1b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 2b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru********************************************************************** 31b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert* Copyright (C) 2004-2015, International Business Machines 4b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Corporation and others. All Rights Reserved. 5b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru********************************************************************** 650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho* file name: uregex.h 7b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* encoding: US-ASCII 8b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* indentation:4 9b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 10b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* created on: 2004mar09 11b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* created by: Andy Heninger 12b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 13b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* ICU Regular Expressions, API for C 14b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru*/ 15b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 16b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 17b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * \file 18b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * \brief C API: Regular Expressions 19b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 20b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p>This is a C wrapper around the C++ RegexPattern and RegexMatcher classes.</p> 21b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 22b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 23b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#ifndef UREGEX_H 24b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define UREGEX_H 25b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#include "unicode/utext.h" 27b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/utypes.h" 28b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 29b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_REGULAR_EXPRESSIONS 30b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#include "unicode/localpointer.h" 32b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/parseerr.h" 33b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 34b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustruct URegularExpression; 35b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 36b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Structure representing a compiled regular expression, plus the results 37b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of a match operation. 38b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 39b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 40b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querutypedef struct URegularExpression URegularExpression; 41b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 42b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 43b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 44b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Constants for Regular Expression Match Modes. 45b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 2.4 46b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 47b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querutypedef enum URegexpFlag{ 48b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 49b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#ifndef U_HIDE_DRAFT_API 50b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** Forces normalization of pattern and strings. 51b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru Not implemented yet, just a placeholder, hence draft. 52b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru @draft ICU 2.4 */ 53b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UREGEX_CANON_EQ = 128, 548393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius#endif /* U_HIDE_DRAFT_API */ 55b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** Enable case insensitive matching. @stable ICU 2.4 */ 56b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UREGEX_CASE_INSENSITIVE = 2, 57b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 58b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** Allow white space and comments within patterns @stable ICU 2.4 */ 59b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UREGEX_COMMENTS = 4, 60b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 61b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** If set, '.' matches line terminators, otherwise '.' matching stops at line end. 62b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 2.4 */ 63b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UREGEX_DOTALL = 32, 64c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 65c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru /** If set, treat the entire pattern as a literal string. 66c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Metacharacters or escape sequences in the input sequence will be given 67103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * no special meaning. 68c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 69103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * The flag UREGEX_CASE_INSENSITIVE retains its impact 70c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * on matching when used in conjunction with this flag. 71c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * The other flags become superfluous. 72103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * 73b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 74c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 75c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UREGEX_LITERAL = 16, 76b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 77b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** Control behavior of "$" and "^" 78b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If set, recognize line terminators within string, 79b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * otherwise, match only at start and end of input string. 80b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 2.4 */ 81b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UREGEX_MULTILINE = 8, 82c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 83c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru /** Unix-only line endings. 84b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * When this mode is enabled, only \\u000a is recognized as a line ending 85c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * in the behavior of ., ^, and $. 86b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 87c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 88c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UREGEX_UNIX_LINES = 1, 89b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 90b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** Unicode word boundaries. 91b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If set, \b uses the Unicode TR 29 definition of word boundaries. 92b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Warning: Unicode word boundaries are quite different from 93b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * traditional regular expression word boundaries. See 94b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * http://unicode.org/reports/tr29/#Word_Boundaries 95b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 2.8 96b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 97c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UREGEX_UWORD = 256, 98c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 99c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru /** Error on Unrecognized backslash escapes. 100c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If set, fail with an error on patterns that contain 101b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * backslash-escaped ASCII letters without a known special 102c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * meaning. If this flag is not set, these 103c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * escaped letters represent themselves. 104b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 105c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 106c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UREGEX_ERROR_ON_UNKNOWN_ESCAPES = 512 107c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 108b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} URegexpFlag; 109b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 110b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 111b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Open (compile) an ICU regular expression. Compiles the regular expression in 112b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * string form into an internal representation using the specified match mode flags. 113b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The resulting regular expression handle can then be used to perform various 114b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * matching operations. 11550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 116b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 117b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pattern The Regular Expression pattern to be compiled. 118b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param patternLength The length of the pattern, or -1 if the pattern is 119b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * NUL terminated. 120b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param flags Flags that alter the default matching behavior for 121b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the regular expression, UREGEX_CASE_INSENSITIVE, for 122b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * example. For default behavior, set this parameter to zero. 123b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * See <code>enum URegexpFlag</code>. All desired flags 124b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * are bitwise-ORed together. 125b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param pe Receives the position (line and column numbers) of any syntax 126b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * error within the source regular expression string. If this 127b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * information is not wanted, pass NULL for this parameter. 128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives error detected by this function. 129b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 131b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 132b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE URegularExpression * U_EXPORT2 133b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_open( const UChar *pattern, 134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t patternLength, 135b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t flags, 136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UParseError *pe, 137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 138103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius 13950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 14050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Open (compile) an ICU regular expression. Compiles the regular expression in 14150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * string form into an internal representation using the specified match mode flags. 14250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The resulting regular expression handle can then be used to perform various 14350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * matching operations. 14450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <p> 14550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The contents of the pattern UText will be extracted and saved. Ownership of the 14650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * UText struct itself remains with the caller. This is to match the behavior of 14750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * uregex_open(). 14850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 14950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param pattern The Regular Expression pattern to be compiled. 15050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param flags Flags that alter the default matching behavior for 15150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * the regular expression, UREGEX_CASE_INSENSITIVE, for 15250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * example. For default behavior, set this parameter to zero. 15350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * See <code>enum URegexpFlag</code>. All desired flags 15450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * are bitwise-ORed together. 155b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param pe Receives the position (line and column numbers) of any syntax 15650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * error within the source regular expression string. If this 15750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * information is not wanted, pass NULL for this parameter. 15850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status Receives error detected by this function. 15950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 160103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 16150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 16254dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE URegularExpression * U_EXPORT2 16350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_openUText(UText *pattern, 16450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho uint32_t flags, 16550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UParseError *pe, 16650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 167103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius 168b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 169b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Open (compile) an ICU regular expression. The resulting regular expression 170b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * handle can then be used to perform various matching operations. 171b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p> 172b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This function is the same as uregex_open, except that the pattern 173b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * is supplied as an 8 bit char * string in the default code page. 174b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 175b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pattern The Regular Expression pattern to be compiled, 176b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * NUL terminated. 177b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param flags Flags that alter the default matching behavior for 178b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the regular expression, UREGEX_CASE_INSENSITIVE, for 179b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * example. For default behavior, set this parameter to zero. 180b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * See <code>enum URegexpFlag</code>. All desired flags 181b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * are bitwise-ORed together. 182b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param pe Receives the position (line and column numbers) of any syntax 183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * error within the source regular expression string. If this 184b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * information is not wanted, pass NULL for this parameter. 185b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The URegularExpression object representing the compiled 187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern. 188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 190b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 191b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_CONVERSION 192b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE URegularExpression * U_EXPORT2 193b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_openC( const char *pattern, 194b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t flags, 195b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UParseError *pe, 196b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 197b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif 198b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 199b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Close the regular expression, recovering all resources (memory) it 203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * was holding. 204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The regular expression to be closed. 206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 207b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE void U_EXPORT2 209b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_close(URegularExpression *regexp); 210b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 21150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#if U_SHOW_CPLUSPLUS_API 21250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 21350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_NAMESPACE_BEGIN 21450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 21550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 21650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * \class LocalURegularExpressionPointer 21750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * "Smart pointer" class, closes a URegularExpression via uregex_close(). 21850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * For most methods see the LocalPointerBase base class. 21950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 22050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @see LocalPointerBase 22150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @see LocalPointer 22227f654740f2a26ad62a5c155af9199af9e69b889claireho * @stable ICU 4.4 22350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 22450294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_DEFINE_LOCAL_OPEN_POINTER(LocalURegularExpressionPointer, URegularExpression, uregex_close); 22550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 22650294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_NAMESPACE_END 22750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 22850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#endif 22950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 230b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 231b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Make a copy of a compiled regular expression. Cloning a regular 232b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * expression is faster than opening a second instance from the source 233b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * form of the expression, and requires less memory. 234b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p> 235b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Note that the current input string and the position of any matched text 236b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * within it are not cloned; only the pattern itself and the 237b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * match mode flags are copied. 238b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p> 239b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Cloning can be particularly useful to threaded applications that perform 240b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * multiple match operations in parallel. Each concurrent RE 241b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * operation requires its own instance of a URegularExpression. 242b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 243b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression to be cloned. 244b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives indication of any errors encountered 245b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return the cloned copy of the compiled regular expression. 246b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 247b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 248b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE URegularExpression * U_EXPORT2 249b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_clone(const URegularExpression *regexp, UErrorCode *status); 250b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 251b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 25250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Returns a pointer to the source form of the pattern for this regular expression. 25350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This function will work even if the pattern was originally specified as a UText. 254b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 255b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 256b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param patLength This output parameter will be set to the length of the 257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern string. A NULL pointer may be used here if the 258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern length is not needed, as would be the case if 259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the pattern is known in advance to be a NUL terminated 260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * string. 261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return a pointer to the pattern string. The storage for the string is 263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * owned by the regular expression object, and must not be 264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * altered or deleted by the application. The returned string 265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * will remain valid until the regular expression is closed. 266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE const UChar * U_EXPORT2 26950294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_pattern(const URegularExpression *regexp, 27050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t *patLength, 27150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 27250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 27350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 27450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Returns the source text of the pattern for this regular expression. 27550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This function will work even if the pattern was originally specified as a UChar string. 27650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 27750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 27850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status Receives errors detected by this function. 27950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return the pattern text. The storage for the text is owned by the regular expression 28050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * object, and must not be altered or deleted. 28150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 282103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 28350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 28454dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 28550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_patternUText(const URegularExpression *regexp, 28650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 287b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 288b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 289b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Get the match mode flags that were specified when compiling this regular expression. 290b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 291b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 292b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The match mode flags 293b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @see URegexpFlag 294b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 295b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 296b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 297b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_flags(const URegularExpression *regexp, 298b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 299b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 300b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 301b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 302b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Set the subject text string upon which the regular expression will look for matches. 303b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This function may be called any number of times, allowing the regular 304b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * expression pattern to be applied to different strings. 305b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p> 306b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Regular expression matching operations work directly on the application's 307b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * string data. No copy is made. The subject string data must not be 308b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * altered after calling this function until after all regular expression 309b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * operations involving this string data are completed. 310b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p> 311b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Zero length strings are permitted. In this case, no subsequent match 312b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * operation will dereference the text string pointer. 313b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 314b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 315b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param text The subject text string. 316b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param textLength The length of the subject text, or -1 if the string 317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * is NUL terminated. 318b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 321b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE void U_EXPORT2 322b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_setText(URegularExpression *regexp, 323b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar *text, 324b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t textLength, 325b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 326b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 32750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 32850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 32950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Set the subject text string upon which the regular expression will look for matches. 33050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This function may be called any number of times, allowing the regular 33150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * expression pattern to be applied to different strings. 33250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <p> 33350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Regular expression matching operations work directly on the application's 33450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * string data; only a shallow clone is made. The subject string data must not be 33550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * altered after calling this function until after all regular expression 33650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * operations involving this string data are completed. 33750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 33850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 33950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param text The subject text string. 34050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status Receives errors detected by this function. 34150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 342103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 34350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 34454dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 34550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_setUText(URegularExpression *regexp, 34650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *text, 34750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 34850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 349b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 350b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Get the subject text that is currently associated with this 35150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * regular expression object. If the input was supplied using uregex_setText(), 35250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * that pointer will be returned. Otherwise, the characters in the input will 35350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * be extracted to a buffer and returned. In either case, ownership remains 35450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * with the regular expression object. 35550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 35650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This function will work even if the input was originally specified as a UText. 357b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 358b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 359b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param textLength The length of the string is returned in this output parameter. 360b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * A NULL pointer may be used here if the 361b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * text length is not needed, as would be the case if 362b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the text is known in advance to be a NUL terminated 363b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * string. 364b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 36550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return Pointer to the subject text string currently associated with 366b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * this regular expression. 367b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 368b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 369b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE const UChar * U_EXPORT2 370b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_getText(URegularExpression *regexp, 371b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t *textLength, 372b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 373103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius 37450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 37550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Get the subject text that is currently associated with this 37650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * regular expression object. 37750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 37850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This function will work even if the input was originally specified as a UChar string. 37950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 38050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 38150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param dest A mutable UText in which to store the current input. 38250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If NULL, a new UText will be created as an immutable shallow clone 38350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * of the actual input string. 38450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status Receives errors detected by this function. 38550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return The subject text currently associated with this regular expression. 38650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If a pre-allocated UText was provided, it will always be used and returned. 38750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 388103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 38950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 39054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 39150294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_getUText(URegularExpression *regexp, 39250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *dest, 39350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 394b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3952e615e9896b12236afe0ff2695e8afc2ee73f961claireho/** 3962e615e9896b12236afe0ff2695e8afc2ee73f961claireho * Set the subject text string upon which the regular expression is looking for matches 3972e615e9896b12236afe0ff2695e8afc2ee73f961claireho * without changing any other aspect of the matching state. 3982e615e9896b12236afe0ff2695e8afc2ee73f961claireho * The new and previous text strings must have the same content. 3992e615e9896b12236afe0ff2695e8afc2ee73f961claireho * 400b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This function is intended for use in environments where ICU is operating on 4012e615e9896b12236afe0ff2695e8afc2ee73f961claireho * strings that may move around in memory. It provides a mechanism for notifying 4022e615e9896b12236afe0ff2695e8afc2ee73f961claireho * ICU that the string has been relocated, and providing a new UText to access the 4032e615e9896b12236afe0ff2695e8afc2ee73f961claireho * string in its new position. 4042e615e9896b12236afe0ff2695e8afc2ee73f961claireho * 405b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Note that the regular expression implementation never copies the underlying text 406b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * of a string being matched, but always operates directly on the original text 407b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * provided by the user. Refreshing simply drops the references to the old text 408b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and replaces them with references to the new. 409b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 410b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Caution: this function is normally used only by very specialized 411b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * system-level code. One example use case is with garbage collection 412b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * that moves the text in memory. 4132e615e9896b12236afe0ff2695e8afc2ee73f961claireho * 4142e615e9896b12236afe0ff2695e8afc2ee73f961claireho * @param regexp The compiled regular expression. 4152e615e9896b12236afe0ff2695e8afc2ee73f961claireho * @param text The new (moved) text string. 4162e615e9896b12236afe0ff2695e8afc2ee73f961claireho * @param status Receives errors detected by this function. 4172e615e9896b12236afe0ff2695e8afc2ee73f961claireho * 41854dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius * @stable ICU 4.8 4192e615e9896b12236afe0ff2695e8afc2ee73f961claireho */ 42054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 4212e615e9896b12236afe0ff2695e8afc2ee73f961clairehouregex_refreshUText(URegularExpression *regexp, 4222e615e9896b12236afe0ff2695e8afc2ee73f961claireho UText *text, 4232e615e9896b12236afe0ff2695e8afc2ee73f961claireho UErrorCode *status); 4242e615e9896b12236afe0ff2695e8afc2ee73f961claireho 425b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 426c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Attempts to match the input string against the pattern. 427c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * To succeed, the match must extend to the end of the string, 428c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * or cover the complete match region. 429c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 430c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If startIndex >= zero the match operation starts at the specified 431c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * index and must extend to the end of the input string. Any region 432c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * that has been specified is reset. 433c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 434c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If startIndex == -1 the match must cover the input region, or the entire 435c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * input string if no region has been set. This directly corresponds to 436c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Matcher.matches() in Java 437b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 438b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 43927f654740f2a26ad62a5c155af9199af9e69b889claireho * @param startIndex The input string (native) index at which to begin matching, or -1 440c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * to match the input Region. 441b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status Receives errors detected by this function. 442b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return TRUE if there is a match 443b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 444b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 445b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE UBool U_EXPORT2 446b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_matches(URegularExpression *regexp, 447b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t startIndex, 448b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 449b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 450b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 45127f654740f2a26ad62a5c155af9199af9e69b889claireho * 64bit version of uregex_matches. 452b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Attempts to match the input string against the pattern. 453b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * To succeed, the match must extend to the end of the string, 454b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * or cover the complete match region. 455b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 456b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If startIndex >= zero the match operation starts at the specified 457b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * index and must extend to the end of the input string. Any region 458b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * that has been specified is reset. 459b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 460b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If startIndex == -1 the match must cover the input region, or the entire 461b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * input string if no region has been set. This directly corresponds to 462b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Matcher.matches() in Java 463b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 464b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 465b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param startIndex The input string (native) index at which to begin matching, or -1 466b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * to match the input Region. 467b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status Receives errors detected by this function. 468b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if there is a match 469103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 47027f654740f2a26ad62a5c155af9199af9e69b889claireho */ 47154dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UBool U_EXPORT2 47227f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_matches64(URegularExpression *regexp, 47327f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t startIndex, 47427f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 47527f654740f2a26ad62a5c155af9199af9e69b889claireho 47627f654740f2a26ad62a5c155af9199af9e69b889claireho/** 477b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Attempts to match the input string, starting from the specified index, against the pattern. 478b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The match may be of any length, and is not required to extend to the end 479b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of the input string. Contrast with uregex_matches(). 480b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 481c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p>If startIndex is >= 0 any input region that was set for this 482c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * URegularExpression is reset before the operation begins. 483c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 484c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p>If the specified starting index == -1 the match begins at the start of the input 485c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * region, or at the start of the full string if no region has been specified. 486c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * This corresponds directly with Matcher.lookingAt() in Java. 487c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 488b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p>If the match succeeds then more information can be obtained via the 489b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <code>uregexp_start()</code>, <code>uregexp_end()</code>, 490b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and <code>uregexp_group()</code> functions.</p> 491b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 492b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 49327f654740f2a26ad62a5c155af9199af9e69b889claireho * @param startIndex The input string (native) index at which to begin matching, or 494c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * -1 to match the Input Region 495b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 496b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return TRUE if there is a match. 497b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 498b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 499b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE UBool U_EXPORT2 500b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_lookingAt(URegularExpression *regexp, 501b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t startIndex, 502b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 503b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 504b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 50527f654740f2a26ad62a5c155af9199af9e69b889claireho * 64bit version of uregex_lookingAt. 506b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Attempts to match the input string, starting from the specified index, against the pattern. 507b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * The match may be of any length, and is not required to extend to the end 508b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * of the input string. Contrast with uregex_matches(). 509b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 510b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p>If startIndex is >= 0 any input region that was set for this 511b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * URegularExpression is reset before the operation begins. 512b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 513b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p>If the specified starting index == -1 the match begins at the start of the input 514b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * region, or at the start of the full string if no region has been specified. 515b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This corresponds directly with Matcher.lookingAt() in Java. 516b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 517b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <p>If the match succeeds then more information can be obtained via the 518b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <code>uregexp_start()</code>, <code>uregexp_end()</code>, 519b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and <code>uregexp_group()</code> functions.</p> 520b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 521b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 522b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param startIndex The input string (native) index at which to begin matching, or 523b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * -1 to match the Input Region 524b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A reference to a UErrorCode to receive any errors. 525b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if there is a match. 526103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 52727f654740f2a26ad62a5c155af9199af9e69b889claireho */ 52854dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UBool U_EXPORT2 52927f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_lookingAt64(URegularExpression *regexp, 53027f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t startIndex, 53127f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 53227f654740f2a26ad62a5c155af9199af9e69b889claireho 53327f654740f2a26ad62a5c155af9199af9e69b889claireho/** 534b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Find the first matching substring of the input string that matches the pattern. 535c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If startIndex is >= zero the search for a match begins at the specified index, 536c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and any match region is reset. This corresponds directly with 537c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Matcher.find(startIndex) in Java. 538c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 539c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If startIndex == -1 the search begins at the start of the input region, 540c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * or at the start of the full string if no region has been specified. 541c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 542b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If a match is found, <code>uregex_start(), uregex_end()</code>, and 543b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <code>uregex_group()</code> will provide more information regarding the match. 544b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 545b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 54627f654740f2a26ad62a5c155af9199af9e69b889claireho * @param startIndex The position (native) in the input string to begin the search, or 547c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * -1 to search within the Input Region. 548b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 549b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return TRUE if a match is found. 550b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 551b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 552b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE UBool U_EXPORT2 553b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_find(URegularExpression *regexp, 554b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t startIndex, 555b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 556b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 557b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 55827f654740f2a26ad62a5c155af9199af9e69b889claireho * 64bit version of uregex_find. 559b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Find the first matching substring of the input string that matches the pattern. 560b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If startIndex is >= zero the search for a match begins at the specified index, 561b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and any match region is reset. This corresponds directly with 562b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Matcher.find(startIndex) in Java. 563b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 564b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If startIndex == -1 the search begins at the start of the input region, 565b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * or at the start of the full string if no region has been specified. 566b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 567b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If a match is found, <code>uregex_start(), uregex_end()</code>, and 568b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * <code>uregex_group()</code> will provide more information regarding the match. 569b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 570b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 571b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param startIndex The position (native) in the input string to begin the search, or 572b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * -1 to search within the Input Region. 573b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A reference to a UErrorCode to receive any errors. 574b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return TRUE if a match is found. 575103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 57627f654740f2a26ad62a5c155af9199af9e69b889claireho */ 57754dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UBool U_EXPORT2 57827f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_find64(URegularExpression *regexp, 57927f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t startIndex, 58027f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 58127f654740f2a26ad62a5c155af9199af9e69b889claireho 58227f654740f2a26ad62a5c155af9199af9e69b889claireho/** 583c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Find the next pattern match in the input string. Begin searching 584c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * the input at the location following the end of he previous match, 585c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * or at the start of the string (or region) if there is no 586c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * previous match. If a match is found, <code>uregex_start(), uregex_end()</code>, and 587b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <code>uregex_group()</code> will provide more information regarding the match. 588b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 589b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 590b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 591b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return TRUE if a match is found. 592b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @see uregex_reset 593b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 594b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 595b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE UBool U_EXPORT2 596b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_findNext(URegularExpression *regexp, 597b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 598b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 599b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 600b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Get the number of capturing groups in this regular expression's pattern. 601b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 602b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 603b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return the number of capture groups 604b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 605b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 606b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 607b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_groupCount(URegularExpression *regexp, 608b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 609b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 6101b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert#ifndef U_HIDE_DRAFT_API 6111b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert/** 6121b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * Get the group number corresponding to a named capture group. 6131b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * The returned number can be used with any function that access 6141b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * capture groups by number. 6151b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6161b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * The function returns an error status if the specified name does not 6171b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * appear in the pattern. 6181b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6191b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param regexp The compiled regular expression. 6201b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param groupName The capture group name. 6211b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param nameLength The length of the name, or -1 if the name is a 6221b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * nul-terminated string. 6231b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param status A pointer to a UErrorCode to receive any errors. 6241b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6251b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @draft ICU 55 6261b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert */ 6271b7d32f919554dda9c193b32188251337bc756f1Fredrik RoubertU_DRAFT int32_t U_EXPORT2 6281b7d32f919554dda9c193b32188251337bc756f1Fredrik Rouberturegex_groupNumberFromName(URegularExpression *regexp, 6291b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert const UChar *groupName, 6301b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int32_t nameLength, 6311b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert UErrorCode *status); 6321b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert 6331b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert 6341b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert/** 6351b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * Get the group number corresponding to a named capture group. 6361b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * The returned number can be used with any function that access 6371b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * capture groups by number. 6381b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6391b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * The function returns an error status if the specified name does not 6401b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * appear in the pattern. 6411b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6421b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param regexp The compiled regular expression. 6431b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param groupName The capture group name, 6441b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * platform invariant characters only. 6451b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param nameLength The length of the name, or -1 if the name is 6461b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * nul-terminated. 6471b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param status A pointer to a UErrorCode to receive any errors. 6481b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * 6491b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @draft ICU 55 6501b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert */ 6511b7d32f919554dda9c193b32188251337bc756f1Fredrik RoubertU_DRAFT int32_t U_EXPORT2 6521b7d32f919554dda9c193b32188251337bc756f1Fredrik Rouberturegex_groupNumberFromCName(URegularExpression *regexp, 6531b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert const char *groupName, 6541b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int32_t nameLength, 6551b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert UErrorCode *status); 6561b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert#endif /* U_HIDE_DRAFT_API */ 6571b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert 658b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** Extract the string for the specified matching expression or subexpression. 659b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #0 is the complete string of matched text. 660b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #1 is the text matched by the first set of capturing parentheses. 661b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 662b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 663b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param groupNum The capture group to extract. Group 0 is the complete 664b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * match. The value of this parameter must be 665b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * less than or equal to the number of capture groups in 666b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the pattern. 667b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param dest Buffer to receive the matching string data 668b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destCapacity Capacity of the dest buffer. 669b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 670b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return Length of matching data, 671b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * or -1 if no applicable match. 672b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 673b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 674b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 675b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_group(URegularExpression *regexp, 676b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t groupNum, 677b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar *dest, 678b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t destCapacity, 679b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 680b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 6811b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert/** Returns a shallow immutable clone of the entire input string with the current index set 6821b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * to the beginning of the requested capture group. The capture group length is also 68327f654740f2a26ad62a5c155af9199af9e69b889claireho * returned via groupLength. 68427f654740f2a26ad62a5c155af9199af9e69b889claireho * Group #0 is the complete string of matched text. 68527f654740f2a26ad62a5c155af9199af9e69b889claireho * Group #1 is the text matched by the first set of capturing parentheses. 68627f654740f2a26ad62a5c155af9199af9e69b889claireho * 68727f654740f2a26ad62a5c155af9199af9e69b889claireho * @param regexp The compiled regular expression. 68827f654740f2a26ad62a5c155af9199af9e69b889claireho * @param groupNum The capture group to extract. Group 0 is the complete 68927f654740f2a26ad62a5c155af9199af9e69b889claireho * match. The value of this parameter must be 69027f654740f2a26ad62a5c155af9199af9e69b889claireho * less than or equal to the number of capture groups in 69127f654740f2a26ad62a5c155af9199af9e69b889claireho * the pattern. 69227f654740f2a26ad62a5c155af9199af9e69b889claireho * @param dest A mutable UText in which to store the current input. 69327f654740f2a26ad62a5c155af9199af9e69b889claireho * If NULL, a new UText will be created as an immutable shallow clone 69427f654740f2a26ad62a5c155af9199af9e69b889claireho * of the entire input string. 6951b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert * @param groupLength The group length of the desired capture group. Output parameter. 69627f654740f2a26ad62a5c155af9199af9e69b889claireho * @param status A reference to a UErrorCode to receive any errors. 69727f654740f2a26ad62a5c155af9199af9e69b889claireho * @return The subject text currently associated with this regular expression. 69827f654740f2a26ad62a5c155af9199af9e69b889claireho * If a pre-allocated UText was provided, it will always be used and returned. 69927f654740f2a26ad62a5c155af9199af9e69b889claireho 70027f654740f2a26ad62a5c155af9199af9e69b889claireho * 701103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 70227f654740f2a26ad62a5c155af9199af9e69b889claireho */ 70354dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 70427f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_groupUText(URegularExpression *regexp, 70527f654740f2a26ad62a5c155af9199af9e69b889claireho int32_t groupNum, 70627f654740f2a26ad62a5c155af9199af9e69b889claireho UText *dest, 70727f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t *groupLength, 70827f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 70927f654740f2a26ad62a5c155af9199af9e69b889claireho 710b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 711b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Returns the index in the input string of the start of the text matched by the 712b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * specified capture group during the previous match operation. Return -1 if 713b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the capture group was not part of the last match. 714b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #0 refers to the complete range of matched text. 715b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #1 refers to the text matched by the first set of capturing parentheses. 716b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 717b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 718b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param groupNum The capture group number 719b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 72027f654740f2a26ad62a5c155af9199af9e69b889claireho * @return the starting (native) position in the input of the text matched 721b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * by the specified group. 722b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 723b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 724b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 725b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_start(URegularExpression *regexp, 726b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t groupNum, 727b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 728b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 729b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 73027f654740f2a26ad62a5c155af9199af9e69b889claireho * 64bit version of uregex_start. 731b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Returns the index in the input string of the start of the text matched by the 732b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * specified capture group during the previous match operation. Return -1 if 733b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * the capture group was not part of the last match. 734b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Group #0 refers to the complete range of matched text. 735b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Group #1 refers to the text matched by the first set of capturing parentheses. 736b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 737b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 738b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param groupNum The capture group number 739b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A reference to a UErrorCode to receive any errors. 740b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return the starting (native) position in the input of the text matched 741b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * by the specified group. 742103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 74327f654740f2a26ad62a5c155af9199af9e69b889claireho */ 74454dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE int64_t U_EXPORT2 74527f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_start64(URegularExpression *regexp, 74627f654740f2a26ad62a5c155af9199af9e69b889claireho int32_t groupNum, 74727f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 74827f654740f2a26ad62a5c155af9199af9e69b889claireho 74927f654740f2a26ad62a5c155af9199af9e69b889claireho/** 750b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Returns the index in the input string of the position following the end 751b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of the text matched by the specified capture group. 752b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Return -1 if the capture group was not part of the last match. 753b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #0 refers to the complete range of matched text. 754b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Group #1 refers to the text matched by the first set of capturing parentheses. 755b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 756b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 757b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param groupNum The capture group number 758b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 75927f654740f2a26ad62a5c155af9199af9e69b889claireho * @return the (native) index of the position following the last matched character. 760b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 761b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 762b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 763b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_end(URegularExpression *regexp, 764b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t groupNum, 765b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 766b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 767b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 76827f654740f2a26ad62a5c155af9199af9e69b889claireho * 64bit version of uregex_end. 769b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Returns the index in the input string of the position following the end 770b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * of the text matched by the specified capture group. 771b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Return -1 if the capture group was not part of the last match. 772b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Group #0 refers to the complete range of matched text. 773b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Group #1 refers to the text matched by the first set of capturing parentheses. 774b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 775b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 776b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param groupNum The capture group number 777b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A reference to a UErrorCode to receive any errors. 778b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return the (native) index of the position following the last matched character. 779103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 78027f654740f2a26ad62a5c155af9199af9e69b889claireho */ 78154dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE int64_t U_EXPORT2 78227f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_end64(URegularExpression *regexp, 78327f654740f2a26ad62a5c155af9199af9e69b889claireho int32_t groupNum, 78427f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 78527f654740f2a26ad62a5c155af9199af9e69b889claireho 78627f654740f2a26ad62a5c155af9199af9e69b889claireho/** 787b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Reset any saved state from the previous match. Has the effect of 788b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * causing uregex_findNext to begin at the specified index, and causing 789b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * uregex_start(), uregex_end() and uregex_group() to return an error 790c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * indicating that there is no match information available. Clears any 791c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * match region that may have been set. 792b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 793b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 79427f654740f2a26ad62a5c155af9199af9e69b889claireho * @param index The position (native) in the text at which a 795b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * uregex_findNext() should begin searching. 796b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 797b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 798b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 799b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE void U_EXPORT2 800b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_reset(URegularExpression *regexp, 801b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t index, 802b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 803103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius 80427f654740f2a26ad62a5c155af9199af9e69b889claireho/** 805b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 64bit version of uregex_reset. 806b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Reset any saved state from the previous match. Has the effect of 807b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * causing uregex_findNext to begin at the specified index, and causing 808b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * uregex_start(), uregex_end() and uregex_group() to return an error 809b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * indicating that there is no match information available. Clears any 810b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * match region that may have been set. 811b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 812b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 813b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param index The position (native) in the text at which a 814b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * uregex_findNext() should begin searching. 815b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A reference to a UErrorCode to receive any errors. 816103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 81727f654740f2a26ad62a5c155af9199af9e69b889claireho */ 81854dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 81927f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_reset64(URegularExpression *regexp, 82027f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t index, 82127f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 82227f654740f2a26ad62a5c155af9199af9e69b889claireho 823b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho/** 824b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Sets the limits of the matching region for this URegularExpression. 825c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * The region is the part of the input string that will be considered when matching. 826c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Invoking this method resets any saved state from the previous match, 827c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * then sets the region to start at the index specified by the start parameter 828c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and end at the index specified by the end parameter. 829c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 830c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Depending on the transparency and anchoring being used (see useTransparentBounds 831c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and useAnchoringBounds), certain constructs such as anchors may behave differently 832c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * at or around the boundaries of the region 833c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 834c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * The function will fail if start is greater than limit, or if either index 835c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * is less than zero or greater than the length of the string being matched. 836c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 837c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 83827f654740f2a26ad62a5c155af9199af9e69b889claireho * @param regionStart The (native) index to begin searches at. 83927f654740f2a26ad62a5c155af9199af9e69b889claireho * @param regionLimit The (native) index to end searches at (exclusive). 840c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 841b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 842c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 84350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 844c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_setRegion(URegularExpression *regexp, 845c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t regionStart, 846c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t regionLimit, 847c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 848c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 849c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 850b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 64bit version of uregex_setRegion. 851b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Sets the limits of the matching region for this URegularExpression. 852b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * The region is the part of the input string that will be considered when matching. 853b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Invoking this method resets any saved state from the previous match, 854b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * then sets the region to start at the index specified by the start parameter 855b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and end at the index specified by the end parameter. 856b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 857b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Depending on the transparency and anchoring being used (see useTransparentBounds 858b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and useAnchoringBounds), certain constructs such as anchors may behave differently 859b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * at or around the boundaries of the region 860b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 861b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * The function will fail if start is greater than limit, or if either index 862b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * is less than zero or greater than the length of the string being matched. 863b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 864b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 865b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regionStart The (native) index to begin searches at. 866b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regionLimit The (native) index to end searches at (exclusive). 867b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A pointer to a UErrorCode to receive any errors. 868103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 86927f654740f2a26ad62a5c155af9199af9e69b889claireho */ 87054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 87127f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_setRegion64(URegularExpression *regexp, 87227f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t regionStart, 87327f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t regionLimit, 87427f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 87527f654740f2a26ad62a5c155af9199af9e69b889claireho 87627f654740f2a26ad62a5c155af9199af9e69b889claireho/** 877b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Set the matching region and the starting index for subsequent matches 878b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * in a single operation. 879b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This is useful because the usual function for setting the starting 880b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * index, urgex_reset(), also resets any region limits. 881b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 882b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 883b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regionStart The (native) index to begin searches at. 884b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regionLimit The (native) index to end searches at (exclusive). 885b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param startIndex The index in the input text at which the next 886b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * match operation should begin. 887b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A pointer to a UErrorCode to receive any errors. 888103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 88927f654740f2a26ad62a5c155af9199af9e69b889claireho */ 89054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 89127f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_setRegionAndStart(URegularExpression *regexp, 89227f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t regionStart, 89327f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t regionLimit, 89427f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t startIndex, 89527f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 89627f654740f2a26ad62a5c155af9199af9e69b889claireho 89727f654740f2a26ad62a5c155af9199af9e69b889claireho/** 898c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Reports the start index of the matching region. Any matches found are limited to 899c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * to the region bounded by regionStart (inclusive) and regionEnd (exclusive). 900c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 901c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 902c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 90327f654740f2a26ad62a5c155af9199af9e69b889claireho * @return The starting (native) index of this matcher's region. 904b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 905c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 90650294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE int32_t U_EXPORT2 907c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_regionStart(const URegularExpression *regexp, 908c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 909c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 91027f654740f2a26ad62a5c155af9199af9e69b889claireho/** 911b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 64bit version of uregex_regionStart. 912b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Reports the start index of the matching region. Any matches found are limited to 913b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * to the region bounded by regionStart (inclusive) and regionEnd (exclusive). 914b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 915b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 916b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A pointer to a UErrorCode to receive any errors. 917b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return The starting (native) index of this matcher's region. 918103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 91927f654740f2a26ad62a5c155af9199af9e69b889claireho */ 92054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE int64_t U_EXPORT2 92127f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_regionStart64(const URegularExpression *regexp, 92227f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 923c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 924c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 925c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Reports the end index (exclusive) of the matching region for this URegularExpression. 926c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Any matches found are limited to to the region bounded by regionStart (inclusive) 927c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and regionEnd (exclusive). 928c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 929c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 930c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 93127f654740f2a26ad62a5c155af9199af9e69b889claireho * @return The ending point (native) of this matcher's region. 932b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 933c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 93450294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE int32_t U_EXPORT2 935c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_regionEnd(const URegularExpression *regexp, 936c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 937c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 938c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 939b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 64bit version of uregex_regionEnd. 940b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Reports the end index (exclusive) of the matching region for this URegularExpression. 941b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Any matches found are limited to to the region bounded by regionStart (inclusive) 942b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and regionEnd (exclusive). 943b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 944b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param regexp The compiled regular expression. 945b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status A pointer to a UErrorCode to receive any errors. 946b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @return The ending point (native) of this matcher's region. 947103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 94827f654740f2a26ad62a5c155af9199af9e69b889claireho */ 94954dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE int64_t U_EXPORT2 95027f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_regionEnd64(const URegularExpression *regexp, 95127f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 95227f654740f2a26ad62a5c155af9199af9e69b889claireho 95327f654740f2a26ad62a5c155af9199af9e69b889claireho/** 954c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Queries the transparency of region bounds for this URegularExpression. 955c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * See useTransparentBounds for a description of transparent and opaque bounds. 956c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * By default, matching boundaries are opaque. 957c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 958c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 959c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 960c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return TRUE if this matcher is using opaque bounds, false if it is not. 961b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 962c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 96350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE UBool U_EXPORT2 964c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_hasTransparentBounds(const URegularExpression *regexp, 965c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 966c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 967c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 968c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 969c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Sets the transparency of region bounds for this URegularExpression. 970c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Invoking this function with an argument of TRUE will set matches to use transparent bounds. 971c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * If the boolean argument is FALSE, then opaque bounds will be used. 972c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 973c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Using transparent bounds, the boundaries of the matching region are transparent 974c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * to lookahead, lookbehind, and boundary matching constructs. Those constructs can 975c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * see text beyond the boundaries of the region while checking for a match. 976c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 977c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * With opaque bounds, no text outside of the matching region is visible to lookahead, 978c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * lookbehind, and boundary matching constructs. 979c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 980c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * By default, opaque bounds are used. 981c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 982c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 983c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param b TRUE for transparent bounds; FALSE for opaque bounds 984c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 985b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 986c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru **/ 98750294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 988c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_useTransparentBounds(URegularExpression *regexp, 989c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UBool b, 990c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 991c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 992c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 993c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 994c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Return true if this URegularExpression is using anchoring bounds. 995c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * By default, anchoring region bounds are used. 996c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 997c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 998c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 999c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return TRUE if this matcher is using anchoring bounds. 1000b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1001c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 100250294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE UBool U_EXPORT2 1003c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_hasAnchoringBounds(const URegularExpression *regexp, 1004c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1005c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1006c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1007c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1008c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Set whether this URegularExpression is using Anchoring Bounds for its region. 1009c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * With anchoring bounds, pattern anchors such as ^ and $ will match at the start 1010c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and end of the region. Without Anchoring Bounds, anchors will only match at 1011c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * the positions they would in the complete text. 1012c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1013c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Anchoring Bounds are the default for regions. 1014c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1015c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1016c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param b TRUE if to enable anchoring bounds; FALSE to disable them. 1017c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 1018b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1019c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 102050294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 1021c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_useAnchoringBounds(URegularExpression *regexp, 1022c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UBool b, 1023c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1024c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1025c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1026c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Return TRUE if the most recent matching operation touched the 1027c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * end of the text being processed. In this case, additional input text could 1028c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * change the results of that match. 1029c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1030c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1031c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 1032c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return TRUE if the most recent match hit the end of input 1033b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1034c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 103550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE UBool U_EXPORT2 1036c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_hitEnd(const URegularExpression *regexp, 1037c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1038c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1039c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1040c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Return TRUE the most recent match succeeded and additional input could cause 1041c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * it to fail. If this function returns false and a match was found, then more input 1042c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * might change the match but the match won't be lost. If a match was not found, 1043c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * then requireEnd has no meaning. 1044c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1045c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1046c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A pointer to a UErrorCode to receive any errors. 1047c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return TRUE if more input could cause the most recent match to no longer match. 1048b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1049c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 105050294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE UBool U_EXPORT2 1051c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_requireEnd(const URegularExpression *regexp, 1052c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1053c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1054c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1055c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1056c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1057b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1058b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1059b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Replaces every substring of the input that matches the pattern 1060b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * with the given replacement string. This is a convenience function that 1061b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * provides a complete find-and-replace-all operation. 1062b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1063b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This method scans the input string looking for matches of the pattern. 1064b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Input that is not part of any match is copied unchanged to the 1065b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destination buffer. Matched regions are replaced in the output 1066b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * buffer by the replacement string. The replacement string may contain 1067b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * references to capture groups; these take the form of $1, $2, etc. 1068b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1069b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 1070b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementText A string containing the replacement text. 1071b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementLength The length of the replacement string, or 1072b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * -1 if it is NUL terminated. 1073b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destBuf A (UChar *) buffer that will receive the result. 1074b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param destCapacity The capacity of the destination buffer. 1075b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1076b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The length of the string resulting from the find 1077b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and replace operation. In the event that the 1078b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destination capacity is inadequate, the return value 1079b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * is still the full length of the untruncated string. 1080b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 1081b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1082b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 1083b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_replaceAll(URegularExpression *regexp, 1084b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar *replacementText, 1085b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t replacementLength, 1086b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar *destBuf, 1087b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t destCapacity, 1088b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 1089b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 109050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 109150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Replaces every substring of the input that matches the pattern 109250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * with the given replacement string. This is a convenience function that 109350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * provides a complete find-and-replace-all operation. 109450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 109550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This method scans the input string looking for matches of the pattern. 109650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Input that is not part of any match is copied unchanged to the 109750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * destination buffer. Matched regions are replaced in the output 109850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * buffer by the replacement string. The replacement string may contain 109950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * references to capture groups; these take the form of $1, $2, etc. 110050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 110150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 110250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param replacement A string containing the replacement text. 110350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param dest A mutable UText that will receive the result. 110450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If NULL, a new UText will be created (which may not be mutable). 110550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status A reference to a UErrorCode to receive any errors. 110650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return A UText containing the results of the find and replace. 110750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If a pre-allocated UText was provided, it will always be used and returned. 110850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 1109103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 111050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 111154dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 111250294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_replaceAllUText(URegularExpression *regexp, 111350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *replacement, 111450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *dest, 111550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 1116b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1117b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1118b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Replaces the first substring of the input that matches the pattern 1119b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * with the given replacement string. This is a convenience function that 1120b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * provides a complete find-and-replace operation. 1121b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1122b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This method scans the input string looking for a match of the pattern. 1123b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * All input that is not part of the match is copied unchanged to the 1124b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destination buffer. The matched region is replaced in the output 1125b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * buffer by the replacement string. The replacement string may contain 1126b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * references to capture groups; these take the form of $1, $2, etc. 1127b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 1129b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementText A string containing the replacement text. 1130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementLength The length of the replacement string, or 1131b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * -1 if it is NUL terminated. 1132b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destBuf A (UChar *) buffer that will receive the result. 1133b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param destCapacity The capacity of the destination buffer. 1134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status a reference to a UErrorCode to receive any errors. 1135b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The length of the string resulting from the find 1136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and replace operation. In the event that the 1137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destination capacity is inadequate, the return value 1138b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * is still the full length of the untruncated string. 1139b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 1140b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1141b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 1142b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_replaceFirst(URegularExpression *regexp, 1143b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar *replacementText, 1144b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t replacementLength, 1145b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar *destBuf, 1146b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t destCapacity, 1147b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 1148b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 114950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 115050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Replaces the first substring of the input that matches the pattern 115150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * with the given replacement string. This is a convenience function that 115250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * provides a complete find-and-replace operation. 115350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 115450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This method scans the input string looking for a match of the pattern. 115550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * All input that is not part of the match is copied unchanged to the 115650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * destination buffer. The matched region is replaced in the output 115750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * buffer by the replacement string. The replacement string may contain 115850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * references to capture groups; these take the form of $1, $2, etc. 115950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 116050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 116150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param replacement A string containing the replacement text. 116250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param dest A mutable UText that will receive the result. 116350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If NULL, a new UText will be created (which may not be mutable). 116450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status A reference to a UErrorCode to receive any errors. 116550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return A UText containing the results of the find and replace. 116650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If a pre-allocated UText was provided, it will always be used and returned. 116750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 1168103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 116950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 117054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 117150294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_replaceFirstUText(URegularExpression *regexp, 117250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *replacement, 117350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *dest, 117450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 117550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 1176b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1177b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Implements a replace operation intended to be used as part of an 1178b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * incremental find-and-replace. 1179b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1180b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p>The input string, starting from the end of the previous match and ending at 1181b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the start of the current match, is appended to the destination string. Then the 1182b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * replacement string is appended to the output string, 1183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * including handling any substitutions of captured text.</p> 1184b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1185b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p>A note on preflight computation of buffersize and error handling: 1186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Calls to uregex_appendReplacement() and uregex_appendTail() are 1187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * designed to be chained, one after another, with the destination 1188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * buffer pointer and buffer capacity updated after each in preparation 1189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to for the next. If the destination buffer is exhausted partway through such a 1190b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * sequence, a U_BUFFER_OVERFLOW_ERROR status will be returned. Normal 1191b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * ICU conventions are for a function to perform no action if it is 1192b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * called with an error status, but for this one case, uregex_appendRepacement() 1193b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * will operate normally so that buffer size computations will complete 1194b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * correctly. 1195b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1196b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <p>For simple, prepackaged, non-incremental find-and-replace 1197b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * operations, see replaceFirst() or replaceAll().</p> 1198b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1199b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The regular expression object. 1200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementText The string that will replace the matched portion of the 1201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * input string as it is copied to the destination buffer. 1202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The replacement text may contain references ($1, for 1203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * example) to capture groups from the match. 1204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param replacementLength The length of the replacement text string, 1205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * or -1 if the string is NUL terminated. 1206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destBuf The buffer into which the results of the 1207b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * find-and-replace are placed. On return, this pointer 1208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * will be updated to refer to the beginning of the 1209b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * unused portion of buffer, leaving it in position for 1210b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a subsequent call to this function. 1211b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destCapacity The size of the output buffer, On return, this 1212b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * parameter will be updated to reflect the space remaining 1213b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * unused in the output buffer. 1214b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1215b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The length of the result string. In the event that 1216b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destCapacity is inadequate, the full length of the 1217b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * untruncated output string is returned. 1218b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1219b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 1220b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1221b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1222b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 1223b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_appendReplacement(URegularExpression *regexp, 122450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const UChar *replacementText, 122550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t replacementLength, 122650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UChar **destBuf, 122750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t *destCapacity, 122850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 122950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 123050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 123150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Implements a replace operation intended to be used as part of an 123250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * incremental find-and-replace. 123350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 123450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <p>The input string, starting from the end of the previous match and ending at 123550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * the start of the current match, is appended to the destination string. Then the 123650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * replacement string is appended to the output string, 123750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * including handling any substitutions of captured text.</p> 123850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 123950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <p>For simple, prepackaged, non-incremental find-and-replace 124050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * operations, see replaceFirst() or replaceAll().</p> 124150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 124250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The regular expression object. 124350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param replacementText The string that will replace the matched portion of the 124450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * input string as it is copied to the destination buffer. 124550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The replacement text may contain references ($1, for 124650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * example) to capture groups from the match. 124750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param dest A mutable UText that will receive the result. Must not be NULL. 124850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status A reference to a UErrorCode to receive any errors. 124950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 1250103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 125150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 125254dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 125350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_appendReplacementUText(URegularExpression *regexp, 125450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *replacementText, 125550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *dest, 125650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 1257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * As the final step in a find-and-replace operation, append the remainder 1260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of the input string, starting at the position following the last match, 1261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to the destination string. <code>uregex_appendTail()</code> is intended 1262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to be invoked after one or more invocations of the 1263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <code>uregex_appendReplacement()</code> function. 1264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The regular expression object. This is needed to 1266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * obtain the input string and with the position 1267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of the last match within it. 1268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destBuf The buffer in which the results of the 1269b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * find-and-replace are placed. On return, the pointer 1270b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * will be updated to refer to the beginning of the 1271b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * unused portion of buffer. 1272b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destCapacity The size of the output buffer, On return, this 1273b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * value will be updated to reflect the space remaining 1274b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * unused in the output buffer. 1275b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1276b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The length of the result string. In the event that 1277b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destCapacity is inadequate, the full length of the 1278b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * untruncated output string is returned. 1279b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1280b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 1281b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1282b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 1283b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_appendTail(URegularExpression *regexp, 1284b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar **destBuf, 1285b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t *destCapacity, 1286b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 1287b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 128850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** 128950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * As the final step in a find-and-replace operation, append the remainder 129050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * of the input string, starting at the position following the last match, 129150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * to the destination string. <code>uregex_appendTailUText()</code> is intended 129250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * to be invoked after one or more invocations of the 129350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <code>uregex_appendReplacementUText()</code> function. 129450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 129550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The regular expression object. This is needed to 129650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * obtain the input string and with the position 129750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * of the last match within it. 129850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param dest A mutable UText that will receive the result. Must not be NULL. 1299b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 1300b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param status Error code 1301b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 130250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return The destination UText. 130350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 1304103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 130550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 130654dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE UText * U_EXPORT2 130750294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_appendTailUText(URegularExpression *regexp, 130827f654740f2a26ad62a5c155af9199af9e69b889claireho UText *dest, 130927f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 1310b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1311b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** 1312b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Split a string into fields. Somewhat like split() from Perl. 1313b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The pattern matches identify delimiters that separate the input 1314b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * into fields. The input data between the matches becomes the 1315b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * fields themselves. 1316b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 1317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Each of the fields is copied from the input string to the destination 131850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * buffer, and NUL terminated. The position of each field within 1319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the destination buffer is returned in the destFields array. 1320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1321b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * If the delimiter pattern includes capture groups, the captured text will 1322b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * also appear in the destination array of output strings, interspersed 1323b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * with the fields. This is similar to Perl, but differs from Java, 1324b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * which ignores the presence of capture groups in the pattern. 1325b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 1326b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Trailing empty fields will always be returned, assuming sufficient 1327b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * destination capacity. This differs from the default behavior for Java 1328b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * and Perl where trailing empty fields are not returned. 1329b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * 1330b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * The number of strings produced by the split operation is returned. 1331b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This count includes the strings from capture groups in the delimiter pattern. 1332b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * This behavior differs from Java, which ignores capture groups. 1333b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1334b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param regexp The compiled regular expression. 1335b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destBuf A (UChar *) buffer to receive the fields that 1336b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * are extracted from the input string. These 1337b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * field pointers will refer to positions within the 1338b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * destination buffer supplied by the caller. Any 1339b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * extra positions within the destFields array will be 1340b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * set to NULL. 1341b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destCapacity The capacity of the destBuf. 1342b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param requiredCapacity The actual capacity required of the destBuf. 1343b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If destCapacity is too small, requiredCapacity will return 1344b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the total capacity required to hold all of the output, and 1345b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a U_BUFFER_OVERFLOW_ERROR will be returned. 1346b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destFields An array to be filled with the position of each 1347b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of the extracted fields within destBuf. 1348b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param destFieldsCapacity The number of elements in the destFields array. 1349b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If the number of fields found is less than destFieldsCapacity, 1350b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the extra destFields elements are set to zero. 1351b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If destFieldsCapacity is too small, the trailing part of the 1352b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * input, including any field delimiters, is treated as if it 1353b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * were the last field - it is copied to the destBuf, and 1354b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * its position is in the destBuf is stored in the last element 1355b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * of destFields. This behavior mimics that of Perl. It is not 1356b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * an error condition, and no error status is returned when all destField 1357b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * positions are used. 1358b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1359b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @return The number of fields into which the input string was split. 1360b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @stable ICU 3.0 1361b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1362b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_STABLE int32_t U_EXPORT2 1363b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruuregex_split( URegularExpression *regexp, 1364b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar *destBuf, 1365b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t destCapacity, 1366b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t *requiredCapacity, 1367b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar *destFields[], 1368b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t destFieldsCapacity, 1369b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode *status); 1370b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 137150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho /** 137250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Split a string into fields. Somewhat like split() from Perl. 137350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The pattern matches identify delimiters that separate the input 137450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * into fields. The input data between the matches becomes the 137550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * fields themselves. 137650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * <p> 137750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The behavior of this function is not very closely aligned with uregex_split(); 137850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * instead, it is based on (and implemented directly on top of) the C++ split method. 137950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 138050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param regexp The compiled regular expression. 138150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param destFields An array of mutable UText structs to receive the results of the split. 138250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If a field is NULL, a new UText is allocated to contain the results for 138350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * that field. This new UText is not guaranteed to be mutable. 138450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param destFieldsCapacity The number of elements in the destination array. 138550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If the number of fields found is less than destCapacity, the 138650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * extra strings in the destination array are not altered. 138750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If the number of destination strings is less than the number 138850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * of fields, the trailing part of the input string, including any 138950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * field delimiters, is placed in the last destination string. 139050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * This behavior mimics that of Perl. It is not an error condition, and no 139150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * error status is returned when all destField positions are used. 139250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @param status A reference to a UErrorCode to receive any errors. 139350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * @return The number of fields into which the input string was split. 139450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * 1395103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 139650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 139754dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE int32_t U_EXPORT2 139850294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouregex_splitUText(URegularExpression *regexp, 139950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UText *destFields[], 140050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t destFieldsCapacity, 140150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UErrorCode *status); 140250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 1403c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1404c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Set a processing time limit for match operations with this URegularExpression. 1405c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1406c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Some patterns, when matching certain strings, can run in exponential time. 1407c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * For practical purposes, the match operation may appear to be in an 1408c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * infinite loop. 1409c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * When a limit is set a match operation will fail with an error if the 1410c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * limit is exceeded. 1411c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1412c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * The units of the limit are steps of the match engine. 1413c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Correspondence with actual processor time will depend on the speed 1414c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * of the processor and the details of the specific pattern, but will 1415c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * typically be on the order of milliseconds. 1416c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1417c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * By default, the matching time is not limited. 1418c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1419c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1420c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1421c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param limit The limit value, or 0 for no limit. 1422c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1423b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1424c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 142550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 1426c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_setTimeLimit(URegularExpression *regexp, 1427c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t limit, 1428c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1429c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1430c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1431c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Get the time limit for for matches with this URegularExpression. 1432c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * A return value of zero indicates that there is no limit. 1433c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1434c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1435c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1436c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return the maximum allowed time for a match, in units of processing steps. 1437b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1438c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 143950294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE int32_t U_EXPORT2 1440c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_getTimeLimit(const URegularExpression *regexp, 1441c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1442c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1443c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1444b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * Set the amount of heap storage available for use by the match backtracking stack. 1445c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1446c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * ICU uses a backtracking regular expression engine, with the backtrack stack 1447c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * maintained on the heap. This function sets the limit to the amount of memory 1448c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * that can be used for this purpose. A backtracking stack overflow will 1449c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * result in an error from the match operation that caused it. 1450c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1451c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * A limit is desirable because a malicious or poorly designed pattern can use 1452c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * excessive memory, potentially crashing the process. A limit is enabled 1453c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * by default. 1454c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * <p> 1455c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1456c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param limit The maximum size, in bytes, of the matching backtrack stack. 145759d709d503bab6e2b61931737e662dd293b40578ccornelius * A value of zero means no limit. 145859d709d503bab6e2b61931737e662dd293b40578ccornelius * The limit must be greater than or equal to zero. 1459c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1460c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1461b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1462c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 146350294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 1464c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_setStackLimit(URegularExpression *regexp, 1465c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t limit, 1466c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1467c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1468c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1469c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Get the size of the heap storage available for use by the back tracking stack. 1470c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1471c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return the maximum backtracking stack size, in bytes, or zero if the 1472c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * stack size is unlimited. 1473b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1474c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 147550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE int32_t U_EXPORT2 1476c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_getStackLimit(const URegularExpression *regexp, 1477c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1478c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1479c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1480c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1481c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Function pointer for a regular expression matching callback function. 1482c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * When set, a callback function will be called periodically during matching 1483c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * operations. If the call back function returns FALSE, the matching 1484c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * operation will be terminated early. 1485c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1486c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Note: the callback function must not call other functions on this 1487c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * URegularExpression. 1488c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1489c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param context context pointer. The callback function will be invoked 1490c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * with the context specified at the time that 1491c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * uregex_setMatchCallback() is called. 1492c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param steps the accumulated processing time, in match steps, 1493c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * for this matching operation. 1494c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @return TRUE to continue the matching operation. 1495c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * FALSE to terminate the matching operation. 1496b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1497c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 1498c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste QueruU_CDECL_BEGIN 1499c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querutypedef UBool U_CALLCONV URegexMatchCallback ( 1500c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru const void *context, 1501c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t steps); 1502c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste QueruU_CDECL_END 1503c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1504c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1505c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Set a callback function for this URegularExpression. 1506c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * During matching operations the function will be called periodically, 1507c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * giving the application the opportunity to terminate a long-running 1508c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * match. 1509c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1510c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1511c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param callback A pointer to the user-supplied callback function. 1512c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param context User context pointer. The value supplied at the 1513c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * time the callback function is set will be saved 1514c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * and passed to the callback each time that it is called. 1515c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1516b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1517c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 151850294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 1519c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_setMatchCallback(URegularExpression *regexp, 1520c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru URegexMatchCallback *callback, 1521c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru const void *context, 1522c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1523c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1524c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1525c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/** 1526c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * Get the callback function for this URegularExpression. 1527c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * 1528c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param regexp The compiled regular expression. 1529b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param callback Out parameter, receives a pointer to the user-supplied 1530c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * callback function. 1531c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param context Out parameter, receives the user context pointer that 1532c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * was set when uregex_setMatchCallback() was called. 1533c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru * @param status A reference to a UErrorCode to receive any errors. 1534b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @stable ICU 4.0 1535c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru */ 153650294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_STABLE void U_EXPORT2 1537c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruuregex_getMatchCallback(const URegularExpression *regexp, 1538c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru URegexMatchCallback **callback, 1539c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru const void **context, 1540c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru UErrorCode *status); 1541c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 154227f654740f2a26ad62a5c155af9199af9e69b889claireho/** 154327f654740f2a26ad62a5c155af9199af9e69b889claireho * Function pointer for a regular expression find callback function. 154427f654740f2a26ad62a5c155af9199af9e69b889claireho * 154527f654740f2a26ad62a5c155af9199af9e69b889claireho * When set, a callback function will be called during a find operation 154627f654740f2a26ad62a5c155af9199af9e69b889claireho * and for operations that depend on find, such as findNext, split and some replace 154727f654740f2a26ad62a5c155af9199af9e69b889claireho * operations like replaceFirst. 154827f654740f2a26ad62a5c155af9199af9e69b889claireho * The callback will usually be called after each attempt at a match, but this is not a 154927f654740f2a26ad62a5c155af9199af9e69b889claireho * guarantee that the callback will be invoked at each character. For finds where the 155027f654740f2a26ad62a5c155af9199af9e69b889claireho * match engine is invoked at each character, this may be close to true, but less likely 155127f654740f2a26ad62a5c155af9199af9e69b889claireho * for more optimized loops where the pattern is known to only start, and the match 155227f654740f2a26ad62a5c155af9199af9e69b889claireho * engine invoked, at certain characters. 155327f654740f2a26ad62a5c155af9199af9e69b889claireho * When invoked, this callback will specify the index at which a match operation is about 155427f654740f2a26ad62a5c155af9199af9e69b889claireho * to be attempted, giving the application the opportunity to terminate a long-running 155527f654740f2a26ad62a5c155af9199af9e69b889claireho * find operation. 155627f654740f2a26ad62a5c155af9199af9e69b889claireho * 155727f654740f2a26ad62a5c155af9199af9e69b889claireho * If the call back function returns FALSE, the find operation will be terminated early. 155827f654740f2a26ad62a5c155af9199af9e69b889claireho * 155927f654740f2a26ad62a5c155af9199af9e69b889claireho * Note: the callback function must not call other functions on this 156027f654740f2a26ad62a5c155af9199af9e69b889claireho * URegularExpression 156127f654740f2a26ad62a5c155af9199af9e69b889claireho * 156227f654740f2a26ad62a5c155af9199af9e69b889claireho * @param context context pointer. The callback function will be invoked 156327f654740f2a26ad62a5c155af9199af9e69b889claireho * with the context specified at the time that 156427f654740f2a26ad62a5c155af9199af9e69b889claireho * uregex_setFindProgressCallback() is called. 156527f654740f2a26ad62a5c155af9199af9e69b889claireho * @param matchIndex the next index at which a match attempt will be attempted for this 156627f654740f2a26ad62a5c155af9199af9e69b889claireho * find operation. If this callback interrupts the search, this is the 156727f654740f2a26ad62a5c155af9199af9e69b889claireho * index at which a find/findNext operation may be re-initiated. 156827f654740f2a26ad62a5c155af9199af9e69b889claireho * @return TRUE to continue the matching operation. 156927f654740f2a26ad62a5c155af9199af9e69b889claireho * FALSE to terminate the matching operation. 1570103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 157127f654740f2a26ad62a5c155af9199af9e69b889claireho */ 157227f654740f2a26ad62a5c155af9199af9e69b889clairehoU_CDECL_BEGIN 157327f654740f2a26ad62a5c155af9199af9e69b889clairehotypedef UBool U_CALLCONV URegexFindProgressCallback ( 157427f654740f2a26ad62a5c155af9199af9e69b889claireho const void *context, 157527f654740f2a26ad62a5c155af9199af9e69b889claireho int64_t matchIndex); 157627f654740f2a26ad62a5c155af9199af9e69b889clairehoU_CDECL_END 157727f654740f2a26ad62a5c155af9199af9e69b889claireho 1578103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius 157927f654740f2a26ad62a5c155af9199af9e69b889claireho/** 158027f654740f2a26ad62a5c155af9199af9e69b889claireho * Set the find progress callback function for this URegularExpression. 158127f654740f2a26ad62a5c155af9199af9e69b889claireho * 158227f654740f2a26ad62a5c155af9199af9e69b889claireho * @param regexp The compiled regular expression. 158327f654740f2a26ad62a5c155af9199af9e69b889claireho * @param callback A pointer to the user-supplied callback function. 158427f654740f2a26ad62a5c155af9199af9e69b889claireho * @param context User context pointer. The value supplied at the 158527f654740f2a26ad62a5c155af9199af9e69b889claireho * time the callback function is set will be saved 158627f654740f2a26ad62a5c155af9199af9e69b889claireho * and passed to the callback each time that it is called. 158727f654740f2a26ad62a5c155af9199af9e69b889claireho * @param status A reference to a UErrorCode to receive any errors. 1588103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 158927f654740f2a26ad62a5c155af9199af9e69b889claireho */ 159054dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 159127f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_setFindProgressCallback(URegularExpression *regexp, 159227f654740f2a26ad62a5c155af9199af9e69b889claireho URegexFindProgressCallback *callback, 159327f654740f2a26ad62a5c155af9199af9e69b889claireho const void *context, 159427f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 159527f654740f2a26ad62a5c155af9199af9e69b889claireho 159627f654740f2a26ad62a5c155af9199af9e69b889claireho/** 159727f654740f2a26ad62a5c155af9199af9e69b889claireho * Get the find progress callback function for this URegularExpression. 159827f654740f2a26ad62a5c155af9199af9e69b889claireho * 159927f654740f2a26ad62a5c155af9199af9e69b889claireho * @param regexp The compiled regular expression. 1600b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho * @param callback Out parameter, receives a pointer to the user-supplied 160127f654740f2a26ad62a5c155af9199af9e69b889claireho * callback function. 160227f654740f2a26ad62a5c155af9199af9e69b889claireho * @param context Out parameter, receives the user context pointer that 160327f654740f2a26ad62a5c155af9199af9e69b889claireho * was set when uregex_setFindProgressCallback() was called. 160427f654740f2a26ad62a5c155af9199af9e69b889claireho * @param status A reference to a UErrorCode to receive any errors. 1605103e9ffba2cba345d0078eb8b8db33249f81840aCraig Cornelius * @stable ICU 4.6 160627f654740f2a26ad62a5c155af9199af9e69b889claireho */ 160754dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusU_STABLE void U_EXPORT2 160827f654740f2a26ad62a5c155af9199af9e69b889clairehouregex_getFindProgressCallback(const URegularExpression *regexp, 160927f654740f2a26ad62a5c155af9199af9e69b889claireho URegexFindProgressCallback **callback, 161027f654740f2a26ad62a5c155af9199af9e69b889claireho const void **context, 161127f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode *status); 1612c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 1613b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif /* !UCONFIG_NO_REGULAR_EXPRESSIONS */ 1614b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif /* UREGEX_H */ 1615