office-gobmx/sal/rtl/uri.cxx
Stephan Bergmann 6a9f342add Introduce rtl::createUriCharClass
...to make those char class array initializations more readable.  (Making the
corresponding variables constexpr is mostly done so that failures in the
provided `unencoded` arguments, like non-ASCII characters or duplicate
character typos, would lead to compile-time errors also for !HAVE_CPP_CONSTEVAL.
And assigning to a sal_Bool std::array needs another hack to avoid false
loplugin:implicitboolconversion warnings.)

Change-Id: Ieb8827f69f55f1212a9428817d5331fcb18ef1d8
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/133058
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2022-04-15 08:39:21 +02:00

759 lines
26 KiB
C++

/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
/*
* This file is part of the LibreOffice project.
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/.
*
* This file incorporates work covered by the following license notice:
*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed
* with this work for additional information regarding copyright
* ownership. The ASF licenses this file to you under the Apache
* License, Version 2.0 (the "License"); you may not use this file
* except in compliance with the License. You may obtain a copy of
* the License at http://www.apache.org/licenses/LICENSE-2.0 .
*/
#include <rtl/character.hxx>
#include <rtl/strbuf.hxx>
#include <rtl/textenc.h>
#include <rtl/textcvt.h>
#include <rtl/uri.h>
#include <rtl/uri.hxx>
#include <rtl/ustrbuf.h>
#include <rtl/ustrbuf.hxx>
#include <rtl/ustring.h>
#include <rtl/ustring.hxx>
#include <sal/types.h>
#include <sal/macros.h>
#include <uri_internal.hxx>
#include <algorithm>
#include <cstddef>
namespace {
sal_Unicode const cEscapePrefix = 0x25; // '%'
int getHexWeight(sal_uInt32 nUtf32)
{
return nUtf32 >= 0x30 && nUtf32 <= 0x39 ? // '0'--'9'
static_cast< int >(nUtf32 - 0x30) :
nUtf32 >= 0x41 && nUtf32 <= 0x46 ? // 'A'--'F'
static_cast< int >(nUtf32 - 0x41 + 10) :
nUtf32 >= 0x61 && nUtf32 <= 0x66 ? // 'a'--'f'
static_cast< int >(nUtf32 - 0x61 + 10) :
-1; // not a hex digit
}
bool isValid(sal_Bool const * pCharClass, sal_uInt32 nUtf32)
{
return nUtf32 < rtl::UriCharClassSize && pCharClass[nUtf32];
}
void writeUnicode(rtl_uString ** pBuffer, sal_Int32 * pCapacity,
sal_Unicode cChar)
{
rtl_uStringbuffer_insert(pBuffer, pCapacity, (*pBuffer)->length, &cChar, 1);
}
}
namespace rtl::uri::detail {
/** Read any of the following:
@li sequence of escape sequences representing character from eCharset,
translated to single UCS4 character; or
@li pair of UTF-16 surrogates, translated to single UCS4 character; or
@li single UTF-16 character, extended to UCS4 character.
*/
sal_uInt32 readUcs4(sal_Unicode const ** pBegin, sal_Unicode const * pEnd,
bool bEncoded, rtl_TextEncoding eCharset,
EscapeType * pType)
{
sal_uInt32 nChar = *(*pBegin)++;
int nWeight1;
int nWeight2;
if (nChar == cEscapePrefix && bEncoded && pEnd - *pBegin >= 2
&& (nWeight1 = getHexWeight((*pBegin)[0])) >= 0
&& (nWeight2 = getHexWeight((*pBegin)[1])) >= 0)
{
*pBegin += 2;
nChar = static_cast< sal_uInt32 >(nWeight1 << 4 | nWeight2);
if (nChar <= 0x7F)
{
*pType = EscapeChar;
}
else if (eCharset == RTL_TEXTENCODING_UTF8)
{
if (nChar >= 0xC0 && nChar <= 0xF4)
{
sal_uInt32 nEncoded;
int nShift;
sal_uInt32 nMin;
if (nChar <= 0xDF)
{
nEncoded = (nChar & 0x1F) << 6;
nShift = 0;
nMin = 0x80;
}
else if (nChar <= 0xEF)
{
nEncoded = (nChar & 0x0F) << 12;
nShift = 6;
nMin = 0x800;
}
else
{
nEncoded = (nChar & 0x07) << 18;
nShift = 12;
nMin = 0x10000;
}
sal_Unicode const * p = *pBegin;
bool bUTF8 = true;
for (; nShift >= 0; nShift -= 6)
{
if (pEnd - p < 3 || p[0] != cEscapePrefix
|| (nWeight1 = getHexWeight(p[1])) < 8
|| nWeight1 > 11
|| (nWeight2 = getHexWeight(p[2])) < 0)
{
bUTF8 = false;
break;
}
p += 3;
nEncoded |= ((nWeight1 & 3) << 4 | nWeight2) << nShift;
}
if (bUTF8 && rtl::isUnicodeScalarValue(nEncoded)
&& nEncoded >= nMin)
{
*pBegin = p;
*pType = EscapeChar;
return nEncoded;
}
}
*pType = EscapeOctet;
}
else
{
OStringBuffer aBuf;
aBuf.append(static_cast< char >(nChar));
rtl_TextToUnicodeConverter aConverter
= rtl_createTextToUnicodeConverter(eCharset);
sal_Unicode const * p = *pBegin;
for (;;)
{
sal_Unicode aDst[2];
sal_uInt32 nInfo;
sal_Size nConverted;
sal_Size nDstSize = rtl_convertTextToUnicode(
aConverter, nullptr, aBuf.getStr(), aBuf.getLength(), aDst,
SAL_N_ELEMENTS( aDst ),
(RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_ERROR
| RTL_TEXTTOUNICODE_FLAGS_MBUNDEFINED_ERROR
| RTL_TEXTTOUNICODE_FLAGS_INVALID_ERROR),
&nInfo, &nConverted);
if (nInfo == 0)
{
assert( nConverted
== sal::static_int_cast< sal_uInt32 >(
aBuf.getLength()));
rtl_destroyTextToUnicodeConverter(aConverter);
*pBegin = p;
*pType = EscapeChar;
assert( nDstSize == 1
|| (nDstSize == 2 && rtl::isHighSurrogate(aDst[0])
&& rtl::isLowSurrogate(aDst[1])));
return nDstSize == 1
? aDst[0] : rtl::combineSurrogates(aDst[0], aDst[1]);
}
if (nInfo == RTL_TEXTTOUNICODE_INFO_SRCBUFFERTOOSMALL
&& pEnd - p >= 3 && p[0] == cEscapePrefix
&& (nWeight1 = getHexWeight(p[1])) >= 0
&& (nWeight2 = getHexWeight(p[2])) >= 0)
{
p += 3;
aBuf.append(static_cast< char >(nWeight1 << 4 | nWeight2));
}
else if (nInfo == RTL_TEXTTOUNICODE_INFO_SRCBUFFERTOOSMALL
&& p != pEnd && *p <= 0x7F)
{
aBuf.append(static_cast< char >(*p++));
}
else
{
assert(
(nInfo & RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOOSMALL)
== 0);
break;
}
}
rtl_destroyTextToUnicodeConverter(aConverter);
*pType = EscapeOctet;
}
return nChar;
}
*pType = EscapeNo;
return rtl::isHighSurrogate(nChar) && *pBegin < pEnd
&& rtl::isLowSurrogate(**pBegin) ?
rtl::combineSurrogates(nChar, *(*pBegin)++) : nChar;
}
}
namespace {
void writeUcs4(rtl_uString ** pBuffer, sal_Int32 * pCapacity, sal_uInt32 nUtf32)
{
rtl_uStringbuffer_insertUtf32(pBuffer, pCapacity, (*pBuffer)->length, nUtf32);
}
void writeEscapeOctet(rtl_uString ** pBuffer, sal_Int32 * pCapacity,
sal_uInt32 nOctet)
{
assert(nOctet <= 0xFF); // bad octet
static sal_Unicode const aHex[16]
= { 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39,
0x41, 0x42, 0x43, 0x44, 0x45, 0x46 }; /* '0'--'9', 'A'--'F' */
writeUnicode(pBuffer, pCapacity, cEscapePrefix);
writeUnicode(pBuffer, pCapacity, aHex[nOctet >> 4]);
writeUnicode(pBuffer, pCapacity, aHex[nOctet & 15]);
}
bool writeEscapeChar(rtl_uString ** pBuffer, sal_Int32 * pCapacity,
sal_uInt32 nUtf32, rtl_TextEncoding eCharset, bool bStrict)
{
assert(rtl::isUnicodeCodePoint(nUtf32));
if (eCharset == RTL_TEXTENCODING_UTF8)
{
if (nUtf32 < 0x80)
{
writeEscapeOctet(pBuffer, pCapacity, nUtf32);
}
else if (nUtf32 < 0x800)
{
writeEscapeOctet(pBuffer, pCapacity, nUtf32 >> 6 | 0xC0);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 & 0x3F) | 0x80);
}
else if (nUtf32 < 0x10000)
{
writeEscapeOctet(pBuffer, pCapacity, nUtf32 >> 12 | 0xE0);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 >> 6 & 0x3F) | 0x80);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 & 0x3F) | 0x80);
}
else
{
writeEscapeOctet(pBuffer, pCapacity, nUtf32 >> 18 | 0xF0);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 >> 12 & 0x3F) | 0x80);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 >> 6 & 0x3F) | 0x80);
writeEscapeOctet(pBuffer, pCapacity, (nUtf32 & 0x3F) | 0x80);
}
}
else
{
rtl_UnicodeToTextConverter aConverter
= rtl_createUnicodeToTextConverter(eCharset);
sal_Unicode aSrc[2];
sal_Size nSrcSize = rtl::splitSurrogates(nUtf32, aSrc);
char aDst[32]; // FIXME random value
sal_uInt32 nInfo;
sal_Size nConverted;
sal_Size nDstSize = rtl_convertUnicodeToText(
aConverter, nullptr, aSrc, nSrcSize, aDst, sizeof aDst,
RTL_UNICODETOTEXT_FLAGS_UNDEFINED_ERROR
| RTL_UNICODETOTEXT_FLAGS_INVALID_ERROR
| RTL_UNICODETOTEXT_FLAGS_FLUSH,
&nInfo, &nConverted);
assert((nInfo & RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL) == 0);
rtl_destroyUnicodeToTextConverter(aConverter);
if (nInfo == 0)
{
assert(nConverted == nSrcSize); // bad rtl_convertUnicodeToText
for (sal_Size i = 0; i < nDstSize; ++i)
{
writeEscapeOctet(pBuffer, pCapacity,
static_cast< unsigned char >(aDst[i]));
// FIXME all octets are escaped, even if there is no need
}
}
else
{
if (bStrict)
return false;
writeUcs4(pBuffer, pCapacity, nUtf32);
}
}
return true;
}
struct Component
{
sal_Unicode const * pBegin;
sal_Unicode const * pEnd;
Component(): pBegin(nullptr), pEnd(nullptr) {}
bool isPresent() const { return pBegin != nullptr; }
sal_Int32 getLength() const;
};
sal_Int32 Component::getLength() const
{
assert(isPresent()); // taking length of non-present component
return static_cast< sal_Int32 >(pEnd - pBegin);
}
struct Components
{
Component aScheme;
Component aAuthority;
Component aPath;
Component aQuery;
Component aFragment;
};
void parseUriRef(rtl_uString const * pUriRef, Components * pComponents)
{
// This algorithm is liberal and accepts various forms of illegal input.
sal_Unicode const * pBegin = pUriRef->buffer;
sal_Unicode const * pEnd = pBegin + pUriRef->length;
sal_Unicode const * pPos = pBegin;
if (pPos != pEnd && rtl::isAsciiAlpha(*pPos))
{
for (sal_Unicode const * p = pPos + 1; p != pEnd; ++p)
{
if (*p == ':')
{
pComponents->aScheme.pBegin = pBegin;
pComponents->aScheme.pEnd = ++p;
pPos = p;
break;
}
if (!rtl::isAsciiAlphanumeric(*p) && *p != '+' && *p != '-'
&& *p != '.')
{
break;
}
}
}
if (pEnd - pPos >= 2 && pPos[0] == '/' && pPos[1] == '/')
{
pComponents->aAuthority.pBegin = pPos;
pPos += 2;
while (pPos != pEnd && *pPos != '/' && *pPos != '?' && *pPos != '#')
{
++pPos;
}
pComponents->aAuthority.pEnd = pPos;
}
pComponents->aPath.pBegin = pPos;
while (pPos != pEnd && *pPos != '?' && * pPos != '#')
{
++pPos;
}
pComponents->aPath.pEnd = pPos;
if (pPos != pEnd && *pPos == '?')
{
pComponents->aQuery.pBegin = pPos++;
while (pPos != pEnd && * pPos != '#')
{
++pPos;
}
pComponents->aQuery.pEnd = pPos;
}
if (pPos != pEnd)
{
assert(*pPos == '#');
pComponents->aFragment.pBegin = pPos;
pComponents->aFragment.pEnd = pEnd;
}
}
void appendPath(
OUStringBuffer & buffer, sal_Int32 bufferStart, bool precedingSlash,
sal_Unicode const * pathBegin, sal_Unicode const * pathEnd)
{
while (precedingSlash || pathBegin != pathEnd)
{
sal_Unicode const * p = pathBegin;
while (p != pathEnd && *p != '/')
{
++p;
}
std::size_t n = p - pathBegin;
if (n == 1 && pathBegin[0] == '.')
{
// input begins with "." -> remove from input (and done):
// i.e., !precedingSlash -> !precedingSlash
// input begins with "./" -> remove from input:
// i.e., !precedingSlash -> !precedingSlash
// input begins with "/." -> replace with "/" in input (and not yet
// done):
// i.e., precedingSlash -> precedingSlash
// input begins with "/./" -> replace with "/" in input:
// i.e., precedingSlash -> precedingSlash
}
else if (n == 2 && pathBegin[0] == '.' && pathBegin[1] == '.')
{
// input begins with ".." -> remove from input (and done):
// i.e., !precedingSlash -> !precedingSlash
// input begins with "../" -> remove from input
// i.e., !precedingSlash -> !precedingSlash
// input begins with "/.." -> replace with "/" in input, and shrink
// output (not yet done):
// i.e., precedingSlash -> precedingSlash
// input begins with "/../" -> replace with "/" in input, and shrink
// output:
// i.e., precedingSlash -> precedingSlash
if (precedingSlash)
{
buffer.truncate(
bufferStart
+ std::max<sal_Int32>(
rtl_ustr_lastIndexOfChar_WithLength(
buffer.getStr() + bufferStart,
buffer.getLength() - bufferStart, '/'),
0));
}
}
else
{
if (precedingSlash)
buffer.append('/');
buffer.append(pathBegin, n);
precedingSlash = p != pathEnd;
}
pathBegin = p + (p == pathEnd ? 0 : 1);
}
}
}
sal_Bool const * SAL_CALL rtl_getUriCharClass(rtl_UriCharClass eCharClass)
SAL_THROW_EXTERN_C()
{
static constexpr std::array<sal_Bool, rtl::UriCharClassSize> aCharClass[] = {
rtl::createUriCharClass(u8""), // None
rtl::createUriCharClass(
u8"!$&'()*+,-./:;=?@[]_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // Uric
rtl::createUriCharClass(
u8"!$&'()*+,-.:;=?@_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // UricNoSlash
rtl::createUriCharClass(
u8"!$&'()*+,-.;=@_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // RelSegment
rtl::createUriCharClass(
u8"!$&'()*+,-.:;=@_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // RegName
rtl::createUriCharClass(
u8"!$&'()*+,-.:;=_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // Userinfo
rtl::createUriCharClass(
u8"!$&'()*+,-.:=@_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"), // Pchar
rtl::createUriCharClass(
u8"!$&'()*+-./:?@_~"
"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz")}; // UnoParamValue
assert(
(eCharClass >= 0
&& (sal::static_int_cast< std::size_t >(eCharClass)
< SAL_N_ELEMENTS(aCharClass)))); // bad eCharClass
return aCharClass[eCharClass].data();
}
void SAL_CALL rtl_uriEncode(rtl_uString * pText, sal_Bool const * pCharClass,
rtl_UriEncodeMechanism eMechanism,
rtl_TextEncoding eCharset, rtl_uString ** pResult)
SAL_THROW_EXTERN_C()
{
assert(!pCharClass[0x25]); // make sure the percent sign is encoded...
sal_Unicode const * p = pText->buffer;
sal_Unicode const * pEnd = p + pText->length;
sal_Int32 nCapacity = 256;
rtl_uString_new_WithLength(pResult, nCapacity);
while (p < pEnd)
{
rtl::uri::detail::EscapeType eType;
sal_uInt32 nUtf32 = rtl::uri::detail::readUcs4(
&p, pEnd,
(eMechanism == rtl_UriEncodeKeepEscapes
|| eMechanism == rtl_UriEncodeCheckEscapes
|| eMechanism == rtl_UriEncodeStrictKeepEscapes),
eCharset, &eType);
switch (eType)
{
case rtl::uri::detail::EscapeNo:
if (isValid(pCharClass, nUtf32)) // implies nUtf32 <= 0x7F
{
writeUnicode(pResult, &nCapacity,
static_cast< sal_Unicode >(nUtf32));
}
else if (!writeEscapeChar(
pResult, &nCapacity, nUtf32, eCharset,
(eMechanism == rtl_UriEncodeStrict
|| eMechanism == rtl_UriEncodeStrictKeepEscapes)))
{
rtl_uString_new(pResult);
return;
}
break;
case rtl::uri::detail::EscapeChar:
if (eMechanism == rtl_UriEncodeCheckEscapes
&& isValid(pCharClass, nUtf32)) // implies nUtf32 <= 0x7F
{
writeUnicode(pResult, &nCapacity,
static_cast< sal_Unicode >(nUtf32));
}
else if (!writeEscapeChar(
pResult, &nCapacity, nUtf32, eCharset,
(eMechanism == rtl_UriEncodeStrict
|| eMechanism == rtl_UriEncodeStrictKeepEscapes)))
{
rtl_uString_new(pResult);
return;
}
break;
case rtl::uri::detail::EscapeOctet:
writeEscapeOctet(pResult, &nCapacity, nUtf32);
break;
}
}
*pResult = rtl_uStringBuffer_makeStringAndClear(pResult, &nCapacity);
}
void SAL_CALL rtl_uriDecode(rtl_uString * pText,
rtl_UriDecodeMechanism eMechanism,
rtl_TextEncoding eCharset, rtl_uString ** pResult)
SAL_THROW_EXTERN_C()
{
switch (eMechanism)
{
case rtl_UriDecodeNone:
rtl_uString_assign(pResult, pText);
break;
case rtl_UriDecodeToIuri:
eCharset = RTL_TEXTENCODING_UTF8;
[[fallthrough]];
default: // rtl_UriDecodeWithCharset, rtl_UriDecodeStrict
{
sal_Unicode const * p = pText->buffer;
sal_Unicode const * pEnd = p + pText->length;
sal_Int32 nCapacity = pText->length;
rtl_uString_new_WithLength(pResult, nCapacity);
while (p < pEnd)
{
rtl::uri::detail::EscapeType eType;
sal_uInt32 nUtf32 = rtl::uri::detail::readUcs4(&p, pEnd, true, eCharset, &eType);
switch (eType)
{
case rtl::uri::detail::EscapeChar:
if (nUtf32 <= 0x7F && eMechanism == rtl_UriDecodeToIuri)
{
writeEscapeOctet(pResult, &nCapacity, nUtf32);
break;
}
[[fallthrough]];
case rtl::uri::detail::EscapeNo:
writeUcs4(pResult, &nCapacity, nUtf32);
break;
case rtl::uri::detail::EscapeOctet:
if (eMechanism == rtl_UriDecodeStrict)
{
rtl_uString_new(pResult);
return;
}
writeEscapeOctet(pResult, &nCapacity, nUtf32);
break;
}
}
*pResult = rtl_uStringBuffer_makeStringAndClear( pResult, &nCapacity );
}
break;
}
}
sal_Bool SAL_CALL rtl_uriConvertRelToAbs(rtl_uString * pBaseUriRef,
rtl_uString * pRelUriRef,
rtl_uString ** pResult,
rtl_uString ** pException)
SAL_THROW_EXTERN_C()
{
// Use the strict parser algorithm from RFC 3986, section 5.2, to turn the
// relative URI into an absolute one:
Components aRelComponents;
parseUriRef(pRelUriRef, &aRelComponents);
OUStringBuffer aBuffer(256);
if (aRelComponents.aScheme.isPresent())
{
aBuffer.append(aRelComponents.aScheme.pBegin,
aRelComponents.aScheme.getLength());
if (aRelComponents.aAuthority.isPresent())
{
aBuffer.append(aRelComponents.aAuthority.pBegin,
aRelComponents.aAuthority.getLength());
}
appendPath(
aBuffer, aBuffer.getLength(), false, aRelComponents.aPath.pBegin,
aRelComponents.aPath.pEnd);
if (aRelComponents.aQuery.isPresent())
{
aBuffer.append(aRelComponents.aQuery.pBegin,
aRelComponents.aQuery.getLength());
}
}
else
{
Components aBaseComponents;
parseUriRef(pBaseUriRef, &aBaseComponents);
if (!aBaseComponents.aScheme.isPresent())
{
rtl_uString_assign(
pException,
(OUString(
"<" + OUString::unacquired(&pBaseUriRef)
+ "> does not start with a scheme component")
.pData));
return false;
}
aBuffer.append(aBaseComponents.aScheme.pBegin,
aBaseComponents.aScheme.getLength());
if (aRelComponents.aAuthority.isPresent())
{
aBuffer.append(aRelComponents.aAuthority.pBegin,
aRelComponents.aAuthority.getLength());
appendPath(
aBuffer, aBuffer.getLength(), false,
aRelComponents.aPath.pBegin, aRelComponents.aPath.pEnd);
if (aRelComponents.aQuery.isPresent())
{
aBuffer.append(aRelComponents.aQuery.pBegin,
aRelComponents.aQuery.getLength());
}
}
else
{
if (aBaseComponents.aAuthority.isPresent())
{
aBuffer.append(aBaseComponents.aAuthority.pBegin,
aBaseComponents.aAuthority.getLength());
}
if (aRelComponents.aPath.pBegin == aRelComponents.aPath.pEnd)
{
aBuffer.append(aBaseComponents.aPath.pBegin,
aBaseComponents.aPath.getLength());
if (aRelComponents.aQuery.isPresent())
{
aBuffer.append(aRelComponents.aQuery.pBegin,
aRelComponents.aQuery.getLength());
}
else if (aBaseComponents.aQuery.isPresent())
{
aBuffer.append(aBaseComponents.aQuery.pBegin,
aBaseComponents.aQuery.getLength());
}
}
else
{
if (*aRelComponents.aPath.pBegin == '/')
{
appendPath(
aBuffer, aBuffer.getLength(), false,
aRelComponents.aPath.pBegin, aRelComponents.aPath.pEnd);
}
else if (aBaseComponents.aAuthority.isPresent()
&& aBaseComponents.aPath.pBegin
== aBaseComponents.aPath.pEnd)
{
appendPath(
aBuffer, aBuffer.getLength(), true,
aRelComponents.aPath.pBegin, aRelComponents.aPath.pEnd);
}
else
{
sal_Int32 n = aBuffer.getLength();
sal_Int32 i = rtl_ustr_lastIndexOfChar_WithLength(
aBaseComponents.aPath.pBegin,
aBaseComponents.aPath.getLength(), '/');
if (i >= 0)
{
appendPath(
aBuffer, n, false, aBaseComponents.aPath.pBegin,
aBaseComponents.aPath.pBegin + i);
}
appendPath(
aBuffer, n, i >= 0, aRelComponents.aPath.pBegin,
aRelComponents.aPath.pEnd);
}
if (aRelComponents.aQuery.isPresent())
{
aBuffer.append(aRelComponents.aQuery.pBegin,
aRelComponents.aQuery.getLength());
}
}
}
}
if (aRelComponents.aFragment.isPresent())
{
aBuffer.append(aRelComponents.aFragment.pBegin,
aRelComponents.aFragment.getLength());
}
rtl_uString_assign(pResult, aBuffer.makeStringAndClear().pData);
return true;
}
/* vim:set shiftwidth=4 softtabstop=4 expandtab: */