def testNoTranslation(self):
t = translators.CategoryNameTranslator(['foo', 'bar'])
tr = t.translateWithReport('baz')
self.assertFalse(tr.changed)
self.assertFalse(tr.didQuotes)
self.assertFalse(tr.didDashes)
self.assertFalse(tr.didManual)
self.assertEqual(tr.oldName, 'baz')
self.assertEqual(tr.newName, 'baz')
def doTestDashTranslation(self, n1, n2):
t = translators.CategoryNameTranslator()
tr = t.translateWithReport(n1)
self.assertTrue(tr.changed)
self.assertFalse(tr.didQuotes)
self.assertTrue(tr.didDashes)
self.assertFalse(tr.didManual)
self.assertEqual(tr.oldName, n1)
self.assertEqual(tr.newName, n2)
def testUnusedManualTranslations(self):
t = translators.CategoryNameTranslator([
'foo',
'bar',
'qwe',
'rty',
'asd',
'fgh',
])
tr = t.translateWithReport('qwe')
self.assertEqual(t.getUnusedManualTranslations(), {'foo', 'asd'})
def getDocumentPriority(documentNumber):
"priority to resolve name collisions for section/category/type codes"
document=next(d for d in documents if d['documentNumber']==documentNumber)
if document['amendmentFlag']==2:
priority=30 # law, published by fincom
elif document['amendmentFlag']==0:
priority=20 # law project, published by fincom
else:
priority=10 # amendment, published by assembly, often contains typos
priority+=document['stageNumber'] # prefer latter stage
return priority
ys2014=dataSets.YearSet(2014,inputDirectory,getDocumentPriority)
ys2015=dataSets.YearSet(2015,inputDirectory,getDocumentPriority)
with open('categoryNameTranslations.txt',encoding='utf8') as file1, open('categoryNameTypos.txt',encoding='utf8') as file2:
translator=translators.CategoryNameTranslator(itertools.chain(file1,file2))
iys=dataSets.InterYearSet([ys2014,ys2015],translator)
print('== unused manual translations ==')
for k in sorted(translator.getUnusedManualTranslations()):
print(k)
### write sql ###
sql=open(outputFilename,'w',encoding='utf8')
sql.write("-- бюджет Санкт-Петербурга на 2014-2017 гг.\n")
def putValue(v):
if type(v) is str:
return "'"+v+"'"
elif type(v) is int:
return str(v)