Useful information for understanding how the dish website works.
The website requires: 1) a db_private directory, in .gitignore by default, that contains a .dishlogin.cnf mysql configuration file pointing to a working mysql installation with a database called g_thedishonscience_dish_website (charset utf8). 2) an apache server setup in a configuration compatible with the example conf file, the-dish.conf 3) a working flask installation, at least python 2.6, and the modules: css-html-js-minify, python-dateutil, xlrd, markdown, mysqlclient (or mysql-python if on 2.x), sqlalchemy, enum34 (if on <py3.4), and flask
The markdown and xlrd are installed as git submodules in subdirectories of cgi-bin and manually added to sys.path in the relevant places, since Stanford's servers do not give us any way to install packages ourselves.
Run the website locally by executing: $ python2 cgi-bin/dishflask.py
On first load, the website will detect that the sql tables it requires are empty, and populate them using the information in WWW/assets/info/blog-teams.json for the team information, and all the directories in the WWW/posts directory that don't throw an error when passed to the "Post" constructor as the posts. This constructor knows to construct post.html from post.md.
In particular, this means that you'll have to fill in old view count data by hand if it exists, and keep the blog_teams.json file up to date with the sql table teams if they change. The SQL association table author_team is defined implicitly based on which authors have written articles for what teams, which is why authors are not listed in blog_teams.json.
A google analytics account exists for the actual dish website, and as of Nov 2016, it is faithfully tracking page views. So if the database must be rebuilt, it is suggested that these page view counts be used to repopulate it, although some several thousand pageviews would be missing if that was done.
If you must change information about the teams on a regular basis, it is suggested that you simply modify the file WWW/assets/info/blog-teams.json directly, then write an update script to update the relevant sql entries from that file.
On each subsequent load after the first, the site will use a global flask session to communicate with the SQL server throughout the request, loading in the versions of the post metadata stored in the sql server, and using the post.html file directly.
In particular, this means that any updates to the post's metadata must be sync'd with the sql server. Because Stanford's servers use an ancient version of Python (<2.7), we currently maintain a separate version of the dishsql module for performing such updates (dishsql_admin.py). This module contains functions for updating various aspects of the SQL database. If you're unsure what to do, you should just run dishsql_admin.initialize_website().
# Production Notes In Stanford's server's chroot'd cgi environment, only python 2.6.6 exists, and it has only a predefined set of packages installed.
This is why we keep a dishsql_admin, that can do things that aren't possible when serving the website.
To install the website for the first time, or after adding a new article folder to posts, simply $ cd cgi-bin $ ../venv-tdos/bin/ipython >>> import dishsql >>> dishsql.initialize_website() where venv-tdos holds a python environment that has the requirements listed above installed. On corn, this can be created with the pyenv-3.4 command currently.