What's the best way to get usable DOM for an AJAX-requested page without loading any related images/scripts/etc?
Backstory:
I want to load a page in background, then perform a sort of data-mining on it (this is a browser extension, so I can't control the pages themselves). I do not want to spend time loading images and running scripts on the background page, since it is only page contents I need.
load data via ajax
strip all the tags containing src and href attributes, or simply change the value of those attributes with data:null. If data also contains inline style you should remove all statements containing a reference to external resources (e.g. background and border images, .htc components, xul bindings, .ico cursor)
append filtered data to the DOM and analyze it
step 2 could be achieved through a regular expression in javascript. e.g.
/* here we are in the ajax "success" callback */
...
data = data.replace(/(src|href|style)=['"]([^'"]+?)['"]/gi,
function(match, attribute) {
return (attribute.toLowerCase() === 'style')
? attribute + '=""' /* remove all inline style */
: attribute + '="data:null"'; /* href and src set to data:null */
})
/* append filtered data */
$(data).appendTo($('body_or_other_element'))
If possible, use jQuery, as I mentioned above. It makes it easy to select portions of the page DOM as needed.
Here are some examples:
You can grab tags href attribute like this: $("a", $(ajax_response)).attr("href");
Title's contents: $("title", $(ajax_response)).html();
You might have to test out the selectors to see which work best, but, I think this would be an easy way of going about this.
Related
I'm testing out CKEditor
I'm trying to get the display in the editor, to match my sites css style for displaying the end result.
What I'm trying to do is style the "wrap code" button to match the css of my site, by adding in a class.
I've seen on this page of the manual, that you can do stuff like this:
config.format_pre = { element: 'pre', attributes: { 'class': 'editorCode' } };
However, doing the same for a code block like so:
config.format_code = { element: 'code', attributes: { 'class': 'someclass' } };
Doesn't actually do anything. Anyone got a pointer on what I might be missing?
I've tested it working on other elements, so I know the config file changes are being picked up.
The one important thing is that every tag which is formatted via config.format_tagname should be also included in config.format_tags. However, this two settings (config.format_tagname and config.format_tags) works only form Block-Level elements (as stated in the manual page you referenced ).
As code element is considered as an inline one by CKEditor (see DTD), it is not possible to use this config here.
However, the easiest way to modify the elements added via Style dropdown is to edit styles.js file which is present in CKEditor directory. The dropdown styles are based on this file, so you can easily modify code element there. You can also define your custom stylesSet.
I have a situation where I am storing dynamic css data about a text object in a database as json. I need to map that same css data into styles in CKEditor. I am successfully able to load the classes into the CKEDITOR styles dropdown by parsing the json into the style set by running:
CKEDITOR.stylesSet.add('myStyles',styleObj);
Unfortunately this does not fully work with the onscreen text because the css does not exists as a file.
I've also successfully generate the css into the head of the dom by appending the dynamically generated css to a style tag. Unfortunately this still does not connect the actual css generated to the CKEDITOR because it is in a separate context.
Does anyone know how I can either connect document level css to the CKEDITOR instance or generate the CSS in a way that CKEDITOR understands? I'd prefer not to write a temporary CSS file to disk for every single user who needs to view the text object.
I figured out the answer to this by using the CKEDITOR.addCss() function.
Instead of trying to load the css into the document head as styles, the process can be much simpler by running CKEDITOR.addCss() function.
The code looks like:
for each css style found in the json:
styleObj.push({name:this.name,element:'p',attributes: { 'class':cssClassName}});
var cssSheetString = '.'+cssClassName+' {font-family:'+this.fontFamily+'; font-size:'+fontSize+'; font-weight:'+this.fontStyle+'; text-decoration:'+textDecoration+'; } ';
CKEDITOR.addCss(cssSheetString);
after the loop ends then also add the styles object:
if(!CKEDITOR.stylesSet.registered.myStyles){
CKEDITOR.stylesSet.add('myStyles',styleObj);
}
Just for posterity. I've seen answers that say this will work
CKEDITOR.on('instanceCreated', function (event) {
event.editor.addCss(styles);
});
but it does not, you have to use
CKEDITOR.on('instanceCreated', function (event) {
CKEDITOR.addCss(styles);
});
also if your styles variable changes you have to destroy and recreate your ckeditor instance with the new styles.
I have links such that when the user clicks on them, the DOM is quckly updated using the methods below.
Basically, I just set the innerHTML document to the text and the page updates.
However I would like html code with other html code when applicable. This is the only place in my .js file that has a significant amount of text. How do I move this?
/*
link - quick dom links - would like to find a way to move this into xhtml where it belongs
*/
function o2(a,b)
{
return document.getElementById(a).innerHTML=b;
}
function l1()
{
........
I would recommend putting all of the possible HTML into your HTML file. Assign a unique id to each element and use CSS to hide them all or all but one by default (using 'display: none'). Then your javascript function can simply change CSS based on which html fragment you need to be visible.
I have a pretty specific scenario where I would like to select all elements with jQuery, make a CSS change, save the elements, then reverse the change I made.
The Goal
I created a jQuery plugin called jQuery.sendFeedback. This plugin allows the user to highlight areas of the screen, as shown in this demo. When they submit their feedback the plugin grabs all the HTML on the page and dumps it into a callback function. Like so:
$('*').each(function ()
{
$(this).width($(this).width());
$(this).height($(this).height());
});
var feedbackInformation = {
subject: $feedbackSubject.val(),
details: $feedbackDetails.val(),
html: '<html>' + $('html').html() + '</html>'
};
if (settings.feedbackSent)
settings.feedbackSent(feedbackInformation);
The callback function accepts this feedback information and makes an AJAX call to store the page HTML on the server (this HTML includes the red box highlights the user drew on the screen). When someone from tech support needs to view the user's "screen shot" they navigate to a page that serves up the stored HTML so the developer can see where the user drew their highlights on the screen.
My original problem was that different screen resolutions made the elements different sizes and the red highlights would highlight the wrong areas as the screen changed. This was fixed pretty easily by selecting all elements on the page and manually setting their height and width to their current height and width when the user takes the snap shot. This makes all the element sizes static, which is perfect.
$('*').each(function ()
{
$(this).width($(this).width());
$(this).height($(this).height());
});
The Problem
The issue with this is that when the plugin is done transmitting this HTML the page currently being viewed now has static heights and widths on every element. This prevents dropdown menus and some other things from operating as they should. I cannot think of an easy way to reverse the change I made to the DOM without refreshing the page (which may very well end up being my only option). I'd prefer not to refresh the page.
Attempted Solution
What I need is a way to manipulate the HTML that I'm sending to the server, but not the DOM. I tried to change the above code to pull out the HTML first, then do the operation on the string containing the HTML (thus not affecting the DOM), but I'm not quite sure what I'm doing here.
var html = '<html>' + $('html').html() + '</html>';
$('*', html).each(function ()
{
$(this).width($(this).width());
$(this).height($(this).height());
});
This did not work. So either I need to be able to manipulate the string of HTML or I need to be able to manipulate the DOM and undo the manipulation afterward. I'm not quite sure what to do here.
Update
I employed the solution that I posted below it is working beautifully now. Now I am wondering if there is a way to statically write all the css for each element to the element, eliminating the need for style sheets to be referenced.
I think you are mostly on the right track by trying to make the modifications to the HTML as a string rather than on the current page for the user.
If you check this post, you might also want to follow the recommendation of creating a temporary <div> on the page, cloning your intended content to the new <div> ensuring it is invisible using "display:none." By also putting a custom Id on the new <div> you can safely apply your static sizing CSS to those elements using more careful selectors. Once you have sent the content to the server, you can blow away the new <div> completely.
Maybe?
After much pain and suffering I figured a crude but effective method for reverting my modifications to the DOM. Though I hadn't gotten around to trying #fdfrye's suggestion of cloning, I will be trying that next to see if there is a mroe elegant solution. In the meantime, here is the new code in case anyone else can benefit from it:
$('*').each(function () {
if ($(this).attr('style'))
$(this).data('oldStyle', $(this).attr('style'));
else
$(this).data('oldStyle', 'none');
$(this).width($(this).width());
$(this).height($(this).height());
});
var html = '<html>' + $('html').html() + '</html>';
$('*').each(function () {
if ($(this).data('oldStyle') != 'none')
$(this).attr('style', $(this).data('oldStyle'));
else
$(this).removeAttr('style');
});
When I'm looping through every element and modifying the css, I log the original value onto the element as data. After I assign the DOM HTML to a variable I then loop through all elements again and restore the style attribute to its original value. If there was no style attribute then I log 'none' to the element data and then remove the style attribute entirely when looping through again.
This is more performance heavy than I wish it was since it loops through all elements twice; it takes a few seconds to finish. Not horrible but it seems like a little much for such a small task. Anyway, it works. I get a string with fixed-sized HTML elements and the DOM goes back to normal as if the plugin never touched it.
We've got a little tool that I built where you can edit a jQuery template in one field and JSON data in another and then hit a button to see the results immediately within the browser.
I really need to expand this though so the designer can edit a full CSS stylesheet within another field and when we render the template, it will have the CSS applied to it. The idea being that once we've got good results we can take the contents of these three fields, put them in files and use them in our project.
I found the jQuery.cssRule plugin but it looks like it's basically abandoned (all the links go nowhere and there's been no development in three years). Is there something better or is it the only game in town?
Note: We're looking for something where someone types traditional CSS stylesheet data in here and that is used immediately for rendering within the page and that can be edited and changed at will with the old rules going away and new ones used in their stead. I'm not looking for something where the designer has to learn jQuery syntax and enter in individual .css("attribute", "value") type calls to jQuery.
Sure, just append a style tag to the head:
$("head").append("<style>p { color: blue; }</style>");
See it in action here.
You can replace the text in a dynamically added style tag using something like this:
$("head").append("<style id='dynamicStylesheet'></style>");
$("#dynamicStylesheet").text(newStyleTextGoesHere);
See this in action here.
The cleanest way to achieve this is by sandboxing your user-generated content into an <iframe>. This way, changes to the CSS won't affect the editor. (For example, input { display:none; } can't break your page.)
Just render out your HTML (including the CSS in the document's <head>, and write it into the <iframe>.
Example:
<iframe id="preview" src="about:blank">
var i = $('#preview')[0];
var doc = i.contentWindow || i.contentDocument;
if (doc.document) doc = doc.document;
doc.open('text/html',true);
doc.write('<!DOCTYPE html><html>...</html>');
doc.close();
If the user should be able to edit a whole stylesheet, not only single style attributes, then you can store the entered stylesheet in a temporary file and load it into your html document using
$('head').append('<link rel="stylesheet" href="temp.css" type="text/css" />');
sounds like you want to write an interpreter for the css? if it is entered by hand in text, then using it later would be as simple as copy and pasting it into a css file.
so if you have a textarea on your page to type in css and want to apply those rules when you press the button, you could use something like this (only pseudocode, needs work):
//for each css id in the text area
$.each($('textarea[name=cssTextArea]').html().split('#'), function({
//now get each property
$.each($(this).split(';'), function(){
$(elem).css({property:value});
});
});
then you could write something to go through each element that your designer typed in, and get the current css rules for it (including those that you applied using some code like the snippet above) and create a css string from that which could then be output or saved in a db. It's a pain and much faffing around with substrings but unfortunately I don't know of a faster or more efficient way.
Hope this atleast gives you some ideas