How it is made ?

This is based on Cython. The main functions are defined into cpython.pxd file and Quvi class is developed into quvi.pyx file. Compilation is made with Cython using the simple command:

python setup.py build_ext

You can also compile .so file in you current directory:

python setup.py build_ext --inplace

To install library on system, you may use as root or by sudo:

python setup.py install

Remember you can use virtualenv to try quvi without impact on your system:

mkdir /tmp/py-virt
virtualenv /tmp/py-virt
source /tmp/py-virt/bin/activate
setup python.py install

This will install quvi into a virtual environment. To deactivate virtual environment, in the same terminal session:

deactivate

Usage

After having compiled quvi, you can use it as a python module. Module defines a class named Quvi. This is a simple usage:

import quvi

#instance:
q = quvi.Quvi()
q.parse("http://a.youtube.url")
print q.get_properties()

You will see a dict definition that have severals information given by libquvi.

It’s possible to use multiprocessing module or threads module to handle several instances. See this example:

import quvi
from multiprocessing import Process

#this function will be called in thread
def getInfo(url):
    q = quvi.Quvi()
    q.parse(url)
    print q.get_properties()

#urls to parse
url = "http://www.youtube.com/watch?v=..."
url2 = "http://www.youtube.com/watch?v=..."

#processes list
processes = []
processes.append( Process(target=getInfo, args=(url,)) )
processes.append( Process(target=getInfo, args=(url2,)) )

#start and join threads
[p.start() for p in processes]
[p.join() for p in processes]

print "done"

Both url will be handle in a thread. So this will be about twice quicker than parsing each url one by one.

Another usecase would be to get the properties of the best format available:

from quvi import Quvi

def get_properties_best_quality(url):
    q = Quvi()

    url = "http://www.youtube.com/watch?v=0gzA6Xzbh1k"
    if q.is_supported(url):
        formats = q.get_formats(url)
        best_format = formats[-1]
        q.set_format(best_format)
        q.parse(url)
        properties = q.get_properties()
        return properties
    return none

And downloading the video:

def get_video(filename, url):
    properties = get_properties_best_quality(url)
    if properties is not None:
        to_dl = properties["mediaurl"]
        filename += properties["filesuffix"]
        urlretrieve(to_dl, filename)

Why this python library ?

Because Quvi command line is really nice and I wanted to get youtube, dailymotion, vimeo (etc…) movies information into my python project. Calling “quvi” command line may be used, but having a real library implementation is the best way to have good performances.

Using Cython is pretty cool

What’s next ?

For now, you can only get media information from a page you ask to parse. I will continue to develop this library to improved and use properties provided by C library (version, nextmediaurl…)

If you want to develop with me, fork the project on GitHub, then process some merge request :)