uregex.cpp | Explore in Territory

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2004-2015, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  uregex.cpp
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_REGULAR_EXPRESSIONS

#include "unicode/regex.h"
#include "unicode/uregex.h"
#include "unicode/unistr.h"
#include "unicode/ustring.h"
#include "unicode/uchar.h"
#include "unicode/uobject.h"
#include "unicode/utf16.h"
#include "cmemory.h"
#include "uassert.h"
#include "uhash.h"
#include "umutex.h"
#include "uvectr32.h"

#include "regextxt.h"

U_NAMESPACE_BEGIN

#define REMAINING_CAPACITY(idx,len) …

struct RegularExpression: public UMemory { … };

static const int32_t REXP_MAGIC = …; // "rexp" in ASCII

RegularExpression::RegularExpression() { … }

RegularExpression::~RegularExpression() { … }

U_NAMESPACE_END

U_NAMESPACE_USE

//----------------------------------------------------------------------------------------
//
//   validateRE    Do boilerplate style checks on API function parameters.
//                 Return true if they look OK.
//----------------------------------------------------------------------------------------
static UBool validateRE(const RegularExpression *re, UBool requiresText, UErrorCode *status) { … }

//----------------------------------------------------------------------------------------
//
//    uregex_open
//
//----------------------------------------------------------------------------------------
U_CAPI URegularExpression *  U_EXPORT2
uregex_open( const  char16_t       *pattern,
                    int32_t         patternLength,
                    uint32_t        flags,
                    UParseError    *pe,
                    UErrorCode     *status) { … }

//----------------------------------------------------------------------------------------
//
//    uregex_openUText
//
//----------------------------------------------------------------------------------------
U_CAPI URegularExpression *  U_EXPORT2
uregex_openUText(UText          *pattern,
                 uint32_t        flags,
                 UParseError    *pe,
                 UErrorCode     *status) { … }

//----------------------------------------------------------------------------------------
//
//    uregex_close
//
//----------------------------------------------------------------------------------------
U_CAPI void  U_EXPORT2
uregex_close(URegularExpression  *re2) { … }


//----------------------------------------------------------------------------------------
//
//    uregex_clone
//
//----------------------------------------------------------------------------------------
U_CAPI URegularExpression * U_EXPORT2
uregex_clone(const URegularExpression *source2, UErrorCode *status)  { … }




//------------------------------------------------------------------------------
//
//    uregex_pattern
//
//------------------------------------------------------------------------------
U_CAPI const char16_t * U_EXPORT2
uregex_pattern(const  URegularExpression *regexp2,
                      int32_t            *patLength,
                      UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_patternUText
//
//------------------------------------------------------------------------------
U_CAPI UText * U_EXPORT2
uregex_patternUText(const URegularExpression *regexp2,
                          UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_flags
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_flags(const URegularExpression *regexp2, UErrorCode *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_setText
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setText(URegularExpression *regexp2,
               const char16_t     *text,
               int32_t             textLength,
               UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_setUText
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setUText(URegularExpression *regexp2,
                UText              *text,
                UErrorCode         *status) { … }



//------------------------------------------------------------------------------
//
//    uregex_getText
//
//------------------------------------------------------------------------------
U_CAPI const char16_t * U_EXPORT2
uregex_getText(URegularExpression *regexp2,
               int32_t            *textLength,
               UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_getUText
//
//------------------------------------------------------------------------------
U_CAPI UText * U_EXPORT2
uregex_getUText(URegularExpression *regexp2,
                UText              *dest,
                UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_refreshUText
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_refreshUText(URegularExpression *regexp2,
                    UText              *text,
                    UErrorCode         *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_matches
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_matches(URegularExpression *regexp2,
               int32_t            startIndex,
               UErrorCode        *status)  { … }

U_CAPI UBool U_EXPORT2
uregex_matches64(URegularExpression *regexp2,
                 int64_t            startIndex,
                 UErrorCode        *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_lookingAt
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_lookingAt(URegularExpression *regexp2,
                 int32_t             startIndex,
                 UErrorCode         *status)  { … }

U_CAPI UBool U_EXPORT2
uregex_lookingAt64(URegularExpression *regexp2,
                   int64_t             startIndex,
                   UErrorCode         *status)  { … }



//------------------------------------------------------------------------------
//
//    uregex_find
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_find(URegularExpression *regexp2,
            int32_t             startIndex,
            UErrorCode         *status)  { … }

U_CAPI UBool U_EXPORT2
uregex_find64(URegularExpression *regexp2,
              int64_t             startIndex,
              UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_findNext
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_findNext(URegularExpression *regexp2,
                UErrorCode         *status)  { … }

//------------------------------------------------------------------------------
//
//    uregex_groupCount
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_groupCount(URegularExpression *regexp2,
                  UErrorCode         *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_groupNumberFromName
//
//------------------------------------------------------------------------------
int32_t
uregex_groupNumberFromName(URegularExpression *regexp2,
                           const char16_t     *groupName,
                           int32_t             nameLength,
                           UErrorCode          *status) { … }

int32_t
uregex_groupNumberFromCName(URegularExpression *regexp2,
                            const char         *groupName,
                            int32_t             nameLength,
                            UErrorCode          *status) { … }

//------------------------------------------------------------------------------
//
//    uregex_group
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_group(URegularExpression *regexp2,
             int32_t             groupNum,
             char16_t           *dest,
             int32_t             destCapacity,
             UErrorCode          *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_groupUText
//
//------------------------------------------------------------------------------
U_CAPI UText * U_EXPORT2
uregex_groupUText(URegularExpression *regexp2,
                  int32_t             groupNum,
                  UText              *dest,
                  int64_t            *groupLength,
                  UErrorCode         *status)  { … }

//------------------------------------------------------------------------------
//
//    uregex_start
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_start(URegularExpression *regexp2,
             int32_t             groupNum,
             UErrorCode          *status)  { … }

U_CAPI int64_t U_EXPORT2
uregex_start64(URegularExpression *regexp2,
               int32_t             groupNum,
               UErrorCode          *status)  { … }

//------------------------------------------------------------------------------
//
//    uregex_end
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_end(URegularExpression   *regexp2,
           int32_t               groupNum,
           UErrorCode           *status)  { … }

U_CAPI int64_t U_EXPORT2
uregex_end64(URegularExpression   *regexp2,
             int32_t               groupNum,
             UErrorCode           *status)  { … }

//------------------------------------------------------------------------------
//
//    uregex_reset
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_reset(URegularExpression    *regexp2,
             int32_t               index,
             UErrorCode            *status)  { … }

U_CAPI void U_EXPORT2
uregex_reset64(URegularExpression    *regexp2,
               int64_t               index,
               UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_setRegion
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setRegion(URegularExpression   *regexp2,
                 int32_t               regionStart,
                 int32_t               regionLimit,
                 UErrorCode           *status)  { … }

U_CAPI void U_EXPORT2
uregex_setRegion64(URegularExpression   *regexp2,
                   int64_t               regionStart,
                   int64_t               regionLimit,
                   UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_setRegionAndStart
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setRegionAndStart(URegularExpression   *regexp2,
                 int64_t               regionStart,
                 int64_t               regionLimit,
                 int64_t               startIndex,
                 UErrorCode           *status)  { … }

//------------------------------------------------------------------------------
//
//    uregex_regionStart
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_regionStart(const  URegularExpression   *regexp2,
                          UErrorCode           *status)  { … }

U_CAPI int64_t U_EXPORT2
uregex_regionStart64(const  URegularExpression   *regexp2,
                            UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_regionEnd
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_regionEnd(const  URegularExpression   *regexp2,
                        UErrorCode           *status)  { … }

U_CAPI int64_t U_EXPORT2
uregex_regionEnd64(const  URegularExpression   *regexp2,
                          UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_hasTransparentBounds
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_hasTransparentBounds(const  URegularExpression   *regexp2,
                                   UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_useTransparentBounds
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_useTransparentBounds(URegularExpression    *regexp2,
                            UBool                  b,
                            UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_hasAnchoringBounds
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_hasAnchoringBounds(const  URegularExpression   *regexp2,
                                 UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_useAnchoringBounds
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_useAnchoringBounds(URegularExpression    *regexp2,
                          UBool                  b,
                          UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_hitEnd
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_hitEnd(const  URegularExpression   *regexp2,
                     UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_requireEnd
//
//------------------------------------------------------------------------------
U_CAPI UBool U_EXPORT2
uregex_requireEnd(const  URegularExpression   *regexp2,
                         UErrorCode           *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_setTimeLimit
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setTimeLimit(URegularExpression   *regexp2,
                    int32_t               limit,
                    UErrorCode           *status) { … }



//------------------------------------------------------------------------------
//
//    uregex_getTimeLimit
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_getTimeLimit(const  URegularExpression   *regexp2,
                           UErrorCode           *status) { … }



//------------------------------------------------------------------------------
//
//    uregex_setStackLimit
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setStackLimit(URegularExpression   *regexp2,
                     int32_t               limit,
                     UErrorCode           *status) { … }



//------------------------------------------------------------------------------
//
//    uregex_getStackLimit
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_getStackLimit(const  URegularExpression   *regexp2,
                            UErrorCode           *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_setMatchCallback
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setMatchCallback(URegularExpression      *regexp2,
                        URegexMatchCallback     *callback,
                        const void              *context,
                        UErrorCode              *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_getMatchCallback
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_getMatchCallback(const URegularExpression    *regexp2,
                        URegexMatchCallback        **callback,
                        const void                 **context,
                        UErrorCode                  *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_setMatchProgressCallback
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_setFindProgressCallback(URegularExpression              *regexp2,
                                URegexFindProgressCallback      *callback,
                                const void                      *context,
                                UErrorCode                      *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_getMatchCallback
//
//------------------------------------------------------------------------------
U_CAPI void U_EXPORT2
uregex_getFindProgressCallback(const URegularExpression          *regexp2,
                                URegexFindProgressCallback        **callback,
                                const void                        **context,
                                UErrorCode                        *status) { … }


//------------------------------------------------------------------------------
//
//    uregex_replaceAll
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_replaceAll(URegularExpression    *regexp2,
                  const char16_t        *replacementText,
                  int32_t                replacementLength,
                  char16_t              *destBuf,
                  int32_t                destCapacity,
                  UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_replaceAllUText
//
//------------------------------------------------------------------------------
U_CAPI UText * U_EXPORT2
uregex_replaceAllUText(URegularExpression    *regexp2,
                       UText                 *replacementText,
                       UText                 *dest,
                       UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_replaceFirst
//
//------------------------------------------------------------------------------
U_CAPI int32_t U_EXPORT2
uregex_replaceFirst(URegularExpression  *regexp2,
                    const char16_t      *replacementText,
                    int32_t              replacementLength,
                    char16_t            *destBuf,
                    int32_t              destCapacity,
                    UErrorCode          *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_replaceFirstUText
//
//------------------------------------------------------------------------------
U_CAPI UText * U_EXPORT2
uregex_replaceFirstUText(URegularExpression  *regexp2,
                         UText                 *replacementText,
                         UText                 *dest,
                         UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_appendReplacement
//
//------------------------------------------------------------------------------

U_NAMESPACE_BEGIN
//
//  Dummy class, because these functions need to be friends of class RegexMatcher,
//               and stand-alone C functions don't work as friends
//
class RegexCImpl { … };

U_NAMESPACE_END



static const char16_t BACKSLASH  = …;
static const char16_t DOLLARSIGN = …;
static const char16_t LEFTBRACKET = …;
static const char16_t RIGHTBRACKET = …;

//
//  Move a character to an output buffer, with bounds checking on the index.
//      Index advances even if capacity is exceeded, for preflight size computations.
//      This little sequence is used a LOT.
//
static inline void appendToBuf(char16_t c, int32_t *idx, char16_t *buf, int32_t bufCapacity) { … }


//
//  appendReplacement, the actual implementation.
//
int32_t RegexCImpl::appendReplacement(RegularExpression    *regexp,
                                      const char16_t        *replacementText,
                                      int32_t                replacementLength,
                                      char16_t             **destBuf,
                                      int32_t               *destCapacity,
                                      UErrorCode            *status)  { … }

//
//   appendReplacement   the actual API function,
//
U_CAPI int32_t U_EXPORT2
uregex_appendReplacement(URegularExpression    *regexp2,
                         const char16_t        *replacementText,
                         int32_t                replacementLength,
                         char16_t             **destBuf,
                         int32_t               *destCapacity,
                         UErrorCode            *status) { … }

//
//   uregex_appendReplacementUText...can just use the normal C++ method
//
U_CAPI void U_EXPORT2
uregex_appendReplacementUText(URegularExpression    *regexp2,
                              UText                 *replText,
                              UText                 *dest,
                              UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    uregex_appendTail
//
//------------------------------------------------------------------------------
int32_t RegexCImpl::appendTail(RegularExpression    *regexp,
                               char16_t             **destBuf,
                               int32_t               *destCapacity,
                               UErrorCode            *status)
{ … }


//
//   appendTail   the actual API function
//
U_CAPI int32_t U_EXPORT2
uregex_appendTail(URegularExpression    *regexp2,
                  char16_t             **destBuf,
                  int32_t               *destCapacity,
                  UErrorCode            *status)  { … }


//
//   uregex_appendTailUText...can just use the normal C++ method
//
U_CAPI UText * U_EXPORT2
uregex_appendTailUText(URegularExpression    *regexp2,
                       UText                 *dest,
                       UErrorCode            *status)  { … }


//------------------------------------------------------------------------------
//
//    copyString     Internal utility to copy a string to an output buffer,
//                   while managing buffer overflow and preflight size
//                   computation.  NUL termination is added to destination,
//                   and the NUL is counted in the output size.
//
//------------------------------------------------------------------------------
#if 0
static void copyString(char16_t     *destBuffer,    //  Destination buffer.
                       int32_t       destCapacity,  //  Total capacity of dest buffer
                       int32_t      *destIndex,     //  Index into dest buffer.  Updated on return.
                                                    //    Update not clipped to destCapacity.
                       const char16_t  *srcPtr,        //  Pointer to source string
                       int32_t       srcLen)        //  Source string len.
{
    int32_t  si;
    int32_t  di = *destIndex;
    char16_t c;

    for (si=0; si<srcLen;  si++) {
        c = srcPtr[si];
        if (di < destCapacity) {
            destBuffer[di] = c;
            di++;
        } else {
            di += srcLen - si;
            break;
        }
    }
    if (di<destCapacity) {
        destBuffer[di] = 0;
    }
    di++;
    *destIndex = di;
}
#endif

//------------------------------------------------------------------------------
//
//    uregex_split
//
//------------------------------------------------------------------------------
int32_t RegexCImpl::split(RegularExpression     *regexp,
                          char16_t              *destBuf,
                          int32_t                destCapacity,
                          int32_t               *requiredCapacity,
                          char16_t              *destFields[],
                          int32_t                destFieldsCapacity,
                          UErrorCode            *status) { … }

//
//   uregex_split   The actual API function
//
U_CAPI int32_t U_EXPORT2
uregex_split(URegularExpression      *regexp2,
             char16_t                *destBuf,
             int32_t                  destCapacity,
             int32_t                 *requiredCapacity,
             char16_t                *destFields[],
             int32_t                  destFieldsCapacity,
             UErrorCode              *status) { … }


//
//   uregex_splitUText...can just use the normal C++ method
//
U_CAPI int32_t U_EXPORT2
uregex_splitUText(URegularExpression    *regexp2,
                  UText                 *destFields[],
                  int32_t                destFieldsCapacity,
                  UErrorCode            *status) { … }


#endif   // !UCONFIG_NO_REGULAR_EXPRESSIONS
chromium/third_party/icu/source/i18n/uregex.cpp
