* Adding methods to enable plotting of custom bboxes
* Adding test for get_pii_bboxes
* Linting fixes and addition of test for add_custom_bboxes
* Adding use of custom bbox into DICOM verification engine
* Fixing tests for the DICOM verification engine
* Adding in ocr kwargs compatibility and updating tests
* Adding example notebook
* Linting fixes
* Remaining linting fix
---------
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Enabling use of ad-hoc recognizers in verifier
* Adding support to standard image verification engine as well
* Linting fix
* Removing redundant init
* Removing unused import
* Adding method to augment words more thoroughly
* Adding unit test
* PR comments changes
* Linting fix
---------
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Enabling allow list approach
* Adding in empty list for allow_list in existing unit tests
* Added unit tests for newly introduced methods
* Adding unit test for allow list functionality
* Linting fixes
* Removing spaces in empty lines
* Fix integration test not accounting for empty space removal
* Updating notebook with more examples and adding ad_hoc_recognizers approach to standard image redactor engine as well
* Linting fixes
* Removing incomplete example code
* Fixing section header numbers
* Removing duplicate comment
---------
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Adding in ability to select redact approach
* Adding redact_approach arg into tests so they pass correctly
* Adding test for _get_analyzer_results
* Linting fixes
* Additional linting fixes
* Making default approach the default
* Linting fix
* Replacing patch.object() with patch() in modified tests
* Fixed mocker patch for new _get_analyzer_results in redact_and_return_bbox
* Removing old assertions brought over from incorrect merge conflict resolution
* Replacing call_count == statements with assert_called_once etc
* Replacing in-test instantiation with passing in mock_engine
* Suggested change from PR comment. redact_approach replaced with use_metadata and ability to pass in ad_hoc_recognizers
* Linting fixes
* Additional linting fixes
* Remove output that is not checked
* Commenting out whole unit test file to see impact on build pipeline hangup
* Only commenting out get_analyzer_results tests
* Only commenting out happy path test for get_analyzer_results
* Changing ad_hoc_recognizers type check to raise only TypeError
* Removing type assertion for exception test
* Reintroduce the happy path test for get_analyzer_results
* Commenting out exception test and keeping in happy path test
* Removing constants from parameterize for exception test
* Adding argument= in call to get_analyzer_results in happy path
* Enable return of bboxes used to redact pixels
* Adding return_bboxes arg values into existing tests
* Adding test for return_bbox==True condition
* Adding test for _save_bbox_json()
* Making argument name more clear
* Creating separate method to return redacted image and bboxes
* Linting fix
* Removing Union return type
* Commenting out DICOM verification engine intergration test to see if that is still the cause of unit test hangup
* Renaming test and removing redundancy in unit test for dicom image redactor
* Fixing duplication of call to a single file likely from main merges
* Removing extra cases for redact() test
* Changing mocked return type from None to an empty list
* Commenting out full unit test for redact to see effect on PR build hangup
* Reintroduce verify integration test and non-parameterized redact test
* Commenting out threshold and expected length test to see impact on PR build hang-up
* Undo comment out of image analyzer engine test
* Commenting out all unit tests for dicom image redactor engine
* Comment out unit test for redact()
* Fixing typing
* Commenting out exception test for redact_and_return_bbox
* Updated how exceptions are handled for redact_and_return_bbox, return all unit tests
* Adding IsADirectoryError exception type
* Commenting out happy path test for redact_and_return_bbox
* Commenting out compressed and icon_image_sequence DICOM test input images for redact_and_return_bbox happy path test
* Commenting out the type assertions in happy path test for redact_and_return_bbox
* Commenting out the call count assertions in happy path for redact_and_return_bbox
* Update type assertion and comment out all mocking and mocking assertions for happy path test for redact_and_return_bbox
* Commenting out all assertions in happy path test for redact_and_return_bbox
* Replacing mocker.patch with mocker.patch.object for all mocked methods in happy path test for redact_and_return_bbox
* Changing all mocker.patch.object calls into mocker.patch for happy path test for redact_and_return_bbox
* Reintroduce assertions for happy path test for redact_and_return_bbox
* Turning off assertions for call count again for happy path for redact_and_return_bbox
* Making assertion for returned bbox type even more explicit for happy path test for redact_and_return_bbox
* Turning off type assertions and turning on mock call count assertions for happy path test for redact_and_return_bbox
* Replacing call count assertions with assert_called_once
* Reintroducing type assertions and changing return_value to include some placeholder mock data instead of being empty dictionaries in list
* Comment out the image type assertion
* Turning on image type assertion and turning off bbox type assertions
* Removing assertion for dict
* Using isinstance instead of type ==
* Removing assertions for bbox type
---------
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Adding in methods for compression and modifying existing unit test for adding redact boxes
* Adding unit tests
* Linting fixes
* Fixing exception type
* Adding in python-gdcm dependency
* Adding python-gdcm to piplock
* Updating pipfile.lock
* Switching from gdcm license to python-gdcm license
* Adding in methods for compression and modifying existing unit test for adding redact boxes
* Adding unit tests
* Linting fixes
* Fixing exception type
* Adding in python-gdcm dependency
* Switching from gdcm license to python-gdcm license
* fix ammend
* comment out new tests
* revert
* Incorporating bug fix that was merged in
* Linting fix
* Temporarily commenting out one integration test to see effect on build pipeline hangup
* Adding _strip_score back in
* Commenting out entire integration test for DICOM Image PII verify engine
* Trying alternate assertion for test_eval_dicom_correctly
* Removing all assertions in test_eval_dicom_correctly
* Explicitly including argument names and adding some temporary print statements
* Using deepcopy of passed-in mock DICOM verify results
* Using deepcopy for the passed-in example instance as well
* Renaming any variables that may have overlap with existing variables in the method being test
* Testing effect of mocking verify_dicom_instance in eval_dicom_instance
* Mocking all calls except get_bboxes_from_ocr_results in test_eval_dicom_correctly
* Removing call to method and assertions
* Commenting out test_verify_correctly
* Adding test_verify_correctly back and reducing ambiguity as much as possible
* Removing ambiguity even more
* Changing _strip_score back to not returning but keeping input arg name change
* Commenting out assertions for test_verify_correctly
* Comment out test_verify_correctly assertions and full test_eval_dicom_correctly test
* Adding assertions back into test_verify_correctly
* Commenting out last assertion in test_verify_correctly
* Reformatting to avoid use of all in assert
* Using frozen set comparison for final assert in test_verify_correctly
* Removing final assertion from test_verify_correctly while keeping test_eval_dicom_correctly in
* Commenting out act and assert in test_verify_correctly while keeping test_eval_dicom_correctly in
* Using module-level mock variables and moving act call to less lines in test_verify_correctly
* Removing all assertions but keeping other sections for both tests
* Adding back in first assertions for each test
* Removing all assertions except fist image type assertion in verify test
* Removing all assertions except image type assertion in test_eval_dicom_correctly
* Remove image type assertions but keep all other assertions in
* Remove image type assertions and all assertions for eval test
* Only keep final assertion in verify test
* Only keep the second assertion in verify and nothing else
* Only keep final assertion in eval test
* Enable all assertions and move helper methods above test methods
* Simplifying assertions
* Removing unused code now that we have simplified assertions
* Reverting back but simplifying asserts another way
* Removing eval integration test since unit test for that covers functionality
* Remvoing analyzer results assertion
* Removing unused import
---------
Co-authored-by: Sharon Hart <sharonh.dev@gmail.com>
* Adding in handling for instances with icon image sequence data
* Linting fix
---------
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Adding exceptions for when DICOM file does not have pixel data
* Updating unit tests to accomodate new exception
* Fixing linting errors
* Adding in test image
* Fixing f-string
* Image redactor - REST API to support web application
* revert port
* lints
* Fix versioning, bump pillow and analyzer
* lower score
* Fix versioning, bump pillow and analyzer
* try to fix test
* add e2e test
Co-authored-by: sharon <sharon.hart@microsoft.com>
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Initial commit with old paths (will need to fix)
* Fixing paths
* Adding in notebook and updating documentation
* Adding in WIP notebook and docs
* Adding example notebook
* Clearing some output to reduce file size
* Linting fix v1
* Linting v2
* Add rounding to integration test to account for minor differences in real testing
* Fixing typo with wrong variable name
* Removing check for exact or rounded conf
* Updating logic for comparing ocr_results
* Minor fixes on DICOM verify code based on PR comments
* Changing results and gt format from dictionary to list of dictionaries
* Updating notebook
* Updating docstring format
* Linting fix
* Changing fixture scope for mock_engine
* Update docs/image-redactor/evaluating_dicom_redaction.md
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Docstring updates
* Improving efficiency of _remove_duplicate_entities
* Add handling for divide by zero in recall and precision calculations
* Making precision and recall calculation functions public
* Moving common fixtures to conftest
* PR comments
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
* Adding new function to get pixel array crop
* Adding unit tests
* Removing unused mocker
* Improve code efficiency of _get_array_corners
* Allow users to specify crop_ratio
* Clarifying and fixing docstrings
Co-authored-by: Omri Mendels <omri374@users.noreply.github.com>
Omri Mendels
Roey Ben Chaim
Sharon Hart
Aivaras Baranauskas
Roberto Aguiar Lima
Jenny Lee
Md Ashhar
Gord Lueck
ayabel
Nile Wilson
dependabot[bot]