WkHtmlToPdf Table Splitting Hack

wkhtmltopdf.tablesplit.js

/**
 * WkHtmlToPdf table splitting hack.
 * 
 * Script to automatically split multiple-pages-spanning HTML tables for PDF 
 * generation using webkit.
 * 
 * To use, you must adjust pdfPage object's contents to reflect your PDF's 
 * page format. 
 * The tables you want to be automatically splitted when the page ends must 
 * have a class name of "splitForPrint" (can be changed).
 * Also, it might be a good idea to update the splitThreshold value if you have 
 * large table rows.
 * 
 * Dependencies: jQuery.
 * 
 * WARNING: WorksForMe(tm)! 
 * If it doesn't work, first check for javascript errors using a webkit browser.
 * 
 * Also, the newest wkhtmltopdf (>= 0.12) fixed this bug, so the script isn't necessary anymore.
 * Use it only if you're stuck with an older version.
 * 
 * Care must be taken if your PDF includes some responsive framework (Bootstrap, Foundation) that makes 
 * use of CSS @media!
 * 
 * @author Florin Stancu <niflostancu@gmail.com>
 * @version 1.1
 * @license http://www.opensource.org/licenses/mit-license.php MIT License
 */


/**
 * PDF page settings.
 * Must have the correct values for the script to work.
 * All numbers must be in inches (as floats)!
 * Use google to convert margins from mm to in ;) 
 * 
 * @type {Object}
 */
var pdfPage = {
	width: 8.26771654, // inches, 210mm
	height: 11.6929134, // inches, 296mm
	margins: {
		top: 1.96850394, left: 0.393700787, 
		right: 0.393700787, bottom: 0.393700787 
	}
};

/**
 * The distance to bottom of which if the element is closer, it should moved on 
 * the next page. Should be at least the element (TR)'s height.
 * 
 * @deprecated Now it is automatically detected from the TR's height, no longer needed.
 * @type {Number}
 */
var splitThreshold = 20;

/**
 * Class name of the tables to automatically split.
 * Should not contain any CSS definitions because it is automatically removed 
 * after the split.
 * 
 * @type {String}
 */
var splitClassName = 'splitForPrint';

/**
 * Set to true to enable visual debugging of the page dimensions via HTML elements / text.
 */
var visualDebug = false;


/**
 * Window load event handler.
 * We use this instead of DOM ready because webkit doesn't load the images yet.
 */
$(window).load(function () {
	// get document resolution
	var dpi = $('<div id="dpi"></div>')
		.css({
			height: '1in', width: '1in',
			top: '-100%', left: '-100%',
			position: 'absolute'
		})
		.appendTo('body')
		.height();
	
	// page height in pixels
	var pageHeight = Math.floor(
		(pdfPage.height - pdfPage.margins.top - pdfPage.margins.bottom) * dpi);
	
	// temporary set body's width and padding to match pdf's size
	var $body = $('body');
	$body.css('width', Math.floor((pdfPage.width - pdfPage.margins.left - pdfPage.margins.right)*dpi)+'px');
	$body.css('padding-left', Math.floor(pdfPage.margins.left*dpi)+'px');
	$body.css('padding-right', Math.floor(pdfPage.margins.right*dpi)+'px');
	//$body.css('padding-top', Math.floor(pdfPage.margins.top*dpi)+'px');
	$body.css('padding-top', 0);
	
	// DEBUG: show the page height (must be an exact fit to the page's content area in order for the script to work)
	if (visualDebug) {
		$body.append('<div id="debug_div" style="position: absolute; top: 0; height:' + (pageHeight - 2) + 'px; ' + 
				'right: 0; border: 1px solid #FF0000; background: blue; color: white;">Test<br />' + pageHeight + '<br /></div>');
		$('#debug_div').append( $('#debug_div').offset().top + '');
	}
	
	/* 
	 * Cycle through all tables and split them in two if necessary.
	 * We need this in a loop for it to work for tables spanning multiple pages:
	 * first, the table is split in two; then, if the second table also spans multiple 
	 * pages, it is also split and so on until there are no more.
	 * Because when modifying the upper tables, the elements' positions will change, 
	 * we need to maintain an offset correction value.
	 * 
	 * This method can be used for all document's elements (not just tables), but the 
	 * overhead would be too big. Use CSS's `page-break-inside: avoid` which works for
	 * divs and many other block elements.
	 */
	var tablesModified = true;
	var offsetCorrection = 0;
	while (tablesModified) {
		tablesModified = false;
		
		$('table.'+splitClassName).each(function(){
			var $t = $(this);
			
			// clone the original table
			var copy = $t.clone();
			copy.find('tbody > tr').remove();
			var $cbody = copy.find('tbody');
			var found = false;
			$t.removeClass(splitClassName); // for optimisation
			
			var newOffsetCorrection = offsetCorrection;
			$('tbody tr', $t).each(function(){
				var $tr = $(this);
				
				// compute element's top position and page's end
				var top = $tr.offset().top;
				var ctop = offsetCorrection + top;
				var pageEnd = (Math.floor(ctop/pageHeight)+1)*pageHeight;
				
				// DEBUG: prints TR's top and the current page end inside its first column
				if (visualDebug) {
					//if (Math.random() > 0.7)
					//	$tr.find('td:first').append('<br /> MULTI!');
					$tr.find('td:first').prepend('<div style="position: absolute;  z-index:2; background: #EEE; padding: 2px;" class="debug">' +
						ctop + ' / ' + pageEnd + '/ off=' + offsetCorrection + ' / h=<span class="tr-height">-</span>px' + '</div>' );
				}
				
				// check whether the current element is close to the page's end. 
				// dynamic threshold
				var threshold = splitThreshold;
				if ($tr.height() > threshold)
					threshold = $tr.height() + 10;
				if (visualDebug)
					$tr.find('.tr-height').text($tr.height());
				
				if (found || (ctop >= (pageEnd - threshold))) {
					// move the element to the cloned table
					$tr.detach().appendTo($cbody);
					if (visualDebug) $tr.find('td .debug').append(' D!');
					
					if (!found) {
						// compute the new offset correction
						newOffsetCorrection += (pageEnd - ctop);
					}
					found = true;
				}
			});
			
			// if the cloned table has no contents...
			if (!found) 
				return;
			
			offsetCorrection = newOffsetCorrection;
			tablesModified = true;
			// add a page-breaking div 
			// (with some whitespace to correctly show table top border)
			var $br = $('<div style="height: 15px;"></div>')
				.css('page-break-before', 'always');
			$br.insertAfter($t);
			copy.insertAfter($br);
		});
	}
	
	// restore body's padding
	$body.css('padding-left', 0);
	$body.css('padding-right', 0);
	$body.css('padding-top', 0);
});

A few questions, if you'll allow me.
Is the variable dpi meant to match the --dpi given to wkhtmltopdf? It seems to be always 96 but the results look equally good with --dpi 100/300/1300
Is this script always meant to be used with the --disable-smart-shrinking switch? I get a different result with or without it and with it seems to produce the desired result.
What version of wkhtmltopdf have you used this with?

My goal is to get a good splitThreshold dynamically for each table so that it works well with very different tables on very long documents with consistent results (10-100 pages, tr height from 20 to 300). I'll fork out my results if I get anything naturally.
Thank you for this great script!

Hey guys please give me an answer on this... sometimes when I split more than one table in a page the code trhow a wrong extra page break... anybody happens the same, there is a fix on this?
Thank you very much... and the split code is very good thanks for that...

Working properly for me

hi, how can i use it ?

i should use run-script option or what ?

thanks for answers :)

Great script, but seems like there is a bug somewhere when using headers and footers.
Can't break the document to the pages properly, especially when I use Landscape A4 format for pdf.

Hi All,

I had some problems with the original (table footers, tables not being populated properly), so I forked+hacked and here is the result!

https://gist.github.com/bluntelk/5573089

comments appreciated!

Wanted to leave my 2cents also.

First, thanks for the idea for the script. Your implementation was nice but it wasn't optimized for large tables.
I've re-written the whole script and it now actually works for 150+ paged tables without delays.
My version also supports showing custom table headers on each new page.

If anyone is interested you can grab it here: https://github.com/AAverin/JSUtils/tree/master/wkhtmltopdfTableSplitHack
If you notice any issues you can add them on my repo page, I'll see what I can do.
Thanks

Hi
pl provide sample working code of working html page...

set var splitThreshold in pixels/inches/something else?

Hi,

I tried this for splitting my tables that needs to be rendered to a pdf. The table is getting splitted when displayed as html but the pdf generated is not having splitted tables. The issue I found is that on runtime the html code comes with splitted < table >tags, but the actual html code has single < table >.. < / table> tag which is sent for table creation.

Any pointers on how to acheive this?

TIA,
Subha

Oh, sorry I didn't respond to those, I didn't get any notification emails from github for any of your comments :(

And sorry it doesn't cover all corner cases, I only used it for a website some time ago, WorksForMe(TM) :D

Also, thanks for further fixing and optimizing the code.

Thanks for sharing!
I did a similar script that does vertical splitting on wide tables instead of horizontal splitting.

I didn't find any satisfying solution for it on the web, so maybe you can find it helpful:
https://gist.github.com/vstefanoxx/574aa61eaf2cc91dd9c9

Here is a working example:
http://jsfiddle.net/mU2Ne/

Didn't used, by +1 for making me figure out why the f* all my code to measure elements on the page was not working properly in wkhtmltopdf: I have to set the body width explicitly.

You're a real saviour! Thanks for the script!