org.w3c.tidy
Class Clean

java.lang.Object
  org.w3c.tidy.Clean

public class Clean

extends java.lang.Object

Clean up misuse of presentation markup. Filters from other formats such as Microsoft Word often make excessive use of presentation markup such as font tags, B, I, and the align attribute. By applying a set of production rules, it is straight forward to transform this to use CSS. Some rules replace some of the children of an element by style properties on the element, e.g.

...

.

...

Such rules are applied to the element's content and then to the element itself until none of the rules more apply. Having applied all the rules to an element, it will have a style attribute with one or more properties. Other rules strip the element they apply to, replacing it by style properties on the contents, e.g.

...

.

... These rules are applied to an element before processing its content and replace the current element by the first element in the exposed content. After applying both sets of rules, you can replace the style attribute by a class value and style rule in the document head. To support this, an association of styles and class names is built. A naive approach is to rely on string matching to test when two property lists are the same. A better approach would be to first sort the properties before matching.

Version:

$Revision: 1.25 $ ($Author: fgiust $)

Author:

Dave Raggett dsr@w3.org , Andy Quick ac.quick@sympatico.ca (translation to Java), Fabrizio Giustina

Field Summary

private int	classNum sequential number for generated css classes.
private TagTable	tt Tag table.

Constructor Summary

Clean(TagTable tagTable)
Instantiates a new Clean.

Method Summary

private void	addAlign(Node node, java.lang.String align) Adds an align style.
private void	addColorRule(Lexer lexer, java.lang.String selector, java.lang.String color) Adds a css rule for color.
private void	addFontColor(Node node, java.lang.String color) Adds a font color style.
private void	addFontFace(Node node, java.lang.String face) Adds a font-family style.
private void	addFontSize(Node node, java.lang.String size) Adds a font size style.
private void	addFontStyles(Node node, AttVal av) Add style properties to node corresponding to the font face, size and color attributes.
private java.lang.String	addProperty(java.lang.String style, java.lang.String property) Creates a string with merged properties.
private void	addStyleProperty(Node node, java.lang.String property) Add style property to element, creating style attribute as needed and adding ; delimiter.
private boolean	blockStyle(Lexer lexer, Node node) Symptom: the only child of a block-level element is a presentation element such as B, I or FONT.
void	bQ2Div(Node node) Replace implicit blockquote by div with an indent taking care to reduce nested blockquotes to a single div with the indent set to match the nesting depth.
(package private) static void	bumpObject(Lexer lexer, Node html) Where appropriate move object elements from head to body.
private boolean	center2Div(Lexer lexer, Node node, Node[] pnode) Symptom:
private void	cleanBodyAttrs(Lexer lexer, Node body) Move presentation attribs from body to style element.
private Node	cleanNode(Lexer lexer, Node node) Applies all matching rules to a node.
void	cleanTree(Lexer lexer, Node doc) Clean an html tree.
void	cleanWord2000(Lexer lexer, Node node) This is a major clean up to strip out all the extra stuff you get when you save as web page from Word 2000.
private StyleProp	createProps(StyleProp prop, java.lang.String style) Create sorted linked list of properties from style string.
private java.lang.String	createPropString(StyleProp props) Create a css property.
private void	createStyleElement(Lexer lexer, Node doc) Create style element using rules from dictionary.
private Node	createStyleProperties(Lexer lexer, Node node, Node[] prepl) Special case: if the current node is destroyed by CleanNode() lower in the tree, this node and its parent no longer exist.
private void	defineStyleRules(Lexer lexer, Node node) Find style attribute in node content, and replace it by corresponding class attribute.
private boolean	dir2Div(Lexer lexer, Node node) Symptom: <dir><li> where <li> is only child.
private void	discardContainer(Node element, Node[] pnode) Used to strip font start and end tags.
void	dropSections(Lexer lexer, Node node) Drop if/endif sections inserted by word2000.
void	emFromI(Node node) Replace i by em and b by strong.
(package private) Node	findEnclosingCell(Node node) Find the enclosing table cell for the given node.
private java.lang.String	findStyle(Lexer lexer, java.lang.String tag, java.lang.String properties) Finds a css style.
private void	fixNodeLinks(Node node) Ensure bidirectional links are consistent.
private boolean	font2Span(Lexer lexer, Node node, Node[] pnode) Replace font elements by span elements, deleting the font element's attributes and replacing them by a single style attribute.
private java.lang.String	fontSize2Name(java.lang.String size) Map a % font size to a named font size.
private java.lang.String	gensymClass(Lexer lexer, java.lang.String tag) Generates a new css class name.
private boolean	inlineStyle(Lexer lexer, Node node, Node[] pnode) If the node has only one b, i, or font child remove the child node and add the appropriate style attributes to parent.
private StyleProp	insertProperty(StyleProp props, java.lang.String name, java.lang.String value) Insert a css style property.
boolean	isWord2000(Node root) Check if the current document is a converted Word document.
void	list2BQ(Node node) Some people use dir or ul without an li to indent the content.
private void	mergeClasses(Node node, Node child) Merge class attributes from 2 nodes.
private boolean	mergeDivs(Lexer lexer, Node node) Symptom: <div><div>...</div></div> Action: merge the two divs.
private java.lang.String	mergeProperties(java.lang.String s1, java.lang.String s2) Create new string that consists of the combined style properties in s1 and s2.
private void	mergeStyles(Node node, Node child) Merge style from 2 nodes.
void	nestedEmphasis(Node node) simplifies ...
private boolean	nestedList(Lexer lexer, Node node, Node[] pnode) Symptom: ...
private boolean	niceBody(Lexer lexer, Node doc) Check deprecated attributes in body tag.
(package private) boolean	noMargins(Node node) Used to hunt for hidden preformatted sections.
private void	normalizeSpaces(Lexer lexer, Node node) Map non-breaking spaces to regular spaces.
Node	pruneSection(Lexer lexer, Node node) node is <![if ...]> prune up to <![endif]>.
void	purgeWord2000Attributes(Node node) Remove word2000 attributes from node.
(package private) boolean	singleSpace(Lexer lexer, Node node) Does element have a single space as its content?
private void	stripOnlyChild(Node node) Used to strip child of node when the node has one and only one child.
Node	stripSpan(Lexer lexer, Node span) Word2000 uses span excessively, so we strip span out.
private void	style2Rule(Lexer lexer, Node node) Find style attribute in node, and replace it by corresponding class attribute.
private void	textAlign(Lexer lexer, Node node) Symptom: <p align=center>.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

classNum

private int classNum

sequential number for generated css classes.

tt

private TagTable tt

Tag table.

Constructor Detail

Clean

public Clean(TagTable tagTable)

Instantiates a new Clean.

Parameters:

tagTable - tag table instance

Method Detail

insertProperty

private StyleProp insertProperty(StyleProp props,
                                 java.lang.String name,
                                 java.lang.String value)

Insert a css style property.

Parameters:

props - StyleProp instance

name - property name

value - property value

Returns:

StyleProp containin the given property

createProps

private StyleProp createProps(StyleProp prop,
                              java.lang.String style)

Create sorted linked list of properties from style string.

Parameters:

prop - StyleProp

style - style string

Returns:

StyleProp with given style

createPropString

private java.lang.String createPropString(StyleProp props)

Create a css property.

Parameters:

props - StyleProp

Returns:

css property as String

addProperty

private java.lang.String addProperty(java.lang.String style,
                                     java.lang.String property)

Creates a string with merged properties.

Parameters:

style - css style

property - css properties

Returns:

merged string

gensymClass

private java.lang.String gensymClass(Lexer lexer,
                                     java.lang.String tag)

Generates a new css class name.

Parameters:

lexer - Lexer

tag - Tag

Returns:

generated css class

findStyle

private java.lang.String findStyle(Lexer lexer,
                                   java.lang.String tag,
                                   java.lang.String properties)

Finds a css style.

Parameters:

lexer - Lexer

tag - tag name

properties - css properties

Returns:

style string

style2Rule

private void style2Rule(Lexer lexer,
                        Node node)

Find style attribute in node, and replace it by corresponding class attribute. Search for class in style dictionary otherwise gensym new class and add to dictionary. Assumes that node doesn't have a class attribute.

Parameters:

lexer - Lexer

node - node with a style attribute

addColorRule

private void addColorRule(Lexer lexer,
                          java.lang.String selector,
                          java.lang.String color)

Adds a css rule for color.

Parameters:

lexer - Lexer

selector - css selector

color - color value

cleanBodyAttrs

private void cleanBodyAttrs(Lexer lexer,
                            Node body)

Move presentation attribs from body to style element.

 background="foo" . body { background-image: url(foo) }
 bgcolor="foo" . body { background-color: foo }
 text="foo" . body { color: foo }
 link="foo" . :link { color: foo }
 vlink="foo" . :visited { color: foo }
 alink="foo" . :active { color: foo }
 

Parameters:

lexer - Lexer

body - body node

niceBody

private boolean niceBody(Lexer lexer,
                         Node doc)

Check deprecated attributes in body tag.

Parameters:

lexer - Lexer

doc - document root node

Returns:

true is the body doesn't contain deprecated attributes, false otherwise.

createStyleElement

private void createStyleElement(Lexer lexer,
                                Node doc)

Create style element using rules from dictionary.

Parameters:

lexer - Lexer

doc - root node

fixNodeLinks

private void fixNodeLinks(Node node)

Ensure bidirectional links are consistent.

Parameters:

node - root node

stripOnlyChild

private void stripOnlyChild(Node node)

Used to strip child of node when the node has one and only one child.

Parameters:

node - parent node

discardContainer

private void discardContainer(Node element,
                              Node[] pnode)

Used to strip font start and end tags.

Parameters:

element - original node

pnode - passed in as array to allow modification. pnode[0] will contain the final node

addStyleProperty

private void addStyleProperty(Node node,
                              java.lang.String property)

Add style property to element, creating style attribute as needed and adding ; delimiter.

Parameters:

node - node

property - property added to node

mergeProperties

private java.lang.String mergeProperties(java.lang.String s1,
                                         java.lang.String s2)

Create new string that consists of the combined style properties in s1 and s2. To merge property lists, we build a linked list of property/values and insert properties into the list in order, merging values for the same property name.

Parameters:

s1 - first property

s2 - second property

Returns:

merged properties

mergeClasses

private void mergeClasses(Node node,
                          Node child)

Merge class attributes from 2 nodes.

Parameters:

node - Node

child - Child node

mergeStyles

private void mergeStyles(Node node,
                         Node child)

Merge style from 2 nodes.

Parameters:

node - Node

child - Child node

fontSize2Name

private java.lang.String fontSize2Name(java.lang.String size)

Map a % font size to a named font size.

Parameters:

size - size in %

Returns:

font size name

addFontFace

private void addFontFace(Node node,
                         java.lang.String face)

Adds a font-family style.

Parameters:

node - Node

face - font face

addFontSize

private void addFontSize(Node node,
                         java.lang.String size)

Adds a font size style.

Parameters:

node - Node

size - font size

addFontColor

private void addFontColor(Node node,
                          java.lang.String color)

Adds a font color style.

Parameters:

node - Node

color - color value

addAlign

private void addAlign(Node node,
                      java.lang.String align)

Adds an align style.

Parameters:

node - Node

align - align value

addFontStyles

private void addFontStyles(Node node,
                           AttVal av)

Add style properties to node corresponding to the font face, size and color attributes.

Parameters:

node - font tag

av - attribute list for node

textAlign

private void textAlign(Lexer lexer,
                       Node node)

Symptom: <p align=center>. Action: <p style="text-align: center">.

Parameters:

lexer - Lexer

node - node with center attribute. Will be modified to use css style.

dir2Div

private boolean dir2Div(Lexer lexer,
                        Node node)

Symptom: <dir><li> where <li> is only child. Action: coerce <dir> <li> to <div> with indent. The clean up rules use the pnode argument to return the next node when the original node has been deleted.

Parameters:

lexer - Lexer

node - dir tag

Returns:

true if a dir tag has been coerced to a div

center2Div

private boolean center2Div(Lexer lexer,
                           Node node,
                           Node[] pnode)

Symptom:

 <center>
 

.

Action: replace <center> by <div style="text-align: center">

Parameters:

lexer - Lexer

node - center tag

pnode - pnode[0] is the same as node, passed in as an array to allow modification

Returns:

true if a center tag has been replaced by a div

mergeDivs

private boolean mergeDivs(Lexer lexer,
                          Node node)

Symptom: <div><div>...</div></div> Action: merge the two divs. This is useful after nested <dir>s used by Word for indenting have been converted to <div>s.

Parameters:

lexer - Lexer

node - first div

Returns:

true if the divs have been merged

nestedList

private boolean nestedList(Lexer lexer,
                           Node node,
                           Node[] pnode)

Symptom:

• ...

Action: discard outer list.

Parameters:

lexer - Lexer

node - Node

pnode - passed in as array to allow modifications.

Returns:

true if nested lists have been found and replaced

blockStyle

private boolean blockStyle(Lexer lexer,
                           Node node)

Symptom: the only child of a block-level element is a presentation element such as B, I or FONT. Action: add style "font-weight: bold" to the block and strip the <b>element, leaving its children. example:

 <p>
 <b><font face="Arial" size="6">Draft Recommended Practice</font></b>
 </p>
 

becomes:

 <p style="font-weight: bold; font-family: Arial; font-size: 6">
 Draft Recommended Practice
 </p>
 

This code also replaces the align attribute by a style attribute. However, to avoid CSS problems with Navigator 4, this isn't done for the elements: caption, tr and table

Parameters:

lexer - Lexer

node - parent node

Returns:

true if the child node has been removed

inlineStyle

private boolean inlineStyle(Lexer lexer,
                            Node node,
                            Node[] pnode)

If the node has only one b, i, or font child remove the child node and add the appropriate style attributes to parent.

Parameters:

lexer - Lexer

node - parent node

pnode - passed as an array to allow modifications

Returns:

true if child node has been stripped, replaced by style attributes.

font2Span

private boolean font2Span(Lexer lexer,
                          Node node,
                          Node[] pnode)

Replace font elements by span elements, deleting the font element's attributes and replacing them by a single style attribute.

Parameters:

lexer - Lexer

node - font tag

pnode - passed as an array to allow modifications

Returns:

true if a font tag has been dropped and replaced by style attributes

cleanNode

private Node cleanNode(Lexer lexer,
                       Node node)

Applies all matching rules to a node.

Parameters:

lexer - Lexer

node - original node

Returns:

cleaned up node

createStyleProperties

private Node createStyleProperties(Lexer lexer,
                                   Node node,
                                   Node[] prepl)

Special case: if the current node is destroyed by CleanNode() lower in the tree, this node and its parent no longer exist. So we must jump back up the CreateStyleProperties() call stack until we have a valid node reference.

Parameters:

lexer - Lexer

node - Node

prepl - passed in as array to allow modifications

Returns:

cleaned Node

defineStyleRules

private void defineStyleRules(Lexer lexer,
                              Node node)

Find style attribute in node content, and replace it by corresponding class attribute.

Parameters:

lexer - Lexer

node - parent node

cleanTree

public void cleanTree(Lexer lexer,
                      Node doc)

Clean an html tree.

Parameters:

lexer - Lexer

doc - root node

nestedEmphasis

public void nestedEmphasis(Node node)

simplifies ... ... etc.

Parameters:

node - root Node

emFromI

public void emFromI(Node node)

Replace i by em and b by strong.

Parameters:

node - root Node

list2BQ

public void list2BQ(Node node)

Some people use dir or ul without an li to indent the content. The pattern to look for is a list with a single implicit li. This is recursively replaced by an implicit blockquote.

Parameters:

node - root Node

bQ2Div

public void bQ2Div(Node node)

Replace implicit blockquote by div with an indent taking care to reduce nested blockquotes to a single div with the indent set to match the nesting depth.

Parameters:

node - root Node

findEnclosingCell

Node findEnclosingCell(Node node)

Find the enclosing table cell for the given node.

Parameters:

node - Node

Returns:

enclosing cell node

pruneSection

public Node pruneSection(Lexer lexer,
                         Node node)

node is <![if ...]> prune up to <![endif]>.

Parameters:

lexer - Lexer

node - Node

Returns:

cleaned up Node

dropSections

public void dropSections(Lexer lexer,
                         Node node)

Drop if/endif sections inserted by word2000.

Parameters:

lexer - Lexer

node - Node root node

purgeWord2000Attributes

public void purgeWord2000Attributes(Node node)

Remove word2000 attributes from node.

Parameters:

node - node to cleanup

stripSpan

public Node stripSpan(Lexer lexer,
                      Node span)

Word2000 uses span excessively, so we strip span out.

Parameters:

lexer - Lexer

span - Node span

Returns:

cleaned node

normalizeSpaces

private void normalizeSpaces(Lexer lexer,
                             Node node)

Map non-breaking spaces to regular spaces.

Parameters:

lexer - Lexer

node - Node

noMargins

boolean noMargins(Node node)

Used to hunt for hidden preformatted sections.

Parameters:

node - checked node

Returns:

true if the node has a "margin-top: 0" or "margin-bottom: 0" style

singleSpace

boolean singleSpace(Lexer lexer,
                    Node node)

Does element have a single space as its content?

Parameters:

lexer - Lexer

node - checked node

Returns:

true if the element has a single space as its content

cleanWord2000

public void cleanWord2000(Lexer lexer,
                          Node node)

This is a major clean up to strip out all the extra stuff you get when you save as web page from Word 2000. It doesn't yet know what to do with VML tags, but these will appear as errors unless you declare them as new tags, such as o:p which needs to be declared as inline.

Parameters:

lexer - Lexer

node - node to clean up

isWord2000

public boolean isWord2000(Node root)

Check if the current document is a converted Word document.

Parameters:

root - root Node

Returns:

true if the document has been geenrated by Microsoft Word.

bumpObject

static void bumpObject(Lexer lexer,
                       Node html)

Where appropriate move object elements from head to body.

Parameters:

lexer - Lexer

html - html node