recgrs.c File Reference

#include <stdio.h>
#include <assert.h>
#include <sys/types.h>
#include <ctype.h>
#include <yaz/log.h>
#include <yaz/oid_db.h>
#include <yaz/diagbib1.h>
#include <yaz/wrbuf.h>
#include <yaz/snprintf.h>
#include <d1_absyn.h>
#include <idzebra/recgrs.h>

Go to the source code of this file.

Data Structures
struct	source_parser

Macros
#define	GRS_MAX_WORD   512
#define	MAX_ATTR_COUNT   50

Functions
static int	sp_lex (struct source_parser *sp)
static int	sp_expr (struct source_parser *sp, data1_node *n, RecWord *wrd)
static int	sp_range (struct source_parser *sp, data1_node *n, RecWord *wrd)
static int	sp_first (struct source_parser *sp, data1_node *n, RecWord *wrd)
static struct source_parser *	source_parser_create (void)
static void	source_parser_destroy (struct source_parser *sp)
static int	sp_parse (struct source_parser *sp, data1_node *n, RecWord *wrd, const char *src)
int	d1_check_xpath_predicate (data1_node *n, struct xpath_predicate *p)
static int	dfa_match_first (struct DFA_state **dfaar, const char *text)
data1_termlist *	xpath_termlist_by_tagpath (char *tagpath, data1_node *n)
static void	index_xpath_attr (char *tag_path, char *name, char *value, char *structure, struct recExtractCtrl *p, RecWord *wrd)
static void	mk_tag_path_full (char *tag_path_full, size_t max, data1_node *n)
static void	index_xpath (struct source_parser *sp, data1_node *n, struct recExtractCtrl *p, int level, RecWord *wrd, char *xpath_index, int xpath_is_start)
static void	index_termlist (struct source_parser *sp, data1_node *par, data1_node *n, struct recExtractCtrl *p, int level, RecWord *wrd)
static int	dumpkeys_r (struct source_parser *sp, data1_node *n, struct recExtractCtrl *p, int level, RecWord *wrd)
static int	dumpkeys (data1_node *n, struct recExtractCtrl *p, RecWord *wrd)
int	grs_extract_tree (struct recExtractCtrl *p, data1_node *n)
static int	grs_extract_sub (void *clientData, struct recExtractCtrl *p, NMEM mem, data1_node *(*grs_read)(struct grs_read_info *))
int	zebra_grs_extract (void *clientData, struct recExtractCtrl *p, data1_node *(*grs_read)(struct grs_read_info *))
static int	process_comp (data1_handle dh, data1_node *n, Z_RecordComposition *c, char **addinfo, ODR o)
static void	zebra_xml_metadata (struct recRetrieveCtrl *p, data1_node *top, NMEM mem)
int	zebra_grs_retrieve (void *clientData, struct recRetrieveCtrl *p, data1_node *(*grs_read)(struct grs_read_info *))

Macro Definition Documentation

GRS_MAX_WORD

#define GRS_MAX_WORD   512

Definition at line 37 of file recgrs.c.

MAX_ATTR_COUNT

#define MAX_ATTR_COUNT   50

Referenced by index_xpath().

Function Documentation

d1_check_xpath_predicate()

int d1_check_xpath_predicate	(	data1_node *	n,
		struct xpath_predicate *	p )

Definition at line 286 of file recgrs.c.

References xpath_predicate::boolean, d1_check_xpath_predicate(), xpath_predicate::left, data1_xattr::name, xpath_predicate::name, data1_xattr::next, xpath_predicate::op, xpath_predicate::relation, xpath_predicate::right, data1_node::tag, data1_node::u, xpath_predicate::u, data1_xattr::value, xpath_predicate::value, xpath_predicate::which, XPATH_PREDICATE_BOOLEAN, and XPATH_PREDICATE_RELATION.

Referenced by d1_check_xpath_predicate(), and xpath_termlist_by_tagpath().

dfa_match_first()

int dfa_match_first ( struct DFA_state ** dfaar, const char * text )

static

Definition at line 348 of file recgrs.c.

References DFA_tran::ch, DFA_state::rule_no, DFA_tran::to, DFA_state::tran_no, and DFA_state::trans.

Referenced by xpath_termlist_by_tagpath().

dumpkeys()

int dumpkeys ( data1_node * n, struct recExtractCtrl * p, RecWord * wrd )

static

Definition at line 877 of file recgrs.c.

References dumpkeys_r(), source_parser_create(), and source_parser_destroy().

Referenced by grs_extract_sub(), and grs_extract_tree().

dumpkeys_r()

int dumpkeys_r ( struct source_parser * sp, data1_node * n, struct recExtractCtrl * p, int level, RecWord * wrd )

static

Definition at line 781 of file recgrs.c.

References data1_node::absyn, data1_node::child, data1_node::data, DATA1N_data, DATA1N_root, DATA1N_tag, DATA1T_numeric, recExtractCtrl::dh, dumpkeys_r(), recExtractCtrl::flagShowRecords, get_parent_tag(), index_termlist(), index_xpath(), data1_element::name, data1_name::name, data1_tagset::name, data1_tag::names, data1_node::next, data1_tag::numeric, data1_node::root, data1_tag::string, data1_element::tag, data1_node::tag, data1_tag::tagset, data1_node::type, data1_tagset::type, data1_node::u, data1_tag::value, data1_node::which, data1_tag::which, ZEBRA_XPATH_CDATA, ZEBRA_XPATH_ELM_BEGIN, and ZEBRA_XPATH_ELM_END.

Referenced by dumpkeys(), and dumpkeys_r().

grs_extract_sub()

int grs_extract_sub ( void * clientData, struct recExtractCtrl * p, NMEM mem, data1_node *(* grs_read )(struct grs_read_info *) )

static

Definition at line 898 of file recgrs.c.

References data1_node::absyn, grs_read_info::clientData, data1_concat_text(), data1_get_encoding(), data1_iconv(), data1_pr_tree(), data1_remove_idzebra_subtree(), grs_read_info::dh, recExtractCtrl::dh, dumpkeys(), recExtractCtrl::init, grs_read_info::mem, data1_absyn::oid, RECCTRL_EXTRACT_EOF, RECCTRL_EXTRACT_ERROR_GENERIC, RECCTRL_EXTRACT_OK, data1_node::root, recExtractCtrl::schemaAdd, grs_read_info::stream, recExtractCtrl::stream, and data1_node::u.

Referenced by zebra_grs_extract().

grs_extract_tree()

int grs_extract_tree	(	struct recExtractCtrl *	p,
		data1_node *	n )

Definition at line 885 of file recgrs.c.

References data1_node::absyn, dumpkeys(), recExtractCtrl::init, data1_absyn::oid, data1_node::root, recExtractCtrl::schemaAdd, and data1_node::u.

Referenced by zebra_extract_explain().

index_termlist()

void index_termlist ( struct source_parser * sp, data1_node * par, data1_node * n, struct recExtractCtrl * p, int level, RecWord * wrd )

static

Definition at line 732 of file recgrs.c.

References recExtractCtrl::dh, recExtractCtrl::flagShowRecords, get_parent_tag(), data1_termlist::index_name, RecWord::index_name, RecWord::index_type, data1_termlist::next, data1_node::parent, data1_termlist::source, sp_parse(), data1_termlist::structure, data1_node::tag, RecWord::term_buf, RecWord::term_len, recExtractCtrl::tokenAdd, and data1_node::u.

Referenced by dumpkeys_r().

index_xpath()

void index_xpath ( struct source_parser * sp, data1_node * n, struct recExtractCtrl * p, int level, RecWord * wrd, char * xpath_index, int xpath_is_start )

static

Definition at line 535 of file recgrs.c.

References data1_node::absyn, data1_node::data, DATA1_XPATH_INDEXING_ENABLE, DATA1N_data, DATA1N_tag, recExtractCtrl::flagShowRecords, data1_termlist::index_name, RecWord::index_name, RecWord::index_type, index_xpath_attr(), MAX_ATTR_COUNT, mk_tag_path_full(), data1_xattr::name, data1_termlist::next, data1_xattr::next, data1_node::root, RecWord::seqno, data1_termlist::source, sp_parse(), data1_termlist::structure, data1_node::tag, RecWord::term_buf, RecWord::term_len, recExtractCtrl::tokenAdd, data1_node::u, data1_xattr::value, data1_node::which, data1_absyn::xpath_indexing, xpath_termlist_by_tagpath(), and ZEBRA_XPATH_ATTR_NAME.

Referenced by dumpkeys_r().

index_xpath_attr()

void index_xpath_attr ( char * tag_path, char * name, char * value, char * structure, struct recExtractCtrl * p, RecWord * wrd )

static

Definition at line 485 of file recgrs.c.

References RecWord::index_name, RecWord::index_type, RecWord::term_buf, RecWord::term_len, recExtractCtrl::tokenAdd, ZEBRA_XPATH_ATTR_CDATA, ZEBRA_XPATH_ELM_BEGIN, and ZEBRA_XPATH_ELM_END.

Referenced by index_xpath().

mk_tag_path_full()

void mk_tag_path_full ( char * tag_path_full, size_t max, data1_node * n )

static

Definition at line 510 of file recgrs.c.

References DATA1N_root, DATA1N_tag, data1_node::parent, data1_node::tag, data1_node::u, and data1_node::which.

Referenced by index_xpath().

process_comp()

int process_comp ( data1_handle dh, data1_node * n, Z_RecordComposition * c, char ** addinfo, ODR o )

static

Definition at line 949 of file recgrs.c.

References data1_node::absyn, data1_doespec1(), data1_getesetbyname(), grs_read_info::dh, data1_node::root, data1_esetname::spec, and data1_node::u.

Referenced by zebra_grs_retrieve().

source_parser_create()

struct source_parser * source_parser_create ( void )

static

Definition at line 257 of file recgrs.c.

References source_parser::nmem.

Referenced by dumpkeys().

source_parser_destroy()

void source_parser_destroy ( struct source_parser * sp )

static

Definition at line 265 of file recgrs.c.

References source_parser::nmem.

Referenced by dumpkeys().

sp_expr()

int sp_expr ( struct source_parser * sp, data1_node * n, RecWord * wrd )

static

Definition at line 171 of file recgrs.c.

References data1_node::data, DATA1N_data, DATA1N_tag, source_parser::len, source_parser::lookahead, data1_xattr::name, data1_xattr::next, source_parser::nmem, sp_expr(), sp_first(), sp_lex(), sp_range(), data1_node::tag, RecWord::term_buf, RecWord::term_len, source_parser::tok, data1_node::u, data1_xattr::value, and data1_node::which.

Referenced by sp_expr(), sp_first(), sp_parse(), and sp_range().

sp_first()

int sp_first ( struct source_parser * sp, data1_node * n, RecWord * wrd )

static

Definition at line 128 of file recgrs.c.

References source_parser::lookahead, source_parser::nmem, sp_expr(), sp_lex(), RecWord::term_buf, and RecWord::term_len.

Referenced by sp_expr().

sp_lex()

int sp_lex ( struct source_parser * sp )

static

Definition at line 47 of file recgrs.c.

References source_parser::len, source_parser::lookahead, source_parser::src, and source_parser::tok.

Referenced by sp_expr(), sp_first(), sp_parse(), and sp_range().

sp_parse()

int sp_parse ( struct source_parser * sp, data1_node * n, RecWord * wrd, const char * src )

static

Definition at line 273 of file recgrs.c.

References source_parser::len, source_parser::lookahead, source_parser::nmem, sp_expr(), sp_lex(), source_parser::src, and source_parser::tok.

Referenced by index_termlist(), and index_xpath().

sp_range()

int sp_range ( struct source_parser * sp, data1_node * n, RecWord * wrd )

static

Definition at line 71 of file recgrs.c.

References source_parser::lookahead, sp_expr(), sp_lex(), RecWord::term_buf, and RecWord::term_len.

Referenced by sp_expr().

xpath_termlist_by_tagpath()

data1_termlist * xpath_termlist_by_tagpath	(	char *	tagpath,
		data1_node *	n )

Definition at line 396 of file recgrs.c.

References data1_node::absyn, d1_check_xpath_predicate(), DATA1N_tag, data1_xpelement::dfa, dfa_match_first(), data1_xpelement::match_next, data1_xpelement::match_state, data1_xpelement::next, data1_node::parent, xpath_location_step::predicate, data1_node::root, DFA::states, data1_xpelement::termlists, data1_node::u, data1_node::which, data1_absyn::xp_elements, data1_xpelement::xpath, and data1_xpelement::xpath_len.

Referenced by index_xpath().

zebra_grs_extract()

int zebra_grs_extract	(	void *	clientData,
		struct recExtractCtrl *	p,
		data1_node *(*	grs_read )(struct grs_read_info *) )

Definition at line 936 of file recgrs.c.

References grs_read_info::clientData, grs_extract_sub(), and grs_read_info::mem.

Referenced by extract_marc(), extract_marcxml(), extract_regx(), and grs_extract_sgml().

zebra_grs_retrieve()

int zebra_grs_retrieve	(	void *	clientData,
		struct recRetrieveCtrl *	p,
		data1_node *(*	grs_read )(struct grs_read_info *) )

Definition at line 1072 of file recgrs.c.

References data1_node::absyn, recRetrieveCtrl::addinfo, grs_read_info::clientData, recRetrieveCtrl::comp, data1_node::data, data1_concat_text(), data1_get_encoding(), data1_get_root_tag(), data1_iconv(), data1_map_record(), data1_mk_tag_data_wd(), data1_nodetobuf(), data1_nodetoexplain(), data1_nodetogr(), data1_nodetoidsgml(), data1_nodetomarc(), data1_nodetosoif(), data1_nodetosummary(), data1_pr_tree(), data1_remove_idzebra_subtree(), data1_set_data_zint(), data1_systag_lookup(), DATA1I_oid, grs_read_info::dh, recRetrieveCtrl::dh, recRetrieveCtrl::diagnostic, recRetrieveCtrl::encoding, recRetrieveCtrl::input_format, recRetrieveCtrl::localno, data1_absyn::maptabs, data1_absyn::marc, grs_read_info::mem, data1_maptab::next, data1_marctab::next, recRetrieveCtrl::odr, data1_absyn::oid, data1_maptab::oid, data1_marctab::oid, recRetrieveCtrl::output_format, process_comp(), recRetrieveCtrl::rec_buf, recRetrieveCtrl::rec_len, recRetrieveCtrl::recordSize, data1_node::root, recRetrieveCtrl::score, grs_read_info::stream, recRetrieveCtrl::stream, data1_node::u, and zebra_xml_metadata().

Referenced by grs_retrieve_sgml(), retrieve_marc(), retrieve_marcxml(), and retrieve_regx().

zebra_xml_metadata()

void zebra_xml_metadata ( struct recRetrieveCtrl * p, data1_node * top, NMEM mem )

static

Definition at line 1035 of file recgrs.c.

References data1_mk_tag(), data1_mk_tag_data_int(), data1_mk_tag_data_text(), data1_mk_tag_data_zint(), data1_mk_text(), recRetrieveCtrl::dh, recRetrieveCtrl::fname, recRetrieveCtrl::localno, grs_read_info::mem, recRetrieveCtrl::recordSize, and recRetrieveCtrl::score.

Referenced by zebra_grs_retrieve().