def testRun(self):
pipeline = RevisionsDiffPipeline('Hello', 'Hello there', {})
pipeline.start_test()
result = pipeline.outputs.default.value
expected = Revision()
expected.wordsAdded = 1
expected.wordsDeleted = 0
expected.wordCount = 2
self.assertEqual(result, expected.to_dict())
def testRun(self, mockUrlOpenMethod):
exportLinks = [
'docs.google.com/123',
'docs.google.com/456',
'docs.google.com/789'
]
mockRevisions = [
Revision(exportLink=exportLinks[0]).to_dict(),
Revision(exportLink=exportLinks[1]).to_dict(),
Revision(exportLink=exportLinks[2]).to_dict()
]
exportedText = {
exportLinks[0]: StringIO('Hello'),
exportLinks[1]: StringIO('Hello world.'),
exportLinks[2]: StringIO('Hello.'),
}
mockUrlOpenMethod.side_effect = lambda url: exportedText[url]
# TODO(michaelcupino): Fix tests
pipeline = RevisionsAnalysisPipeline(mockRevisions, None)
pipeline.start_test()
result = pipeline.outputs.default.value
revision1 = Revision()
revision1.wordsAdded = 1
revision1.wordsDeleted = 0
revision1.wordCount = 1
revision1.exportLink = 'docs.google.com/123'
revision2 = Revision()
revision2.wordsAdded = 1
revision2.wordsDeleted = 0
revision2.wordCount = 2
revision2.exportLink = 'docs.google.com/456'
revision3 = Revision()
revision3.wordsAdded = 0
revision3.wordsDeleted = 1
revision3.wordCount = 1
revision3.exportLink = 'docs.google.com/789'
self.assertEqual([revision1.to_dict(), revision2.to_dict(),
revision3.to_dict()], result)
def run(self, revisionTextA, revisionTextB, revisionDict):
gdiff = diff_match_patch()
revisionDiffs = gdiff.diff_main(revisionTextA, revisionTextB, False)
gdiff.diff_cleanupSemantic(revisionDiffs)
revisionDiffs = filter(self.isRemoveOrAdd, revisionDiffs)
diffWordCount = map(self.countWords, revisionDiffs)
addedWordCount = self.getAddWordCount(diffWordCount)
deletedWordCount = self.getDeletedWordCount(diffWordCount)
revision = Revision(**revisionDict)
revision.wordsAdded = addedWordCount
revision.wordsDeleted = deletedWordCount
revision.wordCount = self.getWordCount(revisionTextB)
return revision.to_dict()
def run(self, revisionTextA, revisionTextB, revisionDict):
gdiff = diff_match_patch()
revisionDiffs = gdiff.diff_main(revisionTextA,
revisionTextB, False)
gdiff.diff_cleanupSemantic(revisionDiffs)
revisionDiffs = filter(self.isRemoveOrAdd, revisionDiffs)
diffWordCount = map(self.countWords, revisionDiffs)
addedWordCount = self.getAddWordCount(diffWordCount)
deletedWordCount = self.getDeletedWordCount(diffWordCount)
revision = Revision(**revisionDict)
revision.wordsAdded = addedWordCount
revision.wordsDeleted = deletedWordCount
revision.wordCount = self.getWordCount(revisionTextB)
return revision.to_dict()