regluit

The first version of the unglue.it codebase was a services-oriented project named "unglu". We decided that "unglu" was too complicated, so we started over and named the new project "regluit". regluit is a Django project that contains four main applications: core, frontend, api and payment that can be deployed and configured on as many ec2 instances that are needed to support traffic. The partitioning between these modules is not as clean as would be ideal. payment is particularly messy because we had to retool it twice because we had to switch from Paypal to Amazon Payments to Stripe.

regluit was originally developed on Django 1.3 (python 2.7) and currently runs on Django 1.8.

Develop

Here are some instructions for setting up regluit for development on an Ubuntu system. If you are on OS X see notes below to install python-setuptools in step 1:

1. aptitude install python-setuptools git python-lxml
2. sudo easy_install virtualenv virtualenvwrapper
3. git clone git@github.com:Gluejar/regluit.git
4. cd regluit
5. mkvirtualenv regluit
6. pip install -r requirements_versioned.pip
7. add2virtualenv ..
8. cp settings/dev.py settings/me.py
9. mkdir settings/keys/
10. cp settings/dummy/* settings/keys/
11. edit settings/me.py with path info, etc. to make things work
12. edit the files in settings/keys filling in account and key information OR if you have the ansible vault password, install ansible, then from the vagrant directory run ansible-playbook create_keys.yml
13. echo 'export DJANGO_SETTINGS_MODULE=regluit.settings.me' >> ~/.virtualenvs/regluit/bin/postactivate
14. deactivate ; workon regluit
15. django-admin.py syncdb --migrate --noinput
16. django-admin.py celeryd --loglevel=INFO start the celery daemon to perform asynchronous tasks like adding related editions, and display logging information in the foreground.`
17. django-admin.py celerybeat -l INFO to start the celerybeat daemon to handle scheduled tasks.
18. django-admin.py runserver 0.0.0.0:8000 (you can change the port number from the default value of 8000)
19. point your browser at http://localhost:8000/

CSS development

1. We are using Less version 2.8 for CSS. http://incident57.com/less/. We use minified CSS.

Production Deployment

OBSOLETE Below are the steps for getting regluit running on EC2 with Apache and mod_wsgi, and talking to an Amazon Relational Data Store instance. Instructions for setting please are slightly different.

1. create an ubuntu ec2 instance (e.g, go http://alestic.com/ to find various ubuntu images)
2. sudo aptitude update
3. sudo aptitude upgrade
4. sudo aptitude install git-core apache libapache2-mod-wsgi mysql-client python-virtualenv python-mysqldb redis-server python-lxml postfix python-dev libmysqlclient-dev
5. sudo mkdir /opt/regluit
6. sudo chown ubuntu:ubuntu /opt/regluit
7. cd /opt
8. git config --global user.name "Raymond Yee"
9. git config --global user.email "rdhyee@gluejar.com"
10. ssh-keygen
11. add ~/.ssh/id\_rsa.pub as a deploy key on github https://github.com/Gluejar/regluit/admin/keys
12. git clone git@github.com:Gluejar/regluit.git
13. cd /opt/regluit
14. create an Amazon RDS instance
15. connect to it, e.g. mysql -u root -h gluejardb.cboagmr25pjs.us-east-1.rds.amazonaws.com -p
16. CREATE DATABASE unglueit CHARSET utf8;
17. GRANT ALL ON unglueit.\* TO ‘unglueit’@’ip-10-244-250-168.ec2.internal’ IDENTIFIED BY 'unglueit' REQUIRE SSL;
18. update settings/prod.py with database credentials
19. virtualenv ENV
20. source ENV/bin/activate
21. pip install -r requirements_versioned.pip
22. echo "/opt/" > ENV/lib/python2.7/site-packages/regluit.pth
23. django-admin.py syncdb --migrate --settings regluit.settings.prod
24. sudo mkdir /var/www/static
25. sudo chown ubuntu:ubuntu /var/www/static
26. django-admin.py collectstatic --settings regluit.settings.prod
27. sudo ln -s /opt/regluit/deploy/regluit.conf /etc/apache2/sites-available/regluit
28. sudo a2ensite regluit
29. sudo a2enmod ssl rewrite
30. cd /home/ubuntu
31. copy SSL server key to /etc/ssl/private/server.key
32. copy SSL certificate to /etc/ssl/certs/server.crt
33. sudo /etc/init.d/apache2 restart
34. sudo adduser --no-create-home celery --disabled-password --disabled-login (just enter return for all?)
35. sudo cp deploy/celeryd /etc/init.d/celeryd
36. sudo chmod 755 /etc/init.d/celeryd
37. sudo cp deploy/celeryd.conf /etc/default/celeryd
38. sudo mkdir /var/log/celery
39. sudo mkdir /var/run/celery
40. sudo chown celery:celery /var/log/celery /var/run/celery
41. sudo /etc/init.d/celeryd start
42. sudo cp deploy/celerybeat /etc/init.d/celerybeat
43. sudo chmod 755 /etc/init.d/celerybeat
44. sudo cp deploy/celerybeat.conf /etc/default/celerybeat
45. sudo mkdir /var/log/celerybeat
46. sudo chown celery:celery /var/log/celerybeat
47. sudo /etc/init.d/celerybeat start

setup to enable ckeditor to work properly

1. mkdir /var/www/static/media/
2. sudo chown ubuntu:www-data /var/www/static/media/

Updating Production

1. Study the latest changes in the master branch, especially keep in mind how it has changed from what's in production.
2. Update the production branch accordingly. If everything in master is ready to be moved into production, you can just merge master into production. Otherwise, you can grab specific parts. (How to do so is something that should probably be described in greater detail.)
3. Login to unglue.it and run /opt/regluit/deploy/update-prod

OS X Developer Notes

To run regluit on OS X you should have XCode installed

Install virtualenvwrapper according to the process at http://blog.praveengollakota.com/47430655:

1. sudo easy\_install pip
2. sudo pip install virtualenv
3. pip install virtualenvwrapper

Edit or create .bashrc in ~ to enable virtualenvwrapper commands:

1. mkdir ~/.virtualenvs

2. Edit .bashrc to include the following lines:

   export WORKON_HOME=$HOME/.virtualenvs source your_path_to_virtualenvwrapper.sh_here

In the above web site, the path to virtualenvwrapper.sh was /Library/Frameworks/Python.framework/Versions/2.7/bin/virtualenvwrapper.sh In Snow Leopard, this may be /usr/local/bin/virtualenvwrapper.sh

Configure Terminal to automatically notice this at startup: Terminal –> Preferences –> Settings –> Shell Click "run command"; add source ~/.bashrc

If you get 'EnvironmentError: mysql_config not found' edit the line ~/.virtualenvs/regluit/build/MySQL-python/setup_posix.py

1. mysql_config.path = "mysql_config" to be (using a path that exists on your system)
2. mysql_config.path = "/usr/local/mysql-5.5.20-osx10.6-x86_64/bin/mysql_config"

You may need to set utf8 in /etc/my.cnf collation-server = utf8_unicode_ci

init-connect='SET NAMES utf8'
character-set-server = utf8

Selenium Install

Download the selenium server: http://selenium.googlecode.com/files/selenium-server-standalone-2.5.0.jar

Start the selenium server: 'java -jar selenium-server-standalone-2.5.0.jar'

MARC Records

For unglued books with existing print edition MARC records

1. Get the MARCXML record for the print edition from the Library of Congress.
   1. Find the book in catalog.loc.gov
   2. Click on the permalink in its record (will look something like lccn.loc.gov/2009009516)
   3. Download MARCXML
2. At /marc/ungluify/ , enter the unglued edition in the Edition field, upload file, choose license
3. The XML record will be automatically...
   • converted to suitable MARCXML and .mrc records, with both direct and via-unglue.it download links
   • written to S3
   • added to a new instance of MARCRecord
   • provided to ungluers at /marc/

For CC/PD books with existing records that link to the ebook edition

1. Use /admin to create a new MARC record instance
2. Upload the MARC records to s3 (or wherever)
3. Add the URLs of the .xml and/or .mrc record(s) to the appropriate field(s)
4. Select the relevant edition
5. Select an appropriate marc_format:
   • use DIRECT if it links directly to the ebook file
   • use UNGLUE if it links to the unglue.it download page
   • if you have records with both DIRECT and UNGLUE links, you'll need two MARCRecord instances
   • if you have both kinds of link, put them in separate records, as marc_format can only take one value
     ungluify_record.py should only be used to modify records of print editions of unglued ebooks. It will not produce appropriate results for CC/PD ebooks.

For unglued ebooks without print edition MARC records, or CC/PD books without ebook MARC records

1. Get a contract cataloger to produce quality records (.xml and .mrc formats)
   • we are using ungas the format for our accession numbers, where is the id of the MARCRecord instance, plus leading zeroes
2. Upload those records to s3 (or wherever)
3. Create a MARCRecord instance in /admin
4. Add the URLs of the .xml and .mrc records to the appropriate fields
5. Select the relevant edition
6. Select an appropriate marc_format:
   • use DIRECT if it links directly to the ebook file
   • use UNGLUE if it links to the unglue.it download page
   • if you have records with both DIRECT and UNGLUE links, you'll need two MARCRecord instances
   • if you have both kinds of link, put them in separate records, as marc_format can only take one value

vagrant / ansible

How to build machines using Vagrant/ansible