For a rails app that works a lot with uploaded image heavy pdf files I'm looking for a way to optimize the file size of uploaded pdf's.
Adobe Acrobat has a 'save as reduced file size pdf' option which often halves the filesize when images are included.
I would like to do a similar action that is triggered after a file upload in my rails app.
Any ideas?
While #lzap's comment may be true, if you still want to give it a shot, you might look at pdftk (PDF Toolkit). Its a library for manipulating and creating PDF files that looks like it offers the ability to compress a given pdf file.
The library can be installed on most major operating systems, so if you have the ability to install it on your host, then simply call:
system("pdftk uncompressed-input.pdf output compressed-outpu.pdf compress")
inside your rails app whenever you want to compress a particular PDF file. I have no idea how long this would take, and if you are compressing many PDF's at the same time, you may want to consider handing off to a background job (without this, Rails will wait until the compression is done before returning anything to the browser, probably causing a timeout error for long running groups of compression calls).
Also, if your file names come from user input, be extra careful to avoid injection attacks.
Related
I'm generating a markdown document using my Rails 4.2 app which includes images that are on the same server (in the public folder).
Using pandoc (pandoc-ruby 1.0.0), I want to convert the document into various formats, especially HTML (to preview it in the browser) and DOCX (to download it).
The preview in the browser works perfect. But when converting to DOCX, the images aren't included. I guess this is due to multiple requests to referenced images while pandoc is generating the document.
I have already experimented with setting allow_concurrency to true, but this didn't solve the problem. Also, it happens on both the development and the production environment (while in development, it takes a long time, and in production it doesn't - maybe due to some differences in timeout limits?).
I have already found a way to solve my problem by not referencing the images using an URL, but by embedding it as base64 string into the document. But this for sure can't be the solution of choice, as it tends to bloat up the HTML document a lot. Also, on production, I already get RuntimeError (Stack space overflow: current size 8388608 bytes) from pretty small embedded images. So I have to find a real solution.
Reference the images by file path instead of url if they are on the same server.
In my app i want the user to be able to download offline map content.
So I (compressed) moved all my tiles into a zip file. (I used 0 compression)
The structure is like that: {z/x/y.jpg}
+0
+-0
+--0.jpg
+1
+-1
+--0.jpg
+2
+-2
+--1.jpg
So basically there are going to be many many files for zoom level 0-15. (about 120.000 tiles for my test-region).
I am using https://github.com/mattconnolly/ZipArchive now but also tried out https://github.com/soffes/ssziparchive before and both are pretty slow. It takes about 5!! minutes on my iPhone 5S for the files to unzip.
Is there any way I can speed things up? What other possibilities rather than downloading the tiles in one big zip file would there be?
Edit:
How can i download the content of the whole folder quickly to my iPhone without the need of unzipping something?
Any help is appreciated!
JPGs rarely compress at all with zip - they are by definition already compressed. What you should do is create your own binary file format, and put whatever metadata you need into it along with the images (which you should encode with a really low quality number, to get their size down).
When you download those files, you can open then, quickly read them into memory, and extract out data or images as needed.
This will be really fast and have virtually no overhead if your extra data is binary (not text).
PS: I just tripped on a PHP Plist class
If anyone is wondering how I was ending up:
For my use-case (MapTiles) I am using MBTiles now instead of zipped images. It's one big database file and super easy to read if using FMDB. No unpacking whatsoever needed...
Even if I was placing the Images all in one binary file without any compression, the "extracting" still took forever!
I am attempting to use Imagemagick to manipulate images that are uploaded by a user. Right now I have a simple set of Imagemagick.convert[ ] commands server side that preform a variety of tasks on the uploaded image. My problem comes from Imagemagick needing the file data to be read into meteor and not from a url. What I end up doing is writing the uploaded file to the /public folder where Imagemagick is able to manipulate the image. However because the list of Imagemagick.convert commands (saving and writing to /public), the application keeps refreshing, breaking up the processes and sending it into an infinite refreshing cycle. I don't think assets is a viable solution, but I need some folder that I can write to in meteor that will not interrupt the various Imagemagick processes through a refresh. I have tried the .folder for a hidden folder, but meteor gives me an error: "You can’t use a name that begins with a dot “.”, because these names are reserved for the system. Please choose another name." Any thoughts?
#Nate I wrote a little example app that solves this problem by using a temporary directory (as others have suggested):
https://github.com/ideaq/image-uploads
My solution gives you:
Easy image uploading in any Meteor app
Images are re-sized to Thumbnail, Mobile Optimised and Full-size Original
Images are uploaded to AWS S3 for CDN delivery (scalability and speed)
A thumbnail of the image is displayed on to the user without refreshing the page
if you found a better way of doing image uploads in meteor, please share! thanks. :-)
Currently image_tag("file.jpg") produces normal image html tag, BUT src="file.jpg**?7485793246**" What are those numbers anyway and how to disable them?
Those are refered to as Asset Timestamps they can be used by the server to cache files. For example lets say you have a file called file.jpg on your server, you can set up your server to tell browsers like firefox to cache the file.jpg so the next time that browser visits your web-page it loads faster because file.jpg was already in memory.
The problem comes when you upload a new file.jpg because even though the image is different, your old users who have the image cached will still see the cached image, that is where asset timestamps come into play. Those numbers represent a timestamp of when the file was updated, so if you replace file.jpg?123456789 with file.jpg?987654321 then the user's browser will not use the cached version.
http://api.rubyonrails.org/classes/ActionView/Helpers/AssetTagHelper.html
long story short, it is only there to help you, and doesn't affect the way the file behaves at all. Users can still download the file and operating systems will see it as being a valid .jpg.
thats your development mode making sure that nothing gets cached so that if you change the image it actually gets to the browser. the production version won't have it.
Solved, add this to environment.rb => ENV['RAILS_ASSET_ID'] = ''
Conclusion: yes, its a good rails stuff, but when you deal with ie6 and PNG images it can break your script. So be careful.
Wondered if anybody knows how customizable Flash swf files are made, where there appears to be a template swf that the user can then input some changes (eg text or image) and receives a newly-compiled swf file with their changes.
Some examples:
- http://flashfreezer.com/landingconfetti/index.html
Constraints:
- user receives a single output swf file that can be played with all their changes included. ie there is no reading from an xml file, or using Flashvars.
Been trying different things for a few weeks with no luck!
There are a number of ways, but generally the most common is to either use a SWF generating library (like PHP's) or through server-side compiling.
Normally, this will be a custom or proprietary library which uses the same language that the serve is running (and there are open-source libraries for this in PHP, Perl, Python, Java, C++... etc). The SWF is generated and served up with the appropriate headers so that the browser knows how to re-direct it. Often this will involve a pre-defined template which is then modified slightly for the new input. Only occasionally does this involve the manipulation of pre-generated SWF directly.
The other option is to have a command line call to the Flash IDE or the Flex compiler (and, technically, this can work for CS3 and CS4, though in a very nasty and hackish way) to generate a new version of the SWF on the fly. This is often slower, but it will generally yield a more finished feel to a product.
You could try Swiffotron. It can modify SWF files and do text replace type things on both text elements and in compiled actionscript.
Here's a swiffotron xml job file that does some text replacing.
And here's a swiffotron XML job file that modifies instances on the stage.
I didn't check the site, but the only way I can think of is to read the requirement details through flash (this can be done through plain html also) and then generate the AS files from their templates and compile them at the server side (using mxmlc or other compilers) and give back the SWF.
I get the impression that you're looking for SwfMill. SwfMill creates a swf based on an XML file that you create/define. You could use SwfMill on the server to generate a swf based on user input.