#include "html_filter.h"

Inheritance diagram for net_instaweb::HtmlFilter:

Public Types
enum	ScriptUsage { kWillInjectScripts, kMayInjectScripts, kNeverInjectsScripts }
	Describes a filter's relationship with scripts. More...

Public Member Functions
virtual void	StartDocument ()=0

virtual void	EndDocument ()=0

virtual void	StartElement (HtmlElement *element)=0

virtual void	EndElement (HtmlElement *element)=0

virtual void	Cdata (HtmlCdataNode *cdata)=0
	Called for CDATA blocks (e.g. <![CDATA[foobar]]>)

virtual void	Comment (HtmlCommentNode *comment)=0
	Called for HTML comments that aren't IE directives (e.g. ).

virtual void	IEDirective (HtmlIEDirectiveNode *directive)=0

virtual void	Characters (HtmlCharactersNode *characters)=0
	Called for raw characters between tags.

virtual void	Directive (HtmlDirectiveNode *directive)=0
	Called for HTML directives (e.g. <!doctype foobar>).

virtual void	Flush ()=0

virtual void	RenderDone ()

virtual void	DetermineEnabled (GoogleString *disabled_reason)=0

bool	is_enabled () const

void	set_is_enabled (bool is_enabled)

virtual bool	CanModifyUrls ()=0

virtual ScriptUsage	GetScriptUsage () const =0

virtual const char *	Name () const =0
	The name of this filter – used for logging and debugging.

Detailed Description

Base-class used to register for HTML Parser Callbacks. Derive from this class and register with HtmlParse::AddFilter to use the HTML Parser.

Member Enumeration Documentation

enum net_instaweb::HtmlFilter::ScriptUsage

Describes a filter's relationship with scripts.

Enumerator

kWillInjectScripts

Indicates that this filter generally needs to inject scripts, and therefore should be disabled in environments where scripts are not allowed, such as amp. The system also DCHECKs of scripts are injected from a filter where CanInjectScripts() is false.

kMayInjectScripts

Indicates that this filter may in some cases inject scripts, but still has value even if scripts are forbidden. For the rare cases where this value is appropriate, the filter must be explicitly verified to function correctly.

kNeverInjectsScripts

Indicates this filter never injects scripts.

Todo:: TODO(jmarantz): Remove kRequiresScriptExecutionFilterSet in rewrite_options.cc, and instead add new enum choices here covering combinations of requiring 'noscript' behavior and their injection behavior.

Member Function Documentation

virtual bool net_instaweb::HtmlFilter::CanModifyUrls ( )

pure virtual

Invoked by the rewrite driver to query whether this filter will rewrite any urls.

Implemented in net_instaweb::EmptyHtmlFilter, net_instaweb::HtmlWriterFilter, and net_instaweb::RewriteFilter.

virtual void net_instaweb::HtmlFilter::DetermineEnabled ( GoogleString * disabled_reason )

pure virtual

Invoked by rewrite driver where all filters should determine whether they are enabled for this request. The re-writer my optionally set disabled_reason to explain why it disabled itself, which will appear in the debug output.

virtual void net_instaweb::HtmlFilter::EndDocument ( )

pure virtual

Note: EndDocument will be called immediately before the last Flush call. (which also means that in the RewriteDriver use it is called before rendering for the last flush window).

virtual void net_instaweb::HtmlFilter::Flush ( )

pure virtual

Notifies the Filter that a flush is occurring. A filter that's generating streamed output should flush at this time. A filter that's mutating elements can mutate any element seen since the most recent flush; once an element is flushed it is already on the wire to its destination and it's too late to mutate. Flush is initiated by an application calling HttpParse::Flush().

Flush() is called after all other handlers during a HttpParse::Flush(), except RenderDone(), which (if in use) happens after Flush().

Implemented in net_instaweb::CombiningFilter, net_instaweb::GoogleAnalyticsFilter, net_instaweb::JsCombineFilter, net_instaweb::CssInlineImportToLinkFilter, net_instaweb::CssCombineFilter, net_instaweb::MetaTagFilter, net_instaweb::ScanFilter, net_instaweb::HtmlWriterFilter, net_instaweb::EmptyHtmlFilter, net_instaweb::CssOutlineFilter, net_instaweb::JsOutlineFilter, net_instaweb::DebugFilter, net_instaweb::FlushHtmlFilter, and net_instaweb::AddHeadFilter.

virtual ScriptUsage net_instaweb::HtmlFilter::GetScriptUsage ( ) const

pure virtual

Note: there is also kRequiresScriptExecutionFilterSet in rewrite_options.cc, which identifies filters that will leave broken pages if javascript is disabled, and hence require noscript handing. The set of filters that CanInjectScripts is larger, as it includes filters that might inject beacons or other optional functionality that is not page-critical.

virtual void net_instaweb::HtmlFilter::IEDirective ( HtmlIEDirectiveNode * directive )

pure virtual

Called for an IE directive; typically used for CSS styling. See http://msdn.microsoft.com/en-us/library/ms537512(VS.85).aspx

Todo:: TODO(mdsteele): Should we try to maintain the nested structure of the conditionals, in the same way that we maintain nesting of elements?

Implemented in net_instaweb::GoogleAnalyticsFilter, net_instaweb::JsCombineFilter, net_instaweb::LoggingFilter, net_instaweb::JavascriptFilter, net_instaweb::CssCombineFilter, net_instaweb::ScanFilter, net_instaweb::HtmlWriterFilter, and net_instaweb::EmptyHtmlFilter.

bool net_instaweb::HtmlFilter::is_enabled ( ) const

inline

Intended to be called from DetermineEnabled implementations in filters. Returns whether a filter is enabled.

virtual void net_instaweb::HtmlFilter::RenderDone ( )

virtual

Notifies a filter that an asynchronous rewrite & render computation phase has finished. This is not used by HtmlParse itself, but only by RewriteDriver for pre-render filters. Happens after the corresponding flush, for every flush window. Default implementation does nothing.

Todo:: TODO(morlovich): Push this down into CommonFilter and convert all the pre-render filters to inherit off it.

Reimplemented in net_instaweb::CssSummarizerBase, net_instaweb::ImageRewriteFilter, and net_instaweb::CriticalSelectorFilter.

void net_instaweb::HtmlFilter::set_is_enabled ( bool is_enabled )

inline

Set whether this filter is enabled or not. Note that a filter may be included in the filter-chain for a configuration, but be disabled for a request based on the request properties, or even due to content (see HtmlParse::set_is_buffered()).

virtual void net_instaweb::HtmlFilter::StartDocument ( )

pure virtual

Starts a new document. Filters should clear their state in this function, as the same Filter instance may be used for multiple HTML documents.

virtual void net_instaweb::HtmlFilter::StartElement ( HtmlElement * element )

pure virtual

When an HTML element is encountered during parsing, each filter's StartElement method is called. The HtmlElement lives for the entire duration of the document.

Todo:: TODO(jmarantz): consider passing handles rather than pointers and reference-counting them instead to save memory on long documents.

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

pagespeed/kernel/html/html_filter.h

Public Types

Public Member Functions

Detailed Description

Member Enumeration Documentation

Member Function Documentation