prototypicalnodefactory.java

html 解析处理代码

            put (ids[i].toUpperCase (Locale.ENGLISH), tag);
    }

    /**
     * Unregister a tag.
     * Unregisters the given tag from every {@link Tag#getIds() id} the tag has.
     * <p><strong>The ids are converted to uppercase to undo the operation
     * of registerTag.</strong>
     * @param tag The tag to unregister.
     */
    public void unregisterTag (Tag tag)
    {
        String[] ids;

        ids = tag.getIds ();
        for (int i = 0; i < ids.length; i++)
            remove (ids[i].toUpperCase (Locale.ENGLISH));
    }

    /**
     * Register all known tags in the tag package.
     * Registers tags from the {@link org.htmlparser.tags tag package} by
     * calling {@link #registerTag(Tag) registerTag()}.
     * @return 'this' nodefactory as a convenience.
     */
    public PrototypicalNodeFactory registerTags ()
    {
        registerTag (new AppletTag ());
        registerTag (new BaseHrefTag ());
        registerTag (new Bullet ());
        registerTag (new BulletList ());
        registerTag (new DefinitionList ());
        registerTag (new DefinitionListBullet ());
        registerTag (new DoctypeTag ());
        registerTag (new FormTag ());
        registerTag (new FrameSetTag ());
        registerTag (new FrameTag ());
        registerTag (new HeadingTag ());
        registerTag (new ImageTag ());
        registerTag (new InputTag ());
        registerTag (new JspTag ());
        registerTag (new LabelTag ());
        registerTag (new LinkTag ());
        registerTag (new MetaTag ());
        registerTag (new ObjectTag ());
        registerTag (new OptionTag ());
        registerTag (new ParagraphTag ());
        registerTag (new ProcessingInstructionTag ());
        registerTag (new ScriptTag ());
        registerTag (new SelectTag ());
        registerTag (new StyleTag ());
        registerTag (new TableColumn ());
        registerTag (new TableHeader ());
        registerTag (new TableRow ());
        registerTag (new TableTag ());
        registerTag (new TextareaTag ());
        registerTag (new TitleTag ());
        registerTag (new Div ());
        registerTag (new Span ());
        registerTag (new BodyTag ());
        registerTag (new HeadTag ());
        registerTag (new Html ());
        

        return (this);
    }

    /**
     * Get the object that is cloned to generate text nodes.
     * @return The prototype for {@link Text} nodes.
     * @see #setTextPrototype
     */
    public Text getTextPrototype ()
    {
        return (mText);
    }

    /**
     * Set the object to be used to generate text nodes.
     * @param text The prototype for {@link Text} nodes.
     * If <code>null</code> the prototype is set to the default
     * ({@link TextNode}).
     * @see #getTextPrototype
     */
    public void setTextPrototype (Text text)
    {
        if (null == text)
            mText = new TextNode (null, 0, 0);
        else
            mText = text;
    }

    /**
     * Get the object that is cloned to generate remark nodes.
     * @return The prototype for {@link Remark} nodes.
     * @see #setRemarkPrototype
     */
    public Remark getRemarkPrototype ()
    {
        return (mRemark);
    }

    /**
     * Set the object to be used to generate remark nodes.
     * @param remark The prototype for {@link Remark} nodes.
     * If <code>null</code> the prototype is set to the default
     * ({@link RemarkNode}).
     * @see #getRemarkPrototype
     */
    public void setRemarkPrototype (Remark remark)
    {
        if (null == remark)
            mRemark = new RemarkNode (null, 0, 0);
        else
            mRemark = remark;
    }

    /**
     * Get the object that is cloned to generate tag nodes.
     * Clones of this object are returned from {@link #createTagNode} when no
     * specific tag is found in the list of registered tags.
     * @return The prototype for {@link Tag} nodes.
     * @see #setTagPrototype
     */
    public Tag getTagPrototype ()
    {
        return (mTag);
    }

    /**
     * Set the object to be used to generate tag nodes.
     * Clones of this object are returned from {@link #createTagNode} when no
     * specific tag is found in the list of registered tags.
     * @param tag The prototype for {@link Tag} nodes.
     * If <code>null</code> the prototype is set to the default
     * ({@link TagNode}).
     * @see #getTagPrototype
     */
    public void setTagPrototype (Tag tag)
    {
        if (null == tag)
            mTag = new TagNode (null, 0, 0, null);
        else
            mTag = tag;
    }

    //
    // NodeFactory interface
    //

    /**
     * Create a new string node.
     * @param page The page the node is on.
     * @param start The beginning position of the string.
     * @param end The ending position of the string.
     * @return A text node comprising the indicated characters from the page.
     */
    public Text createStringNode (Page page, int start, int end)
    {
        Text ret;

        try
        {
            ret = (Text)(getTextPrototype ().clone ());
            ret.setPage (page);
            ret.setStartPosition (start);
            ret.setEndPosition (end);
        }
        catch (CloneNotSupportedException cnse)
        {
            ret = new TextNode (page, start, end);
        }

        return (ret);
    }

    /**
     * Create a new remark node.
     * @param page The page the node is on.
     * @param start The beginning position of the remark.
     * @param end The ending positiong of the remark.
     * @return A remark node comprising the indicated characters from the page.
     */
    public Remark createRemarkNode (Page page, int start, int end)
    {
        Remark ret;

        try
        {
            ret = (Remark)(getRemarkPrototype ().clone ());
            ret.setPage (page);
            ret.setStartPosition (start);
            ret.setEndPosition (end);
        }
        catch (CloneNotSupportedException cnse)
        {
            ret = new RemarkNode (page, start, end);
        }

        return (ret);
    }

    /**
     * Create a new tag node.
     * Note that the attributes vector contains at least one element,
     * which is the tag name (standalone attribute) at position zero.
     * This can be used to decide which type of node to create, or
     * gate other processing that may be appropriate.
     * @param page The page the node is on.
     * @param start The beginning position of the tag.
     * @param end The ending positiong of the tag.
     * @param attributes The attributes contained in this tag.
     * @return A tag node comprising the indicated characters from the page.
     */
    public Tag createTagNode (Page page, int start, int end, Vector attributes)
    {
        Attribute attribute;
        String id;
        Tag prototype;
        Tag ret;

        ret = null;

        if (0 != attributes.size ())
        {
            attribute = (Attribute)attributes.elementAt (0);
            id = attribute.getName ();
            if (null != id)
            {
                try
                {
                    id = id.toUpperCase (Locale.ENGLISH);
                    if (!id.startsWith ("/"))
                    {
                        if (id.endsWith ("/"))
                            id = id.substring (0, id.length () - 1);
                        prototype = (Tag)mBlastocyst.get (id);
                        if (null != prototype)
                        {
                            ret = (Tag)prototype.clone ();
                            ret.setPage (page);
                            ret.setStartPosition (start);
                            ret.setEndPosition (end);
                            ret.setAttributesEx (attributes);
                        }
                    }
                }
                catch (CloneNotSupportedException cnse)
                {
                    // default to creating a generic one
                }
            }
        }
        if (null == ret)
        {   // generate a generic node
            try
            {
                ret = (Tag)getTagPrototype ().clone ();
                ret.setPage (page);
                ret.setStartPosition (start);
                ret.setEndPosition (end);
                ret.setAttributesEx (attributes);
            }
            catch (CloneNotSupportedException cnse)
            {
                ret = new TagNode (page, start, end, attributes);
            }
        }

        return (ret);
    }
}