PyBossa demo application PDF Table Transcribe
PDF Table Transcribe is a demo application for PyBossa that shows how you can crowdsourcing a PDF Table transcription problem.
By using PDF.JS, we have the possibility of rendering almost any PDF that is hosted under an HTTP server and then use a customized form to get the data that we want to extract from it .
In this simple demo application, we load a PDF file in one side of the page, and in the other one an interactive table where the volunteer will be able to transcribe the contents of the table in the PDF page by typing the text in the cells of the table. While this example is really simple, adapting the template to extract specific bits of information from the PDF will be really easy (you will only need to add more HTML input fields with instructions about what you want to extract from the PDF file). The idea is that you could be able for example to extract specific items from the documents, like captions, tabular data, authorship, institutions, etc.
The provided script for creating the tasks is very simple: you only need to tell the script where is the PDF file hosted, the URL, and which pages you want to convert as tasks. By default, this demo explores the 14 pages of the example PDF file.
Re-using the application for your project
You need to install the pybossa-client first (use a virtualenv):
$ pip install -r requirements.txt
Then, you can follow the next steps:
- Create an account in crowdcrafting.org
- Copy under your account profile your API-KEY
- Run python createTasks.py -u http://crowdcrafting.org -k API-KEY -f http://domain/yourpdf -p numPages
- Open with your browser the Applications section and choose the PDF Transcribe app. This will open the presenter for this demo application.
Please, check the full documentation here about how to create an app:
Serving the PDFs using Dropbox Public Folder
Dropvox allows you to store your PDF file for free (check its conditions and term of services) in your public folder. You can use this specific solution for storing and serving the PDF files, in the begining. If you want to have more control, please, check the next section where we explain how you can set up your own server.
In case you want to use the Dropbox solution, you have to follow the next steps:
- Copy the PDF file to your Dropbox public folder.
- Wait until it is uploaded, and then copy its public link
Now you can create the tasks:
$ python createTasks.py -k YOUR-API-KEY -s SERVER -c -f PDF-DROPBOX-PUBLIC-LINK -p NUMBER-PAGES-PDF
And you are done!
Setting up your Apache web server for hosting the PDF files
Usually you will have a set of PDF files that you are currently serving via a web server.
If you use the application as it is, you will see that it does not work loading the PDFs, even though the URL links are fine and the PDF pages are correct in the Google Spreadsheet that you have created. The problem, is that you need to enable CORS in order to get access to your PDF files.
In Enable Cors webpage you can check how you can configure most of the web servers properly, so this application can load the PDF files from other domains without problems. For example, for an Apache web server all you have to do is to enable the module mod_headers:
$ sudo a2enmod headers
Then, open the site config file, i.e. /etc/apache2/sites-enabled/000-default and add the following to the **VirtualHost section:
Header set Access-Control-Allow-Origin "*"
Finally restart the web server and you will be done! The PDFs now should be loaded without problems. Note: you can use .htaccess files too in order to not enable CORS to all your site, or if you prefer place the previous sentence in a Directory or Location, instead of at the level of the VirtualHost section.
Please, see the COPYING file.
The thumbnail has been created using a photo from TempusVolat (license CC BY-NC-SA 2.0).
Special thanks to Miquel Herrera for his JS libraries for the canvas scrolling, and Mozilla Foundation for their PDF.JS library.