mwlib - MediaWiki Parser and Utility Library

Overview

mwlib is a versatile library designed for parsing MediaWiki articles and converting them to various output formats. A notable application of mwlib is in Wikipedia's "Print/export" feature, where it is used to create PDF documents from Wikipedia articles.

Getting Started

Prerequisites

To build mwlib, ensure you have the following software installed:

• Python (version 3.11 or 3.12)
• Ploticus
• re2c
• Perl
• Pillow / PyImage
• ImageMagick
• uv (Python package installer, faster alternative to pip)

Setup a virtual environment for Python 3.11 or 3.12 and activate it.

Installing uv

If you don't have uv installed, you can install it following the instructions at uv's official documentation.

For example, on Unix-like systems:

curl -LsSf https://astral.sh/uv/install.sh | sh

Or using pip:

pip install uv

Installing mwlib

To install all dependencies and the project, run:

$ make install

This will use uv to install all required dependencies.

To build the C extensions and install mwlib in development mode:

$ make build
$ make develop

Documentation

Please visit http://mwlib.readthedocs.org/en/latest/index.html for detailed documentation.

Configuration

OAuth2 Configuration

mwlib supports OAuth2 client_credentials flow for Wikipedia API access. This allows authenticated access to MediaWiki APIs that require OAuth2 authentication, while maintaining compatibility with wikis that don't use OAuth2.

OAuth2 Configuration Options

The following configuration options are available for OAuth2:

• oauth2.client_id: OAuth2 client ID
• oauth2.client_secret: OAuth2 client secret
• oauth2.token_url: URL for obtaining OAuth2 tokens (default: https://meta.wikimedia.org/w/rest.php/oauth2/access_token)
• oauth2.enabled: Whether to use OAuth2 (default: False)

HTTP/2 Configuration Options

mwlib also supports HTTP/2 for improved performance:

• http2.enabled: Whether to use HTTP/2 (default: True)
• http2.auto_detect: Whether to auto-detect HTTP/2 support (default: True)

These configuration options can be set either through environment variables or in a configuration file (mwlib.ini or ~/.mwlibrc). The following table shows the mapping between configuration file options and their corresponding environment variables:

Config File Option	Environment Variable	Description
oauth2.client_id	MWLIB_OAUTH2_CLIENT_ID	OAuth2 client ID
oauth2.client_secret	MWLIB_OAUTH2_CLIENT_SECRET	OAuth2 client secret
oauth2.token_url	MWLIB_OAUTH2_TOKEN_URL	Token endpoint URL
oauth2.enabled	MWLIB_OAUTH2_ENABLED	Enable OAuth2 (true/false)
http2.enabled	MWLIB_HTTP2_ENABLED	Enable HTTP/2 (true/false)
http2.auto_detect	MWLIB_HTTP2_AUTO_DETECT	Auto-detect HTTP/2 support

Example configuration file (mwlib.ini):

[oauth2]
enabled=true
client_id=your-client-id
client_secret=your-client-secret

[http2]
auto_detect=true

Example Usage

You can also set the config parameters directly when instantiating MwApi:

from mwlib.network.sapi import MwApi

# Using OAuth2
api = MwApi(
    apiurl="https://en.wikipedia.org/w/api.php",
    use_oauth2=True
)

The recommended best practice, however, is to configure environment variables:

export MWLIB_OAUTH2_CLIENT_ID="your_client_id"
export MWLIB_OAUTH2_CLIENT_SECRET="your_client_secret"
export MWLIB_OAUTH2_ENABLED="True"

BigQuery Lookup for Image Description Pages

mwlib can use Google BigQuery to look up image description pages (namespace 6) instead of fetching them from the remote MediaWiki API. This significantly reduces the number of API requests and bypasses Wikipedia rate limits for configured domains.

The data originates from the Wikimedia Enterprise API snapshot endpoint, which provides Wikipedia page data as NDJSON files. These snapshots are loaded into a BigQuery table containing pre-extracted metadata: templates (used for license checking), image dimensions, content URLs, and license information.

How it works

1. During ZIP creation (mw-zip), image description page titles for configured domains (default: en.wikipedia.org) are batched and queried from BigQuery.
2. For pages found in BigQuery, templates are stored locally and an early license check is performed — images that fail the license filter are never downloaded.
3. Pages not found in BigQuery fall back to the remote MediaWiki API.
4. Non-configured domains (e.g., Commons) always use the remote API.

Prerequisites

The BigQuery client (and the wme-ingest CLI that loads snapshots into it) is shipped behind the bigquery extra so default installs stay slim. Install it via the extra rather than pinning google-cloud-bigquery directly:

# pip
pip install "mwlib[bigquery]"

# uv (project install)
uv pip install "mwlib[bigquery]"

# uv (developer checkout)
uv sync --extra bigquery

The bigquery extra is also pulled in by the dev dependency group, so uv sync (default groups) installs it for development and test runs.

Set up GCP authentication by pointing to a service account JSON file:

export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"

BigQuery Configuration Options

Config File Option	Environment Variable	Default	Description
bigquery.enabled	MWLIB_BIGQUERY_ENABLED	false	Master switch to enable BigQuery lookups
bigquery.project	MWLIB_BIGQUERY_PROJECT	(required)	GCP project ID
bigquery.dataset	MWLIB_BIGQUERY_DATASET	wme_snapshots	BigQuery dataset name
bigquery.table	MWLIB_BIGQUERY_TABLE	file_pages	BigQuery table name
bigquery.timeout	MWLIB_BIGQUERY_TIMEOUT	30	Query timeout in seconds
bigquery.domains	MWLIB_BIGQUERY_DOMAINS	en.wikipedia.org	Comma-separated domains to route through BigQuery

Example environment variable configuration:

export MWLIB_BIGQUERY_ENABLED="true"
export MWLIB_BIGQUERY_PROJECT="my-gcp-project"
export MWLIB_BIGQUERY_DATASET="wikipedia"
export MWLIB_BIGQUERY_TABLE="file_pages"
export MWLIB_BIGQUERY_DOMAINS="en.wikipedia.org"
export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"

Or in a configuration file (mwlib.ini):

[bigquery]
enabled=true
project=my-gcp-project
dataset=wikipedia
table=file_pages
domains=en.wikipedia.org

When BigQuery is disabled or unavailable (missing credentials, network error, etc.), mwlib falls back to the remote API automatically with no change in behavior.

Loading Data into BigQuery

Use the wme-ingest command to download a Wikimedia Enterprise namespace 6 snapshot and load it into BigQuery:

# Full pipeline: download snapshot + load into BigQuery
wme-ingest --project my-gcp-project --dataset wikipedia

# Load from an already-downloaded tarball
wme-ingest -i /path/to/enwiki_ns6.tar.gz --project my-gcp-project --dataset wikipedia

# List available snapshots
wme-ingest --list

This requires Wikimedia Enterprise API credentials:

export WME_USERNAME="your-username"
export WME_PASSWORD="your-password"

Docker Compose Setup

For users interested in setting up mwlib using Docker Compose, detailed instructions are available at Docker Compose documentation.

License

Copyright (c) 2007-2012 PediaPress GmbH

All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

• Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

• Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

• Neither the name of PediaPress GmbH nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

.. _SpamBayes: http://spambayes.sourceforge.net/