This is a guide to converting bot scripts from version 1 of the Pywikibot framework to version 3.
Most importantly, note that the version 3 framework only supports wikis using MediaWiki v.1.14 or higher software. If you need to access a wiki that uses older software, you should continue using version 1 for this purpose.
The root namespace used in the project has changed from "wikipedia" to "pywikibot". References to wikipedia need to be changed globally to pywikibot. Unless noted in this document, other names have not changed; for example, wikipedia.Page can be replaced by pywikibot.Page throughout any bot. An effort has been made to design the interface to be as backwards- compatible as possible, so that in most cases it should be possible to convert scripts to the new interface simply by changing import statements and doing global search-and-replace on module names, as discussed in this document.
With pywikibot compat branch scripts were importing "wikipedia" or "pagegenerators" libraries; pywikibot is now written as a standard package, and other modules are contained within it (e.g., pywikibot.site contains Site classes). However, most commonly-used names are imported into the pywikibot namespace, so that module names don't need to be used unless specified in the documentation.
Make sure that the directory that contains the "pywikibot" subdirectory (or folder) is in sys.path.
The following changes, at a minimum, need to be made to allow scripts to run:
change "import wikipedia" to "import pywikibot" change "import pagegenerators" to "from pywikibot import pagegenerators" change "import config" to "from pywikibot import config" change "catlib.Category" to "pywikibot.Category" change "catlib.change_category(page, ...)" to "page.change_category(...)" change "query.GetData(request)" to "pywikibot.data.api.request(**request).submit()" change "userlib.User" to "pywikibot.User" change "wikipedia." to "pywikibot."
wikipedia.setAction() no longer works; you must revise the script to pass an explicit edit summary message on each put() call.
There is a helper script which does a lot of changes automatically. Just call it:
pwb.py update_script [<script to update>]
and follow the instructions and hints.
Deprecated methods edit
A lot of object methods have been deprecated; deprecated methods still work, but print a warning message in debug mode. You should follow the warning message and update your script accordingly because deprecated methods might be removed in future releases.
To activate the debugging level just call your script with -debug option like -debug:bot or -debug:*
Note: The -debug option does not warn when importing deprecated modules. A better ability is to set the debugging level in your user-config.py:
debug_log = ['*'] or debug_log = ['bot'] or log += ['deprecation']
Python libraries edit
[Note: the goal will be to package pywikibot with setuptools easy_install, so that these dependencies will be loaded automatically when the package is installed, and users won't need to worry about this...]
To run pywikibot, you will need the requests package:
It may be installed using pip or easy_install.
Page objects edit
The constructor syntax for Pages has been modified; existing calls in the format of Page(site, title) will still work, and this is still the preferred way of creating a Page object from data retrieved from the MediaWiki API (because the API will have parsed and normalized the title). However, for titles input by a user or scraped from wikitext, it is preferred to use the alternative syntax Page(Link(site, wikitext)), where "wikitext" is the string found between [[ and ]] delimiters. The new Link object (more on this below) handles link parsing and interpretation that doesn't require access to the wiki server.
A third syntax allows easy conversion from a Page object to a FilePage or Category, or vice versa: e.g., Category(pageobj) converts a Page to a Category, as long as the page is in the category namespace.
The following methods of the Page object have been obsoleted and no longer work (but these methods don't appear to be used anywhere in the code distributed with the bot framework). The functionality of the two obsolete methods is easily replaced by using standard search-and-replace techniques. If you call them, they will raise an NotImplementedError exception:
The following methods have had their outputs changed:
- getVersionHistory(): Returns a pywikibot.Timestamp object instead of a MediaWiki one
- templates(): Returns a list of Page objects of templates. In compat we have a list of template title strings.
- templatesWithParams(): Returns a list of tuples with two items. The first item is a Page object of the template, the second is a list of parameters. In compat we have a list of tuples with two items. The first item is the template title.
- linkedPages(): Returns a PageGenerator of Page objects of link targets. In compat we have a list of link target strings.
FilePage objects edit
The old ImagePage class has been renamed into FilePage. For FilePage objects, the getFileMd5Sum() method is deprecated; it is recommended to replace it with latest_file_info.sha1 property because MediaWiki now stores the SHA1 hash of images.
Category objects edit
The Category object has been moved from the catlib module to the pywikibot namespace. Any references to "catlib.Category" can be replaced by "pywikibot.Category", but the old form is retained for backwards-compatibility.
For Category objects, the following methods are deprecated:
- subcategoriesList: use, for example, list(self.subcategories()) instead
- articlesList: use, for example, list(self.articles()) instead
- supercategories: use self.categories() instead
- supercategoriesList: use, for example, list(self.categories()) instead
User objects edit
The User object has been moved from the userlib module to the pywikibot namespace. Any references to "userlib.User" can be replaced by "pywikibot.User", but the old form is retained for backwards-compatibility.
The following changes have occurred in the User object:
- contributions(): returns a pywikibot.Timestamp object instead of a Mediawiki one
apispec library and Blocks objects edit
Some apispec functionality could be replaced with other methods:
iso() -> Timestamp.isoformat() uniso() -> Timestamp.strftime('%Y-%m-%d %H:%M:%S') dt() -> Timestamp.totimestampformat() duration() -> BlockEntry.duration()
Blocks.empty() -> (* obsolete parameter cleanup *) Blocks.query() -> site.blocks() or site.logevents('block') Blocks.IPsortkey() -> (* sort key, not needed * ) Blocks.allblocks() -> site.blocks() or site.logevents('block') Blocks.user() -> site.blocks(user=user) Blocks.IP() -> site.blocks(iprange=IP)
Command-line arguments edit
Scripts that supported unnamed arguments as titles of pages on which to work, now require that those titles be written as standard pagegenerators, e.g.:
$ python pwb.py script -page:"A title"
while unlink.py and other scripts that required page titles as main arguments now need only that the titles be wrapped in quotes, as:
$ python pwb.py unlink "A title"
Further Help edit
Ask at T247105 for any Pywikibot compat to core migration support.