fix: dependency and restructure fixes (#343)

* fix: 1355 dependency and restructure fix * Fix: add build.yml for github actions * Fix: run build on push * chore: add pre-commit hooks: isort, black, flake8, pydocstyle (#1412) * style: applied isort, black * fix: flake8 config * fix: removing stale directories * feat: add additional pre-commit flags, updated make * refactor: archived old code * ci: #1411 move from CircleCI to GitHub Actions * refactor: resolve flake8 issues * fix: unify Makefile commands and pre-commit hooks * docs: add requirements.txt for readthedocs * fix: resolve missing import and flake8 issues * test: update stain_texture test fname * test: removing unneeded param for stain fx, updating test * test: update assertion statement * Fix: utils unit test * fix: deprecated numpy type * feat: add scale_factor to dsa bitmask-polygon * test: fixing binary fieldname in parquet file, adding back schema * fix: build readthedocs with poetry * Revert "docs: add requirements.txt for readthedocs" This reverts commit 6424674. Co-authored-by: kohlia <[email protected]>
msk-mind · Dec 22, 2022 · dcfcb8b · dcfcb8b
1 parent a9c06cc
commit dcfcb8b
Show file tree

Hide file tree

Showing 1,125 changed files with 19,397 additions and 272,438 deletions.
diff --git a/.archive/GraphEnum.py b/.archive/GraphEnum.py
@@ -1,97 +1,165 @@
 from enum import Enum
-
+
+
 class NodeType(object):
-	"""
-	NodeType object defines the schema used to populate a graph node.
+    """
+    NodeType object defines the schema used to populate a graph node.
 
-	:param: node_type: node type. e.g. scan
-	:param: name: required node name. e.g. scan-123
-	:param: schema: list of column names. e.g. slide_id
-	"""
-	def __init__(self, node_type, name, schema=[]):
+    :param: node_type: node type. e.g. scan
+    :param: name: required node name. e.g. scan-123
+    :param: schema: list of column names. e.g. slide_id
+    """
 
-		self.type = node_type
-		self.name = name
-		self.schema = schema
+    def __init__(self, node_type, name, schema=[]):
 
+        self.type = node_type
+        self.name = name
+        self.schema = schema
 
-	def get_all_schema(self):
-		"""
-		Name is a required field, but it's still a property of this node.
-		Return the properties including the name property!
-		"""
-		return self.schema + [self.name]
+    def get_all_schema(self):
+        """
+        Name is a required field, but it's still a property of this node.
+        Return the properties including the name property!
+        """
+        return self.schema + [self.name]
 
 
 class Graph(object):
-	"""
-	Graph object that stores src-[relationship]-target information.
-	This object is used get data from existing delta tables to populate the graph database.
+    """
+    Graph object that stores src-[relationship]-target information.
+    This object is used get data from existing delta tables to populate the graph database.
+
+    :param: src: NodeType - source node
+    :param: relationship: str - relationship name e.g. HAS_PX, HAS_RECORD etc
+    :param: target: NodeType - target node
+    """
 
-	:param: src: NodeType - source node
-	:param: relationship: str - relationship name e.g. HAS_PX, HAS_RECORD etc
-	:param: target: NodeType - target node
-	"""
-	def __init__(self, src, relationship, target):
+    def __init__(self, src, relationship, target):
 
-		self.src = src
-		self.relationship = relationship
-		self.target = target
+        self.src = src
+        self.relationship = relationship
+        self.target = target
 
 
 class GraphEnum(Enum):
-	"""
-	Defines Graph relationships. 
-	We assume that all properties come from the table defined in GraphEnum.name
-
-	name: data type
-	value: list of Graphs - to accomodate multiple relationship update
-
-	>>> GraphEnum['DICOM'].value[0].src.type
-	'patient'
-	"""
-	accession_radiology_dicom = NodeType("accession", "metadata.AccessionNumber")
-	accession_radiology = NodeType("accession", "accession_number")
-	scan = NodeType("scan", "metadata.SeriesInstanceUID")
-
-	# radiology
-	DICOM = [Graph(NodeType("patient", "metadata.PatientID"),
-				"HAS_CASE", 
-				accession_radiology_dicom),
-			Graph(accession_radiology_dicom,
-				"HAS_SCAN", 
-				scan), 
-			Graph(scan,
-				"HAS_DATA", 
-				NodeType("DicomSeries", "metadata.SeriesInstanceUID", ["path"]))]
-
-	MHA = [Graph(accession_radiology,
-				"HAS_DATA",
-				NodeType("VolumetricLabel", "scan_annotation_record_uuid", ["path", "label"]))]
-
-	MHD = [Graph(accession_radiology,
-				"HAS_DATA",
-				NodeType("VolumetricImage", "scan_annotation_record_uuid", ["path"]))]
-
-	PNG = [Graph(scan, "HAS_DATA", NodeType("DicomImageSeries", "png_record_uuid",     ["metadata.SeriesInstanceUID", "label"]))]
-	FEATURE = [Graph(scan, "HAS_DATA", NodeType("DicomImageSeries", "feature_record_uuid", ["metadata.SeriesInstanceUID", "label"]))]
-
-	# pathology
-	slide = NodeType("slide", "slide_id", ["sv_project_id","slideviewer_path"])
-	REGIONAL_BITMASK = [Graph(slide,
-					 "HAS_DATA",
-					 NodeType("RegionalAnnotationBitmap", "bmp_record_uuid", ["user","date_updated","latest"]))]
-
-	# regional concat geojson table contains regional geojson table + concatenated jsons
-	REGIONAL_CONCAT_GEOJSON = [Graph(slide,
-								 "HAS_DATA",
-								 NodeType("RegionalAnnotationJSON", "concat_geojson_record_uuid", ["labelset","date_updated","latest"]))]
-
-
-	POINT_RAW_JSON = [Graph(slide,
-					"HAS_DATA",
-					NodeType("PointAnnotation", "sv_json_record_uuid", ["user","date_updated","latest"]))]
-
-	POINT_GEOJSON = [Graph(slide,
-							"HAS_DATA",
-							NodeType("PointAnnotationJson", "geojson_record_uuid", ["labelset","date_updated","latest"]))]
+    """
+    Defines Graph relationships.
+    We assume that all properties come from the table defined in GraphEnum.name
+
+    name: data type
+    value: list of Graphs - to accomodate multiple relationship update
+
+    >>> GraphEnum['DICOM'].value[0].src.type
+    'patient'
+    """
+
+    accession_radiology_dicom = NodeType("accession", "metadata.AccessionNumber")
+    accession_radiology = NodeType("accession", "accession_number")
+    scan = NodeType("scan", "metadata.SeriesInstanceUID")
+
+    # radiology
+    DICOM = [
+        Graph(
+            NodeType("patient", "metadata.PatientID"),
+            "HAS_CASE",
+            accession_radiology_dicom,
+        ),
+        Graph(accession_radiology_dicom, "HAS_SCAN", scan),
+        Graph(
+            scan,
+            "HAS_DATA",
+            NodeType("DicomSeries", "metadata.SeriesInstanceUID", ["path"]),
+        ),
+    ]
+
+    MHA = [
+        Graph(
+            accession_radiology,
+            "HAS_DATA",
+            NodeType(
+                "VolumetricLabel", "scan_annotation_record_uuid", ["path", "label"]
+            ),
+        )
+    ]
+
+    MHD = [
+        Graph(
+            accession_radiology,
+            "HAS_DATA",
+            NodeType("VolumetricImage", "scan_annotation_record_uuid", ["path"]),
+        )
+    ]
+
+    PNG = [
+        Graph(
+            scan,
+            "HAS_DATA",
+            NodeType(
+                "DicomImageSeries",
+                "png_record_uuid",
+                ["metadata.SeriesInstanceUID", "label"],
+            ),
+        )
+    ]
+    FEATURE = [
+        Graph(
+            scan,
+            "HAS_DATA",
+            NodeType(
+                "DicomImageSeries",
+                "feature_record_uuid",
+                ["metadata.SeriesInstanceUID", "label"],
+            ),
+        )
+    ]
+
+    # pathology
+    slide = NodeType("slide", "slide_id", ["sv_project_id", "slideviewer_path"])
+    REGIONAL_BITMASK = [
+        Graph(
+            slide,
+            "HAS_DATA",
+            NodeType(
+                "RegionalAnnotationBitmap",
+                "bmp_record_uuid",
+                ["user", "date_updated", "latest"],
+            ),
+        )
+    ]
+
+    # regional concat geojson table contains regional geojson table + concatenated jsons
+    REGIONAL_CONCAT_GEOJSON = [
+        Graph(
+            slide,
+            "HAS_DATA",
+            NodeType(
+                "RegionalAnnotationJSON",
+                "concat_geojson_record_uuid",
+                ["labelset", "date_updated", "latest"],
+            ),
+        )
+    ]
+
+    POINT_RAW_JSON = [
+        Graph(
+            slide,
+            "HAS_DATA",
+            NodeType(
+                "PointAnnotation",
+                "sv_json_record_uuid",
+                ["user", "date_updated", "latest"],
+            ),
+        )
+    ]
+
+    POINT_GEOJSON = [
+        Graph(
+            slide,
+            "HAS_DATA",
+            NodeType(
+                "PointAnnotationJson",
+                "geojson_record_uuid",
+                ["labelset", "date_updated", "latest"],
+            ),
+        )
+    ]