Skip to content

Latest commit

 

History

History
386 lines (254 loc) · 16 KB

README.md

File metadata and controls

386 lines (254 loc) · 16 KB

wikiteam3

We archive wikis, from Wikipedia to the tiniest wikis

wikiteam3 is an ongoing project to port the legacy wikiteam toolset to Python 3 and PyPI to make it more accessible for today's archivers.

Most of the focus has been on the core dumpgenerator tool, but Python 3 versions of the other wikiteam tools may be added over time.

wikiteam3 Toolset

wikiteam3 is a set of tools for archiving wikis. The tools work on MediaWiki wikis, but the team hopes to expand to other wiki engines. As of 2020, WikiTeam has preserved more than 250,000 wikis, several wikifarms, regular Wikipedia dumps and 34 TB of Wikimedia Commons images.

The main general-purpose module of wikiteam3 is dumpgenerator, which can download XML dumps of MediaWiki sites that can then be parsed or redeployed elsewhere.

Python Environment

wikiteam3 requires Python 3.8 or later (less than 4.0), but you may be able to get it run with earlier versions of Python 3. On recent versions of Linux and macOS Python 3.8 should come preinstalled, but on Windows you will need to install it from python.org.

wikiteam3 has been tested on Linux, macOS, Windows and Android. If you are connecting to Linux or macOS via ssh, you can continue using the bash or zsh command prompt in the same terminal, but if you are starting in a desktop environment and don't already have a preferred Terminal environment you can try one of the following.

NOTE: You may need to update and pre-install dependencies in order for wikiteam3 to work properly. Shell commands for these dependencies appear below each item in the list. (Also note that while installing and running wikiteam3 itself should not require administrative priviliges, installing dependencies usually will.)

  • On desktop Linux you can use the default terminal application such as Konsole or GNOME Terminal.

    Linux Dependencies

    While most Linux distributions will have Python 3 preinstalled, if you are cloning wikiteam3 rather than downloading it directly you may need to install git.

    On Debian, Ubuntu, and the like:

    sudo apt update && sudo apt upgrade && sudo install git

    (On Fedora, Arch, etc., use dnf, pacman, etc., instead.)

  • On macOS you can use the built-in application Terminal, which is found in Applications/Utilities.

    macOS Dependencies

    While macOS will have Python 3 preinstalled, if you are cloning wikiteam3 rather than downloading it directly and you are using an older versions of macOS, you may need to install git.

    If git is not preinstalled, however, macOS will prompt you to install it the first time you run the command. Therefore, to check whether you have git installed or to install git, simply run git (with no arguments) in Terminal:

    git

    If git is already installed, it will print its usage instructions. If git is not preinstalled, the command will pop up a window asking if you want to install Apple's command line developer tools, and clicking "Install" in the popup window will install git.

  • On Windows 10 or Windows 11 you can use Windows Terminal.

    Windows Dependencies

    A simple installation of Python suitable for running scripts and packages is available from the Microsoft Store. Python will then be available from any Command Prompt or PowerShell session. Adding C:\Program Files\Git\usr\bin to the PATH environment variable will add some some useful Linux commands and utilities to Command Prompt.

    If you are already using the Windows Subsystem for Linux, you can follow the Linux instructions above. If you don't want to install a full WSL distribution, Git for Windows provides Bash emulation, so you can use it as a more lightweight option instead. Git Bash also provides some useful Linux commands and utilities.

    When installing Python 3.8 (from python.org), be sure to check "Add Python to PATH" so that installed Python scripts are accessible from any location. If for some reason installed Python scripts, e.g. pip, are not available from any location, you can add Python to the PATH environment variable using the instructions here.

    And while doing so should not be necessary if you follow the instructions further down and install wikiteam3 using pip, if you'd prefer that Windows store installed Python scripts somewhere other than the default Python folder under %appdata%, you can also add your preferred alternative path such as C:\Program Files\Python3\Scripts\ or a subfolder of My Documents. (You will need to restart any terminal sessions in order for this to take effect.)

    Whenever you'd like to run a Bash session, you can open a Bash terminal prompt from any folder in Windows Explorer by right-clicking and choosing the option from the context menu. (For some purposes you may wish to run Bash as an administrator.) This way you can open a Bash prompt and clone the wikiteam3 repository in one location, and subsequently or later open another Bash prompt and run wikiteam3 to dump a wiki wherever else you'd like without having to browse to the directory manually using Bash.

  • On Android you can use Termux.

    Termux Dependencies 
    pkg update && pkg upgrade && pkg install git libxslt python

  • On iOS you can use iSH.

    iSH Dependencies 
    apk update && apk upgrade && apk add git py3-pip

    Note: iSH may automatically quit if your iOS device goes to sleep, and it may lose its status if you switch to another app. You can disable auto-sleep while iSH is running by clicking the gear icon and toggling "Disable Screen Dimming". (You may wish to connect your device to a charger while running iSH.)

Using dumpgenerator

The Python 3 port of the dumpgenerator module of wikiteam3 is largely functional and can be installed from a downloaded or cloned copy of this repository.

There are two versions of these instructions:

  • If you just want to use a version that mostly works
  • If you want to follow my progress and help me test my latest commit

If you run into a problem with the version that mostly works, you can open an Issue. Be sure to include the following:

  1. The operating system you're using
  2. What command you ran that didn't work
  3. What output was printed to your terminal

If you just want to use a version that mostly works

1. Downloading and installing wikiteam3

In whatever folder you use for cloned repositories:

git clone https://github.com/elsiehupp/wikiteam3.git
cd wikiteam3
git checkout --track origin/python3
pip install --force-reinstall dist/*.whl

2. Running dumpgenerator for whatever purpose you need

dumpgenerator [args]

3. Uninstalling the package and deleting the cloned repository when you're done

pip uninstall wikiteam3
rm -r [cloned_wikiteam3_folder]

If you'd like to manually build and install wikiteam3 from a cloned or downloaded copy of this repository, run the following commands from the downloaded base directory:

curl -sSL https://install.python-poetry.org | python3 -
poetry install
poetry build
pip install --force-reinstall dist/*.whl

In either case, to uninstall wikiteam3 run this command (from any local directory):

pip uninstall wikiteam3

If you want to follow my progress and help me test my latest commit

Note: this branch may not actually work at any given time!

1. Install Python Poetry

curl -sSL https://install.python-poetry.org | python3 -

Note: if you get an SSL error, you may need to follow the instructions here.

2. Cloning the repository and switching to the prepare-for-publication branch

git clone [email protected]:elsiehupp/wikiteam3.git

or

git clone https://github.com/elsiehupp/wikiteam3.git

then:

cd wikiteam3
git checkout --track origin/prepare-for-publication

3. Downloading and installing wikiteam3

Note: Re-run the following steps each time to reinstall each time the wikiteam3 branch is updated.

git pull
poetry update && poetry install && poetry build
pip install --force-reinstall dist/*.whl

4. Then, from anywhere, you should be able to run

dumpgenerator [args]

To run the test suite, run:

test-dumpgenerator

5. Uninstalling the package and deleting the cloned repository when you're done

pip uninstall wikiteam3
rm -r [cloned_wikiteam3_folder]

Switching between branches

git checkout python3

or

git checkout prepare-for-publication

Using dumpgenerator (once installed)

After installing wikiteam3 using pip you should be able to use the dumpgenerator command from any local directory.

For basic usage, you can run dumpgenerator in the directory where you'd like the download to be.

For a brief summary of the dumpgenerator command-line options:

dumpgenerator --help

Several examples follow.

Note: the \ and line breaks in the examples below are for legibility in this documentation. dumpgenerator can also be run with the arguments in a single line and separated by a single space each.

Downloading a wiki with complete XML history and images

dumpgenerator \
    http://wiki.domain.org \
    --xml \
    --images

Manually specifying api.php and/or index.php

If the script can't find itself the api.php and/or index.php paths, then you can provide them:

dumpgenerator \
    --api http://wiki.domain.org/w/api.php \
    --xml \
    --images
dumpgenerator \
    --api http://wiki.domain.org/w/api.php \
    --index http://wiki.domain.org/w/index.php \
    --xml \
    --images

If you only want the XML histories, just use --xml. For only the images, just --images. For only the current version of every page, --xml --curonly.

Resuming an incomplete dump

dumpgenerator \
    --api http://wiki.domain.org/w/api.php \
    --xml \
    --images \
    --resume \
    --path=/path/to/incomplete-dump

In the above example, --path is only necessary if the download path is not the default.

dumpgenerator will also ask you if you want to resume if it finds an incomplete dump in the path where it is downloading.

Using launcher

launcher is a way to download a large list of wikis with a single invocation.

Usage:

launcher path-to-apis.txt [--7z-path path-to-7z] [--generator-arg=--arg] ...

launcher will download a complete dump (XML and images) for a list of wikis, then compress the dump into two 7z files: history (containing only metadata and the XML history of the wiki) and wikidump (containing metadata, XML, and images). This is the format that is suitable for upload to a WikiTeam item on the Internet Archive.

launcher will resume incomplete dumps as appropriate and will not attempt to download wikis that have already been downloaded (as determined by the files existing in the working directory).

Each wiki will be stored into files contiaining a stripped version of the url and the date the dump was started.

path-to-apis.txt is a path to a file that contains a list of URLs to api.phps of wikis, one on each line.

By default, a 7z executable is found on PATH. The --7z-path argument can be used to use a specific executable instead.

The --generator-arg argument can be used to pass through arguments to the generator instances that are spawned. For example, one can use --generator-arg=--xmlrevisions to use the modern MediaWiki API for retrieving revisions or --generator-arg=--delay=2 to use a delay of 2 seconds between requests.

Using uploader

uploader is a way to upload a large set of already-generated wiki dumps to the Internet Archive with a single invocation.

Usage:

uploader [-pd] [-pw] [-a] [-c COLLECTION] [-wd WIKIDUMP_DIR] [-u] [-kf KEYSFILE] [-lf LOGFILE] listfile

For the positional parameter listfile, uploader expects a path to a file that contains a list of URLs to api.phps of wikis, one on each line (exactly the same as launcher).

uploader will search a configurable directory for files with the names generated by launcher and upload any that it finds to an Internet Archive item. The item will be created if it does not already exist.

Named arguments:

  • -pd / --prune_directories: After uploading, remove the raw directory generated by launcher
  • -pw / --prune_wikidump: After uploading, remove the wikidump.7z file generated by launcher
  • -c / --collection: Assign the Internet Archive items to the specified collection
  • -a / --admin: Used only if you are an admin of the WikiTeam collection on the Internet Archive
  • -wd / --wikidump_dir: The directory to search for dumps. Defaults to ..
  • -u / --update: Update the metadata on an existing Internet Archive item
  • -kf / --keysfile: Path to a file containing Internet Archive API keys. Should contain two lines: the access key, then the secret key. Defaults to ./keys.txt.
  • -lf / --logfile: Where to store a log of uploaded files (to reduce duplicate work). Defaults to uploader-X.txt, where X is the final part of the listfile path.

Checking dump integrity

If you want to check the XML dump integrity, type this into your command line to count title, page and revision XML tags:

grep -E '<title(.*?)>' *.xml -c;grep -E '<page(.*?)>' *.xml -c;grep "</page>" *.xml -c;grep -E '<revision(.*?)>' *.xml -c;grep "</revision>" *.xml -c

You should see something similar to this (not the actual numbers) - the first three numbers should be the same and the last two should be the same as each other:

  • 580
  • 580
  • 580
  • 5677
  • 5677

If your first three numbers or your last two numbers are different, then, your XML dump is corrupt (it contains one or more unfinished </page> or </revision>). This is not common in small wikis, but large or very large wikis may fail at this due to truncated XML pages while exporting and merging. The solution is to remove the XML dump and re-download, a bit boring, and it can fail again...

WikiTeam Team

WikiTeam is the Archive Team [GitHub] subcommittee on wikis.

It was founded and originally developed by Emilio J. Rodríguez-Posada, a Wikipedia veteran editor and amateur archivist. Thanks to people who have helped, especially to: Federico Leva, Alex Buie, Scott Boyd, Hydriz, Platonides, Ian McEwen, Mike Dupont, balr0g and PiRSquared17.

The Python 3 initiative is currently being led by Elsie Hupp, with contributions from Victor Gambier, Thomas Karcher, and Janet Cobb.