Python API.generator Examples

Example #1

File: list-generators.py Project: kynikos/wiki-scripts

#! /usr/bin/env python3

import os.path

from ws.client import API

api_url = "https://wiki.archlinux.org/api.php"
cookie_path = os.path.expanduser("~/.cache/ArchWiki.cookie")

api = API(api_url, cookie_file=cookie_path, ssl_verify=True)

# allpages
# see list of parameters: https://www.mediawiki.org/wiki/API:Allpages
for page in api.generator(generator="allpages", gaplimit="max"):
    print(page["title"])

# allcategories
# see list of parameters: https://www.mediawiki.org/wiki/API:Allcategories
#for page in api.generator(generator="allcategories", gaclimit="max"):
#    print(page["title"])

# categorymembers
# see list of parameters: https://www.mediawiki.org/wiki/API:Categorymembers
#for page in api.generator(generator="categorymembers", gcmtitle="Category:English", gcmlimit="max"):
#    print(page["title"])

# transclusions
# see list of parameters: https://www.mediawiki.org/wiki/API:Embeddedin
#for page in api.generator(generator="embeddedin", geititle="Template:AUR", geilimit="max"):
#    print(page["title"])

Example #2

File: list-allpages-by-language.py Project: lahwaacz/wiki-scripts

#! /usr/bin/env python3

import os.path
from collections import namedtuple
import itertools

from ws.client import API
import ws.ArchWiki.lang as lang

api_url = "https://wiki.archlinux.org/api.php"
cookie_path = os.path.expanduser("~/.cache/ArchWiki.cookie")

api = API(api_url, cookie_file=cookie_path, ssl_verify=True)

Page = namedtuple("Page", ["title", "langname", "pure"])

pages = []
for page in api.generator(generator="allpages", gaplimit="max", gapfilterredir="nonredirects"):
    pure, langname = lang.detect_language(page["title"])
    pages.append(Page(page["title"], langname, pure))

pages.sort(key=lambda page: (page.langname, page.pure))

groups = itertools.groupby(pages, key=lambda page: page.langname)
for langname, pages in groups:
    print("== {} ==\n".format(langname))
    for page in pages:
        print("* [[:{}|{}]]".format(page.title, page.pure))
    print()

Example #3

File: list-allpages-by-language.py Project: nl6720/wiki-scripts

import os.path
from collections import namedtuple
import itertools

from ws.client import API
import ws.ArchWiki.lang as lang

api_url = "https://wiki.archlinux.org/api.php"
index_url = "https://wiki.arclinux.org/index.php"
session = API.make_session()

api = API(api_url, index_url, session)

Page = namedtuple("Page", ["title", "langname", "pure"])

pages = []
for page in api.generator(generator="allpages",
                          gaplimit="max",
                          gapfilterredir="nonredirects"):
    pure, langname = lang.detect_language(page["title"])
    pages.append(Page(page["title"], langname, pure))

pages.sort(key=lambda page: (page.langname, page.pure))

groups = itertools.groupby(pages, key=lambda page: page.langname)
for langname, pages in groups:
    print("== {} ==\n".format(langname))
    for page in pages:
        print("* [[:{}|{}]]".format(page.title, page.pure))
    print()

Example #4

File: list-interwiki-links.py Project: nl6720/wiki-scripts

#! /usr/bin/env python3

import os.path

from ws.client import API

api_urls = [
    "https://wiki.archlinux.org/api.php",
    "https://wiki.archlinux.de/api.php",
]
index_url = "https://wiki.arclinux.org/index.php"
session = API.make_session()

for api_url in api_urls:
    api = API(api_url, index_url, session)
    prefixes = set()
    for ns in api.site.namespaces:
        if ns < 0:
            continue
        for page in api.generator(generator="allpages",
                                  gaplimit="max",
                                  gapfilterredir="nonredirects",
                                  gapnamespace=ns,
                                  prop="iwlinks",
                                  iwprop="url",
                                  iwlimit="max"):
            for link in page.get("iwlinks", []):
                prefixes.add(link["prefix"])
    print("interwiki prefixes used on {}:".format(api.get_hostname()))
    print(sorted(prefixes))

Example #5

File: list-interwiki-links.py Project: lahwaacz/wiki-scripts

#! /usr/bin/env python3

import os.path

from ws.client import API

api_urls = [
    "https://wiki.archlinux.org/api.php",
    "https://wiki.archlinux.de/api.php",
]
cookie_path = os.path.expanduser("~/.cache/ArchWiki.cookie")

for api_url in api_urls:
    api = API(api_url, cookie_file=cookie_path, ssl_verify=True)
    prefixes = set()
    for ns in api.site.namespaces:
        if ns < 0:
            continue
        for page in api.generator(generator="allpages", gaplimit="max", gapfilterredir="nonredirects", gapnamespace=ns, prop="iwlinks", iwprop="url", iwlimit="max"):
            for link in page.get("iwlinks", []):
                prefixes.add(link["prefix"])
    print("interwiki prefixes used on {}:".format(api.get_hostname()))
    print(sorted(prefixes))