updated gufi and parquet backends to use new artifact handler names

Author: Vedant1

dsi/backends/gufi.py

@@ -15,30 +15,26 @@ class Gufi(Filesystem):
     '''
     GUFI Datastore
     '''
-
     prefix = ""
     index = ""
     dbfile = ""
     table = ""
     column = ""
     isVerbose = False
 
-    """
-    prefix: prefix to GUFI commands
-    index: directory with GUFI indexes
-    dbfile: sqlite db file from DSI
-    table: table name from the DSI db we want to join on
-    column: column name from the DSI db to join on
-    """
-
     def __init__(self, prefix, index, dbfile, table, column, verbose=False):
         '''
-        prefix: prefix to GUFI commands
-        index: directory with GUFI indexes
-        dbfile: sqlite db file from DSI
-        table: table name from the DSI db we want to join on
-        column: column name from the DSI db to join on
-        verbose: print debugging statements or not
+        `prefix`: prefix to GUFI commands
+
+        `index`: directory with GUFI indexes
+
+        `dbfile`: sqlite db file from DSI
+
+        `table`: table name from the DSI db we want to join on
+
+        `column`: column name from the DSI db to join on
+
+        `verbose`: print debugging statements or not
         '''
 
         super().__init__(dbfile)
@@ -55,8 +51,12 @@ def __init__(self, prefix, index, dbfile, table, column, verbose=False):
 
         self.isVerbose = verbose
 
-    # Query GUFI and DSI db
+    # OLD NAME OF query_artifacts(). TO BE DEPRECATED IN FUTURE DSI RELEASE
     def get_artifacts(self, query):
+        return self.query_artifacts(query)
+    
+    # Query GUFI and DSI db
+    def query_artifacts(self, query):
         '''
         Retrieves GUFI's metadata joined with a dsi database
         query: an sql query into the dsi_entries table
@@ -68,7 +68,11 @@ def get_artifacts(self, query):
 
         return resout
 
+    # OLD NAME OF ingest_artifacts(). TO BE DEPRECATED IN FUTURE DSI RELEASE
     def put_artifacts(self, query):
+        return self.ingest_artifacts(query)
+    
+    def ingest_artifacts(self, query):
         pass
 
     # Runs the gufi query command

dsi/backends/parquet.py

@@ -20,14 +20,22 @@ def __init__(self, filename, **kwargs):
         except KeyError:
             self.compression = None
 
+    # OLD NAME OF query_artifacts(). TO BE DEPRECATED IN FUTURE DSI RELEASE
     def get_artifacts(self):
-        """Get Parquet data from filename."""
+        return self.query_artifacts()
+    
+    def query_artifacts(self):
+        """Query Parquet data from filename."""
         table = pq.read_table(self.filename)
         resout = table.to_pydict()
         return resout
 
+    # OLD NAME OF ingest_artifacts(). TO BE DEPRECATED IN FUTURE DSI RELEASE
     def put_artifacts(self, collection):
-        """Put artifacts into file at filename path."""
+        return self.ingest_artifacts(collection)
+    
+    def ingest_artifacts(self, collection):
+        """Ingest artifacts into file at filename path."""
         table = pa.table(collection)
         pq.write_table(table, self.filename, compression=self.compression)
 
@@ -43,7 +51,12 @@ def get_cmd_output(cmd: list) -> str:
             raise Exception(proc.stderr)
         return proc.stdout.strip().decode("utf-8")
 
+    # OLD NAME OF notebook(). TO BE DEPRECATED IN FUTURE DSI RELEASE
     def inspect_artifacts(self, collection, interactive=False):
+        return self.notebook(collection, interactive)
+    
+    def notebook(self, collection, interactive=False):
+        """Generate Jupyter notebook of Parquet data from filename."""
         import nbconvert as nbc
         import nbformat as nbf
 

dsi/backends/tests/test_gufi.py

@@ -11,6 +11,6 @@ def test_artifact_query():
     column = "sample_col"
     store = Gufi(prefix, index, dbpath, table, column, isVerbose)
     sqlstr = "select * from dsi_entries"
-    rows = store.get_artifacts(sqlstr)
+    rows = store.query_artifacts(sqlstr)
     store.close()
     assert len(rows) > 0

dsi/backends/tests/test_parquet.py

@@ -11,19 +11,19 @@ def get_git_root(path):
     git_root = git_repo.git.rev_parse("--show-toplevel")
     return (git_root)
 
-def test_get_artifacts():
+def test_query_artifacts():
     a = Parquet(filename='/'.join([get_git_root('.'), 'dsi/data/wildfiredata.pq']))
-    b = a.get_artifacts()
+    b = a.query_artifacts()
     cnt = 0
     for key in b:
         cnt = cnt + 1
         assert 4 == len(b[key])
     assert 11 == cnt
 
-def test_inspect_artifact():
+def test_notebook():
     a = Parquet(filename='/'.join([get_git_root('.'), 'dsi/data/wildfiredata.pq']))
-    b = a.get_artifacts()
-    a.inspect_artifacts(b)
-    # No error on inspect_artifact return implies success
+    b = a.query_artifacts()
+    a.notebook(b)
+    # No error on notebook return implies success
     assert True