The HTML parser for data queries. More...

#include <htmlpars.hxx>

Inheritance diagram for ScHTMLQueryParser:

Collaboration diagram for ScHTMLQueryParser:

Public Member Functions
	ScHTMLQueryParser (EditEngine pEditEngine, ScDocument pDoc)

virtual	~ScHTMLQueryParser () override

virtual ErrCode	Read (SvStream &rStrm, const OUString &rBaseURL) override

virtual const ScHTMLTable *	GetGlobalTable () const override
	Returns the "global table" which contains the entire HTML document. More...

Public Member Functions inherited from ScHTMLParser
	ScHTMLParser (EditEngine pEditEngine, ScDocument pDoc)
	The destination document. More...

virtual	~ScHTMLParser () override

virtual ErrCode	Read (SvStream &rStrm, const OUString &rBaseURL) override=0

ScHTMLStyles &	GetStyles ()

ScDocument &	GetDoc ()

virtual const ScHTMLTable *	GetGlobalTable () const =0
	Returns the "global table" which contains the entire HTML document. More...

Public Member Functions inherited from ScEEParser
	ScEEParser (EditEngine *)

virtual	~ScEEParser ()

virtual ErrCode	Read (SvStream &, const OUString &rBaseURL)=0

const ColWidthsMap &	GetColWidths () const

ColWidthsMap &	GetColWidths ()

void	GetDimensions (SCCOL &nCols, SCROW &nRows) const

size_t	ListSize () const

ScEEParseEntry *	ListEntry (size_t index)

const ScEEParseEntry *	ListEntry (size_t index) const

Private Types
typedef ::std::unique_ptr< ScHTMLGlobalTable >	ScHTMLGlobalTablePtr

Private Member Functions
void	ProcessToken (const HtmlImportInfo &rInfo)
	Handles all possible tags in the HTML document. More...

void	InsertText (const HtmlImportInfo &rInfo)
	Inserts a text portion into current entry. More...

void	FontOn (const HtmlImportInfo &rInfo)
	Processes the <font> tag. More...

void	MetaOn (const HtmlImportInfo &rInfo)
	Processes the <meta> tag. More...

void	TitleOn ()
	Opens the title of the HTML document (<title> tag). More...

void	TitleOff (const HtmlImportInfo &rInfo)
	Closes the title of the HTML document (</title> tag). More...

void	TableOn (const HtmlImportInfo &rInfo)
	Opens a new table at the current position. More...

void	TableOff (const HtmlImportInfo &rInfo)
	Closes the current table. More...

void	PreOn (const HtmlImportInfo &rInfo)
	Opens a new table based on preformatted text. More...

void	PreOff (const HtmlImportInfo &rInfo)
	Closes the current preformatted text table. More...

void	CloseTable (const HtmlImportInfo &rInfo)
	Closes the current table, regardless on opening tag. More...

void	ParseStyle (std::u16string_view rStrm)

	DECL_LINK (HTMLImportHdl, HtmlImportInfo &, void)

Private Attributes
OUStringBuffer	maTitle

ScHTMLGlobalTablePtr	mxGlobTable
	The title of the document. More...

ScHTMLTable *	mpCurrTable
	Contains the entire imported document. More...

ScHTMLTableId	mnUnusedId
	Pointer to current table (performance). More...

bool	mbTitleOn
	First unused table identifier. More...

Additional Inherited Members
Protected Member Functions inherited from ScEEParser
void	NewActEntry (const ScEEParseEntry *)

Protected Attributes inherited from ScHTMLParser
sal_uInt32	maFontHeights [SC_HTML_FONTSIZES]

ScDocument *	mpDoc

Protected Attributes inherited from ScEEParser
EditEngine *	pEdit

rtl::Reference< SfxItemPool >	pPool

rtl::Reference< SfxItemPool >	pDocPool

std::vector< std::shared_ptr< ScEEParseEntry > >	maList

std::shared_ptr< ScEEParseEntry >	mxActEntry

ColWidthsMap	maColWidths

int	nRtfLastToken

SCCOL	nColCnt

SCROW	nRowCnt

SCCOL	nColMax

SCROW	nRowMax

Detailed Description

The HTML parser for data queries.

Focuses on data import, not on layout.

Builds the table structure correctly, ignores extended formatting like pictures or column widths.

Definition at line 579 of file htmlpars.hxx.

Member Typedef Documentation

◆ ScHTMLGlobalTablePtr

typedef ::std::unique_ptr< ScHTMLGlobalTable > ScHTMLQueryParser::ScHTMLGlobalTablePtr

private

Definition at line 622 of file htmlpars.hxx.

Constructor & Destructor Documentation

◆ ScHTMLQueryParser()

ScHTMLQueryParser::ScHTMLQueryParser	(	EditEngine *	pEditEngine,
		ScDocument *	pDoc
	)

explicit

Definition at line 2772 of file htmlpars.cxx.

References ScEEParser::maList, mnUnusedId, mpCurrTable, mxGlobTable, ScEEParser::pEdit, and ScEEParser::pPool.

◆ ~ScHTMLQueryParser()

ScHTMLQueryParser::~ScHTMLQueryParser ( )

overridevirtual

Definition at line 2782 of file htmlpars.cxx.

Member Function Documentation

◆ CloseTable()

void ScHTMLQueryParser::CloseTable ( const HtmlImportInfo & rInfo )

private

Closes the current table, regardless on opening tag.

Definition at line 3020 of file htmlpars.cxx.

References ScHTMLTable::CloseTable(), and mpCurrTable.

◆ DECL_LINK()

ScHTMLQueryParser::DECL_LINK	(	HTMLImportHdl	,
		HtmlImportInfo &	,
		void
	)

private

◆ FontOn()

void ScHTMLQueryParser::FontOn ( const HtmlImportInfo & rInfo )

private

Processes the <font> tag.

Definition at line 2921 of file htmlpars.cxx.

References ScGlobal::addToken(), ATTR_FONT(), ATTR_FONT_COLOR(), ATTR_FONT_HEIGHT(), FAMILY_DONTKNOW, o3tl::getToken(), ScHTMLParser::maFontHeights, mpCurrTable, nPos, PITCH_DONTKNOW, HtmlImportInfo::pParser, ScHTMLTable::PutItem(), SC_HTML_FONTSIZES, and comphelper::string::strip().

Referenced by ProcessToken().

◆ GetGlobalTable()

const ScHTMLTable * ScHTMLQueryParser::GetGlobalTable ( ) const

overridevirtual

Returns the "global table" which contains the entire HTML document.

Implements ScHTMLParser.

Definition at line 2824 of file htmlpars.cxx.

References mxGlobTable.

◆ InsertText()

void ScHTMLQueryParser::InsertText ( const HtmlImportInfo & rInfo )

private

Inserts a text portion into current entry.

Definition at line 2914 of file htmlpars.cxx.

References HtmlImportInfo::aText, maTitle, mbTitleOn, mpCurrTable, and ScHTMLTable::PutText().

Referenced by ProcessToken(), and TitleOff().

◆ MetaOn()

void ScHTMLQueryParser::MetaOn ( const HtmlImportInfo & rInfo )

private

Processes the <meta> tag.

Definition at line 2963 of file htmlpars.cxx.

References ScDocument::GetDocumentShell(), SfxObjectShell::GetHeaderAttributes(), SfxObjectShell::GetModel(), ScHTMLParser::mpDoc, HTMLParser::ParseMetaOptions(), and HtmlImportInfo::pParser.

Referenced by ProcessToken().

◆ ParseStyle()

void ScHTMLQueryParser::ParseStyle ( std::u16string_view rStrm )

private

Definition at line 3099 of file htmlpars.cxx.

References aStr, ScHTMLParser::GetStyles(), OUStringToOString(), rStrm, and SAL_WARN.

Referenced by ProcessToken().

◆ PreOff()

void ScHTMLQueryParser::PreOff ( const HtmlImportInfo & rInfo )

private

Closes the current preformatted text table.

Definition at line 3015 of file htmlpars.cxx.

References mpCurrTable, and ScHTMLTable::PreOff().

Referenced by ProcessToken().

◆ PreOn()

void ScHTMLQueryParser::PreOn ( const HtmlImportInfo & rInfo )

private

Opens a new table based on preformatted text.

Definition at line 3010 of file htmlpars.cxx.

References mpCurrTable, and ScHTMLTable::PreOn().

Referenced by ProcessToken().

◆ ProcessToken()

void ScHTMLQueryParser::ProcessToken ( const HtmlImportInfo & rInfo )

private

Handles all possible tags in the HTML document.

TODO: store current font size, use following size

TODO: store current font size, use preceding size

Definition at line 2829 of file htmlpars.cxx.

References ScHTMLTable::AnchorOn(), HtmlImportInfo::aText, ATTR_FONT_HEIGHT(), ATTR_FONT_POSTURE(), ATTR_FONT_UNDERLINE(), ATTR_FONT_WEIGHT(), ScHTMLTable::BodyOff(), ScHTMLTable::BodyOn(), ScHTMLTable::BreakOn(), ScHTMLTable::CaptionOff(), ScHTMLTable::CaptionOn(), ScHTMLTable::DataOff(), ScHTMLTable::DataOn(), FontOn(), ScHTMLTable::HeadingOn(), InsertText(), ITALIC_NORMAL, LINESTYLE_SINGLE, ScHTMLParser::maFontHeights, MetaOn(), mpCurrTable, HtmlImportInfo::nToken, ParseStyle(), PreOff(), PreOn(), ScHTMLTable::PutItem(), ScHTMLTable::RowOff(), ScHTMLTable::RowOn(), TableOff(), TableOn(), TitleOff(), TitleOn(), and WEIGHT_BOLD.

◆ Read()

ErrCode ScHTMLQueryParser::Read	(	SvStream &	rStrm,
		const OUString &	rBaseURL
	)

overridevirtual

◆ TableOff()

void ScHTMLQueryParser::TableOff ( const HtmlImportInfo & rInfo )

private

Closes the current table.

Definition at line 3005 of file htmlpars.cxx.

References mpCurrTable, and ScHTMLTable::TableOff().

Referenced by ProcessToken().

◆ TableOn()

void ScHTMLQueryParser::TableOn ( const HtmlImportInfo & rInfo )

private

Opens a new table at the current position.

Definition at line 3000 of file htmlpars.cxx.

References mpCurrTable, and ScHTMLTable::TableOn().

Referenced by ProcessToken().

◆ TitleOff()

void ScHTMLQueryParser::TitleOff ( const HtmlImportInfo & rInfo )

private

Closes the title of the HTML document (</title> tag).

Definition at line 2983 of file htmlpars.cxx.

References ScDocument::GetDocumentShell(), SfxObjectShell::GetModel(), InsertText(), maTitle, mbTitleOn, and ScHTMLParser::mpDoc.

Referenced by ProcessToken().

◆ TitleOn()

void ScHTMLQueryParser::TitleOn ( )

private

Opens the title of the HTML document (<title> tag).

Definition at line 2977 of file htmlpars.cxx.

References maTitle, and mbTitleOn.

Referenced by ProcessToken().

Member Data Documentation

◆ maTitle

OUStringBuffer ScHTMLQueryParser::maTitle

private

Definition at line 624 of file htmlpars.hxx.

Referenced by InsertText(), TitleOff(), and TitleOn().

◆ mbTitleOn

bool ScHTMLQueryParser::mbTitleOn

private

First unused table identifier.

Definition at line 628 of file htmlpars.hxx.

Referenced by InsertText(), TitleOff(), and TitleOn().

◆ mnUnusedId

ScHTMLTableId ScHTMLQueryParser::mnUnusedId

private

Pointer to current table (performance).

Definition at line 627 of file htmlpars.hxx.

Referenced by ScHTMLQueryParser().

◆ mpCurrTable

ScHTMLTable* ScHTMLQueryParser::mpCurrTable

private

Contains the entire imported document.

Definition at line 626 of file htmlpars.hxx.

Referenced by CloseTable(), FontOn(), InsertText(), PreOff(), PreOn(), ProcessToken(), ScHTMLQueryParser(), TableOff(), and TableOn().

◆ mxGlobTable

ScHTMLGlobalTablePtr ScHTMLQueryParser::mxGlobTable

private

The title of the document.

Definition at line 625 of file htmlpars.hxx.

Referenced by GetGlobalTable(), Read(), and ScHTMLQueryParser().

The documentation for this class was generated from the following files:

Public Member Functions

Private Types

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ ScHTMLGlobalTablePtr

Constructor & Destructor Documentation

◆ ScHTMLQueryParser()

◆ ~ScHTMLQueryParser()

Member Function Documentation

◆ CloseTable()

◆ DECL_LINK()

◆ FontOn()

◆ GetGlobalTable()

◆ InsertText()

◆ MetaOn()

◆ ParseStyle()

◆ PreOff()

◆ PreOn()

◆ ProcessToken()

◆ Read()

◆ TableOff()

◆ TableOn()

◆ TitleOff()

◆ TitleOn()

Member Data Documentation

◆ maTitle

◆ mbTitleOn

◆ mnUnusedId

◆ mpCurrTable

◆ mxGlobTable