Readability for Naver blog. Ripped off from brettterpstra.com/share/readability.js

naverreadability.js

var readabilityVersion = "2";
var readStyle='style-ebook';
var readSize='size-medium';
var readMargin='margin-wide';
(function(){
    // removing all existing scripts so they don't cause conflicts...
    var docscripts = document.getElementsByTagName('script');
    for (k=0;k < docscripts.length; k++) {
        if (docscripts[k].src != null && ! docscripts[k].src.match(/readability|[Cc]lippability/)) {
            docscripts[k].parentNode.removeChild(docscripts[k]);
        }
    }
    // let's just load jQuery and get it over with
    var gjs = document.createElement('SCRIPT');
    gjs.type='text/javascript';
    gjs.src = "http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js";
    document.getElementsByTagName('head')[0].appendChild(gjs);
    gjs.onload = gjs.onreadystatechange = function(){

        $('script').each(function(){
            // jQuery gets scripts inside of conditional comments far more easily than I could figure out
            if (! this.src.match(/readability|[Cc]lippability|jquery\.min\.js$/)) { $(this).remove(); }
        });

    };

    var objOverlay = document.createElement("div");
    var objinnerDiv = document.createElement("div");

    objOverlay.id = "readOverlay";
    objinnerDiv.id = "readInner";

    // Apply user-selected styling:
    document.body.className = readStyle;
    objOverlay.className = readStyle;
    objinnerDiv.className = readMargin + " " + readSize;

    objinnerDiv.appendChild(grabArticle());     // Get the article and place it inside the inner Div
    objOverlay.appendChild(objinnerDiv);        // Insert the inner div into the overlay

    // For totally hosed HTML, add body node that can't be found because of bad HTML or something.
    if(document.body == null)
    {
        body = document.createElement("body");
        document.body = body;
    }

    document.body.innerHTML = "";

    // Inserts the new content :

    document.body.insertBefore(objOverlay, document.body.firstChild);
    var o = document.body.firstChild;
    return o.innerHTML;
})();

function getElementsByClassName(classname, node)  {
    if(!node) node = document.getElementsByTagName("body")[0];
    var a = [];
    var re = new RegExp('\\b' + classname + '\\b');
    var els = node.getElementsByTagName("*");
    for(var i=0,j=els.length; i<j; i++) {
        if(re.test(els[i].className))a.push(els[i]);
    }
    return a;
}

function grabArticle() {
    var topDiv = document.getElementById("postListBody");

    var articleContent = document.createElement("DIV");
    var articleTitle = document.createElement("H1");
    var articleFooter = document.createElement("DIV");

    // Replace all doubled-up <BR> tags with <P> tags, and remove fonts.
    var pattern =  new RegExp ("<br/?>[ \r\n\s]*<br/?>", "g");
    topDiv.innerHTML = topDiv.innerHTML.replace(pattern, "</p><p>").replace(/<\/?font[^>]*>/g, '');

        // Grab the title from the <title> tag and inject it as the title.
        articleTitle.innerHTML = document.title;
        articleContent.appendChild(articleTitle);

    // REMOVES ALL STYLESHEETS ...
    for (var k=0;k < document.styleSheets.length; k++) {
        if (document.styleSheets[k].href != null && document.styleSheets[k].href.lastIndexOf("readability") == -1) {
            document.styleSheets[k].disabled = true;
        }
    }
// 
//     var sh = getElementsByClassName("syntaxhighlighter");
//     for (var i=0;i < sh.length;i++) {
//         var bar = getElementsByClassName("toolbar",sh[i]);
//         if (bar.length > 0) {
//             for (var bn=0;bn < bar.length;bn++) {
//                 bar[bn].parentNode.removeChild(bar[bn]);
//             }
//         }
//         var numbers = getElementsByClassName("number",sh[i]);
//         if (numbers.length > 0) {
//             for (var num=0;num < numbers.length;num++) {
//                 numbers[num].parentNode.removeChild(numbers[num]);
//             }
//         }
//     }
// 
//     var dp = getElementsByClassName("dp-highlighter");
//     for (var d=0;d < dp.length;d++) {
//         dp[d].parentNode.removeChild(dp[d]);
//     }
// 
//     var sth = getElementsByClassName("standardLighter");
//     for (d=0;d < sth.length;d++) {
//         sth[d].parentNode.removeChild(sth[d]);
//     }
// 
    // Remove all style tags in head (not doing this on IE) :
    var styleTags = document.getElementsByTagName("style");
    for (var l=0;l < styleTags.length; l++) {
        if (navigator.appName != "Microsoft Internet Explorer")
            styleTags[l].textContent = "";
    }
    topDiv = killCodeSpans(topDiv);         // removes span tags
    cleanStyles(topDiv);                    // Removes all style attributes
    topDiv = killDivs(topDiv);              // Goes in and removes DIV's that have more non <p> stuff than <p> stuff
    topDiv = killBreaks(topDiv);            // Removes any consecutive <br />'s into just one <br />

    // Cleans out junk from the topDiv just in case:
    topDiv = clean(topDiv, "form");
    // topDiv = clean(topDiv, "object");
    topDiv = clean(topDiv, "table", 8);
    topDiv = clean(topDiv, "h1");
    // topDiv = clean(topDiv, "h2");
    topDiv = clean(topDiv, "iframe");

    var garbage = topDiv.getElementsByClassName('blogapp_area');
    for (var i = 0;i < garbage.length;i++) {
        garbage[i].parentNode.removeChild(garbage[i]);
    }
    garbage = topDiv.getElementsByClassName('post-btn');
    for (var i = 0;i < garbage.length;i++) {
        garbage[i].parentNode.removeChild(garbage[i]);
    }

    // Add the footer and contents:
    articleFooter.id = "readFooter";
    articleFooter.innerHTML = "\
        <a href='http://lab.arc90.com/experiments/readability'><img src='http://lab.arc90.com/experiments/readability/images/footer-readability.png'></a>\
        <a href='http://www.arc90.com'><img src='http://lab.arc90.com/experiments/readability/images/footer-arc90.png'></a>\
        <a href='http://www.twitter.com/arc90' class='footer-twitterLink'>Follow us on Twitter &raquo;</a>\
                <div class='footer-right' >\
                        <span class='version'>Readability version " + readabilityVersion + "</span>\
                </div>\
    ";

    articleContent.appendChild(topDiv);
//  articleContent.appendChild(articleFooter);
    document.onkeyup = docOnKeyup;
        return articleContent;
}

function docOnKeyup(ev)
{
    var keyID = null;
    if (navigator.appName == "Microsoft Internet Explorer") {
        keyID = event.keyCode;
    } else {
        keyID = (window.event) ? event.keyCode : ev.keyCode;
    }
    var bgcolor,fgcolor,acolor;
    switch (keyID) {
        case 27: // escape
            document.location.reload(true);
            break;
        case 37: // left arrow
            bgcolor = "#222";
            fgcolor = "#F3EFCE";
            acolor = "#A19F89";
            break;
        case 39: // right arrow
            bgcolor = "#fff";
            fgcolor = "#333";
            acolor = "#276F78";
            break;
        case 46: // delete
            bgcolor = "#eee";
            fgcolor = "#333";
            acolor = "#blue";
            break;
    }
    body = document.getElementById("readOverlay");
    // body.className = body.className.replace('/\blightened\b/','') + " darkened";
    body.style.backgroundColor = bgcolor;
    body.style.color = fgcolor;
    var alinks = body.getElementsByTagName('a');
    for (var lc = 0;lc < alinks.length;lc++) {
        alinks[lc].style.color = acolor;
    }
}


// Get the inner text of a node - cross browser compatibly.
function getInnerText(e) {
    if (navigator.appName == "Microsoft Internet Explorer")
        return e.innerText;
    else
        return e.textContent;
}

// Get character count
function getCharCount ( e,s ) {
    s = s || ",";
    return getInnerText(e).split(s).length;
}

function cleanStyles( e ) {
    e = e || document;
    var cur = e.firstChild;

    // If we had a bad node, there's not much we can do.
    if(!e)
        return;

    // Remove any root styles, if we're able.
    if(typeof e.removeAttribute == 'function')
        e.removeAttribute('style');

    // Go until there are no more child nodes
    while ( cur != null ) {
        if ( cur.nodeType == 1 ) {
            // Remove style attribute(s) :
            cur.removeAttribute("style");
            cleanStyles( cur );
        }
        cur = cur.nextSibling;
    }
}

function killDivs ( e ) {
    var divsList = e.getElementsByTagName( "div" );
    var curDivLength = divsList.length;

    // Gather counts for other typical elements embedded within.
    // Traverse backwards so we can remove nodes at the same time without effecting the traversal.
    for (var i=curDivLength-1; i >= 0; i--) {
        var p = divsList[i].getElementsByTagName("p").length;
        var img = divsList[i].getElementsByTagName("img").length;
        var li = divsList[i].getElementsByTagName("li").length;
        var a = divsList[i].getElementsByTagName("a").length;
        var embed = divsList[i].getElementsByTagName("embed").length;
        var object = divsList[i].getElementsByTagName("object").length;
        var pre = divsList[i].getElementsByTagName("pre").length;
        var code = divsList[i].getElementsByTagName("code").length;
        var divId = divsList[i].id;
        var divClass = divsList[i].className;
        var sphereit = divsList[i].innerHTML.match("<!-- sphereit") == null ? 0 : 1;
        // If the number of commas is less than 10 (bad sign) ...
        if ( getCharCount(divsList[i]) < 10 ) {
            // And the number of non-paragraph elements is more than paragraphs
            // or other ominous signs :
            if (( img > p || li > p || a > p || p == 0 || divId.match("comment") != null || divClass.match("comment") != null || divId.match("share") != null || divClass.match("share") != null) && ( pre == 0 && code == 0 && embed == 0 && object == 0 && sphereit == 0 )) {
                if (!p == 0 && img == 1) { divsList[i].parentNode.removeChild(divsList[i]); }
            }
        }
        var stopwords = ['comment','share','footer','^ad'];
        for (var sw = 0;sw<stopwords.length;sw++) {
          regex = new RegExp(stopwords[sw]);
          if (divId.match(regex) != null || divClass.match(regex) != null) {
              console.log('matched '+stopwords[sw]);
                divsList[i].parentNode.removeChild(divsList[i]);
        }
        }
    // if (divId.match("comment") != null || divClass.match("comment") != null || divId.match("share") != null || divClass.match("share") != null || divClass.match("footer") != null || divId.match("footer") != null || divClass.match(/^ad/) != null || divId.match(/^ad/) != null) {
    //    divsList[i].parentNode.removeChild(divsList[i]);
    // }
    }
    return e;
}

function killBreaks ( e ) {
    e.innerHTML = e.innerHTML.replace(/(<br\s*\/?>(\s|&nbsp;?)*){1,}/g,'<br />');
    return e;
}

function killCodeSpans ( e ) {
    e.innerHTML = e.innerHTML.replace(/<\/?\s?span(?:[^>]+)?>/g,"");
    return e;
}

function clean(e, tags, minWords) {
    var targetList;
    var y;
    if (tags == "table") {
        targetList = e.getElementsByTagName( tags );
        minWords = minWords || 1000000;
        for (y=0; y < targetList.length; y++) {
            // If the text content isn't laden with words, remove the child:
            cells = targetList[y].getElementsByTagName('td').length;
            if (cells < minWords) {
                targetList[y].parentNode.removeChild(targetList[y]);
            }
        }
    } else {
        targetList = e.getElementsByTagName( tags );
        minWords = minWords || 1000000;

        for (y=0; y < targetList.length; y++) {
            // If the text content isn't laden with words, remove the child:
            if (getCharCount(targetList[y], " ") < minWords && targetList[y].tagName != 'pre') {
                targetList[y].parentNode.removeChild(targetList[y]);
            }
        }
    }
    return e;
}

function convert(e,tagId){
    var children,parent,newNode;
    var elems = document.getElementsByTagName(tagId);
    for (y=0; y < elems.length; y++) {
        children = elems[y].childNodes;
        parent = elems[y].parentNode;
        newNode = document.createElement("span");
        newNode.setAttribute("style","font-weight:bold");
        for(var i=0;i<children.length;i++){
            newNode.appendChild(children[i]);
        }
        parent.replaceChild(newNode,elems[y]);
    }
    return e;
}