mozilla
/
pjs
зеркало из https://github.com/mozilla/pjs.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

17275, 16285, 15204: Switch to using nsISaveAsCharset for both charset and entity conversion. r=nhotta

This commit is contained in:
akkana%netscape.com 1999-11-01 21:34:44 +00:00
Родитель ebcfb0c284
Коммит a6e83503f7
4 изменённых файлов: 196 добавлений и 296 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

231
htmlparser/src/nsHTMLContentSinkStream.cpp
Просмотреть файл

@ -29,18 +29,18 @@
#include "nsHTMLContentSinkStream.h" #include "nsHTMLContentSinkStream.h"
#include "nsHTMLTokens.h" #include "nsIParserNode.h"
#include <ctype.h> #include <ctype.h>
#include "nsString.h" #include "nsString.h"
#include "nsIParser.h" #include "nsIParser.h"
#include "nsHTMLEntities.h"
#include "nsCRT.h"
#include "nsIDocumentEncoder.h" // for output flags
#include "nsIUnicodeEncoder.h"
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsISaveAsCharset.h"
#include "nsIEntityConverter.h"
#include "nsCRT.h"
#include "nsIDocumentEncoder.h" // for output flags
#include "nshtmlpars.h"
#include "nsIOutputStream.h" #include "nsIOutputStream.h"
#include "nsFileStream.h" #include "nsFileStream.h"
@ -49,8 +49,7 @@ static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID);
static NS_DEFINE_IID(kIContentSinkIID, NS_ICONTENT_SINK_IID); static NS_DEFINE_IID(kIContentSinkIID, NS_ICONTENT_SINK_IID);
static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID); static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID);
static NS_DEFINE_IID(kIHTMLContentSinkStreamIID, NS_IHTMLCONTENTSINKSTREAM_IID); static NS_DEFINE_IID(kIHTMLContentSinkStreamIID, NS_IHTMLCONTENTSINKSTREAM_IID);
static NS_DEFINE_CID(kSaveAsCharsetCID, NS_SAVEASCHARSET_CID);
static NS_DEFINE_CID(kCharsetConverterManagerCID, NS_ICHARSETCONVERTERMANAGER_CID);
static char* gHeaderComment = "<!-- This page was created by the Gecko output system. -->"; static char* gHeaderComment = "<!-- This page was created by the Gecko output system. -->";
static char* gDocTypeHeader = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">"; static char* gDocTypeHeader = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">";
@ -245,56 +244,6 @@ NS_IMPL_ADDREF(nsHTMLContentSinkStream)
NS_IMPL_RELEASE(nsHTMLContentSinkStream) NS_IMPL_RELEASE(nsHTMLContentSinkStream)
/**
* Inits the encoder instance variable for the sink based on the charset
*
* @update gpk 4/21/99
* @param aCharset
* @return NS_xxx error result
*/
nsresult nsHTMLContentSinkStream::InitEncoder(const nsString& aCharset)
{
nsresult res = NS_OK;
nsICharsetAlias* calias = nsnull;
res = nsServiceManager::GetService(kCharsetAliasCID,
kICharsetAliasIID,
(nsISupports**)&calias);
NS_ASSERTION( nsnull != calias, "cannot find charet alias");
nsAutoString charsetName = aCharset;
if( NS_SUCCEEDED(res) && (nsnull != calias))
{
res = calias->GetPreferred(aCharset, charsetName);
nsServiceManager::ReleaseService(kCharsetAliasCID, calias);
if(NS_FAILED(res))
{
// failed - unknown alias , fallback to ISO-8859-1
charsetName = "ISO-8859-1";
}
nsICharsetConverterManager * ccm = nsnull;
res = nsServiceManager::GetService(kCharsetConverterManagerCID,
nsCOMTypeInfo<nsICharsetConverterManager>::GetIID(),
(nsISupports**)&ccm);
if(NS_SUCCEEDED(res) && (nsnull != ccm))
{
nsIUnicodeEncoder * encoder = nsnull;
res = ccm->GetUnicodeEncoder(&charsetName, &encoder);
if(NS_SUCCEEDED(res) && (nsnull != encoder))
{
NS_IF_RELEASE(mUnicodeEncoder);
mUnicodeEncoder = encoder;
}
nsServiceManager::ReleaseService(kCharsetConverterManagerCID, ccm);
}
}
return res;
}
/** /**
* Construct a content sink stream. * Construct a content sink stream.
* @update gess7/7/98 * @update gess7/7/98
@ -309,11 +258,10 @@ nsHTMLContentSinkStream::nsHTMLContentSinkStream()
mHTMLStackPos = 0; mHTMLStackPos = 0;
mColPos = 0; mColPos = 0;
mIndent = 0; mIndent = 0;
mUnicodeEncoder = nsnull;
mInBody = PR_FALSE; mInBody = PR_FALSE;
mBuffer = nsnull; mBuffer = nsnull;
mBufferSize=0; mBufferSize = 0;
mBufferLength=0; mBufferLength = 0;
} }
NS_IMETHODIMP NS_IMETHODIMP
@ -332,7 +280,7 @@ nsHTMLContentSinkStream::Initialize(nsIOutputStream* aOutStream,
mStream = aOutStream; mStream = aOutStream;
mString = aOutString; mString = aOutString;
if (aCharsetOverride != nsnull) if (aCharsetOverride != nsnull)
mCharsetOverride = *aCharsetOverride; mCharsetOverride.Assign(*aCharsetOverride);
return NS_OK; return NS_OK;
} }
@ -377,85 +325,89 @@ nsHTMLContentSinkStream::EndContext(PRInt32 aPosition)
return NS_OK; return NS_OK;
} }
/**
/* * Initialize the Unicode encoder with our current mCharsetOverride.
* Entities are represented in the dom as single elements.
* Substitute them back into entity for (e.g. &acute;) here.
*/ */
void nsHTMLContentSinkStream::UnicodeToHTMLString(const nsString& aSrc, NS_IMETHODIMP
nsString& aDst) nsHTMLContentSinkStream::InitEncoder()
{ {
PRInt32 length = aSrc.Length(); nsAutoString charsetName = mCharsetOverride;
PRUnichar ch; nsresult res;
nsICharsetAlias* calias = nsnull;
if (mUnicodeEncoder == nsnull) res = nsServiceManager::GetService(kCharsetAliasCID,
InitEncoder(""); kICharsetAliasIID,
(nsISupports**)&calias);
if (length > 0)
{ NS_ASSERTION(nsnull != calias, "cannot find charset alias");
// Convert anything that maps to character entity if(NS_SUCCEEDED(res) && (nsnull != calias))
// to the entity value {
EnsureBufferSize(length); res = calias->GetPreferred(mCharsetOverride, charsetName);
nsServiceManager::ReleaseService(kCharsetAliasCID, calias);
for (PRInt32 i = 0; i < length; i++)
{
ch = aSrc.CharAt(i);
const nsCString& entity = nsHTMLEntities::UnicodeToEntity(ch);
if (0 < entity.Length())
{
aDst.Append('&');
aDst.Append(entity);
aDst.Append(';');
}
else
{
aDst.Append(ch);
}
}
} }
if (NS_FAILED(res))
{
// failed - unknown alias , fallback to ISO-8859-1
charsetName = "ISO-8859-1";
}
res = nsComponentManager::CreateInstance(kSaveAsCharsetCID, NULL,
nsISaveAsCharset::GetIID(),
getter_AddRefs(mUnicodeEncoder));
if (NS_FAILED(res))
return res;
// SaveAsCharset requires a const char* in its first argument:
nsCAutoString charsetCString (charsetName);
res = mUnicodeEncoder->Init(charsetCString,
nsISaveAsCharset::attr_EntityBeforeCharsetConv
| nsISaveAsCharset::attr_FallbackDecimalNCR,
nsIEntityConverter::html40);
return res;
} }
/**
void nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc) * Use mUnicodeEncoder to encode to the buffer;
* this also encodes entities, so it's useful even for the default charset.
*
* @param aSrc - the string to be encoded.
*/
void
nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
nsString htmlstr; char *encodedBuffer = nsnull;
UnicodeToHTMLString(aSrc, htmlstr); nsresult res;
NS_VERIFY(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized"); // Initialize the encoder if we haven't already
if (mUnicodeEncoder == nsnull) if (!mUnicodeEncoder)
return; InitEncoder();
PRInt32 length = htmlstr.Length(); // if (mBuffer)
nsresult result; // nsAllocator::Free(mBuffer);
if (mUnicodeEncoder != nsnull && length > 0) if (mUnicodeEncoder)
{ {
EnsureBufferSize(length); // Call the converter to convert to the target charset.
mBufferLength = mBufferSize; // Convert() takes a char* output param even though it's writing unicode.
res = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &encodedBuffer);
mUnicodeEncoder->Reset(); if (!NS_SUCCEEDED(res))
result = mUnicodeEncoder->Convert(htmlstr.GetUnicode(), &length,
mBuffer, &mBufferLength);
mBuffer[mBufferLength] = 0;
PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp);
#if 0
// Do some conversions to make up for the unicode encoder's foibles:
PRInt32 nbsp = nsHTMLEntities::EntityToUnicode(nsCAutoString("nbsp"));
PRInt32 quot = nsHTMLEntities::EntityToUnicode(nsCAutoString("quot"));
for (PRInt32 i = 0; i < mBufferLength; i++)
{ {
if (mBuffer[i] == quot) #ifdef DEBUG_akkana
mBuffer[i] = '"'; printf("Unicode convert didn't work!\n");
// I don't know why this nbsp mapping was here ...
else if (mBuffer[i] == nbsp)
mBuffer[i] = ' ';
}
#endif #endif
encodedBuffer = aSrc.ToNewUTF8String();
}
}
else
{
#ifdef DEBUG_akkana
printf("Unicode convert didn't work!\n");
#endif
encodedBuffer = aSrc.ToNewCString();
}
mBufferLength = 0;
if (encodedBuffer) {
mBufferLength = nsCRT::strlen(encodedBuffer);
EnsureBufferSize(mBufferLength+1);
nsCRT::memcpy(mBuffer, encodedBuffer, mBufferLength+1);
} }
} }
@ -474,7 +426,7 @@ void nsHTMLContentSinkStream::Write(const nsString& aString)
// Now handle the stream case: // Now handle the stream case:
nsOutputStream out(mStream); nsOutputStream out(mStream);
// If a encoder is being used then convert first convert the input string // If an encoder is being used then convert first convert the input string
if (mUnicodeEncoder) if (mUnicodeEncoder)
{ {
EncodeToBuffer(aString); EncodeToBuffer(aString);
@ -527,9 +479,10 @@ void nsHTMLContentSinkStream::Write(char aData)
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::~nsHTMLContentSinkStream() { nsHTMLContentSinkStream::~nsHTMLContentSinkStream()
NS_IF_RELEASE(mUnicodeEncoder); {
if(mBuffer) delete [] mBuffer; if (mBuffer)
nsAllocator::Free(mBuffer);
} }
@ -984,8 +937,9 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode)
if (tag == eHTMLTag_body) if (tag == eHTMLTag_body)
mInBody = PR_FALSE; mInBody = PR_FALSE;
if ((mDoFormat && BreakAfterClose(tag)) || tag == eHTMLTag_html) if ((mDoFormat && BreakAfterClose(tag))
|| tag == eHTMLTag_body || tag == eHTMLTag_html)
{ {
Write(NS_LINEBREAK); Write(NS_LINEBREAK);
mColPos = 0; mColPos = 0;
@ -1206,9 +1160,8 @@ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
if (key.Equals("charset")) if (key.Equals("charset"))
{ {
if (mCharsetOverride.Length() == 0) if (mCharsetOverride.Length() == 0)
InitEncoder(value); mCharsetOverride.Assign(value);
else InitEncoder();
InitEncoder(mCharsetOverride);
} }
} }
} }

15
htmlparser/src/nsHTMLContentSinkStream.h
Просмотреть файл

@ -39,12 +39,10 @@
#ifndef NS_TXTCONTENTSINK_STREAM #ifndef NS_TXTCONTENTSINK_STREAM
#define NS_TXTCONTENTSINK_STREAM #define NS_TXTCONTENTSINK_STREAM
#include "nsIParserNode.h"
#include "nsIHTMLContentSink.h" #include "nsIHTMLContentSink.h"
#include "nshtmlpars.h"
#include "nsHTMLTokens.h"
#include "nsParserCIID.h" #include "nsParserCIID.h"
#include "nsCOMPtr.h" #include "nsCOMPtr.h"
#include "nsHTMLTokens.h" // for eHTMLTags
#define NS_IHTMLCONTENTSINKSTREAM_IID \ #define NS_IHTMLCONTENTSINKSTREAM_IID \
{0xa39c6bff, 0x15f0, 0x11d2, \ {0xa39c6bff, 0x15f0, 0x11d2, \
@ -54,7 +52,8 @@
class ostream; class ostream;
#endif #endif
class nsIUnicodeEncoder; class nsIParserNode;
class nsISaveAsCharset;
class nsIOutputStream; class nsIOutputStream;
class nsIHTMLContentSinkStream : public nsIHTMLContentSink { class nsIHTMLContentSinkStream : public nsIHTMLContentSink {
@ -149,10 +148,8 @@ protected:
void AddIndent(); void AddIndent();
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
nsresult InitEncoder(const nsString& aCharset);
void UnicodeToHTMLString(const nsString& aSrc, nsString& aDst);
void EncodeToBuffer(const nsString& aString); void EncodeToBuffer(const nsString& aString);
NS_IMETHOD InitEncoder();
void Write(const nsString& aString); void Write(const nsString& aString);
void Write(const char* aCharBuffer); void Write(const char* aCharBuffer);
@ -180,8 +177,8 @@ protected:
PRBool mDoHeader; PRBool mDoHeader;
PRBool mBodyOnly; PRBool mBodyOnly;
nsIUnicodeEncoder* mUnicodeEncoder; nsCOMPtr<nsISaveAsCharset> mUnicodeEncoder;
nsString mCharsetOverride; nsCAutoString mCharsetOverride;
}; };

231
parser/htmlparser/src/nsHTMLContentSinkStream.cpp
Просмотреть файл

@ -29,18 +29,18 @@
#include "nsHTMLContentSinkStream.h" #include "nsHTMLContentSinkStream.h"
#include "nsHTMLTokens.h" #include "nsIParserNode.h"
#include <ctype.h> #include <ctype.h>
#include "nsString.h" #include "nsString.h"
#include "nsIParser.h" #include "nsIParser.h"
#include "nsHTMLEntities.h"
#include "nsCRT.h"
#include "nsIDocumentEncoder.h" // for output flags
#include "nsIUnicodeEncoder.h"
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsISaveAsCharset.h"
#include "nsIEntityConverter.h"
#include "nsCRT.h"
#include "nsIDocumentEncoder.h" // for output flags
#include "nshtmlpars.h"
#include "nsIOutputStream.h" #include "nsIOutputStream.h"
#include "nsFileStream.h" #include "nsFileStream.h"
@ -49,8 +49,7 @@ static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID);
static NS_DEFINE_IID(kIContentSinkIID, NS_ICONTENT_SINK_IID); static NS_DEFINE_IID(kIContentSinkIID, NS_ICONTENT_SINK_IID);
static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID); static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID);
static NS_DEFINE_IID(kIHTMLContentSinkStreamIID, NS_IHTMLCONTENTSINKSTREAM_IID); static NS_DEFINE_IID(kIHTMLContentSinkStreamIID, NS_IHTMLCONTENTSINKSTREAM_IID);
static NS_DEFINE_CID(kSaveAsCharsetCID, NS_SAVEASCHARSET_CID);
static NS_DEFINE_CID(kCharsetConverterManagerCID, NS_ICHARSETCONVERTERMANAGER_CID);
static char* gHeaderComment = "<!-- This page was created by the Gecko output system. -->"; static char* gHeaderComment = "<!-- This page was created by the Gecko output system. -->";
static char* gDocTypeHeader = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">"; static char* gDocTypeHeader = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">";
@ -245,56 +244,6 @@ NS_IMPL_ADDREF(nsHTMLContentSinkStream)
NS_IMPL_RELEASE(nsHTMLContentSinkStream) NS_IMPL_RELEASE(nsHTMLContentSinkStream)
/**
* Inits the encoder instance variable for the sink based on the charset
*
* @update gpk 4/21/99
* @param aCharset
* @return NS_xxx error result
*/
nsresult nsHTMLContentSinkStream::InitEncoder(const nsString& aCharset)
{
nsresult res = NS_OK;
nsICharsetAlias* calias = nsnull;
res = nsServiceManager::GetService(kCharsetAliasCID,
kICharsetAliasIID,
(nsISupports**)&calias);
NS_ASSERTION( nsnull != calias, "cannot find charet alias");
nsAutoString charsetName = aCharset;
if( NS_SUCCEEDED(res) && (nsnull != calias))
{
res = calias->GetPreferred(aCharset, charsetName);
nsServiceManager::ReleaseService(kCharsetAliasCID, calias);
if(NS_FAILED(res))
{
// failed - unknown alias , fallback to ISO-8859-1
charsetName = "ISO-8859-1";
}
nsICharsetConverterManager * ccm = nsnull;
res = nsServiceManager::GetService(kCharsetConverterManagerCID,
nsCOMTypeInfo<nsICharsetConverterManager>::GetIID(),
(nsISupports**)&ccm);
if(NS_SUCCEEDED(res) && (nsnull != ccm))
{
nsIUnicodeEncoder * encoder = nsnull;
res = ccm->GetUnicodeEncoder(&charsetName, &encoder);
if(NS_SUCCEEDED(res) && (nsnull != encoder))
{
NS_IF_RELEASE(mUnicodeEncoder);
mUnicodeEncoder = encoder;
}
nsServiceManager::ReleaseService(kCharsetConverterManagerCID, ccm);
}
}
return res;
}
/** /**
* Construct a content sink stream. * Construct a content sink stream.
* @update gess7/7/98 * @update gess7/7/98
@ -309,11 +258,10 @@ nsHTMLContentSinkStream::nsHTMLContentSinkStream()
mHTMLStackPos = 0; mHTMLStackPos = 0;
mColPos = 0; mColPos = 0;
mIndent = 0; mIndent = 0;
mUnicodeEncoder = nsnull;
mInBody = PR_FALSE; mInBody = PR_FALSE;
mBuffer = nsnull; mBuffer = nsnull;
mBufferSize=0; mBufferSize = 0;
mBufferLength=0; mBufferLength = 0;
} }
NS_IMETHODIMP NS_IMETHODIMP
@ -332,7 +280,7 @@ nsHTMLContentSinkStream::Initialize(nsIOutputStream* aOutStream,
mStream = aOutStream; mStream = aOutStream;
mString = aOutString; mString = aOutString;
if (aCharsetOverride != nsnull) if (aCharsetOverride != nsnull)
mCharsetOverride = *aCharsetOverride; mCharsetOverride.Assign(*aCharsetOverride);
return NS_OK; return NS_OK;
} }
@ -377,85 +325,89 @@ nsHTMLContentSinkStream::EndContext(PRInt32 aPosition)
return NS_OK; return NS_OK;
} }
/**
/* * Initialize the Unicode encoder with our current mCharsetOverride.
* Entities are represented in the dom as single elements.
* Substitute them back into entity for (e.g. &acute;) here.
*/ */
void nsHTMLContentSinkStream::UnicodeToHTMLString(const nsString& aSrc, NS_IMETHODIMP
nsString& aDst) nsHTMLContentSinkStream::InitEncoder()
{ {
PRInt32 length = aSrc.Length(); nsAutoString charsetName = mCharsetOverride;
PRUnichar ch; nsresult res;
nsICharsetAlias* calias = nsnull;
if (mUnicodeEncoder == nsnull) res = nsServiceManager::GetService(kCharsetAliasCID,
InitEncoder(""); kICharsetAliasIID,
(nsISupports**)&calias);
if (length > 0)
{ NS_ASSERTION(nsnull != calias, "cannot find charset alias");
// Convert anything that maps to character entity if(NS_SUCCEEDED(res) && (nsnull != calias))
// to the entity value {
EnsureBufferSize(length); res = calias->GetPreferred(mCharsetOverride, charsetName);
nsServiceManager::ReleaseService(kCharsetAliasCID, calias);
for (PRInt32 i = 0; i < length; i++)
{
ch = aSrc.CharAt(i);
const nsCString& entity = nsHTMLEntities::UnicodeToEntity(ch);
if (0 < entity.Length())
{
aDst.Append('&');
aDst.Append(entity);
aDst.Append(';');
}
else
{
aDst.Append(ch);
}
}
} }
if (NS_FAILED(res))
{
// failed - unknown alias , fallback to ISO-8859-1
charsetName = "ISO-8859-1";
}
res = nsComponentManager::CreateInstance(kSaveAsCharsetCID, NULL,
nsISaveAsCharset::GetIID(),
getter_AddRefs(mUnicodeEncoder));
if (NS_FAILED(res))
return res;
// SaveAsCharset requires a const char* in its first argument:
nsCAutoString charsetCString (charsetName);
res = mUnicodeEncoder->Init(charsetCString,
nsISaveAsCharset::attr_EntityBeforeCharsetConv
| nsISaveAsCharset::attr_FallbackDecimalNCR,
nsIEntityConverter::html40);
return res;
} }
/**
void nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc) * Use mUnicodeEncoder to encode to the buffer;
* this also encodes entities, so it's useful even for the default charset.
*
* @param aSrc - the string to be encoded.
*/
void
nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
nsString htmlstr; char *encodedBuffer = nsnull;
UnicodeToHTMLString(aSrc, htmlstr); nsresult res;
NS_VERIFY(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized"); // Initialize the encoder if we haven't already
if (mUnicodeEncoder == nsnull) if (!mUnicodeEncoder)
return; InitEncoder();
PRInt32 length = htmlstr.Length(); // if (mBuffer)
nsresult result; // nsAllocator::Free(mBuffer);
if (mUnicodeEncoder != nsnull && length > 0) if (mUnicodeEncoder)
{ {
EnsureBufferSize(length); // Call the converter to convert to the target charset.
mBufferLength = mBufferSize; // Convert() takes a char* output param even though it's writing unicode.
res = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &encodedBuffer);
mUnicodeEncoder->Reset(); if (!NS_SUCCEEDED(res))
result = mUnicodeEncoder->Convert(htmlstr.GetUnicode(), &length,
mBuffer, &mBufferLength);
mBuffer[mBufferLength] = 0;
PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp);
#if 0
// Do some conversions to make up for the unicode encoder's foibles:
PRInt32 nbsp = nsHTMLEntities::EntityToUnicode(nsCAutoString("nbsp"));
PRInt32 quot = nsHTMLEntities::EntityToUnicode(nsCAutoString("quot"));
for (PRInt32 i = 0; i < mBufferLength; i++)
{ {
if (mBuffer[i] == quot) #ifdef DEBUG_akkana
mBuffer[i] = '"'; printf("Unicode convert didn't work!\n");
// I don't know why this nbsp mapping was here ...
else if (mBuffer[i] == nbsp)
mBuffer[i] = ' ';
}
#endif #endif
encodedBuffer = aSrc.ToNewUTF8String();
}
}
else
{
#ifdef DEBUG_akkana
printf("Unicode convert didn't work!\n");
#endif
encodedBuffer = aSrc.ToNewCString();
}
mBufferLength = 0;
if (encodedBuffer) {
mBufferLength = nsCRT::strlen(encodedBuffer);
EnsureBufferSize(mBufferLength+1);
nsCRT::memcpy(mBuffer, encodedBuffer, mBufferLength+1);
} }
} }
@ -474,7 +426,7 @@ void nsHTMLContentSinkStream::Write(const nsString& aString)
// Now handle the stream case: // Now handle the stream case:
nsOutputStream out(mStream); nsOutputStream out(mStream);
// If a encoder is being used then convert first convert the input string // If an encoder is being used then convert first convert the input string
if (mUnicodeEncoder) if (mUnicodeEncoder)
{ {
EncodeToBuffer(aString); EncodeToBuffer(aString);
@ -527,9 +479,10 @@ void nsHTMLContentSinkStream::Write(char aData)
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::~nsHTMLContentSinkStream() { nsHTMLContentSinkStream::~nsHTMLContentSinkStream()
NS_IF_RELEASE(mUnicodeEncoder); {
if(mBuffer) delete [] mBuffer; if (mBuffer)
nsAllocator::Free(mBuffer);
} }
@ -984,8 +937,9 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode)
if (tag == eHTMLTag_body) if (tag == eHTMLTag_body)
mInBody = PR_FALSE; mInBody = PR_FALSE;
if ((mDoFormat && BreakAfterClose(tag)) || tag == eHTMLTag_html) if ((mDoFormat && BreakAfterClose(tag))
|| tag == eHTMLTag_body || tag == eHTMLTag_html)
{ {
Write(NS_LINEBREAK); Write(NS_LINEBREAK);
mColPos = 0; mColPos = 0;
@ -1206,9 +1160,8 @@ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
if (key.Equals("charset")) if (key.Equals("charset"))
{ {
if (mCharsetOverride.Length() == 0) if (mCharsetOverride.Length() == 0)
InitEncoder(value); mCharsetOverride.Assign(value);
else InitEncoder();
InitEncoder(mCharsetOverride);
} }
} }
} }

15
parser/htmlparser/src/nsHTMLContentSinkStream.h
Просмотреть файл

@ -39,12 +39,10 @@
#ifndef NS_TXTCONTENTSINK_STREAM #ifndef NS_TXTCONTENTSINK_STREAM
#define NS_TXTCONTENTSINK_STREAM #define NS_TXTCONTENTSINK_STREAM
#include "nsIParserNode.h"
#include "nsIHTMLContentSink.h" #include "nsIHTMLContentSink.h"
#include "nshtmlpars.h"
#include "nsHTMLTokens.h"
#include "nsParserCIID.h" #include "nsParserCIID.h"
#include "nsCOMPtr.h" #include "nsCOMPtr.h"
#include "nsHTMLTokens.h" // for eHTMLTags
#define NS_IHTMLCONTENTSINKSTREAM_IID \ #define NS_IHTMLCONTENTSINKSTREAM_IID \
{0xa39c6bff, 0x15f0, 0x11d2, \ {0xa39c6bff, 0x15f0, 0x11d2, \
@ -54,7 +52,8 @@
class ostream; class ostream;
#endif #endif
class nsIUnicodeEncoder; class nsIParserNode;
class nsISaveAsCharset;
class nsIOutputStream; class nsIOutputStream;
class nsIHTMLContentSinkStream : public nsIHTMLContentSink { class nsIHTMLContentSinkStream : public nsIHTMLContentSink {
@ -149,10 +148,8 @@ protected:
void AddIndent(); void AddIndent();
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
nsresult InitEncoder(const nsString& aCharset);
void UnicodeToHTMLString(const nsString& aSrc, nsString& aDst);
void EncodeToBuffer(const nsString& aString); void EncodeToBuffer(const nsString& aString);
NS_IMETHOD InitEncoder();
void Write(const nsString& aString); void Write(const nsString& aString);
void Write(const char* aCharBuffer); void Write(const char* aCharBuffer);
@ -180,8 +177,8 @@ protected:
PRBool mDoHeader; PRBool mDoHeader;
PRBool mBodyOnly; PRBool mBodyOnly;
nsIUnicodeEncoder* mUnicodeEncoder; nsCOMPtr<nsISaveAsCharset> mUnicodeEncoder;
nsString mCharsetOverride; nsCAutoString mCharsetOverride;
}; };