Allanon 0.2

A Web crawler that visit a predictable set of URLs, and automatically download resources you want from them

Contents

• Introduction

• Documentation

  • Installation

  • Invocation

  • Basic usage (you probably don’t need Allanon at all for this)

  • Main usage (things became interesting now)

  • Extreme usage

  • Naming and storing downloaded resources

• TODO

• Changelog

  • 0.2 (2014-01-02)

  • 0.1 (2013-01-05)

Introduction

Let’s say that you want to access a slow streaming site to see something (obviously: something not protected by copyright).

The streaming site use URLs in that format:

http://legal-streaming-site.org/program-name/season5/episode4/

Every page contains some HTML code like the following:

....
    <div id="video-container">
       ...
       <embed src="http://someotherurl.org/qwerty.flv" ...
       ...
    <div>
...

Let say this is the URL for the episode 4 of the fifth season of your program. You know that this program has 6 seasons with 22 episode each.

As said before: this site is very slow so you prefer downloading episodes in background then watch them later.

To download them you need to watch the HTML inside the page and get some resources (commonly: and FLV file). The best would be download all episode in a single (long running) operation instead of manually doing it.

Allanon will help you exactly in such tasks. You simply need to provide it:

• a simple URL or a dynamic URL pattern

• a query selector for resources inside the page

Quick example (you can keep it single lined):

$ allanon --search="#movie-container embed" \
> "http://legal-streaming-site.org/program-name/season{1:6}/episode{1:22}"

Documentation

Installation

You can use distribute or pip to install the utility in your Python environment.

$ easy_install Allanon

or alternately:

$ pip install Allanon

Invocation

After installing you will be able to run the allanon script from command line. For example: run the following for access the utility help:

$ allanon --help

Basic usage (you probably don’t need Allanon at all for this)

The allanon script accept an URL (or a list of URLs) to be downloaded:

$ allanon http://myhost/folder/image1.jpg http://myhost/folder/image2.jpg ...

Every command line URL given to Allanon can be a simple URL or an URL model like the following:

$ allanon "http://myhost/folder/image{1:50}.jpg"

This will crawl 50 different URLs automatically.

Main usage (things became interesting now)

The allanon script take an additional --search parameter (see the first example given above). When you provide it, you are meaning:

“I don’t want to download those URLs directly, but those URLs contain links to file that I really want”.

The search parameter format must be CSS 3 compatible, like the one supported the famous jQuery library, and it’s based onto the pyquery library. See it’s documentation for more details about what you can look for.

Extreme usage

The --search parameter can be provided multiple times:

$ allanon --search="ul.image-repos a" \
> --search="div.image-containers img" \
> "http://image-repository-sites.org/category{1:30}.html"

When you provide (for example) two different search parameters, you are meaning:

“I don’t want to download resources at given URLs. Those URLs contain links to secondary pages, and inside those pages there’re links to resources I want to download”

Filters are applied in the given order, so:

• Allanon will search inside 30 pages named category1.html, category2.html, …

• inside those pages, Allanon will look for all links inside ul tags with CSS class image-repos and recursively crawl them.

• inside those pages, Allanon will looks for images inside div with class image-containers.

• images will be downloaded.

Potentially you can continue this way, providing a third level of filters, and so on.

Naming and storing downloaded resources

By default Allanon download all files in the current directory so a filename conflict is possible. You can control how/where download, changing dynamically the filename using the --filename option and/or change the directory where to store files with the --directory option.

An example:

$ allanon --filename="%HOST-%INDEX-section%1-version%3-%FULLNAME" \
> "http://foo.org/pdf-repo-{1:10}/file{1:50}.pdf?version={0:3}"

As you seen --filename accept some markers that can be used to better organize resources:

%HOST

Will be replaced with the hostname used in the URL.

%INDEX

Is a progressive from 1 to the number of downloaded resources.

%X

When using dynamic URLs models you can refer to the current number of an URL section.

In this case “%1” is the current “pdf-repo-x” number and “%3” is the “version” parameter value.

%FULLNAME

The original filename (the one used if --filename is not provided).

You can also use the %NAME and %EXTENSION to get only the name of the file (without extension) or simply the extension.

The --directory option can be a simple directory name or a directory path (in unix-like format, for example “foo/bar/baz”).

An example:

$ allanon --directory="/home/keul/%HOST/%1" \
> "http://foo.org/pdf-repo-{1:10}/file{1:50}.pdf" \
> "http://baz.net/pdf-repo-{1:10}/file{1:50}.pdf"

Also the --directory option supports some of the markers: you can use %HOST, %INDEX and %X with the same meaning given above.

TODO

This utility is in alpha stage, a lot of thing can goes wrong when downloading and many features are missing:

• verbosity controls

• bandwidth control

• multi-thread (let’s look at grequests)

• Python 3

If you find other bugs or want to ask for missing features, use the product’s issue tracker.

Changelog

0.2 (2014-01-02)

• Do not crawl or download when on error pages

• Handle duplicate filename when downloading resources: added the --check-duplicate option

• Application specific user agent header (configurable through --user-agent option)

• The --directory option can be a path and so create intermediate directories, and accept markers

• More efficient memory usage

• Show progress bar when getting resources (now requires progress)

• Fixed problem when getting quoted filename from response header

• Added the --timeout option

• Added the --sleep option

0.1 (2013-01-05)

• first release