def getPoisonedLinks(pl):
"""Returns a list of known corrupted links that should be removed if seen
"""
result = []
pywikibot.output(u'getting poisoned links for %s' % pl.title())
dictName, value = date.getAutoFormat(pl.site().language(), pl.title())
if dictName is not None:
pywikibot.output( u'date found in %s' % dictName )
# errors in year BC
if dictName in date.bcFormats:
for fmt in bcDateErrors:
result.append( fmt % value )
# i guess this is like friday the 13th for the years
if value == 398 and dictName == 'yearsBC':
appendFormatedDates(result, dictName, 399)
if dictName == 'yearsBC':
appendFormatedDates(result, 'decadesBC', value)
appendFormatedDates(result, 'yearsAD', value)
if dictName == 'yearsAD':
appendFormatedDates(result, 'decadesAD', value)
appendFormatedDates(result, 'yearsBC', value)
if dictName == 'centuriesBC':
appendFormatedDates(result, 'decadesBC', value * 100 + 1)
if dictName == 'centuriesAD':
appendFormatedDates(result, 'decadesAD', value * 100 + 1)
return result
def getPoisonedLinks(pl):
"""Returns a list of known corrupted links that should be removed if seen
"""
result = []
pywikibot.output(u'getting poisoned links for %s' % pl.title())
dictName, value = date.getAutoFormat(pl.site.language(), pl.title())
if dictName is not None:
pywikibot.output(u'date found in %s' % dictName)
# errors in year BC
if dictName in date.bcFormats:
for fmt in bcDateErrors:
result.append(fmt % value)
# i guess this is like friday the 13th for the years
if value == 398 and dictName == 'yearsBC':
appendFormatedDates(result, dictName, 399)
if dictName == 'yearsBC':
appendFormatedDates(result, 'decadesBC', value)
appendFormatedDates(result, 'yearsAD', value)
if dictName == 'yearsAD':
appendFormatedDates(result, 'decadesAD', value)
appendFormatedDates(result, 'yearsBC', value)
if dictName == 'centuriesBC':
appendFormatedDates(result, 'decadesBC', value * 100 + 1)
if dictName == 'centuriesAD':
appendFormatedDates(result, 'decadesAD', value * 100 + 1)
return result
def isdate(s):
"""returns true iff s is a date or year
"""
dict,val = date.getAutoFormat( pywikibot.getSite().language(), s )
return dict is not None
def translate(page, hints = None, auto = True, removebrackets = False, site = None, family = None):
"""
Please comment your source code! --Daniel
Does some magic stuff. Returns a list of pages.
Goes through all entries in 'hints'. Returns a list of pages.
Entries for single page titles list those pages. Page titles for entries
such as "all:" or "xyz:" or "20:" are first built from the page title of
'page' and then listed. When 'removebrackets' is True, a trailing pair of
brackets and the text between them is removed from the page title.
If 'auto' is true, known year and date page titles are autotranslated
to all known target languages and inserted into the list.
"""
result = []
if site is None and page:
site = page.site()
if family is None and site:
family = site.family
if site:
sitelang = site.language()
if hints:
for h in hints:
if ':' not in h:
# argument given as -hint:xy where xy is a language code
codes = h
newname = ''
else:
codes, newname = h.split(':', 1)
if newname == '':
# if given as -hint:xy or -hint:xy:, assume that there should
# be a page in language xy with the same title as the page
# we're currently working on ...
if page is None:
continue
ns = page.namespace()
if ns:
newname = u'%s:%s' % (family.namespace('_default', ns),
page.titleWithoutNamespace())
else:
# article in the main namespace
newname = page.title()
# ... unless we do want brackets
if removebrackets:
newname = re.sub(re.compile(ur"\W*?\(.*?\)\W*?", re.UNICODE), u" ", newname)
codesplit = codes.split(',')
codes = []
for code in codesplit:
try:
number = int(code)
_join_to_(codes, family.languages_by_size[:number] )
except ValueError:
if code == 'all':
_join_to_(codes, family.languages_by_size )
elif code in family.language_groups:
_join_to_(codes, family.language_groups[code] )
elif code:
_join_to_(codes, [ code ] )
for newcode in codes:
x = None
if newcode in family.langs.keys():
if page is None or \
(newcode != sitelang and
pywikibot.getSite().family.name
not in family.interwiki_forwarded_from):
x = pywikibot.Page(pywikibot.getSite(fam=family, code=newcode), newname)
elif newcode in family.interwiki_forwarded_from:
x = pywikibot.Page(pywikibot.getSite(fam=newcode, code=newcode), newname)
else:
if pywikibot.verbose:
pywikibot.output(u"Ignoring the unknown language code %s" % newcode)
if x:
_join_to_(result, [ x ] )
# Autotranslate dates into all other languages, the rest will come from
# existing interwiki links.
if auto and page:
# search inside all dictionaries for this link
dictName, value = date.getAutoFormat(sitelang, page.title())
if dictName:
if not (dictName == 'yearsBC' and
sitelang in date.maxyearBC and
value > date.maxyearBC[sitelang]) or \
(dictName == 'yearsAD' and
sitelang in date.maxyearAD and
value > date.maxyearAD[sitelang]):
pywikibot.output(
u'TitleTranslate: %s was recognized as %s with value %d'
% (page.title(), dictName, value))
for entryLang, entry in date.formats[dictName].iteritems():
if entryLang != sitelang:
if dictName == 'yearsBC' and \
entryLang in date.maxyearBC and \
value > date.maxyearBC[entryLang]:
pass
elif dictName == 'yearsAD' and \
entryLang in date.maxyearAD and \
value > date.maxyearAD[entryLang]:
pass
else:
newname = entry(value)
x = pywikibot.Page(
pywikibot.getSite(code=entryLang,
fam=family), newname)
_join_to_(result, [ x ] )
return result
def isdate(s):
"""returns true iff s is a date or year
"""
dict, val = date.getAutoFormat(pywikibot.getSite().language(), s)
return dict is not None
def translate(page, hints = None, auto = True, removebrackets = False):
"""
Please comment your source code! --Daniel
Does some magic stuff. Returns a list of pages.
"""
result = []
site = page.site()
if hints:
for h in hints:
if ':' not in h:
# argument given as -hint:xy where xy is a language code
codes = h
newname = ''
else:
codes, newname = h.split(':', 1)
if newname == '':
# if given as -hint:xy or -hint:xy:, assume that there should
# be a page in language xy with the same title as the page
# we're currently working on ...
ns = page.namespace()
if ns:
newname = u'%s:%s' % (site.family.namespace('_default', ns), page.titleWithoutNamespace())
else:
# article in the main namespace
newname = page.title()
# ... unless we do want brackets
if removebrackets:
newname = re.sub(re.compile(ur"\W*?\(.*?\)\W*?", re.UNICODE), u" ", newname)
try:
number = int(codes)
codes = site.family.languages_by_size[:number]
except ValueError:
if codes == 'all':
codes = site.family.languages_by_size
elif codes in site.family.language_groups:
codes = site.family.language_groups[codes]
else:
codes = codes.split(',')
for newcode in codes:
if newcode in site.languages():
if newcode != site.language():
x = wikipedia.Page(site.getSite(code=newcode), newname)
if x not in result:
result.append(x)
else:
if wikipedia.verbose:
wikipedia.output(u"Ignoring unknown language code %s"%newcode)
# Autotranslate dates into all other languages, the rest will come from existing interwiki links.
if auto:
# search inside all dictionaries for this link
dictName, value = date.getAutoFormat( page.site().language(), page.title() )
if dictName:
if not (dictName == 'yearsBC' and date.maxyearBC.has_key(page.site().language()) and value > date.maxyearBC[page.site().language()]) or (dictName == 'yearsAD' and date.maxyearAD.has_key(page.site().language()) and value > date.maxyearAD[page.site().language()]):
wikipedia.output(u'TitleTranslate: %s was recognized as %s with value %d' % (page.title(),dictName,value))
for entryLang, entry in date.formats[dictName].iteritems():
if entryLang != page.site().language():
if dictName == 'yearsBC' and date.maxyearBC.has_key(entryLang) and value > date.maxyearBC[entryLang]:
pass
elif dictName == 'yearsAD' and date.maxyearAD.has_key(entryLang) and value > date.maxyearAD[entryLang]:
pass
else:
newname = entry(value)
x = wikipedia.Page( wikipedia.getSite(code=entryLang, fam=site.family), newname )
if x not in result:
result.append(x) # add new page
return result
def translate(page,
hints=None,
auto=True,
removebrackets=False,
site=None,
family=None):
"""
Goes through all entries in 'hints'. Returns a list of pages.
Entries for single page titles list those pages. Page titles for entries
such as "all:" or "xyz:" or "20:" are first built from the page title of
'page' and then listed. When 'removebrackets' is True, a trailing pair of
brackets and the text between them is removed from the page title.
If 'auto' is true, known year and date page titles are autotranslated
to all known target languages and inserted into the list.
"""
result = []
if site is None and page:
site = page.site
if family is None and site:
family = site.family
if site:
sitelang = site.language()
if hints:
for h in hints:
if ':' not in h:
# argument given as -hint:xy where xy is a language code
codes = h
newname = ''
else:
codes, newname = h.split(':', 1)
if newname == '':
# if given as -hint:xy or -hint:xy:, assume that there should
# be a page in language xy with the same title as the page
# we're currently working on ...
if page is None:
continue
ns = page.namespace()
if ns:
newname = u'%s:%s' % (family.namespace(
'_default', ns), page.title(withNamespace=False))
else:
# article in the main namespace
newname = page.title()
# ... unless we do want brackets
if removebrackets:
newname = re.sub(
re.compile(ur"\W*?\(.*?\)\W*?", re.UNICODE), u" ",
newname)
codesplit = codes.split(',')
codes = []
for code in codesplit:
try:
number = int(code)
_join_to_(codes, family.languages_by_size[:number])
except ValueError:
if code == 'all':
_join_to_(codes, family.languages_by_size)
elif code in family.language_groups:
_join_to_(codes, family.language_groups[code])
elif code:
_join_to_(codes, [code])
for newcode in codes:
x = None
if newcode in family.langs.keys():
if page is None or \
(newcode != sitelang and
pywikibot.getSite().family.name
not in family.interwiki_forwarded_from):
x = pywikibot.Page(
pywikibot.getSite(fam=family, code=newcode),
newname)
elif newcode in family.interwiki_forwarded_from:
x = pywikibot.Page(
pywikibot.getSite(fam=newcode, code=newcode), newname)
else:
if pywikibot.verbose:
pywikibot.output(
u"Ignoring the unknown language code %s" % newcode)
if x:
_join_to_(result, [x])
# Autotranslate dates into all other languages, the rest will come from
# existing interwiki links.
if auto and page:
# search inside all dictionaries for this link
dictName, value = date.getAutoFormat(sitelang, page.title())
if dictName:
if not (dictName == 'yearsBC' and
sitelang in date.maxyearBC and
value > date.maxyearBC[sitelang]) or \
(dictName == 'yearsAD' and
sitelang in date.maxyearAD and
value > date.maxyearAD[sitelang]):
pywikibot.output(
u'TitleTranslate: %s was recognized as %s with value %d' %
(page.title(), dictName, value))
for entryLang, entry in date.formats[dictName].iteritems():
if entryLang != sitelang:
if dictName == 'yearsBC' and \
entryLang in date.maxyearBC and \
value > date.maxyearBC[entryLang]:
pass
elif (dictName == 'yearsAD'
and entryLang in date.maxyearAD
and value > date.maxyearAD[entryLang]):
pass
else:
newname = entry(value)
x = pywikibot.Page(
pywikibot.getSite(code=entryLang, fam=family), newname)
_join_to_(result, [x])
return result
def translate(page, hints=None, auto=True, removebrackets=False):
"""
Please comment your source code! --Daniel
Does some magic stuff. Returns a list of pages.
"""
result = []
site = page.site()
if hints:
for h in hints:
if ':' not in h:
# argument given as -hint:xy where xy is a language code
codes = h
newname = ''
else:
codes, newname = h.split(':', 1)
if newname == '':
# if given as -hint:xy or -hint:xy:, assume that there should
# be a page in language xy with the same title as the page
# we're currently working on ...
ns = page.namespace()
if ns:
newname = u'%s:%s' % (site.family.namespace(
'_default', ns), page.titleWithoutNamespace())
else:
# article in the main namespace
newname = page.title()
# ... unless we do want brackets
if removebrackets:
newname = re.sub(
re.compile(ur"\W*?\(.*?\)\W*?", re.UNICODE), u" ",
newname)
try:
number = int(codes)
codes = site.family.languages_by_size[:number]
except ValueError:
if codes == 'all':
codes = site.family.languages_by_size
elif codes in site.family.language_groups:
codes = site.family.language_groups[codes]
else:
codes = codes.split(',')
for newcode in codes:
if newcode in site.languages():
if newcode != site.language():
x = pywikibot.Page(site.getSite(code=newcode), newname)
if x not in result:
result.append(x)
else:
if pywikibot.verbose:
pywikibot.output(u"Ignoring unknown language code %s" %
newcode)
# Autotranslate dates into all other languages, the rest will come from existing interwiki links.
if auto:
# search inside all dictionaries for this link
dictName, value = date.getAutoFormat(page.site().language(),
page.title())
if dictName:
if not (dictName == 'yearsBC'
and page.site().language() in date.maxyearBC
and value > date.maxyearBC[page.site().language()]) or (
dictName == 'yearsAD'
and page.site().language() in date.maxyearAD
and value > date.maxyearAD[page.site().language()]):
pywikibot.output(
u'TitleTranslate: %s was recognized as %s with value %d' %
(page.title(), dictName, value))
for entryLang, entry in date.formats[dictName].iteritems():
if entryLang != page.site().language():
if dictName == 'yearsBC' and entryLang in date.maxyearBC and value > date.maxyearBC[
entryLang]:
pass
elif dictName == 'yearsAD' and entryLang in date.maxyearAD and value > date.maxyearAD[
entryLang]:
pass
else:
newname = entry(value)
x = pywikibot.Page(
pywikibot.getSite(code=entryLang, fam=site.family),
newname)
if x not in result:
result.append(x) # add new page
return result
