Wikidata:Creating a bot

From Wikidata
Jump to navigation Jump to search
Translate this page
Other languages:

This page explains how to create bots for Wikidata. Please, consider sharing your code, add new examples and any improvements you want.

Requirements[edit]

To create bots you need:

  • Some coding skills (Python, Perl, PHP...)
  • A framework (one of the frameworks below) and some code to run to complete a task
  • A bot account (and approved)
  • A source code editor (Notepad++, Geany, vim, emacs)

Recommendation[edit]

Pywikibot[edit]

In the next sections, you learn how to install, configure and login using pywikibot. You only need to do these first three steps once. Also, there are some basic examples to learn the basics about bot programming.

Installation[edit]

For further details about pywikibot installation, see Manual:Pywikibot/Installation and Wikidata:Pywikibot - Python 3 Tutorial/Setting up Shop
To use pywikibot without installation, see Manual:Pywikibot/PAWS

To install pywikibot:

Configuration[edit]

For further details about pywikibot configuration, see Manual:Pywikibot/user-config.py.

You must configure user-config.py file with the bot username, family project and language. For Wikidata both family and language parameters are the same, wikidata.

Extended content 
mylang = "wikidata"
family = "wikidata"
usernames["wikidata"]["wikidata"] = 'MyBotName'

You can reduce the delay between edits adding: put_throttle = 1

Login[edit]

After you configure the user-config.py file, login as follows: 

$ python login.py

It will ask you for your bot password, insert it and press enter. You should be logged in now.

Example 1: Get data[edit]

This example gets data for the page referring to Douglas Adams. Save the following source code in a file and execute it with python example1.py

Extended content 
import pywikibot
site = pywikibot.Site("en", "wikipedia")
page = pywikibot.Page(site, 'Douglas Adams')
item = pywikibot.ItemPage.fromPage(page)
dictionary = item.get()
print(dictionary)
print(dictionary.keys())
print(item)

item.get() connects to Wikidata and fetches the data. The output is (reformatted for clarity): 

{
    'claims': {
        'P646': [<pywikibot.page.Claim instance at 0x7f1880188b48>],
        'P800': [<pywikibot.page.Claim instance at 0x7f1880188488>, <pywikibot.page.Claim instance at 0x7f1880188368>]
        ...
    }
    'labels': {
        'gu': '\u0aa1\u0a97\u0acd\u0ab2\u0abe\u0ab8 \u0a8f\u0aa1\u0aae\u0acd\u0ab8',
        'scn': 'Douglas Adams',
        ...
    }
    'sitelinks': {
        'fiwiki': 'Douglas Adams',
        'fawiki': '\u062f\u0627\u06af\u0644\u0627\u0633 \u0622\u062f\u0627\u0645\u0632',
        'elwikiquote': '\u039d\u03c4\u03ac\u03b3\u03ba\u03bb\u03b1\u03c2 \u0386\u03bd\u03c4\u03b1\u03bc\u03c2',
        ...
    }
    'descriptions': {
        'eo': 'angla a\u016dtoro de sciencfikcio-romanoj kaj humoristo',
        'en': 'English writer and humorist',
    },
    'aliases': {
        'ru': ['\u0410\u0434\u0430\u043c\u0441, \u0414\u0443\u0433\u043b\u0430\u0441'],
        'fr': ['Douglas Noel Adams', 'Douglas No\xebl Adams'],
        ...
    }
}
['claims', 'labels', 'sitelinks', 'descriptions', 'aliases']
[[wikidata:Q42]]

It prints a dictionary with keys for

  • the set of claims in the page: Property:P646 is the Freebase identifier, Property:P800 is "notable work", etc.
  • the label of the item in many languages
  • the sitelinks for the item, not just Wikipedias in many languages, but also Wikiquote in many languages
  • the item description in many languages
  • the aliases for the item in many languages

Then a list with all the keys for the key-values pairs in the dictionary. Finally, you can see that the Wikidata item about Douglas Adams is Q42.

Alternatives[edit]

The example above gets the ItemPage using the en wikipedia article. Alternatively, we can also get the ItemPage directly:

Extended content 
import pywikibot
site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')

Example 2: Get interwiki links[edit]

After item.get(), for example the sitelinks can be accessed. These are links to all Wikipedias that have the article.

Extended content 
import pywikibot
site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')
item.get()
print(",".join(item.sitelinks))

The output is:

{'fiwiki': 'Douglas Adams', 'eowiki': 'Douglas Adams', 'dewiki': 'Douglas Adams', ...}

With item.iterlinks(), an iterator over all these sitelinks is returned, where each article is given not as plain text as above but already as a Page object for further treatment (e.g., edit the text in the corresponding Wikipedia articles).

Example 4: Set a description[edit]

This example sets an English and a German description for the item about Douglas Adams.

Extended content 
import pywikibot

site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')
item.get()
mydescriptions = {'en': 'English writer and humorist', 'de': 'Keine Panik!'}
item.editDescriptions(mydescriptions, summary='Setting/updating descriptions.')

Setting labels and aliases works accordingly.

Example 6: Set a sitelink[edit]

To set a sitelink, we can either create a corresponding dict corresponding to Example 4 or use Page objects:

Extended content 
import pywikibot

site = pywikibot.Site("en", "wikipedia")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')
page = pywikibot.Page(site, 'Douglas Adams')
item.setSitelink(page, summary='Setting (/updating?) sitelink.')

Example 7: Set a statement[edit]

Statements are set using the Claim class. In the following, we set for Douglas Adams place of birth (P19): Cambridge (Q350).

Extended content 
import pywikibot

site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')
claim = pywikibot.Claim(repo, 'P19')
target = pywikibot.ItemPage(repo, 'Q350')
claim.setTarget(target)
item.addClaim(claim, summary='Adding claim')

For other datatypes, this works similar. In the following, we add claims with string (IMDb ID (P345)) and coordinate (coordinate location (P625)) datatypes (URL is the same as string):

Extended content 
import pywikibot

site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q42')
stringclaim = pywikibot.Claim(repo, 'P345')
stringclaim.setTarget('nm0010930')
item.addClaim(stringclaim, summary='Adding string claim')
coordinateclaim = pywikibot.Claim(repo, 'P625')
coordinate = pywikibot.Coordinate(lat=52.208, lon=0.1225, precision=0.001, site=site)
coordinateclaim.setTarget(coordinate)
item.addClaim(coordinateclaim, summary='Adding coordinate claim')

Example 8: Add a qualifier[edit]

Qualifiers are also represented by the Claim class. In the following, we add the qualifier incertae sedis (P678): family (Q35409) to the Claim "claim". Make sure you add the item before adding the qualifier.

Extended content 
qualifier = pywikibot.Claim(repo, 'P678')
target = pywikibot.ItemPage(repo, "Q35409")
qualifier.setTarget(target)
claim.addQualifier(qualifier, summary='Adding a qualifier.')

Example 9: Add a source[edit]

Also, sources are represented by the Claim class. Unlike for qualifiers, a source may contain more than one Claim. In the following, we add stated in (P248): Integrated Taxonomic Information System (Q82575) with retrieved (P813) March 20, 2014 as source to the Claim "claim". The claim has to be either retrieved from Wikidata or added to an itempage beforehand.

Extended content 
statedin = pywikibot.Claim(repo, 'P248')
itis = pywikibot.ItemPage(repo, "Q82575")
statedin.setTarget(itis)

retrieved = pywikibot.Claim(repo, 'P813')
date = pywikibot.WbTime(year=2014, month=3, day=20)
retrieved.setTarget(date)

claim.addSources([statedin, retrieved], summary='Adding sources.')

Example 10: Page generators[edit]

TODO

Example 11: Get values of sub-properties[edit]

In the following, we get values of sub-properties from branch described by source (P1343) -> Great Soviet Encyclopedia (1969–1978) (Q17378135) -> properties reference URL (P854) and title (P1476).

Extended content 
import pywikibot

site = pywikibot.Site("wikidata", "wikidata")
repo = site.data_repository()
item = pywikibot.ItemPage(repo, 'Q13515')
item.get()

sourcesid = 'P1343'
sourceid = 'Q17378135'
urlid = 'P854'
nameid = 'P1476'

# item.claims['P1343'][1].qualifiers.items():   # This are direct way to get list qualifiers. But '[1]' is hard link to index of list, it will break over time.
if sourcesid in item.claims:
    for source in item.claims[sourcesid]:
        if source.target.id == sourceid:
            s = source.qualifiers
            if urlid in s:     url  = s.get(urlid)[0].target
            if nameid in s:    name = s.get(nameid)[0].target['text']
            print (url, name)

More examples[edit]

Some users share their source codes. Learn more in the next links:

Wikidata Integrator[edit]

WikidataIntegrator is a library for reading and writing to Wikidata/Wikibase. We created it for populating Wikidata with content from authoritative resources on Genes, Proteins, Diseases, Drugs and others. Details on the different tasks can be found on the bot's Wikidata page.

Pywikibot is an existing framework for interacting with the MediaWiki API. The reason why we came up with our own solution is that we need a high integration with the Wikidata SPARQL endpoint in order to ensure data consistency (duplicate checks, consistency checks, correct item selection, etc.). Compared to Pywikibot, WikidataIntegrator currently is not a full Python wrapper for the MediaWiki API but is solely focused on providing an easy means to generate Python-based Wikidata bots.

For more information, documentation, download & installation instructions, see here: https://github.com/SuLab/WikidataIntegrator/

Example Notebook[edit]

An example notebook demonstrating an example bot to add therapeutic areas to drug items, including using fastrun mode, checking references, and removing old statements:

http://public-paws.wmcloud.org/46883698/example%20ema%20bot.ipynb

WikibaseIntegrator[edit]

Forked from Wikidata Integrator by User:Myst in 2020 and has seen several improvements to the API that makes it even easier to create bots using the library.

For more information, documentation, download & installation instructions, see here: https://github.com/LeMyst/WikibaseIntegrator

Example semi-automatic script[edit]

LexUse semi-automatic tool for finding and adding usage examples to lexemes. It's free software written using Python 3 in 2020 Wikidata:LexUse.

Wikibase.NET (Deprecated)[edit]

Wikibase.NET is the api that replaces the now deprecated DotNetDataBot. Api client for the MediaWiki extension Wikibase. They aren't compatible because Wikibase.NET does no longer need the DotNetWikiBot framework.

Download & Installation[edit]

You can download Wikibase.NET from GitHub. Just follow the instructions on that page.

Known issues[edit]

Examples[edit]

Coming not soon...

DotNetDataBot (Deprecated)[edit]

Installation[edit]

Configuration[edit]

After unpacking the package you can see a file called DotNetDataBot.dll and one called DotNetDataBot.xml. The xml document is only for documentation. To use it you have to create a new refer in your project. Then you can write using DotNetDataBot; to import the framework.

Login[edit]

To login you have to create a new Site object with the url of the wiki, your bot's username and its password.

Extended content

C# 

using DotNetDataBot;
public static void Main()
{
    Site wikidata = new Site("http://www.wikidata.org", "User", "Password");
}

Example 1: Get id using wiki page[edit]

You can access the id of an item by searching for using the site and the title of the connected page.

Extended content

C# 

using DotNetDataBot;
public static void Main()
{
    Site site = new Site("http://www.wikidata.org", "User", "Password");
    Item item = new Item(site);
    if (item.itemExists("it", "George Lucas")) // Check if exist on Wikidata
    {
        Console.Write("Q" + item.GetIdBySitelink("it", "George Lucas"));
    }
    else
    {
        Console.Write("Doesn't exist");
    }
}

Example 2: Get interwiki links[edit]

You can get the interwiki links of an item by loading the content and accessing the links field of the object.

Extended content

C# 

using DotNetDataBot;
public static void Main()
{
    Site site = new Site("http://www.wikidata.org", "User", "Password");

    Item item = new Item(site);
    item.id = item.GetIdBySitelink("it", "George Lucas");
    item.Load();
    foreach(KeyValuePair<string, string> link in item.links) 
    {
        Console.Write(link.Key); // lang (eg. en or it)
        Console.Write(link.Value); // page (eg. George_Lucas)
    }
}

Example 3: Set a description[edit]

To set a description, you must call the setDescription function.

Extended content

C# 

using DotNetDataBot;
public static void Main()
{
    Site site = new Site("http://www.wikidata.org", "User", "Password");
    Item item = new Item(site, "Q4115189");
    item.Load();
    if (item.descriptions.ContainsKey("it")) // if alredy exist description in italian
    {
        // Nothing to do
    }
    else
    {
        item.setDescription("it", "description in italian", "Bot: Add italian description");
    }
}

Example 4: Set a label[edit]

It works the same way for setting a label. Just call setLabel.

Extended content

C# 

using DotNetDataBot;
public static void Main()
{
    Site site = new Site("http://www.wikidata.org", "User", "Password");
    Item item = new Item(site, "Q4115189");
    item.Load();
    if (item.labels.ContainsKey("it")) // if alredy exist label in italian
    {
        // Nothing to do
    }
    else
    {
        item.setLabel("it", "label in italian", "Bot: Add italian label");
    }
}

Example 5: Get interwiki links for 100 pages[edit]

This feature is not supported. Just iterate over the list.

Wikibase api for PHP[edit]

This is an api client for Wikibase written in PHP. It can be downloaded from here.

Example 1: Basic example[edit]

Take a look at the source comments to understand how it works.

Extended content 
<?php
/**
 * Basic example for the use of the libary with some small edits
 */
require_once( __DIR__ . "/vendor/autoload.php" );

// Creates some useful objects and logs into the api
$api = new \Mediawiki\Api\MediawikiApi( "http://www.wikidata.org/w/api.php" );
$api->login( new \Mediawiki\Api\ApiUser( 'username', 'password' ) );

$dataValueClasses = array(
    'unknown' => 'DataValues\UnknownValue',
    'string' => 'DataValues\StringValue',
);
$wikidata = new \Wikibase\Api\WikibaseFactory(
    $api,
    new DataValues\Deserializers\DataValueDeserializer( $dataValueClasses ),
    new DataValues\Serializers\DataValueSerializer()
);

// Gets the current revision for item Q777
$revision = $wikidata->newRevisionGetter()->getFromId( 'Q777' );
$item = $revision->getContent()->getData();

// Outputs the current sitelink for enwiki
var_dump( $item->getSiteLink( 'enwiki' ) );

// Sets the de description to 'Foobar'
$item->getFingerprint()->setDescription( 'de', 'Foobar' );

// Saves the item
$wikidata->newRevisionSaver()->save( $revision );

//Log out
$api->logout();


Example 2: Creating claims[edit]

Take a look at the source comments to understand how it works.

Extended content 
<?php
/**
 * Basic example for the use of the library with some small edits.
 */
require_once( __DIR__ . "/vendor/autoload.php" );

// Creates some useful objects and logs into the api
$api = new \Mediawiki\Api\MediawikiApi( "https://www.wikidata.org/w/api.php" );
$api->login( new \Mediawiki\Api\ApiUser( 'username', 'password' ) );

$dataValueClasses = array(
    'unknown' => 'DataValues\UnknownValue',
    'string' => 'DataValues\StringValue',
);
$services = new \Wikibase\Api\WikibaseFactory(
    $api,
    new DataValues\Deserializers\DataValueDeserializer( $dataValueClasses ),
    new DataValues\Serializers\DataValueSerializer()
);

$revision = $services->newRevisionGetter()->getFromId( 'Q777' );
$item = $revision->getContent()->getData();
    
$statementList = $item->getStatements();
if( $statementList->getByPropertyId( \Wikibase\DataModel\Entity\PropertyId::newFromNumber( 1320 ) )->isEmpty() ) {
    $services->newStatementCreator()->create(
       new \Wikibase\DataModel\Snak\PropertyValueSnak(
           \Wikibase\DataModel\Entity\PropertyId::newFromNumber( 1320 ),
           new \DataValues\StringValue( 'New String Value' )
       ),
      'Q777'
   );
}

// Log out
$api->logout();

VBot (no updates since 2017)[edit]

Framework for Wikidata and Wikipedia. Read and write on Wikidata and other Wikimedia project and have a useful list generator to generate list of Wikipedia page and Wikidata entity. Can read also JSON dump of Wikidata.

Overview[edit]

Bot to read and edit Wikidata and Wikipedia.

  • License: CC0 1.0
  • Language C#
  • Can read and write entities with all datatype on Wikidata
  • Can read and write pages on all Wiki project
  • Can read parameter from template on wiki pages
  • Can read JSON dump
  • Can create lists using:
  • Tested with Visual Studio Express 2013 for Windows Desktop.
    • Is necessary to have Newtonsoft.Json. You can install it with NuGet inside Visual Studio
    • Is necessary to add manually a reference to System.Web for "HttpUtility.UrlEncode"

Download[edit]

The framework can be downloaded from GitHub here.

Instruction[edit]

Example 1[edit]

Update en label for all items with instance of (P31): short film (Q24862) that have director (P57) and that have publication date (P577) in 1908. (Use of Wikidata query)

Extended content 
private void CompleteExample()
{
    //Wikidata query
    string strWDQ = "CLAIM[31:24862] AND CLAIM[57] AND BETWEEN[577,+00000001908-00-00T00:00:00Z,+00000001908-12-31T00:00:00Z]";
    ListGenerator lg = new ListGenerator();
    List<string> chunks = lg.WDQ(strWDQ, 50);

    //Connection to Wikipedia
    WikimediaAPI WP = new WikimediaAPI("https://it.wikipedia.org", User, Password);
    Pages PageList = new Pages();
    //Connection to Wikidata
    WikimediaAPI WD = new WikimediaAPI("https://www.wikidata.org", User, Password);
    Entities EntityList = new Entities();

    Dictionary<string, string> Labels = new Dictionary<string, string>();

    foreach (string list in chunks)
    {
        // Load all entity of the chunk
        string strJson = WD.LoadWD(list);
        EntityList = new Entities();
        EntityList = JsonConvert.DeserializeObject<Entities>(strJson, new DatavalueConverter());

        foreach (KeyValuePair<string, Entity> entity in EntityList.entities)
        {
            if (entity.Value.sitelinks.ContainsKey("itwiki"))
            {
                // Load Wikipage
                string Pages = WP.LoadWP(entity.Value.sitelinks["itwiki"].title);
                PageList = JsonConvert.DeserializeObject<Pages>(Pages, new DatavalueConverter());
             
                //Director from template
                string director = Utility.GetTemplateParameter(PageList.query.FirstPageText, "film","Regista").Replace("[","").Replace("]", "");
                Labels = new Dictionary<string, string>();
                if (director=="")
                {
                    Labels.Add("en", "1908 short movie");
                }
                else
                {
                    Labels.Add("en", "1908 short movie directed by " + director);
                }
                // Update Wikidata
                WD.EditEntity(entity.Value.id, null, Labels, null, null, null, "BOT: Update en label");
            }
        }
    }
}

LexData (Python; for Lexicographical data)[edit]

LexData is an easy to use python libary to create and edit Lexemes, Senses and Forms.

Tips[edit]

The documentation of LexData is still a bit lacking so look at existing implementations in MachtSinn or Wikdata Lexeme Forms for ideas how to use it.

If you only want to add statements to Lexemes (not forms or senses) WikibaseIntegrator might be a better choice, as it is more versatile and support a lot of data types.

Installation[edit]

You can install LexData via pip: 

$ pip install LexData

Login[edit]

For all operations you need a WikidataSession. You can create it with your credentials, a bot password or an Edit Token (for example to edit via OAUTH):

Extended content 
repo = LexData.WikidataSession("YourUsername", "YourPassword")


Retrieve a Lexeme[edit]

You can open existing Lexemes and read their content.

Extended content 
L2 = LexData.Lexeme(repo, "L2")

print(L2.claims)
print(L2.forms)
print(L2.senses)
sense1 = L2.senses[0]
print(sense1.claims)

Searching and creating Lexemes[edit]

If you don't know the L-Id of a lexeme you can search for it. And if it doesn't exist you can create it.

Extended content 
# Find an existing Lexeme by lemma, language and grammatical form
L2 = LexData.search_lexemes(repo, "first", en, "Q1084")

# Create a new Lexeme
L2 = LexData.create_lexeme(repo, "first", en, "Q1084")

# Find or create a Lexeme
L2 = LexData.get_or_create_lexeme(repo, "first", en, "Q1084")

Adding information[edit]

You can easily create forms or senses, with or without additional claims:

Extended content 
if len(L2.forms) == 0:
    L2.createForm("firsts", ["Q146786"])

if len(L2.senses) == 0:
    L2.createSense(
        {
            "en": "Element in an ordered list which comes before all others according to the ordering",
            "de": "einer Ordnung folgend das Element vor allen anderen",
        },
        claims={"P5137": ["Q19269277"]},
    )

Using Wikidata's API directly[edit]

The other sections describe how to use bot frameworks to access and update Wikidata information. You can also directly interact with the Wikibase API that Wikidata provides. You need to do this if you're developing your own framework or if you need to do something that a framework doesn't support. The documentation for the Wikibase API can be found at mediawiki.org. You can also play around with it at Special:ApiSandbox, try action=wbgetentities.

Wikibase provides its API as a set of modules for MediaWiki's "action" API. You access this by making HTTP requests to /w/api.php. The default response format is JSON. So for your language of choice, you only need a library to perform HTTP requests and a JSON or XML library to parse the responses.

Example 1: Get Q number[edit]

This example gets the item Q number for the English Wikipedia article about Andromeda Galaxy. The Wikibase API's main "workhorse" module action=wbgetentities provides this information. The HTTP request (using jsonfm format for human-readable JSON output) is simply

https://www.wikidata.org/w/api.php?action=wbgetentities&titles=Andromeda%20Galaxy&sites=enwiki&props=&format=jsonfm&formatversion=2

Try following the link. This requests no additional information about the entity; remove &props= from the URL to see much more information about it. See the generated help for wbgetentities for more parameters you can specify.

Python[edit]

Extended content 
#!/usr/bin/python3
from requests import get

def get_qnumber(wikiarticle, wikisite):
    resp = get('https://www.wikidata.org/w/api.php', {
        'action': 'wbgetentities',
        'titles': wikiarticle,
        'sites': wikisite,
        'props': '',
        'format': 'json'
    }).json()
    return list(resp['entities'])[0]

print(get_qnumber(wikiarticle="Andromeda Galaxy", wikisite="enwiki"))

The output is:

Q2469

Example 2: Get list of items without particular interwiki[edit]

...please contribute if you know how...

See also[edit]

External links[edit]

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Creating_a_bot&oldid=1891907570"