41868 and 45994: Call parser to determine inline vs. block. r=harishd

git-svn-id: svn://10.0.0.236/trunk@77414 18797224-902f-48f8-a5cc-f745e15eee43
This commit is contained in:
akkana%netscape.com
2000-08-28 23:47:11 +00:00
parent e205930fa0
commit b7b56f122f
12 changed files with 500 additions and 636 deletions

View File

@@ -50,20 +50,15 @@
static NS_DEFINE_CID(kSaveAsCharsetCID, NS_SAVEASCHARSET_CID);
static NS_DEFINE_CID(kEntityConverterCID, NS_ENTITYCONVERTER_CID);
static NS_DEFINE_IID(kCParserIID, NS_IPARSER_IID);
static NS_DEFINE_IID(kCParserCID, NS_PARSER_IID);
static char* gHeaderComment = "<!-- This page was created by the Gecko output system. -->";
static char* gDocTypeHeader = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">";
const int gTabSize=2;
static const nsString gMozDirty = NS_ConvertToString("_moz_dirty");
static PRBool IsInline(eHTMLTags aTag);
static PRBool IsBlockLevel(eHTMLTags aTag);
static PRInt32 BreakBeforeOpen(eHTMLTags aTag);
static PRInt32 BreakAfterOpen(eHTMLTags aTag);
static PRInt32 BreakBeforeClose(eHTMLTags aTag);
static PRInt32 BreakAfterClose(eHTMLTags aTag);
static PRBool IndentChildren(eHTMLTags aTag);
/**
* This method gets called as part of our COM-like interfaces.
* Its purpose is to create an interface to parser object
@@ -115,6 +110,7 @@ nsHTMLContentSinkStream::nsHTMLContentSinkStream()
mLowerCaseTags = PR_TRUE;
memset(mHTMLTagStack,0,sizeof(mHTMLTagStack));
memset(mDirtyStack,0,sizeof(mDirtyStack));
mDTD = 0;
mHTMLStackPos = 0;
mColPos = 0;
mIndent = 0;
@@ -166,8 +162,10 @@ nsHTMLContentSinkStream::Initialize(nsIOutputStream* aOutStream,
nsHTMLContentSinkStream::~nsHTMLContentSinkStream()
{
if (mBuffer)
nsMemory::Free(mBuffer);
NS_IF_RELEASE(mDTD);
if (mBuffer)
nsMemory::Free(mBuffer);
}
/**
@@ -1111,6 +1109,159 @@ nsHTMLContentSinkStream::CloseContainer(const nsIParserNode& aNode){
return NS_OK;
}
/**
* Find out from the parser whether a node is a block node.
*/
PRBool nsHTMLContentSinkStream::IsBlockLevel(eHTMLTags aTag)
{
if (!mDTD)
{
nsCOMPtr<nsIParser> parser;
nsresult rv = nsComponentManager::CreateInstance(kCParserCID,
nsnull,
kCParserIID,
(void **)&parser);
if (NS_FAILED(rv)) return rv;
if (!parser) return NS_ERROR_FAILURE;
nsAutoString htmlmime (NS_LITERAL_STRING("text/html"));
rv = parser->CreateCompatibleDTD(&mDTD, 0, eViewNormal,
&htmlmime, eDTDMode_transitional);
/* XXX Note: We output linebreaks for blocks.
I.e. we output linebreaks for "unknown" inline tags.
I just hunted such a bug for <q>, same for <ins>, <col> etc..
Better fallback to inline. /BenB */
if (NS_FAILED(rv) || !mDTD)
return PR_FALSE;
}
// Now we can get the inline status from the DTD:
return mDTD->IsBlockElement(aTag, eHTMLTag_unknown);
}
/**
* **** Pretty Printing Methods ******
*
*/
/**
* Desired line break state before the open tag.
*/
PRBool nsHTMLContentSinkStream::BreakBeforeOpen(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
result = PR_FALSE;
break;
default:
result = IsBlockLevel(aTag);
}
return result;
}
/**
* Desired line break state after the open tag.
*/
PRBool nsHTMLContentSinkStream::BreakAfterOpen(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_body:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_table:
case eHTMLTag_tbody:
case eHTMLTag_style:
case eHTMLTag_br:
result = PR_TRUE;
break;
default:
break;
}
return result;
}
/**
* Desired line break state before the close tag.
*/
PRBool nsHTMLContentSinkStream::BreakBeforeClose(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_head:
case eHTMLTag_body:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_table:
case eHTMLTag_tbody:
case eHTMLTag_style:
result = PR_TRUE;
break;
default:
break;
}
return result;
}
/**
* Desired line break state after the close tag.
*/
PRBool nsHTMLContentSinkStream::BreakAfterClose(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_tr:
case eHTMLTag_th:
case eHTMLTag_td:
case eHTMLTag_pre:
result = PR_TRUE;
break;
default:
result = IsBlockLevel(aTag);
}
return result;
}
/**
* Indent/outdent when the open/close tags are encountered.
* This implies that BreakAfterOpen() and BreakBeforeClose()
* are true no matter what those methods return.
*/
PRBool nsHTMLContentSinkStream::IndentChildren(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_table:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_tbody:
case eHTMLTag_form:
case eHTMLTag_frameset:
result = PR_TRUE;
break;
default:
result = PR_FALSE;
break;
}
return result;
}
/**
* This method gets called when the parser begins the process
@@ -1176,181 +1327,3 @@ nsHTMLContentSinkStream::NotifyError(const nsParserError* aError)
return NS_OK;
}
/////////////////////////////////////////////////////////////
//// Useful static methods
/////////////////////////////////////////////////////////////
static PRBool IsInline(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_a:
case eHTMLTag_address:
case eHTMLTag_big:
case eHTMLTag_blink:
case eHTMLTag_b:
case eHTMLTag_br:
case eHTMLTag_cite:
case eHTMLTag_code:
case eHTMLTag_dfn:
case eHTMLTag_em:
case eHTMLTag_font:
case eHTMLTag_img:
case eHTMLTag_i:
case eHTMLTag_kbd:
case eHTMLTag_keygen:
case eHTMLTag_nobr:
case eHTMLTag_samp:
case eHTMLTag_small:
case eHTMLTag_spacer:
case eHTMLTag_span:
case eHTMLTag_strike:
case eHTMLTag_strong:
case eHTMLTag_sub:
case eHTMLTag_sup:
case eHTMLTag_textarea:
case eHTMLTag_tt:
case eHTMLTag_u:
case eHTMLTag_var:
case eHTMLTag_wbr:
result = PR_TRUE;
break;
default:
break;
}
return result;
}
static PRBool IsBlockLevel(eHTMLTags aTag)
{
return !IsInline(aTag);
}
/**
* **** Pretty Printing Methods ******
*
*/
/**
* Desired line break state before the open tag.
*/
static PRBool BreakBeforeOpen(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
result = PR_FALSE;
break;
default:
result = IsBlockLevel(aTag);
}
return result;
}
/**
* Desired line break state after the open tag.
*/
static PRBool BreakAfterOpen(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_body:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_table:
case eHTMLTag_tbody:
case eHTMLTag_style:
case eHTMLTag_br:
result = PR_TRUE;
break;
default:
break;
}
return result;
}
/**
* Desired line break state before the close tag.
*/
static PRBool BreakBeforeClose(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_head:
case eHTMLTag_body:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_table:
case eHTMLTag_tbody:
case eHTMLTag_style:
result = PR_TRUE;
break;
default:
break;
}
return result;
}
/**
* Desired line break state after the close tag.
*/
static PRBool BreakAfterClose(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_html:
case eHTMLTag_tr:
case eHTMLTag_th:
case eHTMLTag_td:
case eHTMLTag_pre:
result = PR_TRUE;
break;
default:
result = IsBlockLevel(aTag);
}
return result;
}
/**
* Indent/outdent when the open/close tags are encountered.
* This implies that BreakAfterOpen() and BreakBeforeClose()
* are true no matter what those methods return.
*/
static PRBool IndentChildren(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
{
case eHTMLTag_table:
case eHTMLTag_ul:
case eHTMLTag_ol:
case eHTMLTag_tbody:
case eHTMLTag_form:
case eHTMLTag_frameset:
result = PR_TRUE;
break;
default:
result = PR_FALSE;
break;
}
return result;
}

View File

@@ -62,6 +62,7 @@ class ostream;
class nsIParserNode;
class nsIOutputStream;
class nsIDTD;
class nsIHTMLContentSinkStream : public nsIHTMLContentSink {
public:
@@ -102,7 +103,6 @@ class nsHTMLContentSinkStream : public nsIHTMLContentSinkStream
const nsAReadableString* aCharsetOverride,
PRUint32 aFlags);
/*******************************************************************
* The following methods are inherited from nsIContentSink.
* Please see that file for details.
@@ -147,6 +147,13 @@ public:
protected:
PRBool IsBlockLevel(eHTMLTags aTag);
PRInt32 BreakBeforeOpen(eHTMLTags aTag);
PRInt32 BreakAfterOpen(eHTMLTags aTag);
PRInt32 BreakBeforeClose(eHTMLTags aTag);
PRInt32 BreakAfterClose(eHTMLTags aTag);
PRBool IndentChildren(eHTMLTags aTag);
void WriteAttributes(const nsIParserNode& aNode);
void AddStartTag(const nsIParserNode& aNode);
void AddEndTag(const nsIParserNode& aNode);
@@ -170,6 +177,8 @@ protected:
nsIOutputStream* mStream;
nsAWritableString* mString;
nsIDTD* mDTD;
int mTabLevel;
char* mBuffer;
PRInt32 mBufferSize;

View File

@@ -56,6 +56,8 @@
static NS_DEFINE_CID(kCharsetConverterManagerCID, NS_ICHARSETCONVERTERMANAGER_CID);
static NS_DEFINE_CID(kLWBrkCID, NS_LWBRK_CID);
static NS_DEFINE_CID(kPrefServiceCID, NS_PREF_CID);
static NS_DEFINE_IID(kCParserIID, NS_IPARSER_IID);
static NS_DEFINE_IID(kCParserCID, NS_PARSER_IID);
#define PREF_STRUCTS "converter.html2txt.structs"
#define PREF_HEADER_STRATEGY "converter.html2txt.header_strategy"
@@ -73,8 +75,6 @@ const PRInt32 gIndentSizeList = (gTabSize > gOLNumberWidth+3) ? gTabSize: gOLNu
// Indention of non-first lines of ul and ol
const PRInt32 gIndentSizeDD = gTabSize; // Indention of <dd>
static PRBool IsInline(eHTMLTags aTag);
static PRBool IsBlockLevel(eHTMLTags aTag);
static PRInt32 HeaderLevel(eHTMLTags aTag);
static PRInt32 unicharwidth(PRUnichar ucs);
static PRInt32 unicharwidth(const PRUnichar* pwcs, PRInt32 n);
@@ -186,6 +186,7 @@ static const PRUint32 OLStackSize = 100;
nsHTMLToTXTSinkStream::nsHTMLToTXTSinkStream()
{
NS_INIT_REFCNT();
mDTD = 0;
mColPos = 0;
mIndent = 0;
mCiteQuoteLevel = 0;
@@ -236,6 +237,7 @@ nsHTMLToTXTSinkStream::~nsHTMLToTXTSinkStream()
delete[] mBuffer;
delete[] mTagStack;
delete[] mOLStack;
NS_IF_RELEASE(mDTD);
NS_IF_RELEASE(mUnicodeEncoder);
NS_IF_RELEASE(mLineBreaker);
}
@@ -416,6 +418,7 @@ nsHTMLToTXTSinkStream::AddProcessingInstruction(const nsIParserNode& aNode){
NS_IMETHODIMP
nsHTMLToTXTSinkStream::AddDocTypeDecl(const nsIParserNode& aNode, PRInt32 aMode)
{
// Should probably set DTD
return NS_OK;
}
@@ -1603,62 +1606,31 @@ nsHTMLToTXTSinkStream::NotifyError(const nsParserError* aError)
return NS_OK;
}
PRBool IsInline(eHTMLTags aTag)
PRBool nsHTMLToTXTSinkStream::IsBlockLevel(eHTMLTags aTag)
{
PRBool result = PR_FALSE;
switch (aTag)
if (!mDTD)
{
case eHTMLTag_a:
case eHTMLTag_address:
case eHTMLTag_b:
case eHTMLTag_big:
case eHTMLTag_blink:
case eHTMLTag_br:
case eHTMLTag_cite:
case eHTMLTag_code:
case eHTMLTag_dfn:
case eHTMLTag_del:
case eHTMLTag_em:
case eHTMLTag_font:
case eHTMLTag_i:
case eHTMLTag_img:
case eHTMLTag_ins:
case eHTMLTag_kbd:
case eHTMLTag_keygen:
case eHTMLTag_nobr:
case eHTMLTag_q:
case eHTMLTag_samp:
case eHTMLTag_small:
case eHTMLTag_spacer:
case eHTMLTag_span:
case eHTMLTag_strike:
case eHTMLTag_strong:
case eHTMLTag_sub:
case eHTMLTag_sup:
case eHTMLTag_td:
case eHTMLTag_textarea:
case eHTMLTag_th:
case eHTMLTag_tt:
case eHTMLTag_u:
case eHTMLTag_var:
case eHTMLTag_wbr:
result = PR_TRUE;
break;
nsCOMPtr<nsIParser> parser;
nsresult rv = nsComponentManager::CreateInstance(kCParserCID,
nsnull,
kCParserIID,
(void **)&parser);
if (NS_FAILED(rv)) return rv;
if (!parser) return NS_ERROR_FAILURE;
default:
break;
}
return result;
}
PRBool IsBlockLevel(eHTMLTags aTag)
{
return !IsInline(aTag);
nsAutoString htmlmime (NS_LITERAL_STRING("text/html"));
rv = parser->CreateCompatibleDTD(&mDTD, 0, eViewNormal,
&htmlmime, eDTDMode_transitional);
/* XXX Note: We output linebreaks for blocks.
I.e. we output linebreaks for "unknown" inline tags.
I just hunted such a bug for <q>, same for <ins>, <col> etc..
Better fallback to inline. /BenB */
if (NS_FAILED(rv) || !mDTD)
return PR_FALSE;
}
// Now we can get the inline status from the DTD:
return mDTD->IsBlockElement(aTag, eHTMLTag_unknown);
}
/*

View File

@@ -46,6 +46,9 @@
#include "nsHTMLTags.h"
#include "nsParserCIID.h"
#include "nsCOMPtr.h"
#include "nsHTMLTokens.h" // for eHTMLTags
class nsIDTD;
#define NS_IHTMLTOTEXTSINKSTREAM_IID \
{0xa39c6bff, 0x15f0, 0x11d2, \
@@ -157,6 +160,7 @@ protected:
PRBool DoOutput();
PRBool MayWrap();
PRBool IsBlockLevel(eHTMLTags aTag);
protected:
nsIOutputStream* mStream;
@@ -166,6 +170,8 @@ protected:
nsAWritableString* mString;
nsString mCurrentLine;
nsIDTD* mDTD;
PRInt32 mIndent;
// mInIndentString keeps a header that has to be written in the indent.
// That could be, for instance, the bullet in a bulleted list.