michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2011, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  patternprops.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2011mar13
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #ifndef __PATTERNPROPS_H__
michael@0: #define __PATTERNPROPS_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: /**
michael@0:  * Implements the immutable Unicode properties Pattern_Syntax and Pattern_White_Space.
michael@0:  * Hardcodes these properties, does not load data, does not depend on other ICU classes.
michael@0:  * <p>
michael@0:  * Note: Both properties include ASCII as well as non-ASCII, non-Latin-1 code points,
michael@0:  * and both properties only include BMP code points (no supplementary ones).
michael@0:  * Pattern_Syntax includes some unassigned code points.
michael@0:  * <p>
michael@0:  * [:Pattern_White_Space:] =
michael@0:  *   [\u0009-\u000D\ \u0085\u200E\u200F\u2028\u2029]
michael@0:  * <p>
michael@0:  * [:Pattern_Syntax:] =
michael@0:  *   [!-/\:-@\[-\^`\{-~\u00A1-\u00A7\u00A9\u00AB\u00AC\u00AE
michael@0:  *    \u00B0\u00B1\u00B6\u00BB\u00BF\u00D7\u00F7
michael@0:  *    \u2010-\u2027\u2030-\u203E\u2041-\u2053\u2055-\u205E
michael@0:  *    \u2190-\u245F\u2500-\u2775\u2794-\u2BFF\u2E00-\u2E7F
michael@0:  *    \u3001-\u3003\u3008-\u3020\u3030\uFD3E\uFD3F\uFE45\uFE46]
michael@0:  * @author mscherer
michael@0:  */
michael@0: class U_COMMON_API PatternProps {
michael@0: public:
michael@0:     /**
michael@0:      * @return TRUE if c is a Pattern_Syntax code point.
michael@0:      */
michael@0:     static UBool isSyntax(UChar32 c);
michael@0: 
michael@0:     /**
michael@0:      * @return TRUE if c is a Pattern_Syntax or Pattern_White_Space code point.
michael@0:      */
michael@0:     static UBool isSyntaxOrWhiteSpace(UChar32 c);
michael@0: 
michael@0:     /**
michael@0:      * @return TRUE if c is a Pattern_White_Space character.
michael@0:      */
michael@0:     static UBool isWhiteSpace(UChar32 c);
michael@0: 
michael@0:     /**
michael@0:      * Skips over Pattern_White_Space starting at s.
michael@0:      * @return The smallest pointer at or after s with a non-white space character.
michael@0:      */
michael@0:     static const UChar *skipWhiteSpace(const UChar *s, int32_t length);
michael@0: 
michael@0:     /**
michael@0:      * @return s except with leading and trailing Pattern_White_Space removed and length adjusted.
michael@0:      */
michael@0:     static const UChar *trimWhiteSpace(const UChar *s, int32_t &length);
michael@0: 
michael@0:     /**
michael@0:      * Tests whether the string contains a "pattern identifier", that is,
michael@0:      * whether it contains only non-Pattern_White_Space, non-Pattern_Syntax characters.
michael@0:      * @return TRUE if there are no Pattern_White_Space or Pattern_Syntax characters in s.
michael@0:      */
michael@0:     static UBool isIdentifier(const UChar *s, int32_t length);
michael@0: 
michael@0:     /**
michael@0:      * Skips over a "pattern identifier" starting at index s.
michael@0:      * @return The smallest pointer at or after s with
michael@0:      *         a Pattern_White_Space or Pattern_Syntax character.
michael@0:      */
michael@0:     static const UChar *skipIdentifier(const UChar *s, int32_t length);
michael@0: 
michael@0: private:
michael@0:     PatternProps();  // no constructor: all static methods
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif  // __PATTERNPROPS_H__