Gutenberg

Overview

This package contains a variety of scripts to make working with the Project Gutenberg body of public domain texts easier.

The functionality provided by this package includes:

• Downloading texts from Project Gutenberg.
• Cleaning the texts: removing all the crud, leaving just the text behind.
• Making meta-data about the texts easily accessible.

The package has been tested with Python 2.7 and 3.5+.

An HTTP interface to this package exists too. Try it out!

Installation

This project is on PyPI, so I'd recommend that you just install everything from there using your favourite Python package manager.

pip install gutenberg

If you want to install from source or modify the package, you'll need to clone this repository:

git clone https://github.com/c-w/Gutenberg.git

Now, you should probably install the dependencies for the package and verify your checkout by running the tests.

cd Gutenberg

virtualenv --no-site-packages virtualenv
source virtualenv/bin/activate
pip install -r requirements-dev.pip
pip install .

nose2

Alternatively, you can also run the project via Docker:

docker build -t gutenberg -f Dockerfile-py3 .

docker run -it -v /some/mount/path:/data gutenberg python

Python 3

This package depends on BSD-DB. The bsddb module was removed from the Python standard library since version 2.7. This means that if you wish to use gutenberg on Python 3, you will need to manually install BSD-DB.

If getting BSD-DB to run on your platform is difficult, take a look at gutenbergpy which only depends on SQLite or MongoDB.

Linux

On Linux, you can usually install BSD-DB using your distribution's package manager. For example, on Ubuntu, you can use apt-get:

sudo apt-get install libdb++-dev
export BERKELEYDB_DIR=/usr
pip install .

MacOS

On Mac, you can install BSD-DB using homebrew:

brew install berkeley-db4
pip install .

Windows

On Windows, it's easiest to download a pre-compiled version of BSD-DB from pythonlibs which works great for Python 3.5+.

For example, if you have Python 3.5 on a 64-bit version of Windows, you should download bsddb3‑6.2.1‑cp35‑cp35m‑win_amd64.whl.

After you download the wheel, install it and you're good to go:

pip install bsddb3‑6.2.1‑cp35‑cp35m‑win_amd64.whl
pip install .

License conflicts

Since its v6.x releases, BSD-DB switched to the AGPL3 license which is stricter than this project's Apache v2 license. This means that unless you're happy to comply to the terms of the AGPL3 license, you'll have to install an ealier version of BSD-DB (anything between 4.8.30 and 5.x should be fine). If you are happy to use this project under AGPL3 (or if you have a commercial license for BSD-DB), set the following environment variable before attempting to install BSD-DB:

YES_I_HAVE_THE_RIGHT_TO_USE_THIS_BERKELEY_DB_VERSION=1

Apache Jena Fuseki

As an alternative to the BSD-DB backend, this package can also use Apache Jena Fuseki for the metadata store. The Apache Jena Fuseki backend is activated by setting the GUTENBERG_FUSEKI_URL environment variable to the HTTP endpoint at which Fuseki is listening. If the Fuseki server has HTTP basic authentication enabled, the username and password can be provided via the GUTENBERG_FUSEKI_USER and GUTENBERG_FUSEKI_PASSWORD environment variables.

For local development, the Fuseki server can be run via Docker:

docker run \
    --detach \
    --publish 3030:3030 \
    --env ADMIN_PASSWORD=some-password \
    --volume /some/mount/location:/fuseki \
    stain/jena-fuseki:3.6.0 \
    /jena-fuseki/fuseki-server --loc=/fuseki --update /ds

export GUTENBERG_FUSEKI_URL=http://localhost:3030/ds
export GUTENBERG_FUSEKI_USER=admin
export GUTENBERG_FUSEKI_PASSWORD=some-password

Usage

Downloading a text

from gutenberg.acquire import load_etext
from gutenberg.cleanup import strip_headers

text = strip_headers(load_etext(2701)).strip()
print(text)  # prints 'MOBY DICK; OR THE WHALE\n\nBy Herman Melville ...'

python -m gutenberg.acquire.text 2701 moby-raw.txt
python -m gutenberg.cleanup.strip_headers moby-raw.txt moby-clean.txt

Looking up meta-data

A bunch of meta-data about ebooks can be queried:

from gutenberg.query import get_etexts
from gutenberg.query import get_metadata

print(get_metadata('title', 2701))  # prints frozenset([u'Moby Dick; Or, The Whale'])
print(get_metadata('author', 2701)) # prints frozenset([u'Melville, Hermann'])

print(get_etexts('title', 'Moby Dick; Or, The Whale'))  # prints frozenset([2701, ...])
print(get_etexts('author', 'Melville, Hermann'))        # prints frozenset([2701, ...])

You can get a full list of the meta-data that can be queried by calling:

from gutenberg.query import list_supported_metadatas

print(list_supported_metadatas()) # prints (u'author', u'formaturi', u'language', ...)

Before you use one of the gutenberg.query functions you must populate the local metadata cache. This one-off process will take quite a while to complete (18 hours on my machine) but once it is done, any subsequent calls to get_etexts or get_metadata will be very fast. If you fail to populate the cache, the calls will raise an exception.

To populate the cache:

from gutenberg.acquire import get_metadata_cache
cache = get_metadata_cache()
cache.populate()

If you need more fine-grained control over the cache (e.g. where it's stored or which backend is used), you can use the set_metadata_cache function to switch out the backend of the cache before you populate it. For example, to use the Sqlite cache backend instead of the default Sleepycat backend and store the cache at a custom location, you'd do the following:

from gutenberg.acquire import set_metadata_cache
from gutenberg.acquire.metadata import SqliteMetadataCache

cache = SqliteMetadataCache('/my/custom/location/cache.sqlite')
cache.populate()
set_metadata_cache(cache)

Limitations

This project deliberately does not include any natural language processing functionality. Consuming and processing the text is the responsibility of the client; this library merely focuses on offering a simple and easy to use interface to the works in the Project Gutenberg corpus. Any linguistic processing can easily be done client-side e.g. using the TextBlob library.