I'm really excited for this for http://www.sharelatex.com and other similar sites that are actually generating a PDF for you. With native PDF viewers there is no way to interact with the viewer via javascript and even just having the viewer stay on the same page when your reload a document (with minor changes) is impossible. Pdf.js means that we'll be able to do this easily, as well as other cool things like letting letting the user sync between the PDF and source.
Are you guys actually generating PDFs from user input? If so, I'd be really interested in finding out how you do it if you don't mind sharing. I've found a number of apps that allow you to generate PDFs server side but often the installation is near impossible (dependency issues), e rendering is not good enough, or the API makes using it with a web app clunky. I did find a solution (don't recall the name) that actually used Webkit to render web pages and then turned them into PDFs but I was never able to get it to work in an app. It only seemed to work some of the time and only through ssh. But if you can't say, then I can respect that.
I used to use PDF.js for a while (on Linux), until I switched to KParts because it was having difficulty rendering certain kinds of PDF documents. KParts uses the same underlying engine that powers Okular (KDE's default PDF reader.) It renders everything properly and is much faster than PDF.js. It reminded me of Foxit on Windows. KParts might be only available on Linux though...
I use this a lot, and it really does work. It renders everything, and renders it well. The one thing that doesn't work is maps – just too many vectors, and Javascript/Canvas/etc just can't keep up. Otherwise I'm very happy and don't feel nearly as much resentment towards PDFs as I used to.
Yes, I really love PDF.js. I was consider switching to Chrome just because of the (proprietary) pdf viewer. But the Chrome PDF viewer even lacks the concept of pages! I had already kicked out the Adobe Reader plugin but using an external viewer every time makes PDF just so more annoying.
Mozilla is also working on something similar for Flash: https://github.com/mozilla/shumway . I'm a bit sceptical but with the success of PDF.js I can't wait to use it.
1) In Firefox 15, the demo page adds two new options to my right click menu: Rotate clockwise and Rotate Counter-clockwise. Is Firefox recognizing pdf.js (since it appears that they are related) or pdf.js adding menu options? I didn't know JS could do that.
2) Isn't Javascript an embeddable language inside PDFs? I'm pretty sure I read that javascript is used, not necessarily for animations but for run-time dynamic layouts. If that's true, is pdf.js "eval"-ing that javascript?
I've seen custom contextmenus generated on pages[0] that override the built-in context menu but for some reason I've yet to come across a page[1] that adds to the existing context menu until today.
>but for some reason I've yet to come across a page that adds to the existing context menu until today.
the reason is likely because browser compatibility for the context menu js api is pretty minimal [1]. firefox can use it on pages that are only going to be displayed in firefox, but for the internet at large it's pretty useless.
What about Google Docs? Their context menu seems to work cross browser (even on ie9!). Maybe my assumption that they're using HTML5 for this is wrong but if I'm right then what you see on Gdocs is a complete replacement of the regular menu which is pretty cool but also quite annoying for me at times. Please correct me if I'm wrong (which is very likely).
Google docs is not appending stuff to the browser's context menu. they're catching the right-click action and replacing the standard context menu with their own one drawn in HTML.
While I can't imagine myself using it anytime soon, it's clear that web applications are improving at a far faster rate then native applications and, with t large enough, the first derivative means that web will eclipse native.
This seems like an academic exercise at the moment; it's to prove that you can replicate a native experience only.
However, it seems that this could be vastly improved by playing to the strengths of the internet. The only online apps that have beat native ones so far have been because of cloud storage and collaboration. First, use filepicker.io or something so this can open my online files. Second, bake some collaboration into it.
GroupDocs currently provide an app for online annotation and collaboration, including accessing your files from different cloud storage providers, currently Azure & Amazon S3 are supported http://groupdocs.com/apps/annotation/try-it-now
I came across this a few months ago while trying to implement a solution for turning HTML into PDFs server side. This is definitely cool and useful but it's usefulness is limited for now as native PDF readers on the desktop are preferable. Even on iOS the built in reader is nicely done. Chrome on Windows and Mac always opens PDFs in a tab and handles it well I think. That said, this can definitely be of use in Chromebook type situations. I'm sure it'll end up in Firefox OS too which I have hi expectations for. The awesome thing about Firefox OS is that it's all JavaScript and good old fashioned web technologies under the hood so this will fit right in.
So alas, I'm still searching for an easy way to convert HTML to PDF server or client side. I haven't looked at the code yet but I do wonder if one could get that functionality out of this if they wrestled with it enough. (I know there are other ways to turn HTML to PDF but a client or server side script to do so really is the best solution for my situation).
Just as interesting, in my mind, is the inverse library -- jspdf [1] lets you create pdfs in javascript. For automatic document generation, I find I can quickly whip something up in jsbin or jsfiddle that will give me a pdf I can download and do whatever I want with.
i've stumbled upon PDF.js a while ago because i was looking for js tool that allows me to extract data from pdfs... sadly i'm still looking for a good lib to do just that.
This is great, I love it. I can read PDF files without leaving the browser, in any browser. I find it slightly distracting to switch to a third-party application.
Feels like they aren't taking window.devicePixelRatio into account so that they end up with a legitimately retina-sized canvas... My canvases were always blurry like this until I took dpr into account.
Looks like it has basic support for annotations, but is limited in its scope. I'm hoping I can use this to read annotations from PDFs so I can hook it into a journal paper reading/organisation workflow.
1) From a technical perspective, this is damn cool - exceedingly well done. Color me very impressed.
2) I hope I never actually see anyone using this on a website, attempting to make things "easier." Between Scribd and Slideshare, and Adobe trying to force its hideous crash-prone plugins into my browser, there are already enough people making a mess out of what is one of the more well-thought-out aspects of OS X. Give me a link to a PDF, which Preview.app handles in wonderful fashion any day.
3) It would make a sweet browser plugin on browser-in-a-box platforms and other platforms that don't have a nice native implementation (which upon further reading seems to be the goal).
Personally I wish all sites used this instead of scribd/slideshare or opening up external programs.
This will just work on different browsers, different OS etc.
Plus you can send links like this: http://mozilla.github.com/pdf.js/web/viewer.html#page=6&...
To send a user to s specific part of the page. Right now usually you just send the pdf link and have to say, "check out the image on page 6".
Keeping the pdf in the browser and accessible to and from javascript opens up a world of possibilities.
> This will just work on different browsers, different OS etc.
For values of "different" that mean "as long as they all have fast processors and accelerated JavaScript". I tried this in a non-Apple browser on my iPad, which means no JITted Javascript interpreter, and it basically hung. I haven't even attempted it on my smartphone, which does have a native PDF-reading application.
I only noticed it in FF15, but it may have been there for longer. Change pdfjs.disabled to false in about:config (followed by a browser restart) if you want to turn it on. Here's a test file[0] should you wish to test it out.
I've been using pdf.js for a while on my Windows box (where it is enabled by default on the FF beta channel) and it has performed remarkably well. Sometimes it's a bit slower than Adobe Reader or messes up colour or formatting, but on the whole I've quite enjoyed using it.
Sweet! (It voids the non-existent FF warranty though...)
What's better is, it allows the (awesome) extension Pentadactyl to click the buttons in the viewer (since it's part of the browser). Have to click on (or tab to?) to the document itself though, in order to be able to use keyboard navigation on it though :-/
about:config isn't really optimized for usability and there's some advantage to the double negative in that the absence of a value could be treated as a false boolean value.
So if at one point pdfjs should be enabled by default, they can just get rid of the preference in the default preferences file altogether instead of having to ship one where it says enabled: true (because absence of the value would be treated as false)
>Give me a link to a PDF, which Preview.app handles in wonderful fashion any day.
Really? I can't stand that. My download folder ends up cluttered with (inevitably weirdly named) PDF files. Worse, I suddenly am switching between programs and tabs, rather than just tabs, and the two programs have very different search interfaces (and Preview.app's is downright clumsy). Since I rarely look at a PDF if I'm not researching something, navigation and searching are really important considerations.
I'm quite happy with the Chrome PDF viewer though.
For me it put anything in my download folder. It opens in the browser in a reasonable, non-resource-abusive fashion. I'm on Safari 6/Mountain Lion, so I'm not sure what version this happened.
OK, that's new in Mountain Lion apparently (Lion was the OS X that chased me away). I assumed it was the same as in previous versions since you mentioned Preview.app, but it appears that Preview.app isn't involved anymore when you open links in Safari; it just uses a built-in PDF viewer the same way Chrome does.
You said "by default" which is correct but just for the sake of completeness, the built in Chrome PDF reader can easily be added to Chromium on Arch using this package from the AUR: https://aur.archlinux.org/packages.php?ID=44148
I've been using it for almost a year now without issue.
There is also a Google Chrome/Chromium extension also provided by the pdf.js team. (It's not in the webstore, though.)
I've used it since about half a year and I like it a lot. (Not having to download PDFs all the time and having PDFs in tabs alongside the sites I read.)
You can also separately download the proprietary PDF reader of Chrome and use it with Chromium. If you are using Arch Linux, it is available in the repositories.
I found Preview.app always annoying. E.g., the maximize windows button not maximizing the window. Sure it is better than Adobe Reader. But there are quite a few very nice alternative PDF viewers better than it. And pdf.js is great as a browser plugin because opening PDFs won't disturb your browsing any more.
I wouldn't say it's exceedingly well done... yet. It is so slow it is unusable on my OC'd core i5, and can't search pages you haven't looked at yet. The last one is a deal breaker for me. Really the only reason I use chrome for my day to day browser is for the PDF plugin. Once this project matures I'm out.
