Merge branch 'main' into feature/hive-loader

GSK-Biostatistics · Dec 8, 2022 · 6fa2faf · 6fa2faf
2 parents 8f94b38 + a07f607
commit 6fa2faf
Show file tree

Hide file tree

Showing 42 changed files with 148 additions and 39,672 deletions.
diff --git a/README.md b/README.md
@@ -21,4 +21,4 @@ QUERY BUILDERS
 - QueryBuilder - Class to support creation of cypher queries to work with data in Neo4j  
 
 Dependencies:
-- https://mygithub.gsk.com/gsk-tech/neointerface
+- https://github.com/GSK-Biostatistics/neointerface
diff --git a/data_loaders/azure_data_loader.py b/data_loaders/azure_data_loader.py
@@ -46,7 +46,7 @@ def listdir(self, path: str) -> [str]:
         Lists all files (full path) in a directory on Azure blob storage
 
         :param path: A list of strings
-                     # EXAMPLE of list item: "rd/space/test_compound/117106/testing_01/sdtm/gsk_117106_vs.sas7bdat"
+                     # EXAMPLE of list item: "rd/space/test_compound/trial_01/testing_01/sdtm/vs.sas7bdat"
         """
         return [path.name for path in self.fs.get_paths(path=path)]
 

diff --git a/data_loaders/sql_server_data_loader.py b/data_loaders/sql_server_data_loader.py
@@ -15,13 +15,8 @@ def __init__(self,
         self._con = f"jdbc:sqlserver://{sqldburl};databaseName={self.sqldbname};user={self.sqluser};password={self._sqlpassword};"
         super().__init__(*args, **kwargs)
         if self.autoconnect:
-            self.remove_apoc_custom("load_tables_from_sql")
-            self.setup_apoc_custom_load_tables_from_sql()
             self.query("call db.clearQueryCaches()")
 
-            self.create_index("Source Data Row", "_study_")
-            self.create_index("Source Data Row", "_domain_")
-
     def sql_query(self, sqlq:str):
         q = "CALL apoc.load.jdbc($con, $sqlq) YIELD row RETURN row"
         params = {

diff --git a/data_providers/archive/data_provider.py b/data_providers/archive/data_provider.py
diff --git a/data_providers/archive/sdtm_data_provider.py b/data_providers/archive/sdtm_data_provider.py
diff --git a/data_providers/data_provider.py b/data_providers/data_provider.py
@@ -25,7 +25,7 @@ def __init__(self, *args, **kwargs):
 
     def check_schema(self, labels: list, rels: list):
         for label in labels:
-            q = "MATCH " + self.qb.generate_1match_schema_check(label) + " RETURN *"
+            q = "MATCH " + self.qb.generate_1match_schema_check(label=label) + " RETURN *"
             res = self.query(q)
             assert res, f"Class {label} not found in the schema"
         for rel in rels:

diff --git a/examples/refactor_all.py b/examples/refactor_all.py
diff --git a/examples/refactor_sdtm_metadata.py b/examples/refactor_sdtm_metadata.py
diff --git a/install_from_local_repos.bat b/install_from_local_repos.bat