You've already forked DataMate
feature: 增加水印去除/高级匿名化算子 (#151)
* feature: 增加水印去除算子 * feature: clean code * feature: clean code * feature: 增加高级匿名化算子
This commit is contained in:
@@ -59,43 +59,45 @@ VALUES ('26ae585c-8310-4679-adc0-e53215e6e69b', '文本清洗模板', '文本清
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', '图片清洗模板', '图片清洗模板');
|
||||
|
||||
INSERT IGNORE INTO t_operator_instance(instance_id, operator_id, op_index, settings_override)
|
||||
VALUES ('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithShortOrLongLengthFilter', 2, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighRepeatWordRateFilter', 3, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighRepeatPhraseRateFilter', 4, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighSpecialCharRateFilter', 5, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithManySensitiveWordsFilter', 6, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'UnicodeSpaceCleaner', 7, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'ExtraSpaceCleaner', 8, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FullWidthCharacterCleaner', 9, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'InvisibleCharactersCleaner', 10, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'ContentCleaner', 11, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'LegendCleaner', 12, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'EmojiCleaner', 13, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'HtmlTagCleaner', 14, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'TraditionalChineseCleaner', 15, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'GrableCharactersCleaner', 16, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'XMLTagCleaner', 17, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'DuplicateSentencesFilter', 18, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'DuplicateFilesFilter', 19, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'SexualAndViolentWordCleaner', 20, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'PoliticalWordCleaner', 21, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedPhoneNumber', 22, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedCreditCardNumber', 23, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'EmailNumberCleaner', 24, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedIpAddress', 25, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedIdNumber', 26, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedUrlCleaner', 27, null);
|
||||
VALUES ('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithShortOrLongLengthFilter', 1, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighRepeatWordRateFilter', 2, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighRepeatPhraseRateFilter', 3, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithHighSpecialCharRateFilter', 4, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FileWithManySensitiveWordsFilter', 5, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'UnicodeSpaceCleaner', 6, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'ExtraSpaceCleaner', 7, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'FullWidthCharacterCleaner', 8, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'InvisibleCharactersCleaner', 9, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'ContentCleaner', 10, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'LegendCleaner', 11, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'EmojiCleaner', 12, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'HtmlTagCleaner', 13, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'TraditionalChineseCleaner', 14, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'GrableCharactersCleaner', 15, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'XMLTagCleaner', 16, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'DuplicateSentencesFilter', 17, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'DuplicateFilesFilter', 18, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'SexualAndViolentWordCleaner', 19, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'PoliticalWordCleaner', 20, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedPhoneNumber', 21, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedCreditCardNumber', 22, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'EmailNumberCleaner', 23, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedIpAddress', 24, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedIdNumber', 25, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'AnonymizedUrlCleaner', 26, null),
|
||||
('26ae585c-8310-4679-adc0-e53215e6e69b', 'PiiDetector', 27, null);
|
||||
|
||||
INSERT IGNORE INTO t_operator_instance(instance_id, operator_id, op_index, settings_override)
|
||||
VALUES ('4421504e-c6c9-4760-b55a-509d17429597', 'ImgBlurredImagesCleaner', 2, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgDuplicatedImagesCleaner', 3, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSimilarImagesCleaner', 4, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgBrightness', 5, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgContrast', 6, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSaturation', 7, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSharpness', 8, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgDenoise', 9, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgShadowRemove', 10, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgPerspectiveTransformation', 11, null),
|
||||
VALUES ('4421504e-c6c9-4760-b55a-509d17429597', 'ImgBlurredImagesCleaner', 1, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgDuplicatedImagesCleaner', 2, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSimilarImagesCleaner', 3, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgBrightness', 4, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgContrast', 5, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSaturation', 6, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgSharpness', 7, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgDenoise', 8, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgShadowRemove', 9, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgPerspectiveTransformation', 10, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgDirectionCorrect', 11, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgResize', 12, null),
|
||||
('4421504e-c6c9-4760-b55a-509d17429597', 'ImgTypeUnify', 13, null);
|
||||
Reference in New Issue
Block a user