uchar.cpp | Explore in Territory

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
********************************************************************************
*   Copyright (C) 1996-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
********************************************************************************
*
* File UCHAR.C
*
* Modification History:
*
*   Date        Name        Description
*   04/02/97    aliu        Creation.
*   4/15/99     Madhu       Updated all the function definitions for C Implementation
*   5/20/99     Madhu       Added the function u_getVersion()
*   8/19/1999   srl         Upgraded scripts to Unicode3.0 
*   11/11/1999  weiv        added u_isalnum(), cleaned comments
*   01/11/2000  helena      Renamed u_getVersion to u_getUnicodeVersion.
*   06/20/2000  helena      OS/400 port changes; mostly typecast.
******************************************************************************
*/

#include "unicode/utypes.h"
#include "unicode/uchar.h"
#include "unicode/uscript.h"
#include "unicode/udata.h"
#include "uassert.h"
#include "cmemory.h"
#include "ucln_cmn.h"
#include "utrie2.h"
#include "udataswp.h"
#include "uprops.h"
#include "ustr_imp.h"

/* uchar_props_data.h is machine-generated by genprops --csource */
#define INCLUDED_FROM_UCHAR_C
#include "uchar_props_data.h"

/* constants and macros for access to the data ------------------------------ */

/* getting a uint32_t properties word from the data */
#define GET_PROPS(c, result) …

/* API functions ------------------------------------------------------------ */

/* Gets the Unicode character's general category.*/
U_CAPI int8_t U_EXPORT2
u_charType(UChar32 c) { … }

/* Enumerate all code points with their general categories. */
struct _EnumTypeCallback { … };

static uint32_t U_CALLCONV
_enumTypeValue(const void *context, uint32_t value) { … }

static UBool U_CALLCONV
_enumTypeRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { … }

U_CAPI void U_EXPORT2
u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) { … }

/* Checks if ch is a lower case letter.*/
U_CAPI UBool U_EXPORT2
u_islower(UChar32 c) { … }

/* Checks if ch is an upper case letter.*/
U_CAPI UBool U_EXPORT2
u_isupper(UChar32 c) { … }

/* Checks if ch is a title case letter; usually upper case letters.*/
U_CAPI UBool U_EXPORT2
u_istitle(UChar32 c) { … }

/* Checks if ch is a decimal digit. */
U_CAPI UBool U_EXPORT2
u_isdigit(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isxdigit(UChar32 c) { … }

/* Checks if the Unicode character is a letter.*/
U_CAPI UBool U_EXPORT2
u_isalpha(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isUAlphabetic(UChar32 c) { … }

/* Checks if c is a letter or a decimal digit */
U_CAPI UBool U_EXPORT2
u_isalnum(UChar32 c) { … }

/**
 * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
 * @internal
 */
U_CFUNC UBool
u_isalnumPOSIX(UChar32 c) { … }

/* Checks if ch is a unicode character with assigned character type.*/
U_CAPI UBool U_EXPORT2
u_isdefined(UChar32 c) { … }

/* Checks if the Unicode character is a base form character that can take a diacritic.*/
U_CAPI UBool U_EXPORT2
u_isbase(UChar32 c) { … }

/* Checks if the Unicode character is a control character.*/
U_CAPI UBool U_EXPORT2
u_iscntrl(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isISOControl(UChar32 c) { … }

/* Some control characters that are used as space. */
#define IS_THAT_CONTROL_SPACE(c) …

/* Java has decided that U+0085 New Line is not whitespace any more. */
#define IS_THAT_ASCII_CONTROL_SPACE(c) …

/* Checks if the Unicode character is a space character.*/
U_CAPI UBool U_EXPORT2
u_isspace(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isJavaSpaceChar(UChar32 c) { … }

/* Checks if the Unicode character is a whitespace character.*/
U_CAPI UBool U_EXPORT2
u_isWhitespace(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isblank(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isUWhiteSpace(UChar32 c) { … }

/* Checks if the Unicode character is printable.*/
U_CAPI UBool U_EXPORT2
u_isprint(UChar32 c) { … }

/**
 * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
 * Implements UCHAR_POSIX_PRINT.
 * @internal
 */
U_CFUNC UBool
u_isprintPOSIX(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_isgraph(UChar32 c) { … }

/**
 * Checks if c is in
 * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
 * with space=\p{Whitespace} and Control=Cc.
 * Implements UCHAR_POSIX_GRAPH.
 * @internal
 */
U_CFUNC UBool
u_isgraphPOSIX(UChar32 c) { … }

U_CAPI UBool U_EXPORT2
u_ispunct(UChar32 c) { … }

/*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/
U_CAPI UBool U_EXPORT2
u_isIDIgnorable(UChar32 c) { … }

/*Checks if the Unicode character can start a Java identifier.*/
U_CAPI UBool U_EXPORT2
u_isJavaIDStart(UChar32 c) { … }

/*Checks if the Unicode character can be a Java identifier part other than starting the
 * identifier.
 */
U_CAPI UBool U_EXPORT2
u_isJavaIDPart(UChar32 c) { … }

U_CAPI int32_t U_EXPORT2
u_charDigitValue(UChar32 c) { … }

U_CAPI double U_EXPORT2
u_getNumericValue(UChar32 c) { … }

U_CAPI int32_t U_EXPORT2
u_digit(UChar32 ch, int8_t radix) { … }

U_CAPI UChar32 U_EXPORT2
u_forDigit(int32_t digit, int8_t radix) { … }

/* miscellaneous, and support for uprops.cpp -------------------------------- */

U_CAPI void U_EXPORT2
u_getUnicodeVersion(UVersionInfo versionArray) { … }

U_CFUNC uint32_t
u_getMainProperties(UChar32 c) { … }

U_CFUNC uint32_t
u_getUnicodeProperties(UChar32 c, int32_t column) { … }

U_CFUNC int32_t
uprv_getMaxValues(int32_t column) { … }

U_CAPI void U_EXPORT2
u_charAge(UChar32 c, UVersionInfo versionArray) { … }

U_CAPI UScriptCode U_EXPORT2
uscript_getScript(UChar32 c, UErrorCode *pErrorCode) { … }

U_CAPI UBool U_EXPORT2
uscript_hasScript(UChar32 c, UScriptCode sc) UPRV_NO_SANITIZE_UNDEFINED { … }

U_CAPI int32_t U_EXPORT2
uscript_getScriptExtensions(UChar32 c,
                            UScriptCode *scripts, int32_t capacity,
                            UErrorCode *pErrorCode) { … }

U_CAPI UBlockCode U_EXPORT2
ublock_getCode(UChar32 c) { … }

/* property starts for UnicodeSet ------------------------------------------- */

static UBool U_CALLCONV
_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { … }

#define USET_ADD_CP_AND_NEXT(sa, cp) …

U_CFUNC void U_EXPORT2
uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { … }

U_CFUNC void U_EXPORT2
upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { … }
godot/thirdparty/icu4c/common/uchar.cpp
