-/* $Id: rpnsearch.c,v 1.17 2007-10-29 20:07:04 adam Exp $
+/* $Id: rpnsearch.c,v 1.25 2007-12-03 13:04:04 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
static const char **rpn_char_map_handler(void *vp, const char **from, int len)
{
struct rpn_char_map_info *p = (struct rpn_char_map_info *) vp;
- const char **out = zebra_maps_input(p->zm, p->reg_type, from, len, 0);
+ const char **out = zebra_maps_input(p->zm, from, len, 0);
#if 0
if (out && *out)
{
return out;
}
-void rpn_char_map_prepare(struct zebra_register *reg, int reg_type,
+void rpn_char_map_prepare(struct zebra_register *reg, zebra_map_t zm,
struct rpn_char_map_info *map_info)
{
- map_info->zm = reg->zebra_maps;
- map_info->reg_type = reg_type;
- dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler);
+ map_info->zm = zm;
+ if (zebra_maps_is_icu(zm))
+ dict_grep_cmap(reg->dict, 0, 0);
+ else
+ dict_grep_cmap(reg->dict, map_info, rpn_char_map_handler);
}
#define TERM_COUNT
int isam_p_indx;
int trunc_max;
ZebraHandle zh;
- int reg_type;
+ const char *index_type;
ZebraSet termset;
};
const char *index_name;
int len = key_SU_decode(&ord, (const unsigned char *) name);
- zebra_term_untrans (p->zh, p->reg_type, term_tmp, name+len);
+ zebra_term_untrans (p->zh, p->index_type, term_tmp, name+len);
yaz_log(log_level_rpn, "grep: %d %c %s", ord, name[len], term_tmp);
zebraExplain_lookup_ord(p->zh->reg->zei,
ord, 0 /* index_type */, &db, &index_name);
return add_isam_p(name, info, (struct grep_info *) p);
}
-static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src,
+static int term_pre(zebra_map_t zm, const char **src,
const char *ct1, const char *ct2, int first)
{
const char *s1, *s0 = *src;
if (ct2 && strchr(ct2, *s0))
break;
s1 = s0;
- map = zebra_maps_input(zebra_maps, reg_type, &s1, strlen(s1), first);
+ map = zebra_maps_input(zm, &s1, strlen(s1), first);
if (**map != *CHR_SPACE)
break;
s0 = s1;
}
}
+
+static int term_100_icu(zebra_map_t zm,
+ const char **src, WRBUF term_dict, int space_split,
+ char *dst_term)
+{
+ int no = 0;
+ const char *res_buf = 0;
+ size_t res_len = 0;
+ int r = zebra_map_tokenize(zm, *src, strlen(*src),
+ &res_buf, &res_len);
+
+ yaz_log(YLOG_LOG, "term_100_icu r=%d", r);
+ if (r)
+ strcat(dst_term, *src);
+ *src += strlen(*src);
+ while (r)
+ {
+ int i;
+ no++;
+ for (i = 0; i < res_len; i++)
+ {
+ if (strchr(REGEX_CHARS, res_buf[i]))
+ wrbuf_putc(term_dict, '\\');
+ if (res_buf[i] < 32)
+ wrbuf_putc(term_dict, 1);
+ wrbuf_putc(term_dict, res_buf[i]);
+ }
+ r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len);
+ }
+ return no;
+}
+
/* term_100: handle term, where trunc = none(no operators at all) */
-static int term_100(ZebraMaps zebra_maps, const char *index_type,
+static int term_100(zebra_map_t zm,
const char **src, WRBUF term_dict, int space_split,
char *dst_term)
{
const char *space_start = 0;
const char *space_end = 0;
- if (!term_pre(zebra_maps, *index_type, src, NULL, NULL, !space_split))
+ if (zebra_maps_is_icu(zm))
+ return term_100_icu(zm, src, term_dict, space_split, dst_term);
+
+ if (!term_pre(zm, src, NULL, NULL, !space_split))
return 0;
s0 = *src;
while (*s0)
{
const char *s1 = s0;
int q_map_match = 0;
- map = zebra_maps_search(zebra_maps, *index_type, &s0, strlen(s0),
- &q_map_match);
+ map = zebra_maps_search(zm, &s0, strlen(s0), &q_map_match);
if (space_split)
{
if (**map == *CHR_SPACE)
}
/* term_101: handle term, where trunc = Process # */
-static int term_101(ZebraMaps zebra_maps, const char *index_type,
+static int term_101(zebra_map_t zm,
const char **src, WRBUF term_dict, int space_split,
char *dst_term)
{
int i = 0;
int j = 0;
- if (!term_pre(zebra_maps, *index_type, src, "#", "#", !space_split))
+ if (!term_pre(zm, src, "#", "#", !space_split))
return 0;
s0 = *src;
while (*s0)
{
const char *s1 = s0;
int q_map_match = 0;
- map = zebra_maps_search(zebra_maps, *index_type, &s0, strlen(s0),
- &q_map_match);
+ map = zebra_maps_search(zm, &s0, strlen(s0), &q_map_match);
if (space_split && **map == *CHR_SPACE)
break;
}
/* term_103: handle term, where trunc = re-2 (regular expressions) */
-static int term_103(ZebraMaps zebra_maps, const char *index_type,
- const char **src,
+static int term_103(zebra_map_t zm, const char **src,
WRBUF term_dict, int *errors, int space_split,
char *dst_term)
{
const char *s0;
const char **map;
- if (!term_pre(zebra_maps, *index_type, src, "^\\()[].*+?|", "(", !space_split))
+ if (!term_pre(zm, src, "^\\()[].*+?|", "(", !space_split))
return 0;
s0 = *src;
if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
{
const char *s1 = s0;
int q_map_match = 0;
- map = zebra_maps_search(zebra_maps, *index_type, &s0, strlen(s0),
- &q_map_match);
+ map = zebra_maps_search(zm, &s0, strlen(s0), &q_map_match);
if (space_split && **map == *CHR_SPACE)
break;
}
/* term_103: handle term, where trunc = re-1 (regular expressions) */
-static int term_102(ZebraMaps zebra_maps, const char *index_type,
- const char **src,
+static int term_102(zebra_map_t zm, const char **src,
WRBUF term_dict, int space_split, char *dst_term)
{
- return term_103(zebra_maps, index_type, src, term_dict, NULL, space_split,
- dst_term);
+ return term_103(zm, src, term_dict, NULL, space_split, dst_term);
}
/* term_104: handle term, process # and ! */
-static int term_104(ZebraMaps zebra_maps, const char *index_type,
- const char **src, WRBUF term_dict, int space_split,
- char *dst_term)
+static int term_104(zebra_map_t zm, const char **src,
+ WRBUF term_dict, int space_split, char *dst_term)
{
const char *s0;
const char **map;
int i = 0;
int j = 0;
- if (!term_pre(zebra_maps, *index_type, src, "?*#", "?*#", !space_split))
+ if (!term_pre(zm, src, "?*#", "?*#", !space_split))
return 0;
s0 = *src;
while (*s0)
{
const char *s1 = s0;
int q_map_match = 0;
- map = zebra_maps_search(zebra_maps, *index_type, &s0, strlen(s0),
- &q_map_match);
+ map = zebra_maps_search(zm, &s0, strlen(s0), &q_map_match);
if (space_split && **map == *CHR_SPACE)
break;
}
/* term_105/106: handle term, where trunc = Process * and ! and right trunc */
-static int term_105(ZebraMaps zebra_maps, const char *index_type,
- const char **src, WRBUF term_dict, int space_split,
+static int term_105(zebra_map_t zm, const char **src,
+ WRBUF term_dict, int space_split,
char *dst_term, int right_truncate)
{
const char *s0;
int i = 0;
int j = 0;
- if (!term_pre(zebra_maps, *index_type, src, "*!", "*!", !space_split))
+ if (!term_pre(zm, src, "*!", "*!", !space_split))
return 0;
s0 = *src;
while (*s0)
{
const char *s1 = s0;
int q_map_match = 0;
- map = zebra_maps_search(zebra_maps, *index_type, &s0, strlen(s0),
- &q_map_match);
+ map = zebra_maps_search(zm, &s0, strlen(s0), &q_map_match);
if (space_split && **map == *CHR_SPACE)
break;
static int string_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
const char **term_sub, WRBUF term_dict,
const Odr_oid *attributeSet,
- const char *index_type, int space_split, char *term_dst,
+ zebra_map_t zm, int space_split, char *term_dst,
int *error_code)
{
AttrType relation;
switch (relation_value)
{
case 1:
- if (!term_100(zh->reg->zebra_maps, index_type,
- term_sub, term_component,
- space_split, term_dst))
+ if (!term_100(zm, term_sub, term_component, space_split, term_dst))
{
wrbuf_destroy(term_component);
return 0;
wrbuf_putc(term_dict, ')');
break;
case 2:
- if (!term_100(zh->reg->zebra_maps, index_type,
- term_sub, term_component,
- space_split, term_dst))
+ if (!term_100(zm, term_sub, term_component, space_split, term_dst))
{
wrbuf_destroy(term_component);
return 0;
wrbuf_putc(term_dict, ')');
break;
case 5:
- if (!term_100(zh->reg->zebra_maps, index_type,
- term_sub, term_component, space_split, term_dst))
+ if (!term_100(zm, term_sub, term_component, space_split, term_dst))
{
wrbuf_destroy(term_component);
return 0;
wrbuf_putc(term_dict, ')');
break;
case 4:
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub,
- term_component, space_split, term_dst))
+ if (!term_100(zm, term_sub, term_component, space_split, term_dst))
{
wrbuf_destroy(term_component);
return 0;
if (!**term_sub)
return 1;
yaz_log(log_level_rpn, "Relation =");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub,
- term_component, space_split, term_dst))
+ if (!term_100(zm, term_sub, term_component, space_split, term_dst))
{
wrbuf_destroy(term_component);
return 0;
const char *xpath_use,
struct ord_list **ol);
-static ZEBRA_RES term_limits_APT(ZebraHandle zh,
- Z_AttributesPlusTerm *zapt,
- zint *hits_limit_value,
- const char **term_ref_id_str,
- NMEM nmem)
+ZEBRA_RES zebra_term_limits_APT(ZebraHandle zh,
+ Z_AttributesPlusTerm *zapt,
+ zint *hits_limit_value,
+ const char **term_ref_id_str,
+ NMEM nmem)
{
AttrType term_ref_id_attr;
AttrType hits_limit_attr;
WRBUF term_dict = wrbuf_alloc();
*rset = 0;
- term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, stream);
+ zebra_term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str,
+ stream);
grep_info->isam_p_indx = 0;
res = string_term(zh, zapt, term_sub, term_dict,
attributeSet, stream, grep_info,
int relation_error;
char ord_buf[32];
int ord_len, i;
-
+ zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, index_type);
+
*ol = ord_list_create(stream);
- rpn_char_map_prepare(zh->reg, *index_type, &rcmi);
+ rpn_char_map_prepare(zh->reg, zm, &rcmi);
attr_init_APT(&truncation, zapt, 5);
truncation_value = attr_find(&truncation, NULL);
yaz_log(log_level_rpn, "truncation value %d", truncation_value);
case 100: /* do not truncate */
if (!string_relation(zh, zapt, &termp, term_dict,
attributeSet,
- index_type, space_split, term_dst,
+ zm, space_split, term_dst,
&relation_error))
{
if (relation_error)
break;
case 1: /* right truncation */
wrbuf_putc(term_dict, '(');
- if (!term_100(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_100(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 2: /* keft truncation */
wrbuf_puts(term_dict, "(.*");
- if (!term_100(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_100(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 3: /* left&right truncation */
wrbuf_puts(term_dict, "(.*");
- if (!term_100(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_100(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 101: /* process # in term */
wrbuf_putc(term_dict, '(');
- if (!term_101(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_101(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 102: /* Regexp-1 */
wrbuf_putc(term_dict, '(');
- if (!term_102(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_102(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
case 103: /* Regexp-2 */
regex_range = 1;
wrbuf_putc(term_dict, '(');
- if (!term_103(zh->reg->zebra_maps, index_type,
- &termp, term_dict, ®ex_range,
+ if (!term_103(zm, &termp, term_dict, ®ex_range,
space_split, term_dst))
{
*term_sub = 0;
break;
case 104: /* process # and ! in term */
wrbuf_putc(term_dict, '(');
- if (!term_104(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst))
+ if (!term_104(zm, &termp, term_dict, space_split, term_dst))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 105: /* process * and ! in term */
wrbuf_putc(term_dict, '(');
- if (!term_105(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst, 1))
+ if (!term_105(zm, &termp, term_dict, space_split, term_dst, 1))
{
*term_sub = 0;
return ZEBRA_OK;
break;
case 106: /* process * and ! in term */
wrbuf_putc(term_dict, '(');
- if (!term_105(zh->reg->zebra_maps, index_type,
- &termp, term_dict, space_split, term_dst, 0))
+ if (!term_105(zm, &termp, term_dict, space_split, term_dst, 0))
{
*term_sub = 0;
return ZEBRA_OK;
const char *input = wrbuf_cstr(term_dict) + prefix_len;
esc_str(buf, sizeof(buf), input, strlen(input));
}
- yaz_log(log_level_rpn, "dict_lookup_grep: %s",
- wrbuf_cstr(term_dict) + prefix_len);
+ {
+ WRBUF pr_wr = wrbuf_alloc();
+
+ wrbuf_write_escaped(pr_wr, wrbuf_buf(term_dict), wrbuf_len(term_dict));
+ yaz_log(YLOG_LOG, "dict_lookup_grep: %s", wrbuf_cstr(pr_wr));
+ wrbuf_destroy(pr_wr);
+ }
r = dict_lookup_grep(zh->reg->dict, wrbuf_cstr(term_dict), regex_range,
grep_info, &max_pos,
ord_len /* number of "exact" chars */,
static ZEBRA_RES grep_info_prepare(ZebraHandle zh,
Z_AttributesPlusTerm *zapt,
struct grep_info *grep_info,
- int reg_type)
+ const char *index_type)
{
#ifdef TERM_COUNT
grep_info->term_no = 0;
grep_info->isam_p_size = 0;
grep_info->isam_p_buf = NULL;
grep_info->zh = zh;
- grep_info->reg_type = reg_type;
+ grep_info->index_type = index_type;
grep_info->termset = 0;
if (zapt)
{
*num_result_sets = 0;
*term_dst = 0;
- if (grep_info_prepare(zh, zapt, &grep_info, *index_type) == ZEBRA_FAIL)
+ if (grep_info_prepare(zh, zapt, &grep_info, index_type) == ZEBRA_FAIL)
return ZEBRA_FAIL;
while(1)
{
int ord_len;
char *val;
ISAM_P isam_p;
+ zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, index_type);
attr_init_APT(&position, zapt, 3);
position_value = attr_find(&position, NULL);
return ZEBRA_FAIL;
}
- if (!zebra_maps_is_first_in_field(zh->reg->zebra_maps, *index_type))
- {
- zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_POSITION_ATTRIBUTE,
- position_value);
- return ZEBRA_FAIL;
- }
- if (!zh->reg->isamb && !zh->reg->isamc)
+ if (!zebra_maps_is_first_in_field(zm))
{
zebra_setError_zint(zh, YAZ_BIB1_UNSUPP_POSITION_ATTRIBUTE,
position_value);
{
assert(*val == sizeof(ISAM_P));
memcpy(&isam_p, val+1, sizeof(isam_p));
-
- if (zh->reg->isamb)
- *rset = rsisamb_create(rset_nmem, kc, kc->scope,
- zh->reg->isamb, isam_p, 0);
- else if (zh->reg->isamc)
- *rset = rsisamc_create(rset_nmem, kc, kc->scope,
- zh->reg->isamc, isam_p, 0);
+
+ *rset = zebra_create_rset_isam(zh, rset_nmem, kc, kc->scope,
+ isam_p, 0);
}
return ZEBRA_OK;
}
const Odr_oid *attributeSet,
struct grep_info *grep_info,
int *max_pos,
- const char *index_type,
+ zebra_map_t zm,
char *term_dst,
int *error_code)
{
{
case 1:
yaz_log(log_level_rpn, "Relation <");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub, term_num, 1,
- term_dst))
+ if (!term_100(zm, term_sub, term_num, 1, term_dst))
{
wrbuf_destroy(term_num);
return 0;
break;
case 2:
yaz_log(log_level_rpn, "Relation <=");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub, term_num, 1,
- term_dst))
+ if (!term_100(zm, term_sub, term_num, 1, term_dst))
{
wrbuf_destroy(term_num);
return 0;
break;
case 4:
yaz_log(log_level_rpn, "Relation >=");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub, term_num, 1,
- term_dst))
+ if (!term_100(zm, term_sub, term_num, 1, term_dst))
{
wrbuf_destroy(term_num);
return 0;
break;
case 5:
yaz_log(log_level_rpn, "Relation >");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub, term_num, 1,
- term_dst))
+ if (!term_100(zm, term_sub, term_num, 1, term_dst))
{
wrbuf_destroy(term_num);
return 0;
case -1:
case 3:
yaz_log(log_level_rpn, "Relation =");
- if (!term_100(zh->reg->zebra_maps, index_type, term_sub, term_num, 1,
- term_dst))
+ if (!term_100(zm, term_sub, term_num, 1, term_dst))
{
wrbuf_destroy(term_num);
return 0;
int relation_error = 0;
int ord, ord_len, i;
char ord_buf[32];
+ zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, index_type);
*ol = ord_list_create(stream);
- rpn_char_map_prepare(zh->reg, *index_type, &rcmi);
+ rpn_char_map_prepare(zh->reg, zm, &rcmi);
termp = *term_sub;
wrbuf_putc(term_dict, ')');
if (!numeric_relation(zh, zapt, &termp, term_dict,
- attributeSet, grep_info, &max_pos, index_type,
+ attributeSet, grep_info, &max_pos, zm,
term_dst, &relation_error))
{
if (relation_error)
zint hits_limit_value;
const char *term_ref_id_str = 0;
- term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str, stream);
+ zebra_term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str,
+ stream);
yaz_log(log_level_rpn, "APT_numeric t='%s'", termz);
- if (grep_info_prepare(zh, zapt, &grep_info, *index_type) == ZEBRA_FAIL)
+ if (grep_info_prepare(zh, zapt, &grep_info, index_type) == ZEBRA_FAIL)
return ZEBRA_FAIL;
while (1)
{
int ord = zebraExplain_lookup_attr_str(zh->reg->zei,
zinfo_index_category_index,
index_type, xpath_use);
- if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, '0') == ZEBRA_FAIL)
+ if (grep_info_prepare(zh, 0 /* zapt */, &grep_info, "0") == ZEBRA_FAIL)
return rset_create_null(rset_nmem, kc, 0);
if (ord < 0)
static ZEBRA_RES rpn_search_APT(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
const Odr_oid *attributeSet, NMEM stream,
Z_SortKeySpecList *sort_sequence,
- int num_bases, char **basenames,
+ int num_bases, const char **basenames,
NMEM rset_nmem,
RSET *rset,
struct rset_key_control *kc)
const Odr_oid *attributeSet,
NMEM stream, NMEM rset_nmem,
Z_SortKeySpecList *sort_sequence,
- int num_bases, char **basenames,
+ int num_bases, const char **basenames,
RSET **result_sets, int *num_result_sets,
Z_Operator *parent_op,
struct rset_key_control *kc);
const Odr_oid *attributeSet,
NMEM stream, NMEM rset_nmem,
Z_SortKeySpecList *sort_sequence,
- int num_bases, char **basenames,
+ int num_bases, const char **basenames,
RSET *result_set)
{
RSET *result_sets = 0;
const Odr_oid *attributeSet,
NMEM stream, NMEM rset_nmem,
Z_SortKeySpecList *sort_sequence,
- int num_bases, char **basenames,
+ int num_bases, const char **basenames,
RSET **result_sets, int *num_result_sets,
Z_Operator *parent_op,
struct rset_key_control *kc)
projects / idzebra-moved-to-github.git / blobdiff