1b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho/* 2b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho******************************************************************************* 3b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* Copyright (C) 2011, International Business Machines 4b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* Corporation and others. All Rights Reserved. 5b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho******************************************************************************* 6b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* file name: patternprops.h 7b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* encoding: US-ASCII 8b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* tab size: 8 (not used) 9b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* indentation:4 10b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* 11b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* created on: 2011mar13 12b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* created by: Markus W. Scherer 13b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho*/ 14b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 15b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#ifndef __PATTERNPROPS_H__ 16b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#define __PATTERNPROPS_H__ 17b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 18b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#include "unicode/utypes.h" 19b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 20b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoU_NAMESPACE_BEGIN 21b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 22b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho/** 23b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Implements the immutable Unicode properties Pattern_Syntax and Pattern_White_Space. 24b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Hardcodes these properties, does not load data, does not depend on other ICU classes. 25b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p> 26b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Note: Both properties include ASCII as well as non-ASCII, non-Latin-1 code points, 27b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and both properties only include BMP code points (no supplementary ones). 28b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Pattern_Syntax includes some unassigned code points. 29b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p> 30b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * [:Pattern_White_Space:] = 31b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * [\u0009-\u000D\ \u0085\u200E\u200F\u2028\u2029] 32b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p> 33b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * [:Pattern_Syntax:] = 34b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * [!-/\:-@\[-\^`\{-~\u00A1-\u00A7\u00A9\u00AB\u00AC\u00AE 35b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * \u00B0\u00B1\u00B6\u00BB\u00BF\u00D7\u00F7 36b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * \u2010-\u2027\u2030-\u203E\u2041-\u2053\u2055-\u205E 37b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * \u2190-\u245F\u2500-\u2775\u2794-\u2BFF\u2E00-\u2E7F 38b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * \u3001-\u3003\u3008-\u3020\u3030\uFD3E\uFD3F\uFE45\uFE46] 39b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @author mscherer 40b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 41b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoclass U_COMMON_API PatternProps { 42b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehopublic: 43b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 44b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if c is a Pattern_Syntax code point. 45b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 46b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static UBool isSyntax(UChar32 c); 47b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 48b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 49b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if c is a Pattern_Syntax or Pattern_White_Space code point. 50b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 51b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static UBool isSyntaxOrWhiteSpace(UChar32 c); 52b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 53b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 54b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if c is a Pattern_White_Space character. 55b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 56b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static UBool isWhiteSpace(UChar32 c); 57b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 58b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 59b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Skips over Pattern_White_Space starting at s. 60b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return The smallest pointer at or after s with a non-white space character. 61b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 62b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static const UChar *skipWhiteSpace(const UChar *s, int32_t length); 63b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 64b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 65b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return s except with leading and trailing Pattern_White_Space removed and length adjusted. 66b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 67b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static const UChar *trimWhiteSpace(const UChar *s, int32_t &length); 68b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 69b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 70b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Tests whether the string contains a "pattern identifier", that is, 71b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * whether it contains only non-Pattern_White_Space, non-Pattern_Syntax characters. 72b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if there are no Pattern_White_Space or Pattern_Syntax characters in s. 73b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 74b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static UBool isIdentifier(const UChar *s, int32_t length); 75b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 76b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho /** 77b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Skips over a "pattern identifier" starting at index s. 78b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return The smallest pointer at or after s with 79b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * a Pattern_White_Space or Pattern_Syntax character. 80b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho */ 81b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho static const UChar *skipIdentifier(const UChar *s, int32_t length); 82b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 83b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoprivate: 84b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho PatternProps(); // no constructor: all static methods 85b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho}; 86b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 87b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2clairehoU_NAMESPACE_END 88b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho 89b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho#endif // __PATTERNPROPS_H__ 90