return 0;
}
+
#define REGEX_CHARS "^[]{}()|.*+?!$"
#define CCL_CHARS "#?\\"
+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,
+ 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]))
+ {
+ *regex_trunc = 2;
+ strcat(dst_term, "\\");
+ }
+ else if (*z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
+ {
+ *z3958_trunc = 2;
+ 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, ".*");
+ *regex_trunc = 2; /* regex trunc is really needed */
+ }
+ else if (*z3958_trunc)
+ {
+ strcat(dst_term, "?");
+ *z3958_trunc = 2;
+ }
+ 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 && src_str[j] == '#')
+ {
+ if (*regex_trunc)
+ {
+ strcat(dst_term, ".");
+ *regex_trunc = 2; /* regex trunc is really needed */
+ }
+ else if (*z3958_trunc)
+ {
+ strcat(dst_term, "#");
+ *z3958_trunc = 2;
+ }
+ 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]))
+ {
+ *regex_trunc = 2;
+ strcat(dst_term, "\\");
+ }
+ else if (*z3958_trunc && strchr(CCL_CHARS, src_str[j]))
+ {
+ *z3958_trunc = 2;
+ strcat(dst_term, "\\");
+ }
+ strxcat(dst_term, src_str + j, 1);
+ }
+ }
+ return 0;
+}
+
/**
* search_term: Parse CCL search term.
* cclp: CCL Parser
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;
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 is_phrase = 0;
int relation_value = -1;
int position_value = -1;
int structure_value = -1;
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++;
+ this_is_phrase = 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;
}
}
}
}
- /* 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);
{
const char *src_str = cclp->look_token->name;
size_t src_len = cclp->look_token->len;
- int j;
- int quote_mode = 0;
if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
{
strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
cclp->look_token->ws_prefix_len);
}
- for (j = 0; j < src_len; j++)
+ if (append_term(cclp, src_str, src_len, p->u.t.term, ®ex_trunc,
+ &z3958_trunc, truncation_aliases, i == 0, i == no - 1,
+ &left_trunc, &right_trunc))
{
- size_t op_size;
- if (j > 0 && src_str[j-1] == '\\')
- {
- if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
- {
- regex_trunc = 2;
- strcat(p->u.t.term, "\\");
- }
- else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
- {
- z3958_trunc = 2;
- strcat(p->u.t.term, "\\");
- }
- strxcat(p->u.t.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(p->u.t.term, ".*");
- regex_trunc = 2; /* regex trunc is really needed */
- }
- else if (z3958_trunc)
- {
- strcat(p->u.t.term, "?");
- z3958_trunc = 2;
- }
- else if (i == 0 && j == 0)
- left_trunc = 1;
- else if (i == no - 1 && j == src_len - 1)
- right_trunc = 1;
- else
- {
- cclp->error_code = CCL_ERR_TRUNC_NOT_EMBED;
- ccl_rpn_delete(p);
- return NULL;
- }
- }
- else if (!quote_mode && src_str[j] == '#')
- {
- if (regex_trunc)
- {
- strcat(p->u.t.term, ".");
- regex_trunc = 2; /* regex trunc is really needed */
- }
- else if (z3958_trunc)
- {
- strcat(p->u.t.term, "#");
- z3958_trunc = 2;
- }
- else
- {
- cclp->error_code = CCL_ERR_TRUNC_NOT_SINGLE;
- ccl_rpn_delete(p);
- return NULL;
- }
- }
- else if (src_str[j] != '\\')
- {
- if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
- {
- regex_trunc = 2;
- strcat(p->u.t.term, "\\");
- }
- else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
- {
- z3958_trunc = 2;
- strcat(p->u.t.term, "\\");
- }
- strxcat(p->u.t.term, src_str + j, 1);
- }
+ ccl_rpn_delete(p);
+ return NULL;
}
ADVANCE;
}
-
/* make the top node point to us.. */
if (p_top)
{