character conversions (.chr) More...

#include <ctype.h>
#include <string.h>
#include <stdlib.h>
#include <assert.h>
#include <charmap.h>
#include <yaz/yaz-util.h>

Go to the source code of this file.

Data Structures
struct	chrmaptab_info
struct	chr_t_entry
struct	chrwork
struct	chr_equiv_work

Macros
#define	CHR_MAXSTR 1024
#define	CHR_MAXEQUIV 32

Typedefs
typedef unsigned	ucs4_t
typedef struct chrwork	chrwork

Functions
static chr_t_entry *	set_map_string (chr_t_entry root, NMEM nmem, const char from, int len, char to, const char from_0)
static chr_t_entry *	find_entry_x (chr_t_entry t, const char from, int len, int first)
const char **	chr_map_input_x (chrmaptab maptab, const char *from, int len, int first)
const char **	chr_map_input (chrmaptab maptab, const char **from, int len, int first)
const char **	chr_map_q_input (chrmaptab maptab, const char **from, int len, int first)
const char *	chr_map_output (chrmaptab maptab, const char **from, int len)
static int	zebra_ucs4_strlen (ucs4_t *s)
ucs4_t	zebra_prim_w (ucs4_t **s)
static void	fun_addentry (const char s, void data, int num)
static void	fun_addspace (const char s, void data, int num)
static void	fun_addcut (const char s, void data, int num)
static void	fun_mkstring (const char s, void data, int num)
static void	fun_add_equivalent_string (const char s, void data, int num)
static void	fun_add_map (const char s, void data, int num)
static int	scan_to_utf8 (yaz_iconv_t t, ucs4_t from, size_t inlen, char outbuf, size_t outbytesleft)
static int	scan_string (char s_native, yaz_iconv_t t_unicode, yaz_iconv_t t_utf8, void(fun)(const char c, void data, int num), void data, int num)
chrmaptab	chrmaptab_create (const char tabpath, const char name, const char *tabroot)
void	chrmaptab_destroy (chrmaptab tab)

Variables
const unsigned char	CHR_FIELD_BEGIN = '^'
const char *	CHR_UNKNOWN = "\001"
const char *	CHR_SPACE = "\002"
const char *	CHR_CUT = "\003"
const char *	CHR_BASE = "\005"

Detailed Description

character conversions (.chr)

Support module to handle character-conversions into and out of the Zebra dictionary.

Definition in file charmap.c.

Macro Definition Documentation

◆ CHR_MAXEQUIV

#define CHR_MAXEQUIV 32

Definition at line 44 of file charmap.c.

Referenced by fun_add_equivalent_string().

◆ CHR_MAXSTR

#define CHR_MAXSTR 1024

Definition at line 43 of file charmap.c.

Referenced by fun_mkstring().

Typedef Documentation

◆ chrwork

typedef struct chrwork chrwork

◆ ucs4_t

typedef unsigned ucs4_t

Definition at line 37 of file charmap.c.

Function Documentation

◆ chr_map_input()

const char ** chr_map_input	(	chrmaptab	maptab,
		const char **	from,
		int	len,
		int	first )

Definition at line 194 of file charmap.c.

References find_entry_x(), chrmaptab_info::input, and chr_t_entry::target.

Referenced by fun_mkstring(), zebra_maps_input(), and zebra_maps_search().

◆ chr_map_input_x()

const char ** chr_map_input_x	(	chrmaptab	maptab,
		const char **	from,
		int *	len,
		int	first )

Definition at line 184 of file charmap.c.

References find_entry_x(), chrmaptab_info::input, and chr_t_entry::target.

Referenced by tst_string().

◆ chr_map_output()

const char * chr_map_output	(	chrmaptab	maptab,
		const char **	from,
		int	len )

Definition at line 221 of file charmap.c.

References chrmaptab_info::output.

Referenced by zebra_maps_output().

◆ chr_map_q_input()

const char ** chr_map_q_input	(	chrmaptab	maptab,
		const char **	from,
		int	len,
		int	first )

Definition at line 207 of file charmap.c.

References find_entry_x(), chrmaptab_info::q_input, and chr_t_entry::target.

Referenced by zebra_maps_search().

◆ chrmaptab_create()

chrmaptab chrmaptab_create	(	const char *	tabpath,
		const char *	name,
		const char *	tabroot )

Definition at line 513 of file charmap.c.

References chrmaptab_info::base_uppercase, chr_t_entry::children, CHR_BASE, CHR_SPACE, CHR_UNKNOWN, chrmaptab_destroy(), chr_equiv_work::eq, fun_add_equivalent_string(), fun_add_map(), fun_addcut(), fun_addentry(), fun_addspace(), fun_mkstring(), chrmaptab_info::input, chrwork::map, chr_equiv_work::nmem, chrmaptab_info::nmem, chr_equiv_work::no_eq, chrmaptab_info::output, chrmaptab_info::q_input, scan_string(), set_map_string(), chrwork::string, and chr_t_entry::target.

Referenced by tst2(), tst3(), tst_latin1(), tst_utf8(), and zebra_charmap_get().

◆ chrmaptab_destroy()

void chrmaptab_destroy ( chrmaptab tab )

Definition at line 748 of file charmap.c.

References chrmaptab_info::nmem.

Referenced by chrmaptab_create(), tst2(), tst3(), tst_latin1(), tst_utf8(), and zebra_maps_close().

◆ find_entry_x()

chr_t_entry * find_entry_x	(	chr_t_entry *	t,
		const char **	from,
		int *	len,
		int	first )

static

Definition at line 141 of file charmap.c.

References chr_t_entry::children, CHR_FIELD_BEGIN, find_entry_x(), and chr_t_entry::target.

Referenced by chr_map_input(), chr_map_input_x(), chr_map_q_input(), and find_entry_x().

◆ fun_add_equivalent_string()

void fun_add_equivalent_string	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 369 of file charmap.c.

References CHR_MAXEQUIV, chr_equiv_work::eq, chr_equiv_work::nmem, and chr_equiv_work::no_eq.

Referenced by chrmaptab_create().

◆ fun_add_map()

void fun_add_map	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 381 of file charmap.c.

References chrmaptab_info::input, chrwork::map, chrmaptab_info::nmem, set_map_string(), and chrwork::string.

Referenced by chrmaptab_create().

◆ fun_addcut()

void fun_addcut	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 344 of file charmap.c.

References CHR_CUT, chrmaptab_info::input, chrmaptab_info::nmem, and set_map_string().

Referenced by chrmaptab_create().

◆ fun_addentry()

void fun_addentry	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 318 of file charmap.c.

References chrmaptab_info::base_uppercase, chrmaptab_info::input, chrmaptab_info::nmem, chrmaptab_info::output, and set_map_string().

Referenced by chrmaptab_create().

◆ fun_addspace()

void fun_addspace	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 333 of file charmap.c.

References CHR_SPACE, chrmaptab_info::input, chrmaptab_info::nmem, and set_map_string().

Referenced by chrmaptab_create().

◆ fun_mkstring()

void fun_mkstring	(	const char *	s,
		void *	data,
		int	num )

static

Definition at line 354 of file charmap.c.

References chr_map_input(), CHR_MAXSTR, CHR_UNKNOWN, chrwork::map, and chrwork::string.

Referenced by chrmaptab_create().

◆ scan_string()

int scan_string	(	char *	s_native,
		yaz_iconv_t	t_unicode,
		yaz_iconv_t	t_utf8,
		void(*	fun )(const char c, void data, int num),
		void *	data,
		int *	num )

static

Definition at line 421 of file charmap.c.

References begin(), end(), scan_to_utf8(), zebra_prim_w(), and zebra_ucs4_strlen().

Referenced by chrmaptab_create().

◆ scan_to_utf8()

int scan_to_utf8	(	yaz_iconv_t	t,
		ucs4_t *	from,
		size_t	inlen,
		char *	outbuf,
		size_t	outbytesleft )

static

Definition at line 393 of file charmap.c.

Referenced by scan_string().

◆ set_map_string()

chr_t_entry * set_map_string	(	chr_t_entry *	root,
		NMEM	nmem,
		const char *	from,
		int	len,
		char *	to,
		const char *	from_0 )

static

Definition at line 92 of file charmap.c.

References chr_t_entry::children, CHR_UNKNOWN, set_map_string(), and chr_t_entry::target.

Referenced by chrmaptab_create(), fun_add_map(), fun_addcut(), fun_addentry(), fun_addspace(), and set_map_string().

◆ zebra_prim_w()

ucs4_t zebra_prim_w ( ucs4_t ** s )

Definition at line 239 of file charmap.c.

References zebra_ucs4_strlen().

Referenced by scan_string().

◆ zebra_ucs4_strlen()

int zebra_ucs4_strlen ( ucs4_t * s )

static

Definition at line 231 of file charmap.c.

Referenced by scan_string(), and zebra_prim_w().

Variable Documentation

◆ CHR_BASE

const char* CHR_BASE = "\005"

Definition at line 51 of file charmap.c.

Referenced by chrmaptab_create().

◆ CHR_CUT

const char* CHR_CUT = "\003"

Definition at line 50 of file charmap.c.

Referenced by fun_addcut(), and parse_complete_field().

◆ CHR_FIELD_BEGIN

const unsigned char CHR_FIELD_BEGIN = '^'

Definition at line 46 of file charmap.c.

Referenced by find_entry_x().

◆ CHR_SPACE

const char* CHR_SPACE = "\002"

Definition at line 49 of file charmap.c.

Referenced by chrmaptab_create(), extract_add_incomplete_field(), fun_addspace(), parse_complete_field(), snippet_add_incomplete_field(), term_100(), term_101(), term_103(), term_104(), term_105(), term_pre(), trans_scan_term(), and zebra_it_key_str_dump().

◆ CHR_UNKNOWN

const char* CHR_UNKNOWN = "\001"

Definition at line 48 of file charmap.c.

Referenced by chrmaptab_create(), fun_mkstring(), set_map_string(), and zebra_it_key_str_dump().

Data Structures

Macros

Typedefs

Functions

Variables

Detailed Description

Macro Definition Documentation

◆ CHR_MAXEQUIV

◆ CHR_MAXSTR

Typedef Documentation

◆ chrwork

◆ ucs4_t

Function Documentation

◆ chr_map_input()

◆ chr_map_input_x()

◆ chr_map_output()

◆ chr_map_q_input()

◆ chrmaptab_create()

◆ chrmaptab_destroy()

◆ find_entry_x()

◆ fun_add_equivalent_string()

◆ fun_add_map()

◆ fun_addcut()

◆ fun_addentry()

◆ fun_addspace()

◆ fun_mkstring()

◆ scan_string()

◆ scan_to_utf8()

◆ set_map_string()

◆ zebra_prim_w()

◆ zebra_ucs4_strlen()

Variable Documentation

◆ CHR_BASE

◆ CHR_CUT

◆ CHR_FIELD_BEGIN

◆ CHR_SPACE

◆ CHR_UNKNOWN