source: trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/util/HTMLDocAnchorList.java@ 12188

package org.greenstone.gsdl3.gs3build.util;



public class HTMLDocAnchorList extends HTMLBlockList

{ HTMLDoc doc;

    public HTMLDocAnchorList(HTMLDoc doc)

    { super();

        this.doc = doc; // remember the document reference



        super.initialise(doc, anchorStartTags, anchorEndTags, anchorOptEndTags);

    }



    /**
     *  Return the HTMLBlock containing the named local anchor and it's
     *  child (contained) tags.
     *
     *  @param <code>String</code> name of the local anchor to find.
     *  @return <code>HTMLBlock</code> the corresponding block of HTML
     *                  contained by the anchor.  If the local anchor does not
     *                  exist, this value will be <code>null</code>.
     */
    public HTMLBlock localAnchor(String name)

    {   int block;

        String  localname;



        for (block = 0; block < this.size(); block ++)

        { localname = this.tagBlock(block).headTag().idValue("name");

            if (localname == null)

            { continue;

            }

            if (name.equalsIgnoreCase(localname))

            {   return this.tagBlock(block);

            }

        }

        return null;

    }

    private int indexOf(String base, String subString, int from)
    {   int i, j;

        for (i = from; i < base.length() - subString.length(); i ++)
        {   if (Character.toLowerCase(base.charAt(i)) == subString.charAt(0))
            {   j = 0;
                for (j = 1; j < subString.length(); j ++)
                {   if (Character.toLowerCase(base.charAt(i+j)) != subString.charAt(j))
                    {   break;
                    }
                }

                if (j == subString.length())
                {   return i;
                }
            }
        }
        return -1;
    }

    public void flattenLinks()
    {   int                     block;
        String              url;
        char                    endsWith;
        StringBuffer    newContent = new StringBuffer(doc.getContent());
        boolean             changed = false;

//      System.out.println("Flattening");

        for (block = 0; block < this.size(); block ++)
        { HTMLTag tag;
            int         tagPos;
            int         valueStart, valueEnd;

            tag = this.tagBlock(block).headTag();   // get the <A> tag itself

            url = tag.idValue("href");                      // if there is no HREF member, abandon this one
            if (url == null)
            { System.out.println("No href");
                continue;
            }
    
            tagPos  = tag.startPos();
            tagPos  =   indexOf(doc.getContent(), "href", tagPos);  // get the HREF itself
            if (tagPos > tag.endPos() || tagPos < 0)                    // if none in the tag, skip it
            {   System.out.println("Didn't get href");
                continue;
            }
            tagPos += "href".length();                                              // skip past it

            while ( doc.getContent().charAt(tagPos) <= ' ' ||       // skip pass equals, etc.
                            doc.getContent().charAt(tagPos) == '=')
            {   tagPos ++;
            }

            if (doc.getContent().charAt(tagPos) == '"' ||               // get the demarcating whitespace/quotes
                    doc.getContent().charAt(tagPos) == '\'')
            {   endsWith    = doc.getContent().charAt(tagPos);
                tagPos ++;
            }
            else
            {   endsWith  = ' ';
            }
            valueStart = tagPos;

            //          System.out.println("Flattening " + doc.getContent().length() + " " + tagPos);

            while (tagPos < doc.getContent().length() &&
                   doc.getContent().charAt(tagPos) != endsWith) // read it in
            {   if (doc.content.charAt(tagPos) == '?' ||                // replace special characters           
                        doc.content.charAt(tagPos) == '&')
                {   newContent.setCharAt(tagPos, '_');                      //   and note that we've done so
                    changed = true;
                }
                tagPos ++;
            }
            valueEnd = tagPos;

            tagPos = indexOf(doc.getContent(), ".html", valueStart);
            if (tagPos >= 0 && tagPos < valueEnd)
            {   // shuffle around the html bit
                while (tagPos < valueEnd - 5)
                {   newContent.setCharAt(tagPos, newContent.charAt(tagPos + 5));
                    tagPos ++;
                }
                newContent.setCharAt(tagPos ++, '.');
                newContent.setCharAt(tagPos ++, 'h');
                newContent.setCharAt(tagPos ++, 't');
                newContent.setCharAt(tagPos ++, 'm');
                newContent.setCharAt(tagPos ++, 'l');
            }
        }

        if (changed == true)                                                                // replace the document as needsbe
        {   doc.setContent(newContent.toString());
        }
    }

}