mod_grs_regx.c File Reference

#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>
#include <ctype.h>
#include <yaz/tpath.h>
#include <yaz/snprintf.h>
#include <idzebra/util.h>
#include <dfa.h>
#include <idzebra/recgrs.h>

Go to the source code of this file.

Data Structures
struct	regxCode
struct	lexRuleAction
struct	lexRuleInfo
struct	lexRule
struct	lexContext
struct	lexConcatBuf
struct	lexSpec
struct	lexSpecs

Macros
#define	REGX_DEBUG   0
#define	F_WIN_EOF   2000000000
#define	F_WIN_READ   1
#define	REGX_EOF   0
#define	REGX_PATTERN   1
#define	REGX_BODY   2
#define	REGX_BEGIN   3
#define	REGX_END   4
#define	REGX_CODE   5
#define	REGX_CONTEXT   6
#define	REGX_INIT   7
#define	REGX_MAX_ARGS   20

Functions
static char *	f_win_get (struct lexSpec *spec, off_t start_pos, off_t end_pos, int *size)
static int	f_win_advance (struct lexSpec *spec, int *pos)
static void	regxCodeDel (struct regxCode **pp)
static void	regxCodeMk (struct regxCode **pp, const char *buf, int len)
static struct DFA *	lexSpecDFA (void)
static void	actionListDel (struct lexRuleAction **rap)
static struct lexContext *	lexContextCreate (const char *name)
static void	lexContextDestroy (struct lexContext *p)
static struct lexSpec *	lexSpecCreate (const char *name, data1_handle dh)
static void	lexSpecDestroy (struct lexSpec **pp)
static int	readParseToken (const char **cpp, int *len)
static int	actionListMk (struct lexSpec *spec, const char *s, struct lexRuleAction **ap)
int	readOneSpec (struct lexSpec *spec, const char *s)
int	readFileSpec (struct lexSpec *spec)
static void	execData (struct lexSpec *spec, const char *ebuf, int elen, int formatted_text, const char *attribute_str, int attribute_len)
static void	execDataP (struct lexSpec *spec, const char *ebuf, int elen, int formatted_text)
static void	tagDataRelease (struct lexSpec *spec)
static void	variantBegin (struct lexSpec *spec, const char *class_str, int class_len, const char *type_str, int type_len, const char *value_str, int value_len)
static void	tagStrip (const char **tag, int *len)
static void	tagBegin (struct lexSpec *spec, const char *tag, int len)
static void	tagEnd (struct lexSpec *spec, int min_level, const char *tag, int len)
static int	tryMatch (struct lexSpec *spec, int *pptr, int *mptr, struct DFA *dfa, int greedy)
static int	execTok (struct lexSpec *spec, const char **src, const char **tokBuf, int *tokLen)
static char *	regxStrz (const char *src, int len, char *str)
static void	execCode (struct lexSpec *spec, struct regxCode *code)
static int	execAction (struct lexSpec *spec, struct lexRuleAction *ap, int start_ptr, int *pptr)
static int	execRule (struct lexSpec *spec, struct lexContext *context, int ruleNo, int start_ptr, int *pptr)
int	lexNode (struct lexSpec *spec, int *ptr)
static data1_node *	lexRoot (struct lexSpec *spec, off_t offset, const char *context_name)
void	grs_destroy (void *clientData)
void *	grs_init (Res res, RecType recType)
ZEBRA_RES	grs_config (void *clientData, Res res, const char *args)
data1_node *	grs_read_regx (struct grs_read_info *p)
static int	extract_regx (void *clientData, struct recExtractCtrl *ctrl)
static int	retrieve_regx (void *clientData, struct recRetrieveCtrl *ctrl)

Variables
static struct recType	regx_type
RecType	idzebra_filter []

Macro Definition Documentation

F_WIN_EOF

#define F_WIN_EOF   2000000000

Definition at line 45 of file mod_grs_regx.c.

Referenced by execAction(), f_win_advance(), lexNode(), and tryMatch().

F_WIN_READ

#define F_WIN_READ   1

Definition at line 46 of file mod_grs_regx.c.

REGX_BEGIN

#define REGX_BEGIN   3

Definition at line 51 of file mod_grs_regx.c.

Referenced by actionListMk(), readOneSpec(), and readParseToken().

REGX_BODY

#define REGX_BODY   2

Definition at line 50 of file mod_grs_regx.c.

Referenced by actionListMk(), and readParseToken().

REGX_CODE

#define REGX_CODE   5

Definition at line 53 of file mod_grs_regx.c.

Referenced by actionListDel(), actionListMk(), execAction(), readOneSpec(), and readParseToken().

REGX_CONTEXT

#define REGX_CONTEXT   6

Definition at line 54 of file mod_grs_regx.c.

Referenced by readOneSpec(), and readParseToken().

REGX_DEBUG

#define REGX_DEBUG   0

Definition at line 43 of file mod_grs_regx.c.

REGX_END

#define REGX_END   4

Definition at line 52 of file mod_grs_regx.c.

Referenced by actionListMk(), execAction(), readOneSpec(), and readParseToken().

REGX_EOF

#define REGX_EOF   0

Definition at line 48 of file mod_grs_regx.c.

REGX_INIT

#define REGX_INIT   7

Definition at line 55 of file mod_grs_regx.c.

Referenced by actionListMk(), readOneSpec(), and readParseToken().

REGX_MAX_ARGS

#define REGX_MAX_ARGS   20

Definition at line 57 of file mod_grs_regx.c.

Referenced by execAction().

REGX_PATTERN

#define REGX_PATTERN   1

Definition at line 49 of file mod_grs_regx.c.

Referenced by actionListDel(), actionListMk(), execAction(), readOneSpec(), and readParseToken().

Function Documentation

actionListDel()

void actionListDel ( struct lexRuleAction ** rap )

static

Definition at line 247 of file mod_grs_regx.c.

References lexRuleAction::code, lexRuleAction::dfa, dfa_delete(), lexRuleAction::next, lexRuleAction::pattern, REGX_CODE, REGX_PATTERN, regxCodeDel(), lexRuleAction::u, and lexRuleAction::which.

Referenced by lexContextDestroy(), and readOneSpec().

actionListMk()

int actionListMk ( struct lexSpec * spec, const char * s, struct lexRuleAction ** ap )

static

Definition at line 445 of file mod_grs_regx.c.

References debug_dfa_tran, dfa_mkstate(), dfa_parse(), lexSpecDFA(), lexRuleAction::next, readParseToken(), REGX_BEGIN, REGX_BODY, REGX_CODE, REGX_END, REGX_INIT, REGX_PATTERN, regxCodeMk(), and lexRuleAction::which.

Referenced by readOneSpec().

execAction()

int execAction ( struct lexSpec * spec, struct lexRuleAction * ap, int start_ptr, int * pptr )

static

Definition at line 1616 of file mod_grs_regx.c.

References lexSpec::arg_end, lexSpec::arg_no, lexSpec::arg_start, lexRuleAction::body, lexRuleAction::code, lexRuleAction::dfa, execCode(), F_WIN_EOF, lexRuleAction::next, lexRuleAction::pattern, lexSpec::ptr, REGX_CODE, REGX_END, REGX_MAX_ARGS, REGX_PATTERN, lexSpec::stop_flag, tryMatch(), lexRuleAction::u, and lexRuleAction::which.

Referenced by execRule(), and lexRoot().

execCode()

void execCode ( struct lexSpec * spec, struct regxCode * code )

static

Definition at line 1320 of file mod_grs_regx.c.

References lexSpec::arg_no, lexSpec::arg_start, lexSpec::context, lexSpec::context_stack, lexSpec::context_stack_top, lexSpec::d1_level, lexSpec::d1_stack, data1_mk_root(), data1_mk_tag(), lexSpec::dh, execData(), execTok(), lexSpec::m, lexContext::name, lexContext::next, lexSpec::ptr, regxStrz(), lexSpec::stop_flag, regxCode::str, tagBegin(), tagDataRelease(), tagEnd(), and variantBegin().

Referenced by execAction().

execData()

void execData ( struct lexSpec * spec, const char * ebuf, int elen, int formatted_text, const char * attribute_str, int attribute_len )

static

Definition at line 675 of file mod_grs_regx.c.

References lexConcatBuf::buf, lexSpec::concatBuf, lexSpec::d1_level, lexSpec::d1_stack, data1_node::data, data1_mk_node2(), DATA1I_text, DATA1N_data, DATA1N_tag, lexSpec::dh, data1_node::formatted_text, lexSpec::m, lexConcatBuf::max, data1_node::next, data1_xattr::next, data1_node::parent, data1_node::tag, data1_node::u, and data1_node::which.

Referenced by execCode(), and execDataP().

execDataP()

void execDataP ( struct lexSpec * spec, const char * ebuf, int elen, int formatted_text )

static

Definition at line 776 of file mod_grs_regx.c.

References execData(), and data1_node::formatted_text.

Referenced by lexNode().

execRule()

int execRule ( struct lexSpec * spec, struct lexContext * context, int ruleNo, int start_ptr, int * pptr )

static

Definition at line 1689 of file mod_grs_regx.c.

References lexRuleInfo::actionList, execAction(), lexContext::fastRule, and lexContext::ruleNo.

Referenced by lexNode().

execTok()

int execTok ( struct lexSpec * spec, const char ** src, const char ** tokBuf, int * tokLen )

static

Definition at line 1009 of file mod_grs_regx.c.

References lexSpec::arg_end, lexSpec::arg_no, lexSpec::arg_start, and f_win_get().

Referenced by execCode().

extract_regx()

int extract_regx ( void * clientData, struct recExtractCtrl * ctrl )

static

Definition at line 1932 of file mod_grs_regx.c.

References grs_read_regx(), and zebra_grs_extract().

f_win_advance()

int f_win_advance ( struct lexSpec * spec, int * pos )

static

Definition at line 186 of file mod_grs_regx.c.

References lexSpec::f_win_buf, lexSpec::f_win_end, F_WIN_EOF, f_win_get(), and lexSpec::f_win_start.

Referenced by lexNode(), and tryMatch().

f_win_get()

char * f_win_get ( struct lexSpec * spec, off_t start_pos, off_t end_pos, int * size )

static

Definition at line 148 of file mod_grs_regx.c.

References lexSpec::f_win_buf, lexSpec::f_win_end, lexSpec::f_win_rf, lexSpec::f_win_sf, lexSpec::f_win_size, lexSpec::f_win_start, and lexSpec::stream.

Referenced by execTok(), f_win_advance(), and lexNode().

grs_config()

ZEBRA_RES grs_config	(	void *	clientData,
		Res	res,
		const char *	args )

Definition at line 1886 of file mod_grs_regx.c.

References lexSpecs::type, and ZEBRA_OK.

grs_destroy()

void grs_destroy

(

void *

clientData

)

Definition at line 1867 of file mod_grs_regx.c.

References lexSpecDestroy(), and lexSpecs::spec.

grs_init()

void * grs_init	(	Res	res,
		RecType	recType )

Definition at line 1877 of file mod_grs_regx.c.

References lexSpecs::spec, and lexSpecs::type.

grs_read_regx()

data1_node * grs_read_regx

(

struct grs_read_info *

p

)

Definition at line 1894 of file mod_grs_regx.c.

References grs_read_info::clientData, grs_read_info::dh, ZebraRecStream::endf, lexRoot(), lexSpecCreate(), lexSpecDestroy(), grs_read_info::mem, ZebraRecStream::readf, readFileSpec(), ZebraRecStream::seekf, lexSpecs::spec, grs_read_info::stream, ZebraRecStream::tellf, and lexSpecs::type.

Referenced by extract_regx(), and retrieve_regx().

lexContextCreate()

struct lexContext * lexContextCreate ( const char * name )

static

Definition at line 268 of file mod_grs_regx.c.

References lexContext::beginActionList, lexContext::dfa, lexContext::endActionList, lexContext::fastRule, lexContext::initActionList, lexContext::initFlag, lexSpecDFA(), lexContext::name, lexContext::next, lexContext::ruleNo, and lexContext::rules.

Referenced by readOneSpec().

lexContextDestroy()

void lexContextDestroy ( struct lexContext * p )

static

Definition at line 285 of file mod_grs_regx.c.

References lexRuleInfo::actionList, actionListDel(), lexContext::beginActionList, lexContext::dfa, dfa_delete(), lexContext::endActionList, lexContext::fastRule, lexRule::info, lexContext::initActionList, lexContext::name, lexRule::next, and lexContext::rules.

Referenced by lexSpecDestroy().

lexNode()

int lexNode	(	struct lexSpec *	spec,
		int *	ptr )

Definition at line 1699 of file mod_grs_regx.c.

References DFA_tran::ch, lexSpec::context_stack, lexSpec::context_stack_top, lexContext::dfa, execDataP(), execRule(), f_win_advance(), lexSpec::f_win_ef, F_WIN_EOF, f_win_get(), DFA_state::rule_nno, DFA_state::rule_no, DFA::states, lexSpec::stream, DFA_tran::to, DFA_state::tran_no, and DFA_state::trans.

Referenced by lexRoot().

lexRoot()

data1_node * lexRoot ( struct lexSpec * spec, off_t offset, const char * context_name )

static

Definition at line 1823 of file mod_grs_regx.c.

References lexContext::beginActionList, lexSpec::context, lexSpec::context_stack, lexSpec::context_stack_top, lexSpec::d1_level, lexSpec::d1_stack, lexContext::endActionList, execAction(), lexContext::initActionList, lexContext::initFlag, lexNode(), lexContext::name, lexContext::next, lexSpec::stop_flag, and tagDataRelease().

Referenced by grs_read_regx().

lexSpecCreate()

struct lexSpec * lexSpecCreate ( const char * name, data1_handle dh )

static

Definition at line 304 of file mod_grs_regx.c.

References lexConcatBuf::buf, lexSpec::concatBuf, lexSpec::context, lexSpec::context_stack, lexSpec::context_stack_size, lexSpec::d1_level, lexSpec::d1_stack, lexSpec::dh, lexSpec::f_win_buf, lexConcatBuf::max, lexSpec::maxLevel, and lexSpec::name.

Referenced by grs_read_regx().

lexSpecDestroy()

void lexSpecDestroy ( struct lexSpec ** pp )

static

Definition at line 336 of file mod_grs_regx.c.

References lexConcatBuf::buf, lexSpec::concatBuf, lexSpec::context, lexSpec::context_stack, lexSpec::d1_stack, lexSpec::f_win_buf, lexContextDestroy(), lexSpec::maxLevel, lexSpec::name, and lexContext::next.

Referenced by grs_destroy(), and grs_read_regx().

lexSpecDFA()

struct DFA * lexSpecDFA ( void )

static

Definition at line 236 of file mod_grs_regx.c.

References dfa_init(), dfa_parse_cmap_add(), and dfa_parse_cmap_del().

Referenced by actionListMk(), and lexContextCreate().

readFileSpec()

int readFileSpec

(

struct lexSpec *

spec

)

Definition at line 574 of file mod_grs_regx.c.

References lexSpec::context, data1_path_fopen(), debug_dfa_followpos, debug_dfa_tran, debug_dfa_trav, lexContext::dfa, dfa_mkstate(), dfa_verbose, lexSpec::dh, lexContext::fastRule, lexRule::info, lexSpec::lineNo, lexSpec::name, lexContext::next, lexRule::next, lexRuleInfo::no, readOneSpec(), lexContext::ruleNo, and lexContext::rules.

Referenced by grs_read_regx().

readOneSpec()

int readOneSpec	(	struct lexSpec *	spec,
		const char *	s )

Definition at line 507 of file mod_grs_regx.c.

References lexRuleInfo::actionList, actionListDel(), actionListMk(), lexContext::beginActionList, lexSpec::context, lexContext::dfa, dfa_parse(), lexContext::endActionList, lexRule::info, lexContext::initActionList, lexContextCreate(), lexContext::next, lexRule::next, lexRuleInfo::no, readParseToken(), REGX_BEGIN, REGX_CODE, REGX_CONTEXT, REGX_END, REGX_INIT, REGX_PATTERN, lexContext::ruleNo, and lexContext::rules.

Referenced by readFileSpec().

readParseToken()

int readParseToken ( const char ** cpp, int * len )

static

Definition at line 370 of file mod_grs_regx.c.

References REGX_BEGIN, REGX_BODY, REGX_CODE, REGX_CONTEXT, REGX_END, REGX_INIT, and REGX_PATTERN.

Referenced by actionListMk(), and readOneSpec().

regxCodeDel()

void regxCodeDel ( struct regxCode ** pp )

static

Definition at line 205 of file mod_grs_regx.c.

References regxCode::str.

Referenced by actionListDel().

regxCodeMk()

void regxCodeMk ( struct regxCode ** pp, const char * buf, int len )

static

Definition at line 220 of file mod_grs_regx.c.

References regxCode::str.

Referenced by actionListMk().

regxStrz()

char * regxStrz ( const char * src, int len, char * str )

static

Definition at line 1074 of file mod_grs_regx.c.

Referenced by execCode().

retrieve_regx()

int retrieve_regx ( void * clientData, struct recRetrieveCtrl * ctrl )

static

Definition at line 1937 of file mod_grs_regx.c.

References grs_read_regx(), and zebra_grs_retrieve().

tagBegin()

void tagBegin ( struct lexSpec * spec, const char * tag, int len )

static

Definition at line 885 of file mod_grs_regx.c.

References lexSpec::d1_level, lexSpec::d1_stack, data1_mk_tag_n(), lexSpec::dh, data1_node::len, lexSpec::m, data1_node::tag, tagDataRelease(), and tagStrip().

Referenced by execCode().

tagDataRelease()

void tagDataRelease ( struct lexSpec * spec )

static

Definition at line 782 of file mod_grs_regx.c.

References lexConcatBuf::buf, lexSpec::concatBuf, lexSpec::d1_level, lexSpec::d1_stack, data1_node::data, DATA1_LOCALDATA, DATA1I_text, DATA1N_data, data1_node::lbuf, lexSpec::m, data1_node::u, and data1_node::which.

Referenced by execCode(), lexRoot(), tagBegin(), tagEnd(), and variantBegin().

tagEnd()

void tagEnd ( struct lexSpec * spec, int min_level, const char * tag, int len )

static

Definition at line 906 of file mod_grs_regx.c.

References lexSpec::d1_level, lexSpec::d1_stack, DATA1N_tag, data1_node::len, data1_node::tag, tagDataRelease(), tagStrip(), data1_node::u, and data1_node::which.

Referenced by execCode().

tagStrip()

void tagStrip ( const char ** tag, int * len )

static

Definition at line 872 of file mod_grs_regx.c.

References data1_node::len, and data1_node::tag.

Referenced by tagBegin(), and tagEnd().

tryMatch()

int tryMatch ( struct lexSpec * spec, int * pptr, int * mptr, struct DFA * dfa, int greedy )

static

Definition at line 929 of file mod_grs_regx.c.

References DFA_tran::ch, f_win_advance(), F_WIN_EOF, DFA_state::rule_nno, DFA_state::rule_no, DFA::states, DFA_tran::to, DFA_state::tran_no, and DFA_state::trans.

Referenced by execAction().

variantBegin()

void variantBegin ( struct lexSpec * spec, const char * class_str, int class_len, const char * type_str, int type_len, const char * value_str, int value_len )

static

Definition at line 801 of file mod_grs_regx.c.

References data1_node::absyn, lexSpec::d1_level, lexSpec::d1_stack, data1_getvartypeby_absyn(), DATA1_LOCALDATA, DATA1_MAX_SYMBOL, data1_mk_node2(), DATA1N_variant, lexSpec::dh, data1_node::lbuf, lexSpec::m, data1_node::parent, data1_node::root, tagDataRelease(), data1_node::type, data1_node::u, data1_node::value, data1_node::variant, and data1_node::which.

Referenced by execCode().

Variable Documentation

idzebra_filter

RecType idzebra_filter[]

Initial value:

 = {
    &regx_type,
 
 
 
    0,
}

Definition at line 2027 of file mod_grs_regx.c.

regx_type

struct recType regx_type

static

Initial value:

                    = {
    0,
    "grs.regx",
    grs_init,
    grs_config,
    grs_destroy,
    extract_regx,
    retrieve_regx,
}

Definition at line 1942 of file mod_grs_regx.c.