);
ADVANCE;
}
-
/* make the top node point to us.. */
if (p_top)
{
}
ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
}
+ else if (regex_trunc)
+ {
+ ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
+ }
+ else if (z3958_trunc)
+ {
+ ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
+ }
else
{
if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
return search_term_x(cclp, qa, list, 0);
}
+
+static struct ccl_rpn_node *search_terms2(CCL_parser cclp,
+ ccl_qualifier_t *qa)
+{
+ if (KIND == CCL_TOK_LP)
+ {
+ struct ccl_rpn_node *p;
+ ADVANCE;
+ if (!(p = find_spec(cclp, qa)))
+ return NULL;
+ if (KIND != CCL_TOK_RP)
+ {
+ cclp->error_code = CCL_ERR_RP_EXPECTED;
+ ccl_rpn_delete(p);
+ return NULL;
+ }
+ ADVANCE;
+ return p;
+ }
+ else
+ {
+ static int list[] = {
+ CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
+ CCL_TOK_REL, CCL_TOK_SET, -1};
+
+ return search_term_x(cclp, qa, list, 1);
+ }
+}
+
+
+
static
struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
ccl_qualifier_t *ap, char *attset)
if (KIND == CCL_TOK_TERM)
{
size_t i;
+ int quote_mode = 0;
for (i = 0; i<cclp->look_token->len; i++)
{
- if (cclp->look_token->name[i] == '-')
+ if (i > 0 && cclp->look_token->name[i] == '\\')
+ ;
+ else if (cclp->look_token->name[i] == '"')
+ quote_mode = !quote_mode;
+ else if (cclp->look_token->name[i] == '-' && !quote_mode)
break;
}
-
+
if (cclp->look_token->len > 1 && i == 0)
{ /* -xx*/
struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
cclp->look_token->next->kind == CCL_TOK_TERM &&
cclp->look_token->next->len > 1 &&
cclp->look_token->next->name[0] == '-')
-
+
{ /* xx -yy */
/* we _know_ that xx does not have - in it */
struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
ntoken->len = 1;
(ntoken->next->name)++; /* adjust yy */
- (ntoken->next->len)--;
+ (ntoken->next->len)--;
}
}
}
-
+
if (rel == 3 &&
KIND == CCL_TOK_TERM &&
cclp->look_token->next && cclp->look_token->next->len == 1 &&
if (KIND == CCL_TOK_TERM) /* = term - term ? */
{
struct ccl_rpn_node *p2;
-
+
if (!(p2 = search_term(cclp, ap)))
{
ccl_rpn_delete(p1);
ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
return p;
}
- else if (KIND == CCL_TOK_LP)
- {
- ADVANCE;
- if (!(p = find_spec(cclp, ap)))
- return NULL;
- if (KIND != CCL_TOK_RP)
- {
- cclp->error_code = CCL_ERR_RP_EXPECTED;
- ccl_rpn_delete(p);
- return NULL;
- }
- ADVANCE;
- return p;
- }
else
{
if (!(p = search_terms(cclp, ap)))
ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, rel);
return p;
}
- cclp->error_code = CCL_ERR_TERM_EXPECTED;
return NULL;
}
struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
{
char *attset;
- struct ccl_rpn_node *p;
-
+
if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
|| qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
return qualifiers_order(cclp, ap, attset);
return NULL;
}
ADVANCE;
- if (KIND == CCL_TOK_LP)
- {
- ADVANCE;
- if (!(p = find_spec(cclp, ap)))
- {
- return NULL;
- }
- if (KIND != CCL_TOK_RP)
- {
- cclp->error_code = CCL_ERR_RP_EXPECTED;
- ccl_rpn_delete(p);
- return NULL;
- }
- ADVANCE;
- }
- else
- p = search_terms(cclp, ap);
- return p;
+ return search_terms(cclp, ap);
}
/**
- * qualifier_list: Parse CCL qualifiers and search terms.
+ * qualifier_list: Parse CCL qualifiers and search terms.
* cclp: CCL Parser
* la: Token pointer to RELATION token.
* qa: Qualifier attributes already applied.
* return: pointer to node(s); NULL on error.
*/
-static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
+static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
struct ccl_token *la,
ccl_qualifier_t *qa)
{
{
struct ccl_rpn_node *node_sub;
cclp->look_token = la;
-
+
node_sub = qualifier_relation(cclp, ap);
if (!node_sub)
{
}
if (node)
{
- struct ccl_rpn_node *node_this =
+ struct ccl_rpn_node *node_this =
ccl_rpn_node_create(CCL_RPN_OR);
node_this->u.p[0] = node;
node_this->u.p[1] = node_sub;
if (qa)
{
ccl_qualifier_t *qa0 = qa;
-
+
while (*qa0)
ap[i++] = *qa0++;
}
ap[i] = NULL;
-
+
if (!found)
break;
-
+
cclp->look_token = lookahead;
-
+
node_sub = qualifier_relation(cclp, ap);
if (!node_sub)
{
}
if (node)
{
- struct ccl_rpn_node *node_this =
+ struct ccl_rpn_node *node_this =
ccl_rpn_node_create(CCL_RPN_OR);
node_this->u.p[0] = node;
node_this->u.p[1] = node_sub;
static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
{
static int list[] = {
- CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ, CCL_TOK_REL, CCL_TOK_SET, -1};
+ CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
+ CCL_TOK_REL, CCL_TOK_SET, -1};
struct ccl_rpn_node *p1, *p2, *pn;
- p1 = search_term_x(cclp, qa, list, 1);
+ p1 = search_terms2(cclp, qa);
if (!p1)
return NULL;
while (1)
p_prox->u.t.attr_list = 0;
ADVANCE;
- p2 = search_term_x(cclp, qa, list, 1);
+ p2 = search_terms2(cclp, qa);
if (!p2)
{
ccl_rpn_delete(p1);
}
else if (is_term_ok(KIND, list))
{
- p2 = search_term_x(cclp, qa, list, 1);
+ p2 = search_terms2(cclp, qa);
if (!p2)
{
ccl_rpn_delete(p1);
{
struct ccl_rpn_node *p1;
struct ccl_token *lookahead;
- if (KIND == CCL_TOK_LP)
- {
- ADVANCE;
- p1 = find_spec(cclp, qa);
- if (!p1)
- return NULL;
- if (KIND != CCL_TOK_RP)
- {
- cclp->error_code = CCL_ERR_RP_EXPECTED;
- ccl_rpn_delete(p1);
- return NULL;
- }
- ADVANCE;
- return p1;
- }
- else if (KIND == CCL_TOK_SET)
+ if (KIND == CCL_TOK_SET)
{
ADVANCE;
if (KIND == CCL_TOK_EQ)
break;
lookahead = lookahead->next;
}
- if (qa)
+ if (qa || lookahead->kind == CCL_TOK_LP)
return search_terms(cclp, qa);
else
{
}
if (node)
{
- struct ccl_rpn_node *node_this =
+ struct ccl_rpn_node *node_this =
ccl_rpn_node_create(CCL_RPN_OR);
node_this->u.p[0] = node;
node_this->u.p[1] = node_sub;
return p;
}
-struct ccl_rpn_node *ccl_parser_find_token(CCL_parser cclp,
+struct ccl_rpn_node *ccl_parser_find_token(CCL_parser cclp,
struct ccl_token *list)
{
struct ccl_rpn_node *p;