def getLinks(wtext):
#adapted from linkedPages() http://svn.wikimedia.org/svnroot/pywikipedia/trunk/pywikipedia/wikipedia.py
links = []
wtext = wikipedia.removeLanguageLinks(wtext, site)
wtext = wikipedia.removeCategoryLinks(wtext, site)
# remove HTML comments, pre, nowiki, and includeonly sections
# from text before processing
wtext = wikipedia.removeDisabledParts(wtext)
# resolve {{ns:-1}} or {{ns:Help}}
wtext = site.resolvemagicwords(wtext)
for match in Rlink.finditer(wtext):
title = match.group('title')
title = title.replace("_", " ").strip(" ")
if title.startswith("#"): # this is an internal section link
continue
if not site.isInterwikiLink(title):
if title.startswith("#"): # [[#intrasection]] same article
continue
title = title.split('#')[
0] # removing sections [[other article#section|blabla]]
title = '%s%s' % (title[:1].upper(), title[1:]) #first up
title = title.strip()
if title.startswith(":") or title.startswith(
"File:") or title.startswith("Image:") or title.startswith(
"Category:"): # files, cats, etc
continue
if title and title not in links:
links.append(title)
return links
def putAfterTemplate(page, template, toadd, loose=True):
'''
Try to put text after template.
If the template is not found return False if loose is set to False
If loose is set to True: Remove interwiki's, categories, add template, restore categories, restore interwiki's.
Based on cc-by-sa-3.0 code by Dschwen
'''
oldtext = page.get()
newtext = u''
templatePosition = oldtext.find(u'{{%s' % (template, ))
if templatePosition >= 0:
previousChar = u''
currentChar = u''
templatePosition += 2
curly = 1
square = 0
while templatePosition < len(oldtext):
currentChar = oldtext[templatePosition]
if currentChar == u'[' and previousChar == u'[':
square += 1
previousChar = u''
if currentChar == u']' and previousChar == u']':
square -= 1
previousChar = u''
if currentChar == u'{' and previousChar == u'{':
curly += 1
previousChar = u''
if currentChar == u'}' and previousChar == u'}':
curly -= 1
previousChar = u''
previousChar = currentChar
templatePosition += 1
if curly == 0 and square <= 0:
# Found end of template
break
newtext = oldtext[:templatePosition] + u'\n' + toadd + oldtext[
templatePosition:]
else:
if loose:
newtext = oldtext
cats = wikipedia.getCategoryLinks(newtext)
ll = wikipedia.getLanguageLinks(newtext)
nextext = wikipedia.removeLanguageLinks(newtext)
newtext = wikipedia.removeCategoryLinks(newtext)
newtext = newtext + u'\n' + toadd
newtext = wikipedia.replaceCategoryLinks(newtext, cats)
newtext = wikipedia.replaceLanguageLinks(newtext, ll)
return newtext
def putAfterTemplate (page, template, toadd, loose=True):
'''
Try to put text after template.
If the template is not found return False if loose is set to False
If loose is set to True: Remove interwiki's, categories, add template, restore categories, restore interwiki's.
Based on cc-by-sa-3.0 code by Dschwen
'''
oldtext = page.get()
newtext = u''
templatePosition = oldtext.find(u'{{%s' % (template,))
if templatePosition >= 0:
previousChar = u''
currentChar = u''
templatePosition += 2
curly = 1
square = 0
while templatePosition < len(oldtext):
currentChar = oldtext[templatePosition]
if currentChar == u'[' and previousChar == u'[' :
square += 1
previousChar = u''
if currentChar == u']' and previousChar == u']' :
square -= 1
previousChar = u''
if currentChar == u'{' and previousChar == u'{' :
curly += 1
previousChar = u''
if currentChar == u'}' and previousChar == u'}' :
curly -= 1
previousChar = u''
previousChar = currentChar
templatePosition +=1
if curly == 0 and square <= 0 :
# Found end of template
break
newtext = oldtext[:templatePosition] + u'\n' + toadd + oldtext[templatePosition:]
else:
if loose:
newtext = oldtext
cats = wikipedia.getCategoryLinks(newtext)
ll = wikipedia.getLanguageLinks(newtext)
nextext = wikipedia.removeLanguageLinks (newtext)
newtext = wikipedia.removeCategoryLinks(newtext)
newtext = newtext + u'\n' + toadd
newtext = wikipedia.replaceCategoryLinks(newtext, cats)
newtext = wikipedia.replaceLanguageLinks(newtext, ll)
return newtext
def query_percentnames(self):
'''
yields pages that contain internal links where special characters are
encoded as hexadecimal codes, e.g. %F6
'''
Rpercentlink = re.compile('\[\[[^\]]*?%[A-F0-9][A-F0-9][^\]]*?\]\]')
for entry in self.entries():
text = wikipedia.removeLanguageLinks(entry.text)
if Rpercentlink.search(text):
yield entry
def add_text(
page=None,
addText=None,
summary=None,
regexSkip=None,
regexSkipUrl=None,
always=False,
up=False,
putText=True,
oldTextGiven=None,
create=False,
):
if not addText:
raise NoEnoughData("You have to specify what text you want to add!")
if not summary:
summary = wikipedia.translate(wikipedia.getSite(), msg) % addText[:200]
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
starsList = [
u"bueno",
u"cyswllt[ _]erthygl[ _]ddethol",
u"dolen[ _]ed",
u"destacado",
u"destaca[tu]",
u"enllaç[ _]ad",
u"enllaz[ _]ad",
u"leam[ _]vdc",
u"legătură[ _]a[bcf]",
u"liamm[ _]pub",
u"lien[ _]adq",
u"lien[ _]ba",
u"liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt",
u"liên[ _]kết[ _]chọn[ _]lọc",
u"ligam[ _]adq",
u"ligoelstara",
u"ligoleginda",
u"link[ _][afgu]a",
u"link[ _]adq",
u"link[ _]f[lm]",
u"link[ _]km",
u"link[ _]sm",
u"linkfa",
u"na[ _]lotura",
u"nasc[ _]ar",
u"tengill[ _][úg]g",
u"ua",
u"yüm yg",
u"רא",
u"وصلة مقالة جيدة",
u"وصلة مقالة مختارة",
]
errorCount = 0
site = wikipedia.getSite()
# /wiki/ is not always the right path in non-wiki projects
pathWiki = site.family.nicepath(site.lang)
if putText:
wikipedia.output(u"Loading %s..." % page.title())
if oldTextGiven == None:
try:
text = page.get()
except wikipedia.NoPage:
if create:
wikipedia.output(u"%s doesn't exist, creating it!" % page.title())
text = u""
else:
wikipedia.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always) # continue
except wikipedia.IsRedirectPage:
wikipedia.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always) # continue
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl != None:
url = "%s%s" % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
wikipedia.output(u"Exception! regex (or word) used with -exceptUrl is in the page. Skip!")
return (False, False, always) # continue
if regexSkip != None:
result = re.findall(regexSkip, text)
if result != []:
wikipedia.output(u"Exception! regex (or word) used with -except is in the page. Skip!")
return (False, False, always) # continue
# If not up, text put below
if not up:
newtext = text
# Getting the categories
categoriesInside = wikipedia.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = wikipedia.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = wikipedia.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = wikipedia.removeLanguageLinks(newtext, site)
# nn got a message between the categories and the iw's and they want to keep it there, first remove it
if site.language() == u"nn":
newtext = newtext.replace(nn_iw_msg, "")
# Translating the \\n into binary \n
addText = addText.replace("\\n", "\n")
# Adding the text
newtext += u"\n%s" % addText
# Reputting the categories
newtext = wikipedia.replaceCategoryLinks(newtext, categoriesInside, site, True)
# Put the nn iw message back
if site.language() == u"nn":
newtext = newtext + u"\n" + nn_iw_msg
# Dealing the stars' issue
allstars = []
starstext = wikipedia.removeDisabledParts(text)
for star in starsList:
regex = re.compile("(\{\{(?:template:|)%s\|.*?\}\}[\s]*)" % star, re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub("", newtext)
allstars += found
if allstars != []:
newtext = newtext.strip() + "\r\n\r\n"
allstars.sort()
for element in allstars:
newtext += "%s\r\n" % element.strip()
# Adding the interwiki
newtext = wikipedia.replaceLanguageLinks(newtext, interwikiInside, site)
# If instead the text must be added above...
else:
newtext = addText + "\n" + text
if putText and text != newtext:
wikipedia.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<" % page.title())
wikipedia.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if not always:
choice = wikipedia.inputChoice(
u"Do you want to accept these changes?", ["Yes", "No", "All"], ["y", "N", "a"], "N"
)
if choice == "a":
always = True
elif choice == "n":
return (False, False, always)
if always or choice == "y":
try:
if always:
page.put(newtext, summary)
else:
page.put_async(newtext, summary)
except wikipedia.EditConflict:
wikipedia.output(u"Edit conflict! skip!")
return (False, False, always)
except wikipedia.ServerError:
errorCount += 1
if errorCount < 5:
wikipedia.output(u"Server Error! Wait..")
time.sleep(5)
continue
else:
raise wikipedia.ServerError(u"Fifth Server Error!")
except wikipedia.SpamfilterError, e:
wikipedia.output(u"Cannot change %s because of blacklist entry %s" % (page.title(), e.url))
return (False, False, always)
except wikipedia.PageNotSaved, error:
wikipedia.output(u"Error putting page: %s" % error.args)
return (False, False, always)
except wikipedia.LockedPage:
wikipedia.output(u"Skipping %s (locked page)" % page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
def standardizePageFooter(self, text):
"""
Makes sure that interwiki links, categories and star templates are
put to the correct position and into the right order. This combines the
old instances standardizeInterwiki and standardizeCategories
The page footer has the following section in that sequence:
1. categories
2. ## TODO: template beyond categories ##
3. additional information depending on local site policy
4. stars templates for featured and good articles
5. interwiki links
"""
starsList = [
u'bueno',
u'bom interwiki',
u'cyswllt[ _]erthygl[ _]ddethol', u'dolen[ _]ed',
u'destacado', u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligazón[ _]a[bd]',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a', u'link[ _]adq', u'link[ _]f[lm]', u'link[ _]km',
u'link[ _]sm', u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
categories = None
interwikiLinks = None
allstars = []
# The PyWikipediaBot is no longer allowed to touch categories on the
# German Wikipedia. See
# http://de.wikipedia.org/wiki/Hilfe_Diskussion:Personendaten/Archiv/1#Position_der_Personendaten_am_.22Artikelende.22
# ignoring nn-wiki of cause of the comment line above iw section
if not self.template and '{{Personendaten' not in text and \
'{{SORTIERUNG' not in text and '{{DEFAULTSORT' not in text and \
self.site.lang not in ('et', 'it', 'bg', 'ru'):
try:
categories = pywikibot.getCategoryLinks(text, site=self.site)
# there are categories like [[categoy:Foo {{#time:Y...}}]]
except pywikibot.InvalidTitle:
pass
if not self.talkpage: # and pywikibot.calledModuleName() <> 'interwiki':
subpage = False
if self.template:
loc = None
try:
tmpl, loc = moved_links[self.site.lang]
del tmpl
except KeyError:
pass
if loc is not None and loc in self.title:
subpage = True
interwikiLinks = pywikibot.getLanguageLinks(
text, insite=self.site, template_subpage=subpage)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site=self.site)
# Removing the stars' issue
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)'
% star, re.I)
found = regex.findall(starstext)
if found != []:
text = regex.sub('', text)
allstars += found
# Adding categories
if categories:
##Sorting categories in alphabetic order. beta test only on Persian Wikipedia, TODO fix bug for sorting
#if self.site.language() == 'fa':
# categories.sort()
##Taking main cats to top
# for name in categories:
# if re.search(u"(.+?)\|(.{,1}?)",name.title()) or name.title()==name.title().split(":")[0]+title:
# categories.remove(name)
# categories.insert(0, name)
text = pywikibot.replaceCategoryLinks(text, categories,
site=self.site)
# Adding stars templates
if allstars:
text = text.strip() + self.site.family.interwiki_text_separator
allstars.sort()
for element in allstars:
text += '%s\r\n' % element.strip()
if pywikibot.verbose:
pywikibot.output(u'%s' % element.strip())
# Adding the interwiki
if interwikiLinks:
text = pywikibot.replaceLanguageLinks(text, interwikiLinks,
site=self.site,
template=self.template,
template_subpage=subpage)
return text
def standardizePageFooter(self, text):
"""
Makes sure that interwiki links, categories and star templates are
put to the correct position and into the right order.
This combines the old instances standardizeInterwiki and standardizeCategories
The page footer has the following section in that sequence:
1. categories
2. additional information depending on local site policy
3. stars templates for featured and good articles
4. interwiki links
"""
starsList = [
u"bueno",
u"cyswllt[ _]erthygl[ _]ddethol",
u"dolen[ _]ed",
u"destacado",
u"destaca[tu]",
u"enllaç[ _]ad",
u"enllaz[ _]ad",
u"leam[ _]vdc",
u"legătură[ _]a[bcf]",
u"liamm[ _]pub",
u"lien[ _]adq",
u"lien[ _]ba",
u"liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt",
u"liên[ _]kết[ _]chọn[ _]lọc",
u"ligam[ _]adq",
u"ligoelstara",
u"ligoleginda",
u"link[ _][afgu]a",
u"link[ _]adq",
u"link[ _]f[lm]",
u"link[ _]km",
u"link[ _]sm",
u"linkfa",
u"na[ _]lotura",
u"nasc[ _]ar",
u"tengill[ _][úg]g",
u"ua",
u"yüm yg",
u"רא",
u"وصلة مقالة جيدة",
u"وصلة مقالة مختارة",
]
categories = None
interwikiLinks = None
allstars = []
hasCommentLine = False
# The PyWikipediaBot is no longer allowed to touch categories on the German Wikipedia.
# See http://de.wikipedia.org/wiki/Hilfe_Diskussion:Personendaten/Archiv/bis_2006#Position_der_Personendaten_am_.22Artikelende.22
# ignoring nn-wiki of cause of the comment line above iw section
if not self.template and not "{{Personendaten" in text:
categories = pywikibot.getCategoryLinks(text, site=self.site)
if not self.talkpage: # and pywikibot.calledModuleName() <> 'interwiki':
subpage = False
if self.template:
loc = None
try:
tmpl, loc = moved_links[self.site.lang]
del tmpl
except KeyError:
pass
if loc != None and loc in self.title:
subpage = True
interwikiLinks = pywikibot.getLanguageLinks(text, insite=self.site, template_subpage=subpage)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site=self.site)
# Removing the stars' issue
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile("(\{\{(?:template:|)%s\|.*?\}\}[\s]*)" % star, re.I)
found = regex.findall(starstext)
if found != []:
if pywikibot.verbose:
print found
text = regex.sub("", text)
allstars += found
# nn got a message between the categories and the iw's
# and they want to keep it there, first remove it
if self.site.language() == "nn":
regex = re.compile(
"(<!-- ?interwiki \(no(?:/nb)?, ?sv, ?da first; then other languages alphabetically by name\) ?-->)"
)
found = regex.findall(text)
if found:
if pywikibot.verbose:
print found
hasCommentLine = True
text = regex.sub("", text)
# Adding categories
if categories:
text = pywikibot.replaceCategoryLinks(text, categories, site=self.site)
# Put the nn iw message back
if self.site.language() == "nn" and not self.talkpage and (interwikiLinks or hasCommentLine):
text = text + "\r\n\r\n" + nn_iw_msg
# Adding stars templates
if allstars:
text = text.strip() + self.site.family.interwiki_text_separator
allstars.sort()
for element in allstars:
text += "%s\r\n" % element.strip()
if pywikibot.verbose:
pywikibot.output(u"%s" % element.strip())
# Adding the interwiki
if interwikiLinks:
text = pywikibot.replaceLanguageLinks(
text, interwikiLinks, site=self.site, template=self.template, template_subpage=subpage
)
return text
def add_text(page=None, addText=None, summary=None, regexSkip=None,
regexSkipUrl=None, always=False, up=False, putText=True,
oldTextGiven=None, reorderEnabled=True, create=False):
if not addText:
raise NoEnoughData('You have to specify what text you want to add!')
if not summary:
summary = i18n.twtranslate(pywikibot.getSite(), 'add_text-adding',
{'adding': addText[:200]})
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
errorCount = 0
site = pywikibot.getSite()
pathWiki = site.family.nicepath(site.lang)
if putText:
pywikibot.output(u'Loading %s...' % page.title())
if oldTextGiven is None:
try:
text = page.get()
except pywikibot.NoPage:
if create:
pywikibot.output(u"%s doesn't exist, creating it!"
% page.title())
text = u''
else:
pywikibot.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always)
except pywikibot.IsRedirectPage:
pywikibot.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always)
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl is not None:
url = '%s%s' % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -exceptUrl is in the page. Skip!
Match was: %s''' % result)
return (False, False, always)
if regexSkip is not None:
result = re.findall(regexSkip, text)
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -except is in the page. Skip!
Match was: %s''' % result)
return (False, False, always)
# If not up, text put below
if not up:
newtext = text
# Translating the \\n into binary \n
addText = addText.replace('\\n', config.line_separator)
if (reorderEnabled):
# Getting the categories
categoriesInside = pywikibot.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = pywikibot.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = pywikibot.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = pywikibot.removeLanguageLinks(newtext, site)
# Adding the text
newtext += u"%s%s" % (config.line_separator, addText)
# Reputting the categories
newtext = pywikibot.replaceCategoryLinks(newtext,
categoriesInside, site,
True)
# Dealing the stars' issue
allstars = []
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)'
% star, re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub('', newtext)
allstars += found
if allstars != []:
newtext = newtext.strip() + config.line_separator * 2
allstars.sort()
for element in allstars:
newtext += '%s%s' % (element.strip(), config.LS)
# Adding the interwiki
newtext = pywikibot.replaceLanguageLinks(newtext, interwikiInside,
site)
else:
newtext += u"%s%s" % (config.line_separator, addText)
else:
newtext = addText + config.line_separator + text
if putText and text != newtext:
pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<"
% page.title())
pywikibot.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if not always:
choice = pywikibot.inputChoice(
u'Do you want to accept these changes?',
['Yes', 'No', 'All', 'open in Browser'],
['y', 'n', 'a', 'b'], 'n')
if choice == 'a':
always = True
elif choice == 'n':
return (False, False, always)
elif choice == 'b':
webbrowser.open("http://%s%s" % (
page.site.hostname(),
page.site.nice_get_address(page.title())
))
pywikibot.input("Press Enter when finished in browser.")
if always or choice == 'y':
try:
if always:
page.put(newtext, summary,
minorEdit=page.namespace() != 3)
else:
page.put_async(newtext, summary,
minorEdit=page.namespace() != 3)
except pywikibot.EditConflict:
pywikibot.output(u'Edit conflict! skip!')
return (False, False, always)
except pywikibot.ServerError:
errorCount += 1
if errorCount < config.maxretries:
pywikibot.output(u'Server Error! Wait..')
time.sleep(5)
continue
else:
raise pywikibot.ServerError(u'Fifth Server Error!')
except pywikibot.SpamfilterError as e:
pywikibot.output(
u'Cannot change %s because of blacklist entry %s'
% (page.title(), e.url))
return (False, False, always)
except pywikibot.PageNotSaved as error:
pywikibot.output(u'Error putting page: %s' % error.args)
return (False, False, always)
except pywikibot.LockedPage:
pywikibot.output(u'Skipping %s (locked page)'
% page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
else:
return (text, newtext, always)
def add_text(page = None, addText = None, summary = None, regexSkip = None,
regexSkipUrl = None, always = False, up = False, putText = True,
oldTextGiven = None, reorderEnabled = True, create=False):
if not addText:
raise NoEnoughData('You have to specify what text you want to add!')
if not summary:
summary = i18n.twtranslate(pywikibot.getSite(), 'add_text-adding',
{'adding': addText[:200]})
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
starsList = [
u'bueno',
u'cyswllt[ _]erthygl[ _]ddethol', u'dolen[ _]ed',
u'destacado', u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a', u'link[ _]adq', u'link[ _]f[lm]', u'link[ _]km',
u'link[ _]sm', u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
errorCount = 0
site = pywikibot.getSite()
# /wiki/ is not always the right path in non-wiki projects
pathWiki = site.family.nicepath(site.lang)
if putText:
pywikibot.output(u'Loading %s...' % page.title())
if oldTextGiven == None:
try:
text = page.get()
except pywikibot.NoPage:
if create:
pywikibot.output(u"%s doesn't exist, creating it!"
% page.title())
text = u''
else:
pywikibot.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always) # continue
except pywikibot.IsRedirectPage:
pywikibot.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always) # continue
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl != None:
url = '%s%s' % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -exceptUrl is in the page. Skip!
Match was: %s''' % result)
return (False, False, always) # continue
if regexSkip != None:
result = re.findall(regexSkip, text)
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -except is in the page. Skip!
Match was: %s''' % result)
return (False, False, always) # continue
# If not up, text put below
if not up:
newtext = text
# Translating the \\n into binary \n
addText = addText.replace('\\n', '\n')
if (reorderEnabled):
# Getting the categories
categoriesInside = pywikibot.getCategoryLinks(newtext, site)
# Deleting the categories
#newtext = pywikibot.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = pywikibot.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = pywikibot.removeLanguageLinks(newtext, site)
# nn got a message between the categories and the iw's
# and they want to keep it there, first remove it
hasCommentLine = False
if (site.language()==u'nn'):
regex = re.compile('(<!-- ?interwiki \(no(?:/nb)?, ?sv, ?da first; then other languages alphabetically by name\) ?-->)')
found = regex.findall(newtext)
if found:
hasCommentLine = True
newtext = regex.sub('', newtext)
# Adding the text
newtext += u"\n%s" % addText
# Reputting the categories
#newtext = pywikibot.replaceCategoryLinks(newtext,
#categoriesInside, site, True)
#Put the nn iw message back
if site.language()==u'nn' and (interwikiInside or hasCommentLine):
newtext = newtext + u'\r\n\r\n' + nn_iw_msg
# Dealing the stars' issue
allstars = []
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)' % star,
re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub('', newtext)
allstars += found
if allstars != []:
newtext = newtext.strip()+'\r\n\r\n'
allstars.sort()
for element in allstars:
newtext += '%s\r\n' % element.strip()
# Adding the interwiki
newtext = pywikibot.replaceLanguageLinks(newtext, interwikiInside, site)
else:
# Adding the text
newtext += u"\n%s" % addText
# If instead the text must be added above...
else:
newtext = addText + '\n' + text
if putText and text != newtext:
pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<"
% page.title())
pywikibot.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if not always:
choice = pywikibot.inputChoice(
u'Do you want to accept these changes?',
['Yes', 'No', 'All'], ['y', 'N', 'a'], 'N')
if choice == 'a':
always = True
elif choice == 'n':
return (False, False, always)
if always or choice == 'y':
try:
if always:
page.put(newtext, summary)
else:
page.put_async(newtext, summary)
except pywikibot.EditConflict:
pywikibot.output(u'Edit conflict! skip!')
return (False, False, always)
except pywikibot.ServerError:
errorCount += 1
if errorCount < 5:
pywikibot.output(u'Server Error! Wait..')
time.sleep(5)
continue
else:
raise pywikibot.ServerError(u'Fifth Server Error!')
except pywikibot.SpamfilterError, e:
pywikibot.output(
u'Cannot change %s because of blacklist entry %s'
% (page.title(), e.url))
return (False, False, always)
except pywikibot.PageNotSaved, error:
pywikibot.output(u'Error putting page: %s' % error.args)
return (False, False, always)
except pywikibot.LockedPage:
pywikibot.output(u'Skipping %s (locked page)'
% page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
def standardizePageFooter(self, text):
"""
Makes sure that interwiki links, categories and star templates are
put to the correct position and into the right order. This combines the
old instances standardizeInterwiki and standardizeCategories
The page footer has the following section in that sequence:
1. categories
2. additional information depending on local site policy
3. stars templates for featured and good articles
4. interwiki links
"""
starsList = [
u'bueno',
u'bom interwiki',
u'cyswllt[ _]erthygl[ _]ddethol', u'dolen[ _]ed',
u'destacado', u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a', u'link[ _]adq', u'link[ _]f[lm]', u'link[ _]km',
u'link[ _]sm', u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
categories = None
interwikiLinks = None
allstars = []
hasCommentLine = False
# The PyWikipediaBot is no longer allowed to touch categories on the
# German Wikipedia. See
# http://de.wikipedia.org/wiki/Hilfe_Diskussion:Personendaten/Archiv/1#Position_der_Personendaten_am_.22Artikelende.22
# ignoring nn-wiki of cause of the comment line above iw section
if not self.template and not '{{Personendaten' in text:
categories = pywikibot.getCategoryLinks(text, site = self.site)
if not self.talkpage:# and pywikibot.calledModuleName() <> 'interwiki':
subpage = False
if self.template:
loc = None
try:
tmpl, loc = moved_links[self.site.lang]
del tmpl
except KeyError:
pass
if loc != None and loc in self.title:
subpage = True
interwikiLinks = pywikibot.getLanguageLinks(
text, insite=self.site, template_subpage=subpage)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site = self.site)
# Removing the stars' issue
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)'
% star, re.I)
found = regex.findall(starstext)
if found != []:
if pywikibot.verbose:
print found
text = regex.sub('', text)
allstars += found
# nn got a message between the categories and the iw's
# and they want to keep it there, first remove it
if self.site.lang in msg_interwiki:
iw_msg = msg_interwiki[self.site.lang]
if isinstance(iw_msg, tuple):
iw_reg = iw_msg[1]
iw_msg = iw_msg[0]
else:
iw_reg = u'(%s)' % iw_msg
regex = re.compile(iw_reg)
found = regex.findall(text)
if found:
if pywikibot.verbose:
print found
hasCommentLine = True
text = regex.sub('', text)
# Adding categories
if categories:
text = pywikibot.replaceCategoryLinks(text, categories,
site=self.site)
# Put the iw message back
if not self.talkpage and \
((interwikiLinks or hasCommentLine) and
self.site.language() == 'nn' or
(interwikiLinks and hasCommentLine) and
self.site.language() == 'fr'):
text = text + '\r\n\r\n' + iw_msg
# Adding stars templates
if allstars:
text = text.strip()+self.site.family.interwiki_text_separator
allstars.sort()
for element in allstars:
text += '%s\r\n' % element.strip()
if pywikibot.verbose:
pywikibot.output(u'%s' %element.strip())
# Adding the interwiki
if interwikiLinks:
text = pywikibot.replaceLanguageLinks(text, interwikiLinks,
site=self.site,
template=self.template,
template_subpage=subpage)
return text
def featuredWithInterwiki(fromsite, tosite, template_on_top, pType, quiet):
if fromsite.lang not in cache:
cache[fromsite.lang] = {}
if tosite.lang not in cache[fromsite.lang]:
cache[fromsite.lang][tosite.lang] = {}
cc = cache[fromsite.lang][tosite.lang]
if nocache:
cc = {}
templatelist = getTemplateList(tosite.lang, pType)
findtemplate = '(' + '|'.join(templatelist) + ')'
re_Link_FA = re.compile(ur"\{\{%s\|%s\}\}"
% (findtemplate.replace(u' ', u'[ _]'),
fromsite.lang), re.IGNORECASE)
gen = featuredArticles(fromsite, pType)
gen = PreloadingGenerator(gen)
for a in gen:
if a.title() < afterpage:
continue
if u"/" in a.title() and a.namespace() != 0:
pywikibot.output(u"%s is a subpage" % a.title())
continue
if a.title() in cc:
pywikibot.output(u"(cached) %s -> %s"
% (a.title(), cc[a.title()]))
continue
if a.isRedirectPage():
a = a.getRedirectTarget()
try:
if not a.exists():
pywikibot.output(u"source page doesn't exist: %s" % a.title())
continue
atrans = findTranslated(a, tosite, quiet)
if pType != 'former':
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
pywikibot.output(u"(already done)")
else:
# insert just before interwiki
if (not interactive or
pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]'
% (a.title(), atrans.title())) in ['Y', 'y']):
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(site, 'featured-' + pType,
{'page': unicode(a)}))
### Moving {{Link FA|xx}} to top of interwikis ###
if template_on_top:
# Getting the interwiki
iw = pywikibot.getLanguageLinks(text, site)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site)
text += u"\r\n{{%s|%s}}\r\n" % (templatelist[0],
fromsite.lang)
# Adding the interwiki
text = pywikibot.replaceLanguageLinks(text,
iw, site)
# Placing {{Link FA|xx}} right next to
# corresponding interwiki
else:
text = (text[:m.end()] +
(u" {{%s|%s}}" % (templatelist[0],
fromsite.lang)) +
text[m.end():])
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!'
% atrans.title())
cc[a.title()] = atrans.title()
else:
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
# insert just before interwiki
if (not interactive or
pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]'
% (a.title(), atrans.title())) in ['Y', 'y']):
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(site, 'featured-former',
{'page': unicode(a)}))
text = re.sub(re_Link_FA, '', text)
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!'
% atrans.title())
else:
pywikibot.output(u"(already done)")
cc[a.title()] = atrans.title()
except pywikibot.PageNotSaved:
pywikibot.output(u"Page not saved")
def featuredWithInterwiki(fromsite, tosite, template_on_top, pType, quiet, dry = False):
if not fromsite.lang in cache:
cache[fromsite.lang]={}
if not tosite.lang in cache[fromsite.lang]:
cache[fromsite.lang][tosite.lang]={}
cc=cache[fromsite.lang][tosite.lang]
if nocache:
cc={}
templatelist = getTemplateList(tosite.lang, pType)
findtemplate = '(' + '|'.join(templatelist) + ')'
re_Link_FA=re.compile(ur"\{\{%s\|%s\}\}" % (findtemplate.replace(u' ', u'[ _]'), fromsite.lang), re.IGNORECASE)
re_this_iw=re.compile(ur"\[\[%s:[^]]+\]\]" % fromsite.lang)
arts=featuredArticles(fromsite, pType)
pairs=[]
for a in arts:
if a.title()<afterpage:
continue
if u"/" in a.title() and a.namespace() != 0:
wikipedia.output(u"%s is a subpage" % a.title())
continue
if a.title() in cc:
wikipedia.output(u"(cached) %s -> %s"%(a.title(), cc[a.title()]))
continue
if a.isRedirectPage():
a=a.getRedirectTarget()
try:
if not a.exists():
wikipedia.output(u"source page doesn't exist: %s" % a.title())
continue
atrans = findTranslated(a, tosite, quiet)
if pType!='former':
if atrans:
text=atrans.get()
m=re_Link_FA.search(text)
if m:
wikipedia.output(u"(already done)")
else:
# insert just before interwiki
if (not interactive or
wikipedia.input(u'Connecting %s -> %s. Proceed? [Y/N]'%(a.title(), atrans.title())) in ['Y','y']
):
m=re_this_iw.search(text)
if not m:
wikipedia.output(u"no interwiki record, very strange")
continue
site = wikipedia.getSite()
if pType == 'good':
comment = wikipedia.setAction(wikipedia.translate(site, msg_good) % (fromsite.lang, a.title()))
elif pType == 'list':
comment = wikipedia.setAction(wikipedia.translate(site, msg_lists) % (fromsite.lang, a.title()))
else:
comment = wikipedia.setAction(wikipedia.translate(site, msg) % (fromsite.lang, a.title()))
### Moving {{Link FA|xx}} to top of interwikis ###
if template_on_top == True:
# Getting the interwiki
iw = wikipedia.getLanguageLinks(text, site)
# Removing the interwiki
text = wikipedia.removeLanguageLinks(text, site)
text += u"\r\n{{%s|%s}}\r\n"%(templatelist[0], fromsite.lang)
# Adding the interwiki
text = wikipedia.replaceLanguageLinks(text, iw, site)
### Placing {{Link FA|xx}} right next to corresponding interwiki ###
else:
text=(text[:m.end()]
+ (u" {{%s|%s}}" % (templatelist[0], fromsite.lang))
+ text[m.end():])
if not dry:
try:
atrans.put(text, comment)
except wikipedia.LockedPage:
wikipedia.output(u'Page %s is locked!' % atrans.title())
cc[a.title()]=atrans.title()
else:
if atrans:
text=atrans.get()
m=re_Link_FA.search(text)
if m:
# insert just before interwiki
if (not interactive or
wikipedia.input(u'Connecting %s -> %s. Proceed? [Y/N]'%(a.title(), atrans.title())) in ['Y','y']
):
m=re_this_iw.search(text)
if not m:
wikipedia.output(u"no interwiki record, very strange")
continue
site = wikipedia.getSite()
comment = wikipedia.setAction(wikipedia.translate(site, msg_former) % (fromsite.lang, a.title()))
name=templatelist[0]
name2=name[0].lower()+name[1:]
text=text.replace(u"{{%s|%s}}" %(name, fromsite.lang),'',1)
text=text.replace(u"{{%s|%s}}" %(name2, fromsite.lang),'',1)
if not dry:
try:
atrans.put(text, comment)
except wikipedia.LockedPage:
wikipedia.output(u'Page %s is locked!' % atrans.title())
else:
wikipedia.output(u"(already done)")
cc[a.title()]=atrans.title()
except wikipedia.PageNotSaved, e:
wikipedia.output(u"Page not saved")
def add_text(
page=None,
addText=None,
summary=None,
regexSkip=None,
regexSkipUrl=None,
always=False,
up=False,
putText=True,
oldTextGiven=None,
reorderEnabled=True,
create=False,
):
if not addText:
raise NoEnoughData("You have to specify what text you want to add!")
if not summary:
summary = i18n.twtranslate(pywikibot.getSite(), "add_text-adding", {"adding": addText[:200]})
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
errorCount = 0
site = pywikibot.getSite()
pathWiki = site.family.nicepath(site.lang)
if putText:
pywikibot.output(u"Loading %s..." % page.title())
if oldTextGiven is None:
try:
text = page.get()
except pywikibot.NoPage:
if create:
pywikibot.output(u"%s doesn't exist, creating it!" % page.title())
text = u""
else:
pywikibot.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always)
except pywikibot.IsRedirectPage:
pywikibot.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always)
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl is not None:
url = "%s%s" % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
pywikibot.output(
u"""Exception! regex (or word) used with -exceptUrl is in the page. Skip!
Match was: %s"""
% result
)
return (False, False, always)
if regexSkip is not None:
result = re.findall(regexSkip, text)
if result != []:
pywikibot.output(
u"""Exception! regex (or word) used with -except is in the page. Skip!
Match was: %s"""
% result
)
return (False, False, always)
# If not up, text put below
if not up:
newtext = text
# Translating the \\n into binary \n
addText = addText.replace("\\n", config.line_separator)
if reorderEnabled:
# Getting the categories
categoriesInside = pywikibot.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = pywikibot.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = pywikibot.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = pywikibot.removeLanguageLinks(newtext, site)
# Adding the text
newtext += u"%s%s" % (config.line_separator, addText)
# Reputting the categories
newtext = pywikibot.replaceCategoryLinks(newtext, categoriesInside, site, True)
# Dealing the stars' issue
allstars = []
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile("(\{\{(?:template:|)%s\|.*?\}\}[\s]*)" % star, re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub("", newtext)
allstars += found
if allstars != []:
newtext = newtext.strip() + config.line_separator * 2
allstars.sort()
for element in allstars:
newtext += "%s%s" % (element.strip(), config.LS)
# Adding the interwiki
newtext = pywikibot.replaceLanguageLinks(newtext, interwikiInside, site)
else:
newtext += u"%s%s" % (config.line_separator, addText)
else:
newtext = addText + config.line_separator + text
if putText and text != newtext:
pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<" % page.title())
pywikibot.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if not always:
choice = pywikibot.inputChoice(
u"Do you want to accept these changes?",
["Yes", "No", "All", "open in Browser"],
["y", "n", "a", "b"],
"n",
)
if choice == "a":
always = True
elif choice == "n":
return (False, False, always)
elif choice == "b":
webbrowser.open("http://%s%s" % (page.site.hostname(), page.site.nice_get_address(page.title())))
pywikibot.input("Press Enter when finished in browser.")
if always or choice == "y":
try:
if always:
page.put(newtext, summary, minorEdit=page.namespace() != 3)
else:
page.put_async(newtext, summary, minorEdit=page.namespace() != 3)
except pywikibot.EditConflict:
pywikibot.output(u"Edit conflict! skip!")
return (False, False, always)
except pywikibot.ServerError:
errorCount += 1
if errorCount < config.maxretries:
pywikibot.output(u"Server Error! Wait..")
time.sleep(5)
continue
else:
raise pywikibot.ServerError(u"Fifth Server Error!")
except pywikibot.SpamfilterError as e:
pywikibot.output(u"Cannot change %s because of blacklist entry %s" % (page.title(), e.url))
return (False, False, always)
except pywikibot.PageNotSaved as error:
pywikibot.output(u"Error putting page: %s" % error.args)
return (False, False, always)
except pywikibot.LockedPage:
pywikibot.output(u"Skipping %s (locked page)" % page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
else:
return (text, newtext, always)
def featuredbot(arts, cc, tosite, template_on_top, pType, quiet, dry):
templatelist = getTemplateList(tosite.lang, pType)
findtemplate = '(' + '|'.join(templatelist) + ')'
re_Link_FA=re.compile(ur"\{\{%s\|%s\}\}"
% (findtemplate.replace(u' ', u'[ _]'),
fromsite.lang), re.IGNORECASE)
re_this_iw=re.compile(ur"\[\[%s:[^]]+\]\]" % fromsite.lang)
pairs=[]
for a in arts:
if a.title() < afterpage:
continue
if u"/" in a.title() and a.namespace() != 0:
pywikibot.output(u"%s is a subpage" % a.title())
continue
if a.title() in cc:
pywikibot.output(u"(cached) %s -> %s"%(a.title(), cc[a.title()]))
continue
if a.isRedirectPage():
a=a.getRedirectTarget()
try:
if not a.exists():
pywikibot.output(u"source page doesn't exist: %s" % a.title())
continue
atrans = findTranslated(a, tosite, quiet)
if pType!='former':
if atrans:
text=atrans.get()
m=re_Link_FA.search(text)
if m:
pywikibot.output(u"(already done)")
else:
# insert just before interwiki
if (not interactive or
pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]'
% (a.title(), atrans.title())) in ['Y','y']
):
m=re_this_iw.search(text)
if not m:
pywikibot.output(
u"no interwiki record, very strange")
continue
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(
site, 'featured-' + pType,
{'page': a.title(
asLink=True, forceInterwiki=True)}))
### Moving {{Link FA|xx}} to top of interwikis ###
if template_on_top == True:
# Getting the interwiki
iw = pywikibot.getLanguageLinks(text, site)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site)
text += u"\r\n{{%s|%s}}\r\n" % (templatelist[0],
fromsite.lang)
# Adding the interwiki
text = pywikibot.replaceLanguageLinks(text,
iw, site)
### Placing {{Link FA|xx}} right next to corresponding interwiki ###
else:
text=(text[:m.end()]
+ (u" {{%s|%s}}" % (templatelist[0],
fromsite.lang))
+ text[m.end():])
if not dry:
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!'
% atrans.title())
cc[a.title()]=atrans.title()
else:
if atrans:
text=atrans.get()
m=re_Link_FA.search(text)
if m:
# insert just before interwiki
if (not interactive or
pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]'
% (a.title(), atrans.title())) in ['Y','y']
):
m=re_this_iw.search(text)
if not m:
pywikibot.output(
u"no interwiki record, very strange")
continue
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(
site, 'featured-former',
{'page': a.title(
asLink=True, forceInterwiki=True)}))
text = re.sub(re_Link_FA,'',text)
if not dry:
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!'
% atrans.title())
else:
pywikibot.output(u"(already done)")
cc[a.title()]=atrans.title()
except pywikibot.PageNotSaved, e:
pywikibot.output(u"Page not saved")
def featuredbot(arts, cc, tosite, template_on_top, pType, quiet, dry):
templatelist = getTemplateList(tosite.lang, pType)
findtemplate = '(' + '|'.join(templatelist) + ')'
re_Link_FA = re.compile(
ur"\{\{%s\|%s\}\}" %
(findtemplate.replace(u' ', u'[ _]'), fromsite.lang), re.IGNORECASE)
re_this_iw = re.compile(ur"\[\[%s:[^]]+\]\]" % fromsite.lang)
pairs = []
for a in arts:
if a.title() < afterpage:
continue
if u"/" in a.title() and a.namespace() != 0:
pywikibot.output(u"%s is a subpage" % a.title())
continue
if a.title() in cc:
pywikibot.output(u"(cached) %s -> %s" % (a.title(), cc[a.title()]))
continue
if a.isRedirectPage():
a = a.getRedirectTarget()
try:
if not a.exists():
pywikibot.output(u"source page doesn't exist: %s" % a.title())
continue
atrans = findTranslated(a, tosite, quiet)
if pType != 'former':
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
pywikibot.output(u"(already done)")
else:
# insert just before interwiki
if (not interactive or pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]' %
(a.title(), atrans.title())) in ['Y', 'y']):
m = re_this_iw.search(text)
if not m:
pywikibot.output(
u"no interwiki record, very strange")
continue
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(
site, 'featured-' + pType, {
'page':
a.title(asLink=True,
forceInterwiki=True)
}))
### Moving {{Link FA|xx}} to top of interwikis ###
if template_on_top == True:
# Getting the interwiki
iw = pywikibot.getLanguageLinks(text, site)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(
text, site)
text += u"\r\n{{%s|%s}}\r\n" % (
templatelist[0], fromsite.lang)
# Adding the interwiki
text = pywikibot.replaceLanguageLinks(
text, iw, site)
### Placing {{Link FA|xx}} right next to corresponding interwiki ###
else:
text = (text[:m.end()] +
(u" {{%s|%s}}" %
(templatelist[0], fromsite.lang)) +
text[m.end():])
if not dry:
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!' %
atrans.title())
cc[a.title()] = atrans.title()
else:
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
# insert just before interwiki
if (not interactive or pywikibot.input(
u'Connecting %s -> %s. Proceed? [Y/N]' %
(a.title(), atrans.title())) in ['Y', 'y']):
m = re_this_iw.search(text)
if not m:
pywikibot.output(
u"no interwiki record, very strange")
continue
site = pywikibot.getSite()
comment = pywikibot.setAction(
i18n.twtranslate(
site, 'featured-former', {
'page':
a.title(asLink=True,
forceInterwiki=True)
}))
text = re.sub(re_Link_FA, '', text)
if not dry:
try:
atrans.put(text, comment)
except pywikibot.LockedPage:
pywikibot.output(u'Page %s is locked!' %
atrans.title())
else:
pywikibot.output(u"(already done)")
cc[a.title()] = atrans.title()
except pywikibot.PageNotSaved, e:
pywikibot.output(u"Page not saved")
def featuredWithInterwiki(fromsite,
tosite,
template_on_top,
pType,
quiet,
dry=False):
if not fromsite.lang in cache:
cache[fromsite.lang] = {}
if not tosite.lang in cache[fromsite.lang]:
cache[fromsite.lang][tosite.lang] = {}
cc = cache[fromsite.lang][tosite.lang]
if nocache:
cc = {}
templatelist = getTemplateList(tosite.lang, pType)
findtemplate = '(' + '|'.join(templatelist) + ')'
re_Link_FA = re.compile(
ur"\{\{%s\|%s\}\}" %
(findtemplate.replace(u' ', u'[ _]'), fromsite.lang), re.IGNORECASE)
re_this_iw = re.compile(ur"\[\[%s:[^]]+\]\]" % fromsite.lang)
arts = featuredArticles(fromsite, pType)
pairs = []
for a in arts:
if a.title() < afterpage:
continue
if u"/" in a.title() and a.namespace() != 0:
wikipedia.output(u"%s is a subpage" % a.title())
continue
if a.title() in cc:
wikipedia.output(u"(cached) %s -> %s" % (a.title(), cc[a.title()]))
continue
if a.isRedirectPage():
a = a.getRedirectTarget()
try:
if not a.exists():
wikipedia.output(u"source page doesn't exist: %s" % a.title())
continue
atrans = findTranslated(a, tosite, quiet)
if pType != 'former':
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
wikipedia.output(u"(already done)")
else:
# insert just before interwiki
if (not interactive or wikipedia.input(
u'Connecting %s -> %s. Proceed? [Y/N]' %
(a.title(), atrans.title())) in ['Y', 'y']):
m = re_this_iw.search(text)
if not m:
wikipedia.output(
u"no interwiki record, very strange")
continue
site = wikipedia.getSite()
if pType == 'good':
comment = wikipedia.setAction(
wikipedia.translate(site, msg_good) %
(fromsite.lang, a.title()))
elif pType == 'list':
comment = wikipedia.setAction(
wikipedia.translate(site, msg_lists) %
(fromsite.lang, a.title()))
else:
comment = wikipedia.setAction(
wikipedia.translate(site, msg) %
(fromsite.lang, a.title()))
### Moving {{Link FA|xx}} to top of interwikis ###
if template_on_top == True:
# Getting the interwiki
iw = wikipedia.getLanguageLinks(text, site)
# Removing the interwiki
text = wikipedia.removeLanguageLinks(
text, site)
text += u"\r\n{{%s|%s}}\r\n" % (
templatelist[0], fromsite.lang)
# Adding the interwiki
text = wikipedia.replaceLanguageLinks(
text, iw, site)
### Placing {{Link FA|xx}} right next to corresponding interwiki ###
else:
text = (text[:m.end()] +
(u" {{%s|%s}}" %
(templatelist[0], fromsite.lang)) +
text[m.end():])
if not dry:
try:
atrans.put(text, comment)
except wikipedia.LockedPage:
wikipedia.output(u'Page %s is locked!' %
atrans.title())
cc[a.title()] = atrans.title()
else:
if atrans:
text = atrans.get()
m = re_Link_FA.search(text)
if m:
# insert just before interwiki
if (not interactive or wikipedia.input(
u'Connecting %s -> %s. Proceed? [Y/N]' %
(a.title(), atrans.title())) in ['Y', 'y']):
m = re_this_iw.search(text)
if not m:
wikipedia.output(
u"no interwiki record, very strange")
continue
site = wikipedia.getSite()
comment = wikipedia.setAction(
wikipedia.translate(site, msg_former) %
(fromsite.lang, a.title()))
name = templatelist[0]
name2 = name[0].lower() + name[1:]
text = text.replace(
u"{{%s|%s}}" % (name, fromsite.lang), '', 1)
text = text.replace(
u"{{%s|%s}}" % (name2, fromsite.lang), '', 1)
if not dry:
try:
atrans.put(text, comment)
except wikipedia.LockedPage:
wikipedia.output(u'Page %s is locked!' %
atrans.title())
else:
wikipedia.output(u"(already done)")
cc[a.title()] = atrans.title()
except wikipedia.PageNotSaved, e:
wikipedia.output(u"Page not saved")
def add_text(page=None, addText=None, summary=None, regexSkip=None,
regexSkipUrl=None, always=False, up=False, putText=True,
oldTextGiven=None, reorderEnabled=True, create=False):
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
starsList = [
u'bueno',
u'bom interwiki',
u'cyswllt[ _]erthygl[ _]ddethol', u'dolen[ _]ed',
u'destacado', u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a', u'link[ _]adq', u'link[ _]f[lm]', u'link[ _]km',
u'link[ _]sm', u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
errorCount = 0
site = pywikibot.getSite()
pathWiki = site.family.nicepath(site.lang)
site = pywikibot.getSite()
if oldTextGiven is None:
try:
text = page.get()
except pywikibot.NoPage:
if create:
pywikibot.output(u"%s doesn't exist, creating it!"
% page.title())
text = u''
else:
pywikibot.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always)
except pywikibot.IsRedirectPage:
pywikibot.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always)
else:
text = oldTextGiven
# If not up, text put below
if not up:
newtext = text
# Translating the \\n into binary \n
addText = addText.replace('\\n', '\n')
if (reorderEnabled):
# Getting the categories
categoriesInside = pywikibot.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = pywikibot.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = pywikibot.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = pywikibot.removeLanguageLinks(newtext, site)
# Adding the text
newtext += u"\n%s" % addText
# Reputting the categories
newtext = pywikibot.replaceCategoryLinks(newtext,
categoriesInside, site,
True)
# Dealing the stars' issue
allstars = []
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)'
% star, re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub('', newtext)
allstars += found
if allstars != []:
newtext = newtext.strip() + '\r\n\r\n'
allstars.sort()
for element in allstars:
newtext += '%s\r\n' % element.strip()
# Adding the interwiki
newtext = pywikibot.replaceLanguageLinks(newtext, interwikiInside,
site)
else:
newtext += u"\n%s" % addText
else:
newtext = addText + '\n' + text
if putText and text != newtext:
pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<"
% page.title())
#pywikibot.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if always or choice == 'y':
try:
pass
if always:
page.put(newtext, summary,
minorEdit=False)
else:
page.put_async(newtext, summary,
minorEdit=False)
except pywikibot.EditConflict:
pywikibot.output(u'Edit conflict! skip!')
return (False, False, always)
except pywikibot.ServerError:
errorCount += 1
if errorCount < 5:
pywikibot.output(u'Server Error! Wait..')
time.sleep(5)
continue
else:
raise pywikibot.ServerError(u'Fifth Server Error!')
except pywikibot.SpamfilterError, e:
pywikibot.output(
u'Cannot change %s because of blacklist entry %s'
% (page.title(), e.url))
return (False, False, always)
except pywikibot.PageNotSaved, error:
pywikibot.output(u'Error putting page: %s' % error.args)
return (False, False, always)
except pywikibot.LockedPage:
pywikibot.output(u'Skipping %s (locked page)'
% page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
wikipedia.setAction(msg[msglang]+': '+wrd)
thispl = wikipedia.PageLink(wikipedia.mylang, wrd)
if solve_redirect:
try:
alternatives.append(str(thispl.getRedirectTo()))
except wikipedia.IsNotRedirectPage:
print "The specified page is not a redirect."
sys.exit(1)
elif getalternatives:
try:
thistxt = thispl.get()
except wikipedia.IsRedirectPage,arg:
thistxt = wikipedia.PageLink(wikipedia.mylang, str(arg)).get()
thistxt = wikipedia.removeLanguageLinks(thistxt)
w=r'([^\]\|]*)'
Rlink = re.compile(r'\[\['+w+r'(\|'+w+r')?\]\]')
for a in Rlink.findall(thistxt):
alternatives.append(a[0])
alternatives = unique(alternatives)
# sort possible choices
alternatives.sort()
# print choices on screen
for i in range(len(alternatives)):
print "%3d" % i, repr(alternatives[i])
def treat(refpl, thispl):
try:
def standardizePageFooter(self, text):
"""
Makes sure that interwiki links, categories and star templates are
put to the correct position and into the right order.
This combines the old instances standardizeInterwiki and standardizeCategories
The page footer has the following section in that sequence:
1. categories
2. additional information depending on local site policy
3. stars templates for featured and good articles
4. interwiki links
"""
starsList = [
u'bueno',
u'cyswllt[ _]erthygl[ _]ddethol',
u'dolen[ _]ed',
u'destacado',
u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a',
u'link[ _]adq',
u'link[ _]f[lm]',
u'link[ _]km',
u'link[ _]sm',
u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
categories = None
interwikiLinks = None
allstars = []
hasCommentLine = False
# The PyWikipediaBot is no longer allowed to touch categories on the German Wikipedia.
# See http://de.wikipedia.org/wiki/Hilfe_Diskussion:Personendaten/Archiv/bis_2006#Position_der_Personendaten_am_.22Artikelende.22
# ignoring nn-wiki of cause of the comment line above iw section
if not self.template and not '{{Personendaten' in text:
categories = pywikibot.getCategoryLinks(text, site=self.site)
if not self.talkpage: # and pywikibot.calledModuleName() <> 'interwiki':
subpage = False
if self.template:
loc = None
try:
tmpl, loc = moved_links[self.site.lang]
del tmpl
except KeyError:
pass
if loc != None and loc in self.title:
subpage = True
interwikiLinks = pywikibot.getLanguageLinks(
text, insite=self.site, template_subpage=subpage)
# Removing the interwiki
text = pywikibot.removeLanguageLinks(text, site=self.site)
# Removing the stars' issue
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile(
'(\{\{(?:template:|)%s\|.*?\}\}[\s]*)' % star, re.I)
found = regex.findall(starstext)
if found != []:
if pywikibot.verbose:
print found
text = regex.sub('', text)
allstars += found
# nn got a message between the categories and the iw's
# and they want to keep it there, first remove it
if self.site.language() == 'nn':
regex = re.compile(
'(<!-- ?interwiki \(no(?:/nb)?, ?sv, ?da first; then other languages alphabetically by name\) ?-->)'
)
found = regex.findall(text)
if found:
if pywikibot.verbose:
print found
hasCommentLine = True
text = regex.sub('', text)
# Adding categories
if categories:
text = pywikibot.replaceCategoryLinks(text,
categories,
site=self.site)
# Put the nn iw message back
if self.site.language() == 'nn' and not self.talkpage and (
interwikiLinks or hasCommentLine):
text = text + '\r\n\r\n' + nn_iw_msg
# Adding stars templates
if allstars:
text = text.strip() + self.site.family.interwiki_text_separator
allstars.sort()
for element in allstars:
text += '%s\r\n' % element.strip()
if pywikibot.verbose:
pywikibot.output(u'%s' % element.strip())
# Adding the interwiki
if interwikiLinks:
text = pywikibot.replaceLanguageLinks(text,
interwikiLinks,
site=self.site,
template=self.template,
template_subpage=subpage)
return text
def add_text(page = None, addText = None, summary = None, regexSkip = None, regexSkipUrl = None,
always = False, up = False, putText = True, oldTextGiven = None):
if not addText:
raise NoEnoughData('You have to specify what text you want to add!')
if not summary:
summary = wikipedia.setAction(wikipedia.translate(wikipedia.getSite(), msg) % addText)
# When a page is tagged as "really well written" it has a star in the interwiki links.
# This is a list of all the templates used (in regex format) to make the stars appear.
starsList = ['link[ _]fa', 'link[ _]adq', 'enllaç[ _]ad',
'link[ _]ua', 'legătură[ _]af', 'destacado',
'ua', 'liên k[ _]t[ _]chọn[ _]lọc']
errorCount = 0
site = wikipedia.getSite()
# /wiki/ is not always the right path in non-wiki projects
pathWiki = site.family.nicepath(site.lang)
if putText:
wikipedia.output(u'Loading %s...' % page.title())
if oldTextGiven == None:
try:
text = page.get()
except wikipedia.NoPage:
wikipedia.output(u"%s doesn't exist, skip!" % page.title())
return (False, always) # continue
except wikipedia.IsRedirectPage:
wikipedia.output(u"%s is a redirect, skip!" % page.title())
return (False, always) # continue
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl != None:
url = '%s%s' % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
wikipedia.output(u'Exception! regex (or word) used with -exceptUrl is in the page. Skip!')
return (False, always) # continue
if regexSkip != None:
result = re.findall(regexSkip, text)
if result != []:
wikipedia.output(u'Exception! regex (or word) used with -except is in the page. Skip!')
return (False, always) # continue
# If not up, text put below
if not up:
newtext = text
# Getting the categories
categoriesInside = wikipedia.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = wikipedia.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = wikipedia.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = wikipedia.removeLanguageLinks(newtext, site)
#nn got a message between the categories and the iw's and they want to keep it there, first remove it
if (site.language()==u'nn'):
newtext = newtext.replace(nn_iw_msg, '')
# Translating the \\n into binary \n
addText = addText.replace('\\n', '\n')
# Adding the text
newtext += u"\n%s" % addText
# Reputting the categories
newtext = wikipedia.replaceCategoryLinks(newtext, categoriesInside, site, True)
#Put the nn iw message back
if (site.language()==u'nn'):
newtext = newtext + u'\n' + nn_iw_msg
# Dealing the stars' issue
starsListInPage = list()
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}\n)' % star, re.I)
risultato = regex.findall(newtext)
if risultato != []:
newtext = regex.sub('', newtext)
for element in risultato:
newtext += '\n%s' % element
# Adding the interwiki
newtext = wikipedia.replaceLanguageLinks(newtext, interwikiInside, site)
# If instead the text must be added above...
else:
newtext = addText + '\n' + text
if putText and text != newtext:
wikipedia.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<" % page.title())
wikipedia.showDiff(text, newtext)
choice = ''
# Let's put the changes.
while 1:
# If someone load it as module, maybe it's not so useful to put the text in the page
if putText:
if not always:
choice = wikipedia.inputChoice(u'Do you want to accept these changes?', ['Yes', 'No', 'All'], ['y', 'N', 'a'], 'N')
if choice == 'a':
always = True
if choice == 'n':
return (False, always)
if choice == 'y' or always:
try:
if always:
page.put(newtext, summary)
else:
page.put_async(newtext, summary)
except wikipedia.EditConflict:
wikipedia.output(u'Edit conflict! skip!')
return (False, always)
except wikipedia.ServerError:
errorCount += 1
if errorCount < 5:
wikipedia.output(u'Server Error! Wait..')
time.sleep(3)
continue
else:
raise wikipedia.ServerError(u'Fifth Server Error!')
except wikipedia.SpamfilterError, e:
wikipedia.output(u'Cannot change %s because of blacklist entry %s' % (page.title(), e.url))
return (False, always)
except wikipedia.PageNotSaved, error:
wikipedia.output(u'Error putting page: %s' % error.args)
return (False, always)
except wikipedia.LockedPage:
wikipedia.output(u'Skipping %s (locked page)' % page.title())
return (False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, always)
def add_text(page = None, addText = None, summary = None, regexSkip = None,
regexSkipUrl = None, always = False, up = False, putText = True,
oldTextGiven = None, reorderEnabled = True, create=False):
if not addText:
raise NoEnoughData('You have to specify what text you want to add!')
if not summary:
summary = i18n.twtranslate(pywikibot.getSite(), 'add_text-adding',
{'adding': addText[:200]})
# When a page is tagged as "really well written" it has a star in the
# interwiki links. This is a list of all the templates used (in regex
# format) to make the stars appear.
starsList = [
u'bueno',
u'bom interwiki',
u'cyswllt[ _]erthygl[ _]ddethol', u'dolen[ _]ed',
u'destacado', u'destaca[tu]',
u'enllaç[ _]ad',
u'enllaz[ _]ad',
u'leam[ _]vdc',
u'legătură[ _]a[bcf]',
u'liamm[ _]pub',
u'lien[ _]adq',
u'lien[ _]ba',
u'liên[ _]kết[ _]bài[ _]chất[ _]lượng[ _]tốt',
u'liên[ _]kết[ _]chọn[ _]lọc',
u'ligam[ _]adq',
u'ligoelstara',
u'ligoleginda',
u'link[ _][afgu]a', u'link[ _]adq', u'link[ _]f[lm]', u'link[ _]km',
u'link[ _]sm', u'linkfa',
u'na[ _]lotura',
u'nasc[ _]ar',
u'tengill[ _][úg]g',
u'ua',
u'yüm yg',
u'רא',
u'وصلة مقالة جيدة',
u'وصلة مقالة مختارة',
]
errorCount = 0
site = pywikibot.getSite()
# /wiki/ is not always the right path in non-wiki projects
pathWiki = site.family.nicepath(site.lang)
if putText:
pywikibot.output(u'Loading %s...' % page.title())
if oldTextGiven == None:
try:
text = page.get()
except pywikibot.NoPage:
if create:
pywikibot.output(u"%s doesn't exist, creating it!"
% page.title())
text = u''
else:
pywikibot.output(u"%s doesn't exist, skip!" % page.title())
return (False, False, always) # continue
except pywikibot.IsRedirectPage:
pywikibot.output(u"%s is a redirect, skip!" % page.title())
return (False, False, always) # continue
else:
text = oldTextGiven
# Understand if the bot has to skip the page or not
# In this way you can use both -except and -excepturl
if regexSkipUrl != None:
url = '%s%s' % (pathWiki, page.urlname())
result = re.findall(regexSkipUrl, site.getUrl(url))
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -exceptUrl is in the page. Skipping!
Match was: %s''' % result)
return (False, False, always) # continue
if regexSkip != None:
result = re.findall(regexSkip, text)
if result != []:
pywikibot.output(
u'''Exception! regex (or word) used with -except is in the page. Skipping!
Match was: %s''' % result)
return (False, False, always) # continue
# If not up, text put below
if not up:
newtext = text
# Translating the \\n into binary \n
addText = addText.replace('\\n', '\n')
if (reorderEnabled):
# Getting the categories
categoriesInside = pywikibot.getCategoryLinks(newtext, site)
# Deleting the categories
newtext = pywikibot.removeCategoryLinks(newtext, site)
# Getting the interwiki
interwikiInside = pywikibot.getLanguageLinks(newtext, site)
# Removing the interwiki
newtext = pywikibot.removeLanguageLinks(newtext, site)
# nn got a message between the categories and the iw's
# and they want to keep it there, first remove it
hasCommentLine = False
if (site.language()==u'nn'):
regex = re.compile('(<!-- ?interwiki \(no(?:/nb)?, ?sv, ?da first; then other languages alphabetically by name\) ?-->)')
found = regex.findall(newtext)
if found:
hasCommentLine = True
newtext = regex.sub('', newtext)
# Adding the text
newtext += u"\n%s" % addText
# Reputting the categories
newtext = pywikibot.replaceCategoryLinks(newtext,
categoriesInside, site, True)
#Put the nn iw message back
if site.language()==u'nn' and (interwikiInside or hasCommentLine):
newtext = newtext + u'\r\n\r\n' + nn_iw_msg
# Dealing the stars' issue
allstars = []
starstext = pywikibot.removeDisabledParts(text)
for star in starsList:
regex = re.compile('(\{\{(?:template:|)%s\|.*?\}\}[\s]*)' % star,
re.I)
found = regex.findall(starstext)
if found != []:
newtext = regex.sub('', newtext)
allstars += found
if allstars != []:
newtext = newtext.strip()+'\r\n\r\n'
allstars.sort()
for element in allstars:
newtext += '%s\r\n' % element.strip()
# Adding the interwiki
newtext = pywikibot.replaceLanguageLinks(newtext, interwikiInside, site)
else:
# Adding the text
newtext += u"\n%s" % addText
# If instead the text must be added above...
else:
newtext = addText + '\n' + text
if putText and text != newtext:
pywikibot.output(u"\n\n>>> \03{lightpurple}%s\03{default} <<<"
% page.title())
pywikibot.showDiff(text, newtext)
# Let's put the changes.
while True:
# If someone load it as module, maybe it's not so useful to put the
# text in the page
if putText:
if not always:
choice = pywikibot.inputChoice(
u'Do you want to accept these changes?',
['Yes', 'No', 'All', 'open in Browser'], ['y', 'N', 'a', 'b'], 'N')
if choice == 'a':
always = True
elif choice == 'n':
return (False, False, always)
elif choice == 'b':
webbrowser.open("http://%s%s" % (
page.site().hostname(),
page.site().nice_get_address(page.title())
))
pywikibot.input("Press Enter when finished in browser.")
if always or choice == 'y':
try:
if always:
page.put(newtext, summary)
else:
page.put_async(newtext, summary)
except pywikibot.EditConflict:
pywikibot.output(u'Edit conflict! skip!')
return (False, False, always)
except pywikibot.ServerError:
errorCount += 1
if errorCount < 5:
pywikibot.output(u'Server Error! Wait..')
time.sleep(5)
continue
else:
raise pywikibot.ServerError(u'Fifth Server Error!')
except pywikibot.SpamfilterError, e:
pywikibot.output(
u'Cannot change %s because of blacklist entry %s'
% (page.title(), e.url))
return (False, False, always)
except pywikibot.PageNotSaved, error:
pywikibot.output(u'Error putting page: %s' % error.args)
return (False, False, always)
except pywikibot.LockedPage:
pywikibot.output(u'Skipping %s (locked page)'
% page.title())
return (False, False, always)
else:
# Break only if the errors are one after the other...
errorCount = 0
return (True, True, always)
id=u""
iw=u""
for i in m:
if not id and not iw:
id=i.group(1)
iw=i.group(2)
if not id or not iw:
continue
p=wikipedia.Page(wikipedia.Site(id, "wikipedia"), iw)
try:
if p.exists() and not p.isRedirectPage() and not p.isDisambig():
wiws=p.interwiki()
wiws.append(p)
wiws.sort()
nuevo=wikipedia.removeLanguageLinks(ctext, p.site())
nuevo+=u"\n"
for i in wiws:
nuevo+=u"\n[[%s:%s]]" % (i.site().lang, i.title())
if nuevo!=ctext and len(nuevo)>len(ctext)+10:
wikipedia.showDiff(ctext, nuevo)
page.put(nuevo, u"BOT - Updating interwikis")
else:
wikipedia.output(u"Los interwikis ya estan actualizados")
except:
pass
