Magic-Mirror/gecko-dev
1
0
Fork 0
mirror of https://github.com/mozilla/gecko-dev.git synced 2025-05-13 08:57:27 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Added support to encode the document as either a nsString or an nsIOutputStream.

Browse Source 
In the case of the stream, the document is encoded using the document defined charset
or the user may override the charset.

Added constructors for creating a sink using a nsIOutputStream or an nsString.

Removed all HTML and TEXT encoding which relied on references to ostream. Using John McMullen's
platform independent streaming code in its place. :)
This commit is contained in:
kostello%netscape.com 1999-05-03 18:41:23 +00:00
parent d4a48d70d3
commit dbc32c5326
8 changed files with 960 additions and 934 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

567
htmlparser/src/nsHTMLContentSinkStream.cpp
View File

@ -42,6 +42,8 @@
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsICharsetConverterManager.h"
#include "nsIOutputStream.h"
#include "nsFileStream.h"
static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID); static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID);
@ -51,7 +53,6 @@ static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID);
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\">";
const int gTabSize=2; const int gTabSize=2;
static char gBuffer[1024];
static const char* UnicodeToEntity(PRInt32 aCode); static const char* UnicodeToEntity(PRInt32 aCode);
@ -332,18 +333,23 @@ NS_IMPL_RELEASE(nsHTMLContentSinkStream)
/** /**
* This method is defined in nsIParser. It is used to * Create an new sink
* cause the COM-like construction of an nsParser.
* *
* @update gess 4/8/98 * @update gpk 05/01/99
* @param nsIParser** ptr to newly instantiated parser
* @return NS_xxx error result * @return NS_xxx error result
*/ */
NS_HTMLPARS nsresult NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult, NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
PRBool aDoFormat, nsIOutputStream* aOutStream,
PRBool aDoHeader) { const nsString* aCharsetOverride,
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(aDoFormat,aDoHeader); PRBool aDoFormat,
PRBool aDoHeader)
{
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(aOutStream,
nsnull,
aCharsetOverride,
aDoFormat,
aDoHeader);
if (nsnull == it) { if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY; return NS_ERROR_OUT_OF_MEMORY;
} }
@ -352,6 +358,32 @@ NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
} }
/**
* Create an new sink
*
* @update gpk 05/01/99
* @return NS_xxx error result
*/
NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString,
PRBool aDoFormat,
PRBool aDoHeader)
{
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(nsnull,
aOutString,
nsnull,
aDoFormat,
aDoHeader);
if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY;
}
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
}
/** /**
* Inits the encoder instance variable for the sink based on the charset * Inits the encoder instance variable for the sink based on the charset
@ -409,30 +441,12 @@ nsresult nsHTMLContentSinkStream::InitEncoder(const nsString& aCharset)
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::nsHTMLContentSinkStream(PRBool aDoFormat,PRBool aDoHeader) { nsHTMLContentSinkStream::nsHTMLContentSinkStream(nsIOutputStream* aOutStream,
nsString* aOutString,
const nsString* aCharsetOverride,
PRBool aDoFormat,
PRBool aDoHeader) {
NS_INIT_REFCNT(); NS_INIT_REFCNT();
mOutput=&cout;
mLowerCaseTags = PR_TRUE;
memset(mHTMLTagStack,0,sizeof(mHTMLTagStack));
mHTMLStackPos = 0;
mColPos = 0;
mIndent = 0;
mDoFormat = aDoFormat;
mDoHeader = aDoHeader;
mBuffer = nsnull;
mBufferSize = 0;
mUnicodeEncoder = nsnull;
}
/**
* Construct a content sink stream.
* @update gess7/7/98
* @param
* @return
*/
nsHTMLContentSinkStream::nsHTMLContentSinkStream(ostream& aStream,PRBool aDoFormat,PRBool aDoHeader) {
NS_INIT_REFCNT();
mOutput = &aStream;
mLowerCaseTags = PR_TRUE; mLowerCaseTags = PR_TRUE;
memset(mHTMLTagStack,0,sizeof(mHTMLTagStack)); memset(mHTMLTagStack,0,sizeof(mHTMLTagStack));
mHTMLStackPos = 0; mHTMLStackPos = 0;
@ -443,6 +457,10 @@ nsHTMLContentSinkStream::nsHTMLContentSinkStream(ostream& aStream,PRBool aDoForm
mBuffer = nsnull; mBuffer = nsnull;
mBufferSize = 0; mBufferSize = 0;
mUnicodeEncoder = nsnull; mUnicodeEncoder = nsnull;
mStream = aOutStream;
mString = aOutString;
if (aCharsetOverride != nsnull)
mCharsetOverride = *aCharsetOverride;
} }
@ -498,82 +516,119 @@ void nsHTMLContentSinkStream::EnsureBufferSize(PRInt32 aNewSize)
} }
} }
void nsHTMLContentSinkStream::UnicodeToHTMLString(const nsString& aSrc)
void nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
PRInt32 length = aSrc.Length();
PRUnichar ch; NS_ASSERTION(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized");
const char* entity = nsnull;
PRUint32 offset = 0;
PRUint32 addedLength = 0;
nsAutoString data;
if (mUnicodeEncoder == nsnull) if (mUnicodeEncoder == nsnull)
InitEncoder(""); return;
if (length > 0) #define CH_NBSP 160
PRInt32 length = aSrc.Length();
nsresult result;
if (mUnicodeEncoder != nsnull && length > 0)
{ {
// Step 1. Convert anything that maps to character entity to
// the entity value
EnsureBufferSize(length); EnsureBufferSize(length);
for (PRInt32 i = 0; i < length; i++) mBufferLength = mBufferSize;
{
ch = aSrc.CharAt(i);
entity = UnicodeToEntity(ch);
if (entity)
{
nsAutoString temp(entity);
temp.ToLowerCase();
data.Append('&');
data.Append(temp);
data.Append(';');
}
else
{
data.Append(ch);
}
}
// Step 2. Run the result through the converter
length = data.Length();
EnsureBufferSize(length);
PRInt32 bufferLength = mBufferSize;
mUnicodeEncoder->Reset(); mUnicodeEncoder->Reset();
nsresult result = mUnicodeEncoder->Convert(data.GetUnicode(), &length, mBuffer, &bufferLength); result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &mBufferLength);
mBuffer[bufferLength] = 0; mBuffer[mBufferLength] = 0;
PRInt32 temp = bufferLength; PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result)) if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp); result = mUnicodeEncoder->Finish(mBuffer,&temp);
}
for (PRInt32 i = 0; i < mBufferLength; i++)
{
if (mBuffer[i] == char(CH_NBSP))
mBuffer[i] = ' ';
}
}
} }
void nsHTMLContentSinkStream::Write(const nsString& aString)
{
// If a encoder is being used then convert first convert the input string
if (mUnicodeEncoder != nsnull)
{
EncodeToBuffer(aString);
if (mStream != nsnull)
{
nsOutputStream out(mStream);
out.write(mBuffer,mBufferLength);
}
if (mString != nsnull)
{
mString->Append(mBuffer);
}
}
else
{
if (mStream != nsnull)
{
nsOutputStream out(mStream);
const PRUnichar* unicode = aString.GetUnicode();
PRUint32 length = aString.Length();
out.write(unicode,length);
}
else
{
mString->Append(aString);
}
}
}
void nsHTMLContentSinkStream::Write(const char* aData)
{
if (mStream)
{
nsOutputStream out(mStream);
out << aData;
}
if (mString)
{
mString->Append(aData);
}
}
void nsHTMLContentSinkStream::Write(char aData)
{
if (mStream)
{
nsOutputStream out(mStream);
out << aData;
}
if (mString)
{
mString->Append(aData);
}
}
/** /**
* *
* @update gess7/7/98 * @update 04/30/99 gpk
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::~nsHTMLContentSinkStream() { nsHTMLContentSinkStream::~nsHTMLContentSinkStream() {
NS_IF_RELEASE(mUnicodeEncoder); NS_IF_RELEASE(mUnicodeEncoder);
mOutput=0; //we don't own the stream we're given; just forget it.
} }
/**
*
* @update gess7/22/98
* @param
* @return
*/
NS_IMETHODIMP_(void)
nsHTMLContentSinkStream::SetOutputStream(ostream& aStream){
mOutput=&aStream;
}
/** /**
* *
@ -581,7 +636,7 @@ nsHTMLContentSinkStream::SetOutputStream(ostream& aStream){
* @param * @param
* @return * @return
*/ */
void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream& aStream) { void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode) {
int theCount=aNode.GetAttributeCount(); int theCount=aNode.GetAttributeCount();
if(theCount) { if(theCount) {
int i=0; int i=0;
@ -600,14 +655,20 @@ void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream
EnsureBufferSize(key.Length()); EnsureBufferSize(key.Length());
key.ToCString(mBuffer,mBufferSize); key.ToCString(mBuffer,mBufferSize);
aStream << " " << mBuffer << char(kEqual); // send to ouput " [KEY]="
Write(' ');
Write(mBuffer);
Write(char(kEqual));
mColPos += 1 + strlen(mBuffer) + 1; mColPos += 1 + strlen(mBuffer) + 1;
const nsString& value=aNode.GetValueAt(i); const nsString& value=aNode.GetValueAt(i);
UnicodeToHTMLString(value);
// send to ouput "\"[VALUE]\""
Write('\"');
Write(value);
Write('\"');
aStream << "\"" << mBuffer << "\"";
mColPos += 1 + strlen(mBuffer) + 1; mColPos += 1 + strlen(mBuffer) + 1;
} }
} }
@ -615,85 +676,31 @@ void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream
} }
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void OpenTagWithAttributes(const char* theTag,const nsIParserNode& aNode,int tab,ostream& aStream,PRBool aNewline) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << theTag;
// WriteAttributes(aNode,aStream);
aStream << (char)kGreaterThan;
if(aNewline)
aStream << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void OpenTag(const char* theTag,int tab,ostream& aStream,PRBool aNewline) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << theTag << (char)kGreaterThan;
if(aNewline)
aStream << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void CloseTag(const char* theTag,int tab,ostream& aStream) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << (char)kForwardSlash << theTag << (char)kGreaterThan << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void WritePair(eHTMLTags aTag,const nsString& theContent,int tab,ostream& aStream) {
const char* titleStr = GetTagName(aTag);
OpenTag(titleStr,tab,aStream,PR_FALSE);
theContent.ToCString(gBuffer,sizeof(gBuffer)-1);
aStream << gBuffer;
CloseTag(titleStr,0,aStream);
}
/** /**
* This method gets called by the parser when it encounters * This method gets called by the parser when it encounters
* a title tag and wants to set the document title in the sink. * a title tag and wants to set the document title in the sink.
* *
* @update 4/1/98 gess * @update 04/30/99 gpk
* @param nsString reference to new title value * @param nsString reference to new title value
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::SetTitle(const nsString& aValue){ nsHTMLContentSinkStream::SetTitle(const nsString& aValue){
if(mOutput) { const char* tagName = GetTagName(eHTMLTag_title);
WritePair(eHTMLTag_title,aValue,1+mTabLevel,*mOutput); Write(kLessThan);
} Write(tagName);
Write(kGreaterThan);
Write(aValue);
Write(kLessThan);
Write(kForwardSlash);
Write(tagName);
Write(kGreaterThan);
return NS_OK; return NS_OK;
} }
@ -701,17 +708,15 @@ nsHTMLContentSinkStream::SetTitle(const nsString& aValue){
/** /**
* This method is used to open the outer HTML container. * This method is used to open the outer HTML container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){
if(mOutput) {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
if (tag == eHTMLTag_html) if (tag == eHTMLTag_html)
AddStartTag(aNode,*mOutput); AddStartTag(aNode);
}
return NS_OK; return NS_OK;
} }
@ -719,18 +724,15 @@ nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){
/** /**
* This method is used to close the outer HTML container. * This method is used to close the outer HTML container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_html)
if (tag == eHTMLTag_html) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
mOutput->flush();
}
return NS_OK; return NS_OK;
} }
@ -738,17 +740,15 @@ nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){
/** /**
* This method is used to open the only HEAD container. * This method is used to open the only HEAD container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_head)
if (tag == eHTMLTag_head) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -756,17 +756,15 @@ nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){
/** /**
* This method is used to close the only HEAD container. * This method is used to close the only HEAD container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_head)
if (tag == eHTMLTag_head) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -774,17 +772,15 @@ nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){
/** /**
* This method is used to open the main BODY container. * This method is used to open the main BODY container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_body)
if (tag == eHTMLTag_body) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -792,17 +788,15 @@ nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){
/** /**
* This method is used to close the main BODY container. * This method is used to close the main BODY container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_body)
if (tag == eHTMLTag_body) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -816,11 +810,9 @@ nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_form)
if (tag == eHTMLTag_form) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -834,11 +826,9 @@ nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_form)
if (tag == eHTMLTag_form) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -851,11 +841,9 @@ nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_map)
if (tag == eHTMLTag_map) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -869,12 +857,10 @@ nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_map)
if (tag == eHTMLTag_map) AddEndTag(aNode);
AddEndTag(aNode,*mOutput); return NS_OK;
}
return NS_OK;
} }
@ -887,11 +873,9 @@ nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_frameset)
if (tag == eHTMLTag_frameset) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -905,27 +889,26 @@ nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseFrameset(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseFrameset(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_frameset)
if (tag == eHTMLTag_frameset) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
void nsHTMLContentSinkStream::AddIndent(ostream& aStream) void nsHTMLContentSinkStream::AddIndent()
{ {
nsString padding(" ");
for (PRInt32 i = mIndent; --i >= 0; ) for (PRInt32 i = mIndent; --i >= 0; )
{ {
aStream << " "; Write(padding);
mColPos += 2; mColPos += 2;
} }
} }
void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& aStream) void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode)
{ {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
const nsString& name = aNode.GetText(); const nsString& name = aNode.GetText();
@ -942,37 +925,41 @@ void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& a
if (mColPos != 0 && BreakBeforeOpen(tag)) if (mColPos != 0 && BreakBeforeOpen(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
if (PermitWSBeforeOpen(tag)) if (PermitWSBeforeOpen(tag))
AddIndent(aStream); AddIndent();
EnsureBufferSize(tagName.Length());
tagName.ToCString(mBuffer,mBufferSize);
Write(kLessThan);
Write(mBuffer);
tagName.ToCString(gBuffer,sizeof(gBuffer)-1);
aStream << (char)kLessThan << gBuffer;
mColPos += 1 + tagName.Length(); mColPos += 1 + tagName.Length();
if (tag == eHTMLTag_style) if (tag == eHTMLTag_style)
{ {
aStream << (char)kGreaterThan << endl; Write(">\n");
const nsString& data = aNode.GetSkippedContent(); const nsString& data = aNode.GetSkippedContent();
PRInt32 size = data.Length(); PRInt32 size = data.Length();
char* buffer = new char[size+1]; char* buffer = new char[size+1];
data.ToCString(buffer,size+1); data.ToCString(buffer,size+1);
aStream << buffer; Write(buffer);
delete[] buffer; delete[] buffer;
} }
else else
{ {
WriteAttributes(aNode,aStream); WriteAttributes(aNode);
aStream << (char)kGreaterThan; Write(kGreaterThan);
mColPos += 1; mColPos += 1;
} }
if (BreakAfterOpen(tag)) if (BreakAfterOpen(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
@ -983,7 +970,7 @@ void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& a
void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aStream) void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode)
{ {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
// const nsString& name = aNode.GetText(); // const nsString& name = aNode.GetText();
@ -1010,19 +997,25 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aSt
{ {
if (mColPos != 0) if (mColPos != 0)
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
AddIndent(aStream); AddIndent();
} }
tagName.ToCString(gBuffer,sizeof(gBuffer)-1); EnsureBufferSize(tagName.Length());
aStream << (char)kLessThan << (char)kForwardSlash << gBuffer << (char)kGreaterThan; tagName.ToCString(mBuffer,mBufferSize);
mColPos += 1 + 1 + strlen(gBuffer) + 1;
Write(kLessThan);
Write(kForwardSlash);
Write(mBuffer);
Write(kGreaterThan);
mColPos += 1 + 1 + strlen(mBuffer) + 1;
if (BreakAfterClose(tag)) if (BreakAfterClose(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown; mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown;
@ -1040,7 +1033,7 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aSt
* @return * @return
*/ */
nsresult nsresult
nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
eHTMLTags type = (eHTMLTags)aNode.GetNodeType(); eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = eHTMLTag_unknown; eHTMLTags tag = eHTMLTag_unknown;
if (mHTMLStackPos > 0) if (mHTMLStackPos > 0)
@ -1062,14 +1055,16 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
type == eHTMLTag_meta || type == eHTMLTag_meta ||
type == eHTMLTag_style) type == eHTMLTag_style)
{ {
AddStartTag(aNode,aStream); AddStartTag(aNode);
mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown; mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown;
} }
else if (type == eHTMLTag_entity) else if (type == eHTMLTag_entity)
{ {
const nsString& entity = aNode.GetText(); const nsString& entity = aNode.GetText();
UnicodeToHTMLString(entity); EncodeToBuffer(entity);
aStream << '&' << mBuffer << ';'; Write('&');
Write(mBuffer);
Write(';');
mColPos += entity.Length() + 2; mColPos += entity.Length() + 2;
} }
else if (type == eHTMLTag_text) else if (type == eHTMLTag_text)
@ -1077,8 +1072,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
if ((mDoFormat == PR_FALSE) || preformatted == PR_TRUE) if ((mDoFormat == PR_FALSE) || preformatted == PR_TRUE)
{ {
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
else else
@ -1092,8 +1086,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
// than the max then just add it // than the max then just add it
if (mColPos + length < mMaxColumn) if (mColPos + length < mMaxColumn)
{ {
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
else else
@ -1115,8 +1108,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
// if there is no break than just add it // if there is no break than just add it
if (index == kNotFound) if (index == kNotFound)
{ {
UnicodeToHTMLString(str); Write(str);
aStream << mBuffer;
mColPos += str.Length(); mColPos += str.Length();
done = PR_TRUE; done = PR_TRUE;
} }
@ -1128,8 +1120,8 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
first.Truncate(index); first.Truncate(index);
UnicodeToHTMLString(first); Write(first);
aStream << mBuffer << endl; Write('\n');
mColPos = 0; mColPos = 0;
// cut the string from the beginning to the index // cut the string from the beginning to the index
@ -1145,8 +1137,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
if ((mDoFormat == PR_FALSE) || preformatted || IgnoreWS(tag) == PR_FALSE) if ((mDoFormat == PR_FALSE) || preformatted || IgnoreWS(tag) == PR_FALSE)
{ {
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
} }
@ -1155,8 +1146,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
if ((mDoFormat == PR_FALSE) || preformatted) if ((mDoFormat == PR_FALSE) || preformatted)
{ {
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos = 0; mColPos = 0;
} }
} }
@ -1215,25 +1205,26 @@ nsHTMLContentSinkStream::AddComment(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
if(mOutput)
const nsString& name = aNode.GetText();
if (name.Equals("XIF_DOC_INFO"))
{ {
const nsString& name = aNode.GetText(); PRInt32 count=aNode.GetAttributeCount();
if (name.Equals("XIF_DOC_INFO")) for(PRInt32 i=0;i<count;i++)
{ {
PRInt32 count=aNode.GetAttributeCount(); const nsString& key=aNode.GetKeyAt(i);
for(PRInt32 i=0;i<count;i++) const nsString& value=aNode.GetValueAt(i);
{
const nsString& key=aNode.GetKeyAt(i);
const nsString& value=aNode.GetValueAt(i);
if (key.Equals("charset")) if (key.Equals("charset"))
{
if (mCharsetOverride.Length() == 0)
InitEncoder(value); InitEncoder(value);
} }
} }
else }
{ else
AddStartTag(aNode,*mOutput); {
} AddStartTag(aNode);
} }
return NS_OK; return NS_OK;
} }
@ -1242,37 +1233,17 @@ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
/** /**
* This method is used to close a generic container. * This method is used to close a generic container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseContainer(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseContainer(const nsIParserNode& aNode){
if(mOutput) { AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
/**
* This method is used to add a leaf to the currently
* open container.
*
* @update 07/12/98 gpk
* @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful.
*/
NS_IMETHODIMP
nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
nsresult result = NS_OK;
if(mOutput) {
result = AddLeaf(aNode,*mOutput);
}
return result;
}
/** /**
* This method gets called when the parser begins the process * This method gets called when the parser begins the process
* of building the content model via the content sink. * of building the content model via the content sink.
@ -1282,10 +1253,12 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::WillBuildModel(void){ nsHTMLContentSinkStream::WillBuildModel(void){
mTabLevel=-1; mTabLevel=-1;
if(mDoHeader && mOutput) { if(mDoHeader) {
(*mOutput) << gHeaderComment << endl; Write(gHeaderComment);
(*mOutput) << gDocTypeHeader << endl; Write('\n');
} Write(gDocTypeHeader);
Write('\n');
}
return NS_OK; return NS_OK;
} }

52
htmlparser/src/nsHTMLContentSinkStream.h
View File

@ -54,23 +54,24 @@ class ostream;
#endif #endif
class nsIUnicodeEncoder; class nsIUnicodeEncoder;
class nsIOutputStream;
class nsHTMLContentSinkStream : public nsIHTMLContentSink { class nsHTMLContentSinkStream : public nsIHTMLContentSink {
public: public:
/**
* Standard constructor
* @update gess7/7/98
*/
nsHTMLContentSinkStream(PRBool aDoFormat = PR_TRUE, PRBool aDoHeader = PR_TRUE);
/** /**
* Constructor with associated stream. If you use this, it means that you want * Constructor with associated stream. If you use this, it means that you want
* this class to emits its output to the stream you provide. * this class to emits its output to the stream you provide.
* @update gess7/7/98 * @update gpk 04/30/99
* @param aStream -- ref to stream where you want output sent * @param aOutStream -- stream where you want output sent
* @param aOutStream -- ref to string where you want output sent
*/ */
nsHTMLContentSinkStream(ostream& aStream,PRBool aDoFormat = PR_TRUE, PRBool aDoHeader = PR_TRUE); nsHTMLContentSinkStream(nsIOutputStream* aOutStream,
nsString* aOutString,
const nsString* aCharsetOverride,
PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE);
/** /**
* virtual destructor * virtual destructor
@ -78,8 +79,6 @@ class nsHTMLContentSinkStream : public nsIHTMLContentSink {
*/ */
virtual ~nsHTMLContentSinkStream(); virtual ~nsHTMLContentSinkStream();
NS_IMETHOD_(void) SetOutputStream(ostream& aStream);
// nsISupports // nsISupports
NS_DECL_ISUPPORTS NS_DECL_ISUPPORTS
@ -127,22 +126,26 @@ public:
protected: protected:
nsresult AddLeaf(const nsIParserNode& aNode, ostream& aStream); void WriteAttributes(const nsIParserNode& aNode);
void WriteAttributes(const nsIParserNode& aNode,ostream& aStream); void AddStartTag(const nsIParserNode& aNode);
void AddStartTag(const nsIParserNode& aNode, ostream& aStream); void AddEndTag(const nsIParserNode& aNode);
void AddEndTag(const nsIParserNode& aNode, ostream& aStream); void AddIndent();
void AddIndent(ostream& aStream);
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
void UnicodeToHTMLString(const nsString& aSrc);
nsresult InitEncoder(const nsString& aCharset); nsresult InitEncoder(const nsString& aCharset);
void EncodeToBuffer(const nsString& aString);
void Write(const nsString& aString);
void Write(const char* aCharBuffer);
void Write(char aChar);
protected: protected:
ostream* mOutput; nsIOutputStream* mStream;
nsString* mString;
int mTabLevel; int mTabLevel;
PRInt32 mIndent; PRInt32 mIndent;
@ -155,13 +158,24 @@ protected:
PRBool mDoHeader; PRBool mDoHeader;
char* mBuffer; char* mBuffer;
PRInt32 mBufferSize; PRInt32 mBufferLength; // The length of the data in the buffer
PRInt32 mBufferSize; // The actual size of the buffer, regardless of the data
nsIUnicodeEncoder* mUnicodeEncoder; nsIUnicodeEncoder* mUnicodeEncoder;
nsString mCharsetOverride;
}; };
extern NS_HTMLPARS nsresult extern NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult, NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsIOutputStream* aOutStream,
const nsString* aCharsetOverride=nsnull,
PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE);
extern NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString,
PRBool aDoFormat = PR_TRUE, PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE); PRBool aDoHeader = PR_TRUE);

285
htmlparser/src/nsHTMLToTXTSinkStream.cpp
View File

@ -39,6 +39,8 @@
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsICharsetConverterManager.h"
#include "nsIOutputStream.h"
#include "nsFileStream.h"
static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID); 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);
@ -60,7 +62,19 @@ static PRBool IsBlockLevel(eHTMLTags aTag);
nsresult nsHTMLToTXTSinkStream::InitEncoder(const nsString& aCharset) nsresult nsHTMLToTXTSinkStream::InitEncoder(const nsString& aCharset)
{ {
nsresult res = NS_OK; nsresult res = NS_OK;
// If the converter is ucs2, then do not use a converter
nsString ucs2("ucs2");
if (aCharset.Equals(ucs2))
{
NS_IF_RELEASE(mUnicodeEncoder);
return res;
}
nsICharsetAlias* calias = nsnull; nsICharsetAlias* calias = nsnull;
res = nsServiceManager::GetService(kCharsetAliasCID, res = nsServiceManager::GetService(kCharsetAliasCID,
@ -141,58 +155,71 @@ NS_IMPL_RELEASE(nsHTMLToTXTSinkStream)
/** /**
* This method is defined in nsIParser. It is used to * This method creates a new sink, it sets the stream used
* cause the COM-like construction of an nsParser. * for the sink to aStream
* *
* @update gpk02/03/99 * @update gpk 04/30/99
* @param nsIParser** ptr to newly instantiated parser
* @return NS_xxx error result
*/ */
NS_HTMLPARS nsresult NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult) { NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(); nsIOutputStream* aStream,
const nsString* aCharsetOverride) {
NS_ASSERTION(aStream != nsnull, "a valid stream is required");
nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(aStream,nsnull);
if (nsnull == it) { if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY; return NS_ERROR_OUT_OF_MEMORY;
} }
if (aCharsetOverride != nsnull)
it->SetCharsetOverride(aCharsetOverride);
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult); return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
} }
/** /**
* Construct a content sink stream. * This method creates a new sink, it sets the stream used
* @update gpk02/03/99 * for the sink to aStream
* @param *
* @return * @update gpk 04/30/99
*/ */
nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream() { NS_HTMLPARS nsresult
NS_INIT_REFCNT(); NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
mOutput=&cout; nsString* aString) {
mColPos = 0;
mIndent = 0; NS_ASSERTION(aString != nsnull, "a valid stream is required");
mDoOutput = PR_FALSE; nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(nsnull,aString);
mBufferSize = 0; if (nsnull == it) {
mBuffer = nsnull; return NS_ERROR_OUT_OF_MEMORY;
mUnicodeEncoder = nsnull; }
nsString ucs2("ucs2");
it->SetCharsetOverride(&ucs2);
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
} }
/** /**
* Construct a content sink stream. * Construct a content sink stream.
* @update gpk02/03/99 * @update gpk02/03/99
* @param * @param
* @return * @return
*/ */
nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(ostream& aStream) { nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(nsIOutputStream* aStream, nsString* aString) {
NS_INIT_REFCNT(); NS_INIT_REFCNT();
mOutput = &aStream; mStream = aStream;
mColPos = 0; mColPos = 0;
mIndent = 0; mIndent = 0;
mDoOutput = PR_FALSE; mDoOutput = PR_FALSE;
mBufferSize = 0; mBufferSize = 0;
mBufferLength = 0;
mBuffer = nsnull; mBuffer = nsnull;
mUnicodeEncoder = nsnull; mUnicodeEncoder = nsnull;
mStream = aStream;
mString = aString;
} }
/** /**
* *
* @update gpk02/03/99 * @update gpk02/03/99
@ -200,7 +227,6 @@ nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(ostream& aStream) {
* @return * @return
*/ */
nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() { nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() {
mOutput=0; //we don't own the stream we're given; just forget it.
delete [] mBuffer; delete [] mBuffer;
NS_IF_RELEASE(mUnicodeEncoder); NS_IF_RELEASE(mUnicodeEncoder);
} }
@ -208,50 +234,24 @@ nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() {
/** /**
* *
* @update gpk02/03/99 * @update gpk04/30/99
* @param * @param
* @return * @return
*/ */
NS_IMETHODIMP_(void)
nsHTMLToTXTSinkStream::SetOutputStream(ostream& aStream){ NS_IMETHODIMP
mOutput=&aStream; nsHTMLToTXTSinkStream::SetCharsetOverride(const nsString* aCharset)
{
if (aCharset)
{
mCharsetOverride = *aCharset;
InitEncoder(mCharsetOverride);
}
return NS_OK;
} }
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void OpenTagWithAttributes(const char* theTag,const nsIParserNode& aNode,int tab,ostream& aStream,PRBool aNewline) {
}
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void OpenTag(const char* theTag,int tab,ostream& aStream,PRBool aNewline) {
}
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void CloseTag(const char* theTag,int tab,ostream& aStream) {
}
/** /**
* This method gets called by the parser when it encounters * This method gets called by the parser when it encounters
* a title tag and wants to set the document title in the sink. * a title tag and wants to set the document title in the sink.
@ -463,38 +463,38 @@ void nsHTMLToTXTSinkStream::EnsureBufferSize(PRInt32 aNewSize)
mBufferSize = 2*aNewSize+1; // make the twice as large mBufferSize = 2*aNewSize+1; // make the twice as large
mBuffer = new char[mBufferSize]; mBuffer = new char[mBufferSize];
mBuffer[0] = 0; mBuffer[0] = 0;
mBufferLength = 0;
} }
} }
void nsHTMLToTXTSinkStream::UnicodeToTXTString(const nsString& aSrc) void nsHTMLToTXTSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
NS_ASSERTION(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized");
if (mUnicodeEncoder == nsnull)
return;
#define CH_NBSP 160 #define CH_NBSP 160
PRInt32 length = aSrc.Length(); PRInt32 length = aSrc.Length();
nsresult result; nsresult result;
PRInt32 bufferLength;
if (mUnicodeEncoder == nsnull) if (mUnicodeEncoder != nsnull && length > 0)
InitEncoder("");
if (length > 0)
{ {
EnsureBufferSize(length); EnsureBufferSize(length);
bufferLength = mBufferSize; mBufferLength = mBufferSize;
mUnicodeEncoder->Reset(); mUnicodeEncoder->Reset();
result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &bufferLength); result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &mBufferLength);
mBuffer[bufferLength] = 0; mBuffer[mBufferLength] = 0;
PRInt32 temp = bufferLength; PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result)) if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp); result = mUnicodeEncoder->Finish(mBuffer,&temp);
for (PRInt32 i = 0; i < bufferLength; i++) for (PRInt32 i = 0; i < mBufferLength; i++)
{ {
if (mBuffer[i] == char(CH_NBSP)) if (mBuffer[i] == char(CH_NBSP))
mBuffer[i] = ' '; mBuffer[i] = ' ';
@ -504,76 +504,47 @@ void nsHTMLToTXTSinkStream::UnicodeToTXTString(const nsString& aSrc)
} }
NS_IMETHODIMP
nsHTMLToTXTSinkStream::GetStringBuffer(nsString & aStrBuffer)
{
aStrBuffer = mStrBuffer;
return NS_OK;
}
/** /**
* This gets called by the parser when you want to add * Write places the contents of aString into either the output stream
* a leaf node to the current container in the content * or the output string.
* model. * When going to the stream, all data is run through the encoder
* *
* @updated gpk 06/18/98 * @updated gpk02/03/99
* @param * @param
* @return * @return
*/ */
nsresult void nsHTMLToTXTSinkStream::Write(const nsString& aString)
nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream)
{ {
eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
const nsString& text = aNode.GetText();
if (mDoOutput == PR_FALSE) // If a encoder is being used then convert first convert the input string
return NS_OK; if (mUnicodeEncoder != nsnull)
if (type == eHTMLTag_text) {
UnicodeToTXTString(text);
aStream << mBuffer;
mStrBuffer.Append(mBuffer);
mColPos += text.Length();
}
else if (type == eHTMLTag_entity)
{ {
const nsString& text = aNode.GetText(); EncodeToBuffer(aString);
UnicodeToTXTString(text); if (mStream != nsnull)
PRInt32 entity = NS_EntityToUnicode(mBuffer);
if (entity < 256)
{ {
char ch = (char)entity; nsOutputStream out(mStream);
aStream << ch; out.write(mBuffer,mBufferLength);
mColPos++; }
if (mString != nsnull)
{
mString->Append(mBuffer);
} }
} }
else if (type == eHTMLTag_whitespace) else
{ {
if (PR_TRUE) if (mStream != nsnull)
{ {
const nsString& text = aNode.GetText(); nsOutputStream out(mStream);
UnicodeToTXTString(text); const PRUnichar* unicode = aString.GetUnicode();
aStream << mBuffer; PRUint32 length = aString.Length();
mStrBuffer.Append(mBuffer); out.write(unicode,length);
mColPos += text.Length(); }
else
{
mString->Append(aString);
} }
} }
else if (type == eHTMLTag_br)
{
if (PR_TRUE)
{
aStream << endl;
mStrBuffer.Append("\n");
mColPos += 1;
}
}
return NS_OK;
} }
@ -628,7 +599,10 @@ nsHTMLToTXTSinkStream::OpenContainer(const nsIParserNode& aNode){
const nsString& value=aNode.GetValueAt(i); const nsString& value=aNode.GetValueAt(i);
if (key.Equals("charset")) if (key.Equals("charset"))
InitEncoder(value); {
if (mCharsetOverride.Length() == 0)
InitEncoder(value);
}
} }
} }
@ -657,9 +631,8 @@ nsHTMLToTXTSinkStream::CloseContainer(const nsIParserNode& aNode){
{ {
if (mColPos != 0) if (mColPos != 0)
{ {
if (mOutput) nsString temp("\n");
*mOutput << endl; Write(temp);
mStrBuffer.Append("\n");
mColPos = 0; mColPos = 0;
} }
} }
@ -677,11 +650,53 @@ nsHTMLToTXTSinkStream::CloseContainer(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode){ nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode){
nsresult result = NS_OK; eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
if(mOutput) {
result = AddLeaf(aNode,*mOutput); const nsString& text = aNode.GetText();
if (mDoOutput == PR_FALSE)
return NS_OK;
if (type == eHTMLTag_text) {
Write(text);
mColPos += text.Length();
}
else if (type == eHTMLTag_entity)
{
const nsString& text = aNode.GetText();
EncodeToBuffer(text);
PRUnichar entity = NS_EntityToUnicode(mBuffer);
nsString temp;
temp.Append(entity);
Write(temp);
mColPos++;
} }
return result; else if (type == eHTMLTag_whitespace)
{
if (PR_TRUE)
{
const nsString& text = aNode.GetText();
Write(text);
mColPos += text.Length();
}
}
else if (type == eHTMLTag_br)
{
nsString temp("\n");
Write(text);
mColPos++;
}
else if (type == eHTMLTag_newline)
{
nsString temp("\n");
Write(text);
mColPos++;
}
return NS_OK;
} }

43
htmlparser/src/nsHTMLToTXTSinkStream.h
View File

@ -47,12 +47,9 @@
{0xa39c6bff, 0x15f0, 0x11d2, \ {0xa39c6bff, 0x15f0, 0x11d2, \
{0x80, 0x41, 0x0, 0x10, 0x4b, 0x98, 0x3f, 0xd4}} {0x80, 0x41, 0x0, 0x10, 0x4b, 0x98, 0x3f, 0xd4}}
#ifndef XP_MAC
class ostream;
#endif
class nsIUnicodeEncoder; class nsIUnicodeEncoder;
class nsIOutputStream;
class nsHTMLToTXTSinkStream : public nsIHTMLContentSink { class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
public: public:
@ -61,8 +58,7 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
* Standard constructor * Standard constructor
* @update gpk02/03/99 * @update gpk02/03/99
*/ */
nsHTMLToTXTSinkStream(); nsHTMLToTXTSinkStream(nsIOutputStream* aOutStream, nsString* aOutString);
nsHTMLToTXTSinkStream(ostream& aStream);
/** /**
* virtual destructor * virtual destructor
@ -70,8 +66,8 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
*/ */
virtual ~nsHTMLToTXTSinkStream(); virtual ~nsHTMLToTXTSinkStream();
NS_IMETHOD_(void) SetOutputStream(ostream& aStream); NS_IMETHOD SetCharsetOverride(const nsString* aCharset);
NS_IMETHOD GetStringBuffer(nsString & aStrBuffer);
// nsISupports // nsISupports
NS_DECL_ISUPPORTS NS_DECL_ISUPPORTS
@ -113,32 +109,47 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
NS_IMETHOD BeginContext(PRInt32 aPosition); NS_IMETHOD BeginContext(PRInt32 aPosition);
NS_IMETHOD EndContext(PRInt32 aPosition); NS_IMETHOD EndContext(PRInt32 aPosition);
protected: protected:
nsresult AddLeaf(const nsIParserNode& aNode, ostream& aStream);
void WriteAttributes(const nsIParserNode& aNode,ostream& aStream);
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
void UnicodeToTXTString(const nsString& aSrc);
nsresult InitEncoder(const nsString& aCharset); nsresult InitEncoder(const nsString& aCharset);
void Write(const nsString& aString);
void EncodeToBuffer(const nsString& aString);
protected: protected:
ostream* mOutput; nsIOutputStream* mStream;
nsString* mString;
PRInt32 mIndent; PRInt32 mIndent;
PRInt32 mColPos; PRInt32 mColPos;
PRBool mDoOutput; PRBool mDoOutput;
char* mBuffer;
PRInt32 mBufferSize;
nsString mStrBuffer; char* mBuffer;
PRInt32 mBufferLength; // The length of the data in the buffer
PRInt32 mBufferSize; // The actual size of the buffer, regardless of the data
nsIUnicodeEncoder* mUnicodeEncoder; nsIUnicodeEncoder* mUnicodeEncoder;
nsString mCharsetOverride;
}; };
extern NS_HTMLPARS nsresult extern NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult); NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsIOutputStream* aOutStream,
const nsString* aCharsetOverride=nsnull);
extern NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString);
#endif #endif

567
parser/htmlparser/src/nsHTMLContentSinkStream.cpp
View File

@ -42,6 +42,8 @@
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsICharsetConverterManager.h"
#include "nsIOutputStream.h"
#include "nsFileStream.h"
static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID); static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID);
@ -51,7 +53,6 @@ static NS_DEFINE_IID(kIHTMLContentSinkIID, NS_IHTML_CONTENT_SINK_IID);
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\">";
const int gTabSize=2; const int gTabSize=2;
static char gBuffer[1024];
static const char* UnicodeToEntity(PRInt32 aCode); static const char* UnicodeToEntity(PRInt32 aCode);
@ -332,18 +333,23 @@ NS_IMPL_RELEASE(nsHTMLContentSinkStream)
/** /**
* This method is defined in nsIParser. It is used to * Create an new sink
* cause the COM-like construction of an nsParser.
* *
* @update gess 4/8/98 * @update gpk 05/01/99
* @param nsIParser** ptr to newly instantiated parser
* @return NS_xxx error result * @return NS_xxx error result
*/ */
NS_HTMLPARS nsresult NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult, NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
PRBool aDoFormat, nsIOutputStream* aOutStream,
PRBool aDoHeader) { const nsString* aCharsetOverride,
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(aDoFormat,aDoHeader); PRBool aDoFormat,
PRBool aDoHeader)
{
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(aOutStream,
nsnull,
aCharsetOverride,
aDoFormat,
aDoHeader);
if (nsnull == it) { if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY; return NS_ERROR_OUT_OF_MEMORY;
} }
@ -352,6 +358,32 @@ NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
} }
/**
* Create an new sink
*
* @update gpk 05/01/99
* @return NS_xxx error result
*/
NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString,
PRBool aDoFormat,
PRBool aDoHeader)
{
nsHTMLContentSinkStream* it = new nsHTMLContentSinkStream(nsnull,
aOutString,
nsnull,
aDoFormat,
aDoHeader);
if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY;
}
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
}
/** /**
* Inits the encoder instance variable for the sink based on the charset * Inits the encoder instance variable for the sink based on the charset
@ -409,30 +441,12 @@ nsresult nsHTMLContentSinkStream::InitEncoder(const nsString& aCharset)
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::nsHTMLContentSinkStream(PRBool aDoFormat,PRBool aDoHeader) { nsHTMLContentSinkStream::nsHTMLContentSinkStream(nsIOutputStream* aOutStream,
nsString* aOutString,
const nsString* aCharsetOverride,
PRBool aDoFormat,
PRBool aDoHeader) {
NS_INIT_REFCNT(); NS_INIT_REFCNT();
mOutput=&cout;
mLowerCaseTags = PR_TRUE;
memset(mHTMLTagStack,0,sizeof(mHTMLTagStack));
mHTMLStackPos = 0;
mColPos = 0;
mIndent = 0;
mDoFormat = aDoFormat;
mDoHeader = aDoHeader;
mBuffer = nsnull;
mBufferSize = 0;
mUnicodeEncoder = nsnull;
}
/**
* Construct a content sink stream.
* @update gess7/7/98
* @param
* @return
*/
nsHTMLContentSinkStream::nsHTMLContentSinkStream(ostream& aStream,PRBool aDoFormat,PRBool aDoHeader) {
NS_INIT_REFCNT();
mOutput = &aStream;
mLowerCaseTags = PR_TRUE; mLowerCaseTags = PR_TRUE;
memset(mHTMLTagStack,0,sizeof(mHTMLTagStack)); memset(mHTMLTagStack,0,sizeof(mHTMLTagStack));
mHTMLStackPos = 0; mHTMLStackPos = 0;
@ -443,6 +457,10 @@ nsHTMLContentSinkStream::nsHTMLContentSinkStream(ostream& aStream,PRBool aDoForm
mBuffer = nsnull; mBuffer = nsnull;
mBufferSize = 0; mBufferSize = 0;
mUnicodeEncoder = nsnull; mUnicodeEncoder = nsnull;
mStream = aOutStream;
mString = aOutString;
if (aCharsetOverride != nsnull)
mCharsetOverride = *aCharsetOverride;
} }
@ -498,82 +516,119 @@ void nsHTMLContentSinkStream::EnsureBufferSize(PRInt32 aNewSize)
} }
} }
void nsHTMLContentSinkStream::UnicodeToHTMLString(const nsString& aSrc)
void nsHTMLContentSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
PRInt32 length = aSrc.Length();
PRUnichar ch; NS_ASSERTION(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized");
const char* entity = nsnull;
PRUint32 offset = 0;
PRUint32 addedLength = 0;
nsAutoString data;
if (mUnicodeEncoder == nsnull) if (mUnicodeEncoder == nsnull)
InitEncoder(""); return;
if (length > 0) #define CH_NBSP 160
PRInt32 length = aSrc.Length();
nsresult result;
if (mUnicodeEncoder != nsnull && length > 0)
{ {
// Step 1. Convert anything that maps to character entity to
// the entity value
EnsureBufferSize(length); EnsureBufferSize(length);
for (PRInt32 i = 0; i < length; i++) mBufferLength = mBufferSize;
{
ch = aSrc.CharAt(i);
entity = UnicodeToEntity(ch);
if (entity)
{
nsAutoString temp(entity);
temp.ToLowerCase();
data.Append('&');
data.Append(temp);
data.Append(';');
}
else
{
data.Append(ch);
}
}
// Step 2. Run the result through the converter
length = data.Length();
EnsureBufferSize(length);
PRInt32 bufferLength = mBufferSize;
mUnicodeEncoder->Reset(); mUnicodeEncoder->Reset();
nsresult result = mUnicodeEncoder->Convert(data.GetUnicode(), &length, mBuffer, &bufferLength); result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &mBufferLength);
mBuffer[bufferLength] = 0; mBuffer[mBufferLength] = 0;
PRInt32 temp = bufferLength; PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result)) if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp); result = mUnicodeEncoder->Finish(mBuffer,&temp);
}
for (PRInt32 i = 0; i < mBufferLength; i++)
{
if (mBuffer[i] == char(CH_NBSP))
mBuffer[i] = ' ';
}
}
} }
void nsHTMLContentSinkStream::Write(const nsString& aString)
{
// If a encoder is being used then convert first convert the input string
if (mUnicodeEncoder != nsnull)
{
EncodeToBuffer(aString);
if (mStream != nsnull)
{
nsOutputStream out(mStream);
out.write(mBuffer,mBufferLength);
}
if (mString != nsnull)
{
mString->Append(mBuffer);
}
}
else
{
if (mStream != nsnull)
{
nsOutputStream out(mStream);
const PRUnichar* unicode = aString.GetUnicode();
PRUint32 length = aString.Length();
out.write(unicode,length);
}
else
{
mString->Append(aString);
}
}
}
void nsHTMLContentSinkStream::Write(const char* aData)
{
if (mStream)
{
nsOutputStream out(mStream);
out << aData;
}
if (mString)
{
mString->Append(aData);
}
}
void nsHTMLContentSinkStream::Write(char aData)
{
if (mStream)
{
nsOutputStream out(mStream);
out << aData;
}
if (mString)
{
mString->Append(aData);
}
}
/** /**
* *
* @update gess7/7/98 * @update 04/30/99 gpk
* @param * @param
* @return * @return
*/ */
nsHTMLContentSinkStream::~nsHTMLContentSinkStream() { nsHTMLContentSinkStream::~nsHTMLContentSinkStream() {
NS_IF_RELEASE(mUnicodeEncoder); NS_IF_RELEASE(mUnicodeEncoder);
mOutput=0; //we don't own the stream we're given; just forget it.
} }
/**
*
* @update gess7/22/98
* @param
* @return
*/
NS_IMETHODIMP_(void)
nsHTMLContentSinkStream::SetOutputStream(ostream& aStream){
mOutput=&aStream;
}
/** /**
* *
@ -581,7 +636,7 @@ nsHTMLContentSinkStream::SetOutputStream(ostream& aStream){
* @param * @param
* @return * @return
*/ */
void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream& aStream) { void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode) {
int theCount=aNode.GetAttributeCount(); int theCount=aNode.GetAttributeCount();
if(theCount) { if(theCount) {
int i=0; int i=0;
@ -600,14 +655,20 @@ void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream
EnsureBufferSize(key.Length()); EnsureBufferSize(key.Length());
key.ToCString(mBuffer,mBufferSize); key.ToCString(mBuffer,mBufferSize);
aStream << " " << mBuffer << char(kEqual); // send to ouput " [KEY]="
Write(' ');
Write(mBuffer);
Write(char(kEqual));
mColPos += 1 + strlen(mBuffer) + 1; mColPos += 1 + strlen(mBuffer) + 1;
const nsString& value=aNode.GetValueAt(i); const nsString& value=aNode.GetValueAt(i);
UnicodeToHTMLString(value);
// send to ouput "\"[VALUE]\""
Write('\"');
Write(value);
Write('\"');
aStream << "\"" << mBuffer << "\"";
mColPos += 1 + strlen(mBuffer) + 1; mColPos += 1 + strlen(mBuffer) + 1;
} }
} }
@ -615,85 +676,31 @@ void nsHTMLContentSinkStream::WriteAttributes(const nsIParserNode& aNode,ostream
} }
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void OpenTagWithAttributes(const char* theTag,const nsIParserNode& aNode,int tab,ostream& aStream,PRBool aNewline) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << theTag;
// WriteAttributes(aNode,aStream);
aStream << (char)kGreaterThan;
if(aNewline)
aStream << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void OpenTag(const char* theTag,int tab,ostream& aStream,PRBool aNewline) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << theTag << (char)kGreaterThan;
if(aNewline)
aStream << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void CloseTag(const char* theTag,int tab,ostream& aStream) {
int i=0;
for(i=0;i<tab*gTabSize;i++)
aStream << " ";
aStream << (char)kLessThan << (char)kForwardSlash << theTag << (char)kGreaterThan << endl;
}
/**
*
* @update gess7/5/98
* @param
* @return
*/
static
void WritePair(eHTMLTags aTag,const nsString& theContent,int tab,ostream& aStream) {
const char* titleStr = GetTagName(aTag);
OpenTag(titleStr,tab,aStream,PR_FALSE);
theContent.ToCString(gBuffer,sizeof(gBuffer)-1);
aStream << gBuffer;
CloseTag(titleStr,0,aStream);
}
/** /**
* This method gets called by the parser when it encounters * This method gets called by the parser when it encounters
* a title tag and wants to set the document title in the sink. * a title tag and wants to set the document title in the sink.
* *
* @update 4/1/98 gess * @update 04/30/99 gpk
* @param nsString reference to new title value * @param nsString reference to new title value
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::SetTitle(const nsString& aValue){ nsHTMLContentSinkStream::SetTitle(const nsString& aValue){
if(mOutput) { const char* tagName = GetTagName(eHTMLTag_title);
WritePair(eHTMLTag_title,aValue,1+mTabLevel,*mOutput); Write(kLessThan);
} Write(tagName);
Write(kGreaterThan);
Write(aValue);
Write(kLessThan);
Write(kForwardSlash);
Write(tagName);
Write(kGreaterThan);
return NS_OK; return NS_OK;
} }
@ -701,17 +708,15 @@ nsHTMLContentSinkStream::SetTitle(const nsString& aValue){
/** /**
* This method is used to open the outer HTML container. * This method is used to open the outer HTML container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){
if(mOutput) {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
if (tag == eHTMLTag_html) if (tag == eHTMLTag_html)
AddStartTag(aNode,*mOutput); AddStartTag(aNode);
}
return NS_OK; return NS_OK;
} }
@ -719,18 +724,15 @@ nsHTMLContentSinkStream::OpenHTML(const nsIParserNode& aNode){
/** /**
* This method is used to close the outer HTML container. * This method is used to close the outer HTML container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_html)
if (tag == eHTMLTag_html) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
mOutput->flush();
}
return NS_OK; return NS_OK;
} }
@ -738,17 +740,15 @@ nsHTMLContentSinkStream::CloseHTML(const nsIParserNode& aNode){
/** /**
* This method is used to open the only HEAD container. * This method is used to open the only HEAD container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_head)
if (tag == eHTMLTag_head) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -756,17 +756,15 @@ nsHTMLContentSinkStream::OpenHead(const nsIParserNode& aNode){
/** /**
* This method is used to close the only HEAD container. * This method is used to close the only HEAD container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_head)
if (tag == eHTMLTag_head) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -774,17 +772,15 @@ nsHTMLContentSinkStream::CloseHead(const nsIParserNode& aNode){
/** /**
* This method is used to open the main BODY container. * This method is used to open the main BODY container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_body)
if (tag == eHTMLTag_body) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -792,17 +788,15 @@ nsHTMLContentSinkStream::OpenBody(const nsIParserNode& aNode){
/** /**
* This method is used to close the main BODY container. * This method is used to close the main BODY container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_body)
if (tag == eHTMLTag_body) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -816,11 +810,9 @@ nsHTMLContentSinkStream::CloseBody(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_form)
if (tag == eHTMLTag_form) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -834,11 +826,9 @@ nsHTMLContentSinkStream::OpenForm(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_form)
if (tag == eHTMLTag_form) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -851,11 +841,9 @@ nsHTMLContentSinkStream::CloseForm(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_map)
if (tag == eHTMLTag_map) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -869,12 +857,10 @@ nsHTMLContentSinkStream::OpenMap(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_map)
if (tag == eHTMLTag_map) AddEndTag(aNode);
AddEndTag(aNode,*mOutput); return NS_OK;
}
return NS_OK;
} }
@ -887,11 +873,9 @@ nsHTMLContentSinkStream::CloseMap(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_frameset)
if (tag == eHTMLTag_frameset) AddStartTag(aNode);
AddStartTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
@ -905,27 +889,26 @@ nsHTMLContentSinkStream::OpenFrameset(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseFrameset(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseFrameset(const nsIParserNode& aNode){
if(mOutput) { eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); if (tag == eHTMLTag_frameset)
if (tag == eHTMLTag_frameset) AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
void nsHTMLContentSinkStream::AddIndent(ostream& aStream) void nsHTMLContentSinkStream::AddIndent()
{ {
nsString padding(" ");
for (PRInt32 i = mIndent; --i >= 0; ) for (PRInt32 i = mIndent; --i >= 0; )
{ {
aStream << " "; Write(padding);
mColPos += 2; mColPos += 2;
} }
} }
void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& aStream) void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode)
{ {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
const nsString& name = aNode.GetText(); const nsString& name = aNode.GetText();
@ -942,37 +925,41 @@ void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& a
if (mColPos != 0 && BreakBeforeOpen(tag)) if (mColPos != 0 && BreakBeforeOpen(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
if (PermitWSBeforeOpen(tag)) if (PermitWSBeforeOpen(tag))
AddIndent(aStream); AddIndent();
EnsureBufferSize(tagName.Length());
tagName.ToCString(mBuffer,mBufferSize);
Write(kLessThan);
Write(mBuffer);
tagName.ToCString(gBuffer,sizeof(gBuffer)-1);
aStream << (char)kLessThan << gBuffer;
mColPos += 1 + tagName.Length(); mColPos += 1 + tagName.Length();
if (tag == eHTMLTag_style) if (tag == eHTMLTag_style)
{ {
aStream << (char)kGreaterThan << endl; Write(">\n");
const nsString& data = aNode.GetSkippedContent(); const nsString& data = aNode.GetSkippedContent();
PRInt32 size = data.Length(); PRInt32 size = data.Length();
char* buffer = new char[size+1]; char* buffer = new char[size+1];
data.ToCString(buffer,size+1); data.ToCString(buffer,size+1);
aStream << buffer; Write(buffer);
delete[] buffer; delete[] buffer;
} }
else else
{ {
WriteAttributes(aNode,aStream); WriteAttributes(aNode);
aStream << (char)kGreaterThan; Write(kGreaterThan);
mColPos += 1; mColPos += 1;
} }
if (BreakAfterOpen(tag)) if (BreakAfterOpen(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
@ -983,7 +970,7 @@ void nsHTMLContentSinkStream::AddStartTag(const nsIParserNode& aNode, ostream& a
void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aStream) void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode)
{ {
eHTMLTags tag = (eHTMLTags)aNode.GetNodeType(); eHTMLTags tag = (eHTMLTags)aNode.GetNodeType();
// const nsString& name = aNode.GetText(); // const nsString& name = aNode.GetText();
@ -1010,19 +997,25 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aSt
{ {
if (mColPos != 0) if (mColPos != 0)
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
AddIndent(aStream); AddIndent();
} }
tagName.ToCString(gBuffer,sizeof(gBuffer)-1); EnsureBufferSize(tagName.Length());
aStream << (char)kLessThan << (char)kForwardSlash << gBuffer << (char)kGreaterThan; tagName.ToCString(mBuffer,mBufferSize);
mColPos += 1 + 1 + strlen(gBuffer) + 1;
Write(kLessThan);
Write(kForwardSlash);
Write(mBuffer);
Write(kGreaterThan);
mColPos += 1 + 1 + strlen(mBuffer) + 1;
if (BreakAfterClose(tag)) if (BreakAfterClose(tag))
{ {
aStream << endl; Write('\n');
mColPos = 0; mColPos = 0;
} }
mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown; mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown;
@ -1040,7 +1033,7 @@ void nsHTMLContentSinkStream::AddEndTag(const nsIParserNode& aNode, ostream& aSt
* @return * @return
*/ */
nsresult nsresult
nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
eHTMLTags type = (eHTMLTags)aNode.GetNodeType(); eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
eHTMLTags tag = eHTMLTag_unknown; eHTMLTags tag = eHTMLTag_unknown;
if (mHTMLStackPos > 0) if (mHTMLStackPos > 0)
@ -1062,14 +1055,16 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
type == eHTMLTag_meta || type == eHTMLTag_meta ||
type == eHTMLTag_style) type == eHTMLTag_style)
{ {
AddStartTag(aNode,aStream); AddStartTag(aNode);
mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown; mHTMLTagStack[--mHTMLStackPos] = eHTMLTag_unknown;
} }
else if (type == eHTMLTag_entity) else if (type == eHTMLTag_entity)
{ {
const nsString& entity = aNode.GetText(); const nsString& entity = aNode.GetText();
UnicodeToHTMLString(entity); EncodeToBuffer(entity);
aStream << '&' << mBuffer << ';'; Write('&');
Write(mBuffer);
Write(';');
mColPos += entity.Length() + 2; mColPos += entity.Length() + 2;
} }
else if (type == eHTMLTag_text) else if (type == eHTMLTag_text)
@ -1077,8 +1072,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
if ((mDoFormat == PR_FALSE) || preformatted == PR_TRUE) if ((mDoFormat == PR_FALSE) || preformatted == PR_TRUE)
{ {
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
else else
@ -1092,8 +1086,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
// than the max then just add it // than the max then just add it
if (mColPos + length < mMaxColumn) if (mColPos + length < mMaxColumn)
{ {
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
else else
@ -1115,8 +1108,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
// if there is no break than just add it // if there is no break than just add it
if (index == kNotFound) if (index == kNotFound)
{ {
UnicodeToHTMLString(str); Write(str);
aStream << mBuffer;
mColPos += str.Length(); mColPos += str.Length();
done = PR_TRUE; done = PR_TRUE;
} }
@ -1128,8 +1120,8 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
first.Truncate(index); first.Truncate(index);
UnicodeToHTMLString(first); Write(first);
aStream << mBuffer << endl; Write('\n');
mColPos = 0; mColPos = 0;
// cut the string from the beginning to the index // cut the string from the beginning to the index
@ -1145,8 +1137,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
if ((mDoFormat == PR_FALSE) || preformatted || IgnoreWS(tag) == PR_FALSE) if ((mDoFormat == PR_FALSE) || preformatted || IgnoreWS(tag) == PR_FALSE)
{ {
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos += text.Length(); mColPos += text.Length();
} }
} }
@ -1155,8 +1146,7 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream){
if ((mDoFormat == PR_FALSE) || preformatted) if ((mDoFormat == PR_FALSE) || preformatted)
{ {
const nsString& text = aNode.GetText(); const nsString& text = aNode.GetText();
UnicodeToHTMLString(text); Write(text);
aStream << mBuffer;
mColPos = 0; mColPos = 0;
} }
} }
@ -1215,25 +1205,26 @@ nsHTMLContentSinkStream::AddComment(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
if(mOutput)
const nsString& name = aNode.GetText();
if (name.Equals("XIF_DOC_INFO"))
{ {
const nsString& name = aNode.GetText(); PRInt32 count=aNode.GetAttributeCount();
if (name.Equals("XIF_DOC_INFO")) for(PRInt32 i=0;i<count;i++)
{ {
PRInt32 count=aNode.GetAttributeCount(); const nsString& key=aNode.GetKeyAt(i);
for(PRInt32 i=0;i<count;i++) const nsString& value=aNode.GetValueAt(i);
{
const nsString& key=aNode.GetKeyAt(i);
const nsString& value=aNode.GetValueAt(i);
if (key.Equals("charset")) if (key.Equals("charset"))
{
if (mCharsetOverride.Length() == 0)
InitEncoder(value); InitEncoder(value);
} }
} }
else }
{ else
AddStartTag(aNode,*mOutput); {
} AddStartTag(aNode);
} }
return NS_OK; return NS_OK;
} }
@ -1242,37 +1233,17 @@ nsHTMLContentSinkStream::OpenContainer(const nsIParserNode& aNode){
/** /**
* This method is used to close a generic container. * This method is used to close a generic container.
* *
* @update 07/12/98 gpk * @update 04/30/99 gpk
* @param nsIParserNode reference to parser node interface * @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful. * @return PR_TRUE if successful.
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::CloseContainer(const nsIParserNode& aNode){ nsHTMLContentSinkStream::CloseContainer(const nsIParserNode& aNode){
if(mOutput) { AddEndTag(aNode);
AddEndTag(aNode,*mOutput);
}
return NS_OK; return NS_OK;
} }
/**
* This method is used to add a leaf to the currently
* open container.
*
* @update 07/12/98 gpk
* @param nsIParserNode reference to parser node interface
* @return PR_TRUE if successful.
*/
NS_IMETHODIMP
nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
nsresult result = NS_OK;
if(mOutput) {
result = AddLeaf(aNode,*mOutput);
}
return result;
}
/** /**
* This method gets called when the parser begins the process * This method gets called when the parser begins the process
* of building the content model via the content sink. * of building the content model via the content sink.
@ -1282,10 +1253,12 @@ nsHTMLContentSinkStream::AddLeaf(const nsIParserNode& aNode){
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLContentSinkStream::WillBuildModel(void){ nsHTMLContentSinkStream::WillBuildModel(void){
mTabLevel=-1; mTabLevel=-1;
if(mDoHeader && mOutput) { if(mDoHeader) {
(*mOutput) << gHeaderComment << endl; Write(gHeaderComment);
(*mOutput) << gDocTypeHeader << endl; Write('\n');
} Write(gDocTypeHeader);
Write('\n');
}
return NS_OK; return NS_OK;
} }

52
parser/htmlparser/src/nsHTMLContentSinkStream.h
View File

@ -54,23 +54,24 @@ class ostream;
#endif #endif
class nsIUnicodeEncoder; class nsIUnicodeEncoder;
class nsIOutputStream;
class nsHTMLContentSinkStream : public nsIHTMLContentSink { class nsHTMLContentSinkStream : public nsIHTMLContentSink {
public: public:
/**
* Standard constructor
* @update gess7/7/98
*/
nsHTMLContentSinkStream(PRBool aDoFormat = PR_TRUE, PRBool aDoHeader = PR_TRUE);
/** /**
* Constructor with associated stream. If you use this, it means that you want * Constructor with associated stream. If you use this, it means that you want
* this class to emits its output to the stream you provide. * this class to emits its output to the stream you provide.
* @update gess7/7/98 * @update gpk 04/30/99
* @param aStream -- ref to stream where you want output sent * @param aOutStream -- stream where you want output sent
* @param aOutStream -- ref to string where you want output sent
*/ */
nsHTMLContentSinkStream(ostream& aStream,PRBool aDoFormat = PR_TRUE, PRBool aDoHeader = PR_TRUE); nsHTMLContentSinkStream(nsIOutputStream* aOutStream,
nsString* aOutString,
const nsString* aCharsetOverride,
PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE);
/** /**
* virtual destructor * virtual destructor
@ -78,8 +79,6 @@ class nsHTMLContentSinkStream : public nsIHTMLContentSink {
*/ */
virtual ~nsHTMLContentSinkStream(); virtual ~nsHTMLContentSinkStream();
NS_IMETHOD_(void) SetOutputStream(ostream& aStream);
// nsISupports // nsISupports
NS_DECL_ISUPPORTS NS_DECL_ISUPPORTS
@ -127,22 +126,26 @@ public:
protected: protected:
nsresult AddLeaf(const nsIParserNode& aNode, ostream& aStream); void WriteAttributes(const nsIParserNode& aNode);
void WriteAttributes(const nsIParserNode& aNode,ostream& aStream); void AddStartTag(const nsIParserNode& aNode);
void AddStartTag(const nsIParserNode& aNode, ostream& aStream); void AddEndTag(const nsIParserNode& aNode);
void AddEndTag(const nsIParserNode& aNode, ostream& aStream); void AddIndent();
void AddIndent(ostream& aStream);
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
void UnicodeToHTMLString(const nsString& aSrc);
nsresult InitEncoder(const nsString& aCharset); nsresult InitEncoder(const nsString& aCharset);
void EncodeToBuffer(const nsString& aString);
void Write(const nsString& aString);
void Write(const char* aCharBuffer);
void Write(char aChar);
protected: protected:
ostream* mOutput; nsIOutputStream* mStream;
nsString* mString;
int mTabLevel; int mTabLevel;
PRInt32 mIndent; PRInt32 mIndent;
@ -155,13 +158,24 @@ protected:
PRBool mDoHeader; PRBool mDoHeader;
char* mBuffer; char* mBuffer;
PRInt32 mBufferSize; PRInt32 mBufferLength; // The length of the data in the buffer
PRInt32 mBufferSize; // The actual size of the buffer, regardless of the data
nsIUnicodeEncoder* mUnicodeEncoder; nsIUnicodeEncoder* mUnicodeEncoder;
nsString mCharsetOverride;
}; };
extern NS_HTMLPARS nsresult extern NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult, NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsIOutputStream* aOutStream,
const nsString* aCharsetOverride=nsnull,
PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE);
extern NS_HTMLPARS nsresult
NS_New_HTML_ContentSinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString,
PRBool aDoFormat = PR_TRUE, PRBool aDoFormat = PR_TRUE,
PRBool aDoHeader = PR_TRUE); PRBool aDoHeader = PR_TRUE);

285
parser/htmlparser/src/nsHTMLToTXTSinkStream.cpp
View File

@ -39,6 +39,8 @@
#include "nsICharsetAlias.h" #include "nsICharsetAlias.h"
#include "nsIServiceManager.h" #include "nsIServiceManager.h"
#include "nsICharsetConverterManager.h" #include "nsICharsetConverterManager.h"
#include "nsIOutputStream.h"
#include "nsFileStream.h"
static NS_DEFINE_IID(kISupportsIID, NS_ISUPPORTS_IID); 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);
@ -60,7 +62,19 @@ static PRBool IsBlockLevel(eHTMLTags aTag);
nsresult nsHTMLToTXTSinkStream::InitEncoder(const nsString& aCharset) nsresult nsHTMLToTXTSinkStream::InitEncoder(const nsString& aCharset)
{ {
nsresult res = NS_OK; nsresult res = NS_OK;
// If the converter is ucs2, then do not use a converter
nsString ucs2("ucs2");
if (aCharset.Equals(ucs2))
{
NS_IF_RELEASE(mUnicodeEncoder);
return res;
}
nsICharsetAlias* calias = nsnull; nsICharsetAlias* calias = nsnull;
res = nsServiceManager::GetService(kCharsetAliasCID, res = nsServiceManager::GetService(kCharsetAliasCID,
@ -141,58 +155,71 @@ NS_IMPL_RELEASE(nsHTMLToTXTSinkStream)
/** /**
* This method is defined in nsIParser. It is used to * This method creates a new sink, it sets the stream used
* cause the COM-like construction of an nsParser. * for the sink to aStream
* *
* @update gpk02/03/99 * @update gpk 04/30/99
* @param nsIParser** ptr to newly instantiated parser
* @return NS_xxx error result
*/ */
NS_HTMLPARS nsresult NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult) { NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(); nsIOutputStream* aStream,
const nsString* aCharsetOverride) {
NS_ASSERTION(aStream != nsnull, "a valid stream is required");
nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(aStream,nsnull);
if (nsnull == it) { if (nsnull == it) {
return NS_ERROR_OUT_OF_MEMORY; return NS_ERROR_OUT_OF_MEMORY;
} }
if (aCharsetOverride != nsnull)
it->SetCharsetOverride(aCharsetOverride);
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult); return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
} }
/** /**
* Construct a content sink stream. * This method creates a new sink, it sets the stream used
* @update gpk02/03/99 * for the sink to aStream
* @param *
* @return * @update gpk 04/30/99
*/ */
nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream() { NS_HTMLPARS nsresult
NS_INIT_REFCNT(); NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
mOutput=&cout; nsString* aString) {
mColPos = 0;
mIndent = 0; NS_ASSERTION(aString != nsnull, "a valid stream is required");
mDoOutput = PR_FALSE; nsHTMLToTXTSinkStream* it = new nsHTMLToTXTSinkStream(nsnull,aString);
mBufferSize = 0; if (nsnull == it) {
mBuffer = nsnull; return NS_ERROR_OUT_OF_MEMORY;
mUnicodeEncoder = nsnull; }
nsString ucs2("ucs2");
it->SetCharsetOverride(&ucs2);
return it->QueryInterface(kIHTMLContentSinkIID, (void **)aInstancePtrResult);
} }
/** /**
* Construct a content sink stream. * Construct a content sink stream.
* @update gpk02/03/99 * @update gpk02/03/99
* @param * @param
* @return * @return
*/ */
nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(ostream& aStream) { nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(nsIOutputStream* aStream, nsString* aString) {
NS_INIT_REFCNT(); NS_INIT_REFCNT();
mOutput = &aStream; mStream = aStream;
mColPos = 0; mColPos = 0;
mIndent = 0; mIndent = 0;
mDoOutput = PR_FALSE; mDoOutput = PR_FALSE;
mBufferSize = 0; mBufferSize = 0;
mBufferLength = 0;
mBuffer = nsnull; mBuffer = nsnull;
mUnicodeEncoder = nsnull; mUnicodeEncoder = nsnull;
mStream = aStream;
mString = aString;
} }
/** /**
* *
* @update gpk02/03/99 * @update gpk02/03/99
@ -200,7 +227,6 @@ nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream(ostream& aStream) {
* @return * @return
*/ */
nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() { nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() {
mOutput=0; //we don't own the stream we're given; just forget it.
delete [] mBuffer; delete [] mBuffer;
NS_IF_RELEASE(mUnicodeEncoder); NS_IF_RELEASE(mUnicodeEncoder);
} }
@ -208,50 +234,24 @@ nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream() {
/** /**
* *
* @update gpk02/03/99 * @update gpk04/30/99
* @param * @param
* @return * @return
*/ */
NS_IMETHODIMP_(void)
nsHTMLToTXTSinkStream::SetOutputStream(ostream& aStream){ NS_IMETHODIMP
mOutput=&aStream; nsHTMLToTXTSinkStream::SetCharsetOverride(const nsString* aCharset)
{
if (aCharset)
{
mCharsetOverride = *aCharset;
InitEncoder(mCharsetOverride);
}
return NS_OK;
} }
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void OpenTagWithAttributes(const char* theTag,const nsIParserNode& aNode,int tab,ostream& aStream,PRBool aNewline) {
}
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void OpenTag(const char* theTag,int tab,ostream& aStream,PRBool aNewline) {
}
/**
*
* @update gpk02/03/99
* @param
* @return
*/
static
void CloseTag(const char* theTag,int tab,ostream& aStream) {
}
/** /**
* This method gets called by the parser when it encounters * This method gets called by the parser when it encounters
* a title tag and wants to set the document title in the sink. * a title tag and wants to set the document title in the sink.
@ -463,38 +463,38 @@ void nsHTMLToTXTSinkStream::EnsureBufferSize(PRInt32 aNewSize)
mBufferSize = 2*aNewSize+1; // make the twice as large mBufferSize = 2*aNewSize+1; // make the twice as large
mBuffer = new char[mBufferSize]; mBuffer = new char[mBufferSize];
mBuffer[0] = 0; mBuffer[0] = 0;
mBufferLength = 0;
} }
} }
void nsHTMLToTXTSinkStream::UnicodeToTXTString(const nsString& aSrc) void nsHTMLToTXTSinkStream::EncodeToBuffer(const nsString& aSrc)
{ {
NS_ASSERTION(mUnicodeEncoder != nsnull,"The unicode encoder needs to be initialized");
if (mUnicodeEncoder == nsnull)
return;
#define CH_NBSP 160 #define CH_NBSP 160
PRInt32 length = aSrc.Length(); PRInt32 length = aSrc.Length();
nsresult result; nsresult result;
PRInt32 bufferLength;
if (mUnicodeEncoder == nsnull) if (mUnicodeEncoder != nsnull && length > 0)
InitEncoder("");
if (length > 0)
{ {
EnsureBufferSize(length); EnsureBufferSize(length);
bufferLength = mBufferSize; mBufferLength = mBufferSize;
mUnicodeEncoder->Reset(); mUnicodeEncoder->Reset();
result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &bufferLength); result = mUnicodeEncoder->Convert(aSrc.GetUnicode(), &length, mBuffer, &mBufferLength);
mBuffer[bufferLength] = 0; mBuffer[mBufferLength] = 0;
PRInt32 temp = bufferLength; PRInt32 temp = mBufferLength;
if (NS_SUCCEEDED(result)) if (NS_SUCCEEDED(result))
result = mUnicodeEncoder->Finish(mBuffer,&temp); result = mUnicodeEncoder->Finish(mBuffer,&temp);
for (PRInt32 i = 0; i < bufferLength; i++) for (PRInt32 i = 0; i < mBufferLength; i++)
{ {
if (mBuffer[i] == char(CH_NBSP)) if (mBuffer[i] == char(CH_NBSP))
mBuffer[i] = ' '; mBuffer[i] = ' ';
@ -504,76 +504,47 @@ void nsHTMLToTXTSinkStream::UnicodeToTXTString(const nsString& aSrc)
} }
NS_IMETHODIMP
nsHTMLToTXTSinkStream::GetStringBuffer(nsString & aStrBuffer)
{
aStrBuffer = mStrBuffer;
return NS_OK;
}
/** /**
* This gets called by the parser when you want to add * Write places the contents of aString into either the output stream
* a leaf node to the current container in the content * or the output string.
* model. * When going to the stream, all data is run through the encoder
* *
* @updated gpk 06/18/98 * @updated gpk02/03/99
* @param * @param
* @return * @return
*/ */
nsresult void nsHTMLToTXTSinkStream::Write(const nsString& aString)
nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode, ostream& aStream)
{ {
eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
const nsString& text = aNode.GetText();
if (mDoOutput == PR_FALSE) // If a encoder is being used then convert first convert the input string
return NS_OK; if (mUnicodeEncoder != nsnull)
if (type == eHTMLTag_text) {
UnicodeToTXTString(text);
aStream << mBuffer;
mStrBuffer.Append(mBuffer);
mColPos += text.Length();
}
else if (type == eHTMLTag_entity)
{ {
const nsString& text = aNode.GetText(); EncodeToBuffer(aString);
UnicodeToTXTString(text); if (mStream != nsnull)
PRInt32 entity = NS_EntityToUnicode(mBuffer);
if (entity < 256)
{ {
char ch = (char)entity; nsOutputStream out(mStream);
aStream << ch; out.write(mBuffer,mBufferLength);
mColPos++; }
if (mString != nsnull)
{
mString->Append(mBuffer);
} }
} }
else if (type == eHTMLTag_whitespace) else
{ {
if (PR_TRUE) if (mStream != nsnull)
{ {
const nsString& text = aNode.GetText(); nsOutputStream out(mStream);
UnicodeToTXTString(text); const PRUnichar* unicode = aString.GetUnicode();
aStream << mBuffer; PRUint32 length = aString.Length();
mStrBuffer.Append(mBuffer); out.write(unicode,length);
mColPos += text.Length(); }
else
{
mString->Append(aString);
} }
} }
else if (type == eHTMLTag_br)
{
if (PR_TRUE)
{
aStream << endl;
mStrBuffer.Append("\n");
mColPos += 1;
}
}
return NS_OK;
} }
@ -628,7 +599,10 @@ nsHTMLToTXTSinkStream::OpenContainer(const nsIParserNode& aNode){
const nsString& value=aNode.GetValueAt(i); const nsString& value=aNode.GetValueAt(i);
if (key.Equals("charset")) if (key.Equals("charset"))
InitEncoder(value); {
if (mCharsetOverride.Length() == 0)
InitEncoder(value);
}
} }
} }
@ -657,9 +631,8 @@ nsHTMLToTXTSinkStream::CloseContainer(const nsIParserNode& aNode){
{ {
if (mColPos != 0) if (mColPos != 0)
{ {
if (mOutput) nsString temp("\n");
*mOutput << endl; Write(temp);
mStrBuffer.Append("\n");
mColPos = 0; mColPos = 0;
} }
} }
@ -677,11 +650,53 @@ nsHTMLToTXTSinkStream::CloseContainer(const nsIParserNode& aNode){
*/ */
NS_IMETHODIMP NS_IMETHODIMP
nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode){ nsHTMLToTXTSinkStream::AddLeaf(const nsIParserNode& aNode){
nsresult result = NS_OK; eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
if(mOutput) {
result = AddLeaf(aNode,*mOutput); const nsString& text = aNode.GetText();
if (mDoOutput == PR_FALSE)
return NS_OK;
if (type == eHTMLTag_text) {
Write(text);
mColPos += text.Length();
}
else if (type == eHTMLTag_entity)
{
const nsString& text = aNode.GetText();
EncodeToBuffer(text);
PRUnichar entity = NS_EntityToUnicode(mBuffer);
nsString temp;
temp.Append(entity);
Write(temp);
mColPos++;
} }
return result; else if (type == eHTMLTag_whitespace)
{
if (PR_TRUE)
{
const nsString& text = aNode.GetText();
Write(text);
mColPos += text.Length();
}
}
else if (type == eHTMLTag_br)
{
nsString temp("\n");
Write(text);
mColPos++;
}
else if (type == eHTMLTag_newline)
{
nsString temp("\n");
Write(text);
mColPos++;
}
return NS_OK;
} }

43
parser/htmlparser/src/nsHTMLToTXTSinkStream.h
View File

@ -47,12 +47,9 @@
{0xa39c6bff, 0x15f0, 0x11d2, \ {0xa39c6bff, 0x15f0, 0x11d2, \
{0x80, 0x41, 0x0, 0x10, 0x4b, 0x98, 0x3f, 0xd4}} {0x80, 0x41, 0x0, 0x10, 0x4b, 0x98, 0x3f, 0xd4}}
#ifndef XP_MAC
class ostream;
#endif
class nsIUnicodeEncoder; class nsIUnicodeEncoder;
class nsIOutputStream;
class nsHTMLToTXTSinkStream : public nsIHTMLContentSink { class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
public: public:
@ -61,8 +58,7 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
* Standard constructor * Standard constructor
* @update gpk02/03/99 * @update gpk02/03/99
*/ */
nsHTMLToTXTSinkStream(); nsHTMLToTXTSinkStream(nsIOutputStream* aOutStream, nsString* aOutString);
nsHTMLToTXTSinkStream(ostream& aStream);
/** /**
* virtual destructor * virtual destructor
@ -70,8 +66,8 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
*/ */
virtual ~nsHTMLToTXTSinkStream(); virtual ~nsHTMLToTXTSinkStream();
NS_IMETHOD_(void) SetOutputStream(ostream& aStream); NS_IMETHOD SetCharsetOverride(const nsString* aCharset);
NS_IMETHOD GetStringBuffer(nsString & aStrBuffer);
// nsISupports // nsISupports
NS_DECL_ISUPPORTS NS_DECL_ISUPPORTS
@ -113,32 +109,47 @@ class nsHTMLToTXTSinkStream : public nsIHTMLContentSink {
NS_IMETHOD BeginContext(PRInt32 aPosition); NS_IMETHOD BeginContext(PRInt32 aPosition);
NS_IMETHOD EndContext(PRInt32 aPosition); NS_IMETHOD EndContext(PRInt32 aPosition);
protected: protected:
nsresult AddLeaf(const nsIParserNode& aNode, ostream& aStream);
void WriteAttributes(const nsIParserNode& aNode,ostream& aStream);
void EnsureBufferSize(PRInt32 aNewSize); void EnsureBufferSize(PRInt32 aNewSize);
void UnicodeToTXTString(const nsString& aSrc);
nsresult InitEncoder(const nsString& aCharset); nsresult InitEncoder(const nsString& aCharset);
void Write(const nsString& aString);
void EncodeToBuffer(const nsString& aString);
protected: protected:
ostream* mOutput; nsIOutputStream* mStream;
nsString* mString;
PRInt32 mIndent; PRInt32 mIndent;
PRInt32 mColPos; PRInt32 mColPos;
PRBool mDoOutput; PRBool mDoOutput;
char* mBuffer;
PRInt32 mBufferSize;
nsString mStrBuffer; char* mBuffer;
PRInt32 mBufferLength; // The length of the data in the buffer
PRInt32 mBufferSize; // The actual size of the buffer, regardless of the data
nsIUnicodeEncoder* mUnicodeEncoder; nsIUnicodeEncoder* mUnicodeEncoder;
nsString mCharsetOverride;
}; };
extern NS_HTMLPARS nsresult extern NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult); NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsIOutputStream* aOutStream,
const nsString* aCharsetOverride=nsnull);
extern NS_HTMLPARS nsresult
NS_New_HTMLToTXT_SinkStream(nsIHTMLContentSink** aInstancePtrResult,
nsString* aOutString);
#endif #endif