You can work on the project using your own machine, or some other environment. However, the course staff can only support the enviroment we recommend. You will also need to make sure that the URL you submit for part 3 works when the staff visits and grades it.
The following documentation may be helpful for both learning Python and Flask:
- Java to Python Cheatsheet
- Python 2 tutorial
- Python 3 tutorial
- Learn Python The Hard Way
- Flask documentation
- Flask Tutorial
- Jinja Template documentation
- Jinja Tutorial
If your application has users, and you'd like to implement login/logout pages with password authentication, check:
- Flask Quickstart: Sessions
- Creating a login page
- Note: do not follow the "Connecting to your database" section of this tutorial, as it uses ORM. Remember that you are not allowed to use ORM, and your code must issue SQL queries instead.
One drawback of using a cloud computing platform is that it is difficult to open GUI text editors such as Sublime Text to write your code. We recommend setting up a version control system for your project, such as git on GitHub, so your team can share code. This way, you can code on your desktop, commit your changes, and pull the updated changes on your cloud virtual machine.
We will use the Flask Python webserver in this course. It is a lightweight webserver that requires a minimal amount of understanding of how the webserver framework is implemented.
To use it, follow the steps in Python Flask Skeleton for Google App Engine to create Python applications using the Flask framework on App Engine.
We strongly recommend reading the following documentations:
- General Flask Documentation
- Jinja Templates: this makes it easy to send data (e.g., arrays, dictionaries) to your HTML code and dynamically render them.
We use a python package called SQLAlchemy to simplify our work for connecting to the database.
For example, server.py contains the following code to load useful functions from
the package:
# import useful functions from the package
from sqlalchemy import *
SQLAlchemy is able to connect to many different types of DBMSes such as
SQLite, PostgreSQL, MySQL, Oracle and other databases. Each such DBMS
is called an "engine". The create_engine() function sets up the configuration
to specify which type of DBMS we want to connect to, and what their parameters are.
engine = create_engine(DATABASEURI)
Given an engine, we can then connect to it (this is similar to how psql connects
to the staff database).
conn = engine.connect()
At this point, the conn connection object can be used to
execute queries to the database. This is basically what psql
is doing under the covers!
cursor = conn.execute("select 1")
The execute function takes a SQL query string as input, and
returns a cursor object. You can think of this as an iterator
over the result relation. This means you can run select *
on a million row table, and not run out of memory. Instead of
sending the entire result at once. Instead, this
object lets you treat the result as an iterator and call .next()
on it, or loop through it. See the documentation for a detailed description.
# this fetches the first row if called right after
# the execute function above. It also moves the
# iterator to the next result row.
record = cursor.fetchone()
# this will fetch the next record, or None if
# there are no more results.
second_record = cursor.fetchone()
# this loops through the results of the cursor one by one
for row in cursor:
print list(row)
The above description is a way to directly write and run SQL queries as strings, and directly manipulate the result relations. SQLAlchemy is also an Object Relational Mapper that provides an interface that hides SQL query strings and result sets from you. Instead you access and manipulate tables in the database as if they were normal Python objects.
In this project, you will directly write and run SQL queries, and will not use any ORM functionality.
-
Fork this repository so you have your own copy that you can edit. You will submit a link to the repository. (click the Fork button on the top right corner of this page)
-
Clone it to your VM (or your local machine, if you have Python installed and want to run locally):
git clone git@github.com:[YOUR_GITHUB_USERNAME]/project1.git -
Edit your files
-
Use the following commands to add and checkpoint (commit) your changes locally
git add --help git add <new files to store in git> git commit -m "a sentence describing your changes" -
When everything has been committed you can
pushall the committed changes so GitHub.com has a copygit push -
If you cloned the repository on another machine (say the VM), then you can download and apply those changes from GitHub.com
git pull
Some notes
- Your life will be easier by setting up SSH keys
and cloning the
git://....versions of repositories. That way GitHub won't keep asking for your password when runninggitcommands. However, if don't know what this means, stick to the original HTTP version. - Most errors you will encounter can be solved by consulting a search engine.
You will deploy your application to your Google App Engine virtual machine.
Also, you'll need to open the firewall so you can access your web application. This is a one-time setup.
-
Write down the external IP of your virtual machine, but remember that it changes every time you restart it.
-
Perform some default installations and scaffolding for the web-app. Setup Instructions.
-
Copy your code to the Google App Engine virtual machine as per instructions above or on GitHub's help pages.
-
Click on the SSH button on the Google App Engine dashboard to access your virtual machine and enter the "test" virtualenv.
-
Run the python server with the defaults, which will listen for requests on port 8111. Run with
--helpif you need helpcd project1/webserver python server.py --debug -
Go to
http://<IP ADDRESS>:8111/in your browser to check that it worked.
You will need this URL when presenting the project to your mentor.
Note: This is just a suggestion. Since it is impossible to support setting up Python on everyone's personal computers, we can't really help debug issues that aren't happening on an Google App Engine VM. Your best bet is google, office hours, or asking your fellow students on the discussion board.
It is much more convenient to be able to test your application on your laptop or local computer, and run it on the Google App Engine VM when you are happy with the code. You can do this by following the virtualenv setup commands from HW0 on your own computer. Once you have the correct virtualenv set up, you can run the the web server with:
To run the webserver, go into the webserver/ directory and run (make sure you have enabled the virtualenv environment)
python server.py --debug
It should print something like:
running on 0.0.0.0:8111
* Running on http://0.0.0.0:8111/
The 0.0.0.0 listens to any IPv4 address on the machine. The 8111 after the : is the port number.
So if this is running on your laptop, you can open you web browser to http://localhost:8111.
You can specify a custom port by passing a host and port as arguments:
python server.py --debug 0.0.0.0 8888
To see its command line options, use the --help flag
python server.py --help
If you run the server with the --debug flag, it will automatically pick up changes when you reload the page, which is more convenient than restarting the server each time. It additionally will display detailed errors in the web browser, instead of only on the console.
The following are optional instructions on how to keep servers running. You'll need it after your project is complete, so staff can access your application to run additional tests if needed.
There are several ways to keep the server running after you have logged out of the VM. Note that these are all poor man's techniques.
-
nohup. the HUP signal is how the terminal warns a process of user logout. the nohup command ensures that the process ignores this signal, allowing it to continue running. the "&" character at the end of the command tells the terminal to detach this process from the terminal.
nohup python server.py 0.0.0.0 8008 &You can kill the process explicitly by getting the process ID and using the
killcommand:ps -A | grep python kill <the ID of the python process> -
tmux is a remote terminal manager. You can think of the terminal as two parts --
the client that you interact with by typing characters and pressing ENTER, and a server that actually reads those commands and runs processes in response.
Usually when you login to a VM, the client and server are tied together in a single process, so that when you logout the client and server both die. TMUX on the other hand explicitly starts two processes -- the server process that continues to run after you log out, and a client process that connects to the server process. This way, even if you disconnect, only the client dies. When you re-connect, you can re-attach to the server process and resume your terminal session! This is what I do.# install tmux sudo apt-get install tmux # run tmux tmux # it will open a terminal python server.py 0.0.0.0 8008 # don't press ctrl-c, just close your window.
Tmux is quite powerful -- come ask me directly or post to the discussion board if you are curious about its other functionalities. GNU Screen is an alternative to tmux.