Ricordisamoa has uploaded a new change for review.
https://gerrit.wikimedia.org/r/126213
Change subject: create a new WikidataBot class for cacheSources() and
getSource() for claimit.py, coordinate_import.py and harvest_template.py
......................................................................
create a new WikidataBot class for cacheSources() and getSource()
for claimit.py, coordinate_import.py and harvest_template.py
bug: 62438
this also allows for non-Wikipedia sources
in claimit.py and coordinate_import.py
as with change I29df011b82fd8687d4c38def62d0c0ea3a1b7994
bug: 62991
Change-Id: Ic05987911f0878d42a06cf5af842134bd9d615e8
---
M pywikibot/bot.py
M scripts/claimit.py
M scripts/coordinate_import.py
M scripts/harvest_template.py
4 files changed, 45 insertions(+), 85 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/pywikibot/core
refs/changes/13/126213/1
diff --git a/pywikibot/bot.py b/pywikibot/bot.py
index 3aa8189..7beda63 100644
--- a/pywikibot/bot.py
+++ b/pywikibot/bot.py
@@ -21,6 +21,7 @@
import os.path
import sys
import re
+import json
_logger = "bot"
@@ -815,3 +816,32 @@
if choice != 'n':
page.put(newtext, async=(choice == 'a'))
+
+
+class WikidataBot:
+ """
+ Generic Wikidata Bot to be subclassed
+ used in claimit.py, coordinate_import.py and harvest_template.py
+ """
+
+ def cacheSources(self):
+ """
+ Fetches the sources from the onwiki list
+ and stores it internally
+ """
+ page = pywikibot.Page(self.repo, u'List of wikis/python', ns=4)
+ self.source_values = json.loads(page.get())
+ for family_code, family in self.source_values.iteritems():
+ for source_lang in family:
+ self.source_values[family_code][source_lang] =
pywikibot.ItemPage(self.repo,
+
family[source_lang])
+
+ def getSource(self, site):
+ """
+ Get the source for the specified site,
+ if possible
+ """
+ if site.family.name in self.source_values and site.code in
self.source_values[site.family.name]:
+ source = pywikibot.Claim(self.repo, 'P143')
+
source.setTarget(self.source_values.get(site.family.name).get(site.code))
+ return source
diff --git a/scripts/claimit.py b/scripts/claimit.py
index db8ff17..64beeb2 100755
--- a/scripts/claimit.py
+++ b/scripts/claimit.py
@@ -50,19 +50,18 @@
"""
#
# (C) Legoktm, 2013
-# (C) Pywikibot team, 2013
+# (C) Pywikibot team, 2013-2014
#
# Distributed under the terms of the MIT license.
#
__version__ = '$Id$'
#
-import json
import pywikibot
-from pywikibot import pagegenerators
+from pywikibot import pagegenerators, WikidataBot
-class ClaimRobot:
+class ClaimRobot(WikidataBot):
"""
A bot to add Wikidata claims
"""
@@ -79,28 +78,6 @@
self.exists_arg = exists_arg
self.repo = pywikibot.Site().data_repository()
self.cacheSources()
-
- def getSource(self, lang):
- """
- Get the source for the specified language,
- if possible
- """
- if lang in self.source_values:
- source = pywikibot.Claim(self.repo, 'p143')
- source.setTarget(self.source_values.get(lang))
- return source
-
- def cacheSources(self):
- """
- Fetches the sources from the onwiki list
- and stores it internally
- """
- page = pywikibot.Page(self.repo, u'Wikidata:List of wikis/python')
- self.source_values = json.loads(page.get())
- self.source_values = self.source_values['wikipedia']
- for source_lang in self.source_values:
- self.source_values[source_lang] = pywikibot.ItemPage(self.repo,
-
self.source_values[source_lang])
def run(self):
"""
@@ -147,7 +124,7 @@
% (claim.getID(), claim.getTarget()))
item.addClaim(claim)
# A generator might yield pages from multiple languages
- source = self.getSource(page.site.language())
+ source = self.getSource(page.site)
if source:
claim.addSource(source, bot=True)
# TODO FIXME: We need to check that we aren't adding a
diff --git a/scripts/coordinate_import.py b/scripts/coordinate_import.py
index f89f55f..d732c22 100644
--- a/scripts/coordinate_import.py
+++ b/scripts/coordinate_import.py
@@ -18,18 +18,17 @@
"""
#
# (C) Multichill 2014
-# (C) Pywikibot team, 2013
+# (C) Pywikibot team, 2013-2014
#
# Distributed under the terms of MIT License.
#
__version__ = '$Id$'
#
-import json
import pywikibot
-from pywikibot import pagegenerators
+from pywikibot import pagegenerators, WikidataBot
-class coordImportRobot:
+class CoordImportRobot(WikidataBot):
"""
A bot to import coordinates to Wikidata
"""
@@ -40,31 +39,8 @@
"""
self.generator = pagegenerators.PreloadingGenerator(generator)
- self.site = pywikibot.Site()
self.repo = pywikibot.Site().data_repository()
self.cacheSources()
-
- def getSource(self, lang):
- """
- Get the source for the specified language,
- if possible
- """
- if lang in self.source_values:
- source = pywikibot.Claim(self.repo, 'p143')
- source.setTarget(self.source_values.get(lang))
- return source
-
- def cacheSources(self):
- """
- Fetches the sources from the onwiki list
- and stores it internally
- """
- page = pywikibot.Page(self.repo, u'Wikidata:List of wikis/python')
- self.source_values = json.loads(page.get())
- self.source_values = self.source_values['wikipedia']
- for source_lang in self.source_values:
- self.source_values[source_lang] = pywikibot.ItemPage(self.repo,
-
self.source_values[source_lang])
def run(self):
"""
@@ -88,7 +64,7 @@
pywikibot.output(u'Adding %s, %s to %s' %
(coordinate.lat, coordinate.lon, item.title()))
item.addClaim(newclaim)
- source = self.getSource(page.site.language())
+ source = self.getSource(page.site)
if source:
newclaim.addSource(source, bot=True)
@@ -102,7 +78,7 @@
generator = gen.getCombinedGenerator()
- coordbot = coordImportRobot(generator)
+ coordbot = CoordImportRobot(generator)
coordbot.run()
if __name__ == "__main__":
diff --git a/scripts/harvest_template.py b/scripts/harvest_template.py
index 79adb7c..ec85415 100755
--- a/scripts/harvest_template.py
+++ b/scripts/harvest_template.py
@@ -16,7 +16,7 @@
"""
#
# (C) Multichill, Amir, 2013
-# (C) Pywikibot team, 2013
+# (C) Pywikibot team, 2013-2014
#
# Distributed under the terms of MIT License.
#
@@ -24,14 +24,13 @@
#
import re
-import json
import pywikibot
-from pywikibot import pagegenerators as pg
+from pywikibot import pagegenerators as pg, WikidataBot
docuReplacements = {'¶ms;': pywikibot.pagegenerators.parameterHelp}
-class HarvestRobot:
+class HarvestRobot(WikidataBot):
"""
A bot to add Wikidata claims
"""
@@ -50,28 +49,6 @@
self.repo = pywikibot.Site().data_repository()
self.cacheSources()
- def getSource(self, site):
- """
- Get the source for the specified site,
- if possible
- """
- if site.family.name in self.source_values and site.code in
self.source_values[site.family.name]:
- source = pywikibot.Claim(self.repo, 'P143')
-
source.setTarget(self.source_values.get(site.family.name).get(site.code))
- return source
-
- def cacheSources(self):
- """
- Fetches the sources from the onwiki list
- and stores it internally
- """
- page = pywikibot.Page(self.repo, u'List of wikis/python', ns=4)
- self.source_values = json.loads(page.get())
- for family_code, family in self.source_values.iteritems():
- for source_lang in family:
- self.source_values[family_code][source_lang] =
pywikibot.ItemPage(self.repo,
-
family[source_lang])
-
def run(self):
"""
Starts the robot.
@@ -79,7 +56,7 @@
self.templateTitles = self.getTemplateSynonyms(self.templateTitle)
for page in self.generator:
try:
- self.procesPage(page)
+ self.processPage(page)
except Exception as e:
pywikibot.exception(tb=True)
@@ -97,9 +74,9 @@
titles.append(temp.title(withNamespace=False))
return titles
- def procesPage(self, page):
+ def processPage(self, page):
"""
- Proces a single page
+ Process a single page
"""
item = pywikibot.ItemPage.fromPage(page)
pywikibot.output('Processing %s' % page)
--
To view, visit https://gerrit.wikimedia.org/r/126213
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: Ic05987911f0878d42a06cf5af842134bd9d615e8
Gerrit-PatchSet: 1
Gerrit-Project: pywikibot/core
Gerrit-Branch: master
Gerrit-Owner: Ricordisamoa <ricordisa...@live.it>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
