#1017

opteryx/managers/expression/__init__.py

@@ -30,85 +30,13 @@
 from opteryx.models.node import Node
 from opteryx.third_party.pyarrow_ops.ops import filter_operations
 
+from .format_expression import format_expression
+
 # These are bit-masks
 LOGICAL_TYPE: int = int("00010000", 2)
 INTERNAL_TYPE: int = int("00100000", 2)
 
 
-def format_expression(root):
-    """
-    TODO:
-        - change so literals are called `column_1`, `column_2`
-        - change so expressions are called `1_times_7`
-    """
-
-    if root is None:
-        return "null"
-
-    if isinstance(root, list):
-        return [format_expression(item) for item in root]
-
-    node_type = root.node_type
-    _map: dict = {}
-
-    # LITERAL TYPES
-    if node_type == NodeType.LITERAL:
-        literal_type = root.type
-        if literal_type == OrsoTypes.VARCHAR:
-            return "'" + root.value.replace("'", "'") + "'"
-        if literal_type == OrsoTypes.TIMESTAMP:
-            return "'" + str(root) + "'"
-        if literal_type == OrsoTypes.INTERVAL:
-            return "<INTERVAL>"
-        return str(root.value)
-    # INTERAL IDENTIFIERS
-    if node_type & INTERNAL_TYPE == INTERNAL_TYPE:
-        if node_type in (NodeType.FUNCTION, NodeType.AGGREGATOR):
-            if root.value == "CASE":
-                con = [format_expression(a) for a in root.parameters[0].value]
-                vals = [format_expression(a) for a in root.parameters[1].value]
-                return "CASE " + "".join([f"WHEN {c} THEN {v} " for c, v in zip(con, vals)]) + "END"
-            distinct = "DISTINCT " if root.distinct else ""
-            order = ""
-            if root.order:
-                order = f" ORDER BY {', '.join(item[0].value + (' DESC' if not item[1] else '') for item in (root.order or []))}"
-            if root.value == "ARRAY_AGG":
-                limit = f" LIMIT {root.limit}" if root.limit else ""
-                return f"{root.value.upper()}({distinct}{format_expression(root.expression)}{order}{limit})"
-            return f"{root.value.upper()}({distinct}{','.join([format_expression(e) for e in root.parameters])}{order})"
-        if node_type == NodeType.WILDCARD:
-            return "*"
-        if node_type == NodeType.BINARY_OPERATOR:
-            _map = {
-                "StringConcat": "||",
-                "Plus": "+",
-                "Minus": "-",
-                "Multiply": "*",
-                "Divide": "/",
-                "MyIntegerDivide": "div",
-            }
-            return f"{format_expression(root.left)} {_map.get(root.value, root.value).upper()} {format_expression(root.right)}"
-    if node_type == NodeType.COMPARISON_OPERATOR:
-        _map = {"Eq": "=", "Lt": "<", "Gt": ">", "NotEq": "!=", "BitwiseOr": "|"}
-        return f"{format_expression(root.left)} {_map.get(root.value, root.value).upper()} {format_expression(root.right)}"
-    if node_type == NodeType.UNARY_OPERATOR:
-        _map = {"IsNull": "%s IS NULL", "IsNotNull": "%s IS NOT NULL"}
-        return _map.get(root.value, root.value + "(%s)").replace(
-            "%s", format_expression(root.centre)
-        )
-    if node_type == NodeType.NOT:
-        return f"NOT {format_expression(root.centre)}"
-    if node_type in (NodeType.AND, NodeType.OR, NodeType.XOR):
-        _map = {
-            NodeType.AND: "AND",
-            NodeType.OR: "OR",
-            NodeType.XOR: "XOR",
-        }  # type:ignore
-        return f"{format_expression(root.left)} {_map[node_type]} {format_expression(root.right)}"
-
-    return str(root.value)
-
-
 class NodeType(int, Enum):
     """
     The types of Nodes we will see.

opteryx/managers/expression/format_expression.py

@@ -0,0 +1,83 @@
+from enum import Enum
+
+import numpy
+import pyarrow
+from orso.types import OrsoTypes
+from pyarrow import Table
+
+from opteryx.functions import FUNCTIONS
+from opteryx.functions.binary_operators import binary_operations
+from opteryx.functions.unary_operations import UNARY_OPERATIONS
+from opteryx.models.node import Node
+from opteryx.third_party.pyarrow_ops.ops import filter_operations
+
+
+def format_expression(root):
+    # circular imports
+    from . import NodeType, INTERNAL_TYPE, LOGICAL_TYPE
+
+    if root is None:
+        return "null"
+
+    if isinstance(root, list):
+        return [format_expression(item) for item in root]
+
+    node_type = root.node_type
+    _map: dict = {}
+
+    # LITERAL TYPES
+    if node_type == NodeType.LITERAL:
+        literal_type = root.type
+        if literal_type == OrsoTypes.VARCHAR:
+            return "'" + root.value.replace("'", "'") + "'"
+        if literal_type == OrsoTypes.TIMESTAMP:
+            return "'" + str(root) + "'"
+        if literal_type == OrsoTypes.INTERVAL:
+            return "<INTERVAL>"
+        return str(root.value)
+    # INTERAL IDENTIFIERS
+    if node_type & INTERNAL_TYPE == INTERNAL_TYPE:
+        if node_type in (NodeType.FUNCTION, NodeType.AGGREGATOR):
+            if root.value == "CASE":
+                con = [format_expression(a) for a in root.parameters[0].value]
+                vals = [format_expression(a) for a in root.parameters[1].value]
+                return "CASE " + "".join([f"WHEN {c} THEN {v} " for c, v in zip(con, vals)]) + "END"
+            distinct = "DISTINCT " if root.distinct else ""
+            order = ""
+            if root.order:
+                order = f" ORDER BY {', '.join(item[0].value + (' DESC' if not item[1] else '') for item in (root.order or []))}"
+            if root.value == "ARRAY_AGG":
+                limit = f" LIMIT {root.limit}" if root.limit else ""
+                return f"{root.value.upper()}({distinct}{format_expression(root.expression)}{order}{limit})"
+            return f"{root.value.upper()}({distinct}{','.join([format_expression(e) for e in root.parameters])}{order})"
+        if node_type == NodeType.WILDCARD:
+            return "*"
+        if node_type == NodeType.BINARY_OPERATOR:
+            _map = {
+                "StringConcat": "||",
+                "Plus": "+",
+                "Minus": "-",
+                "Multiply": "*",
+                "Divide": "/",
+                "MyIntegerDivide": "div",
+            }
+            return f"{format_expression(root.left)} {_map.get(root.value, root.value).upper()} {format_expression(root.right)}"
+    if node_type == NodeType.COMPARISON_OPERATOR:
+        _map = {"Eq": "=", "Lt": "<", "Gt": ">", "NotEq": "!=", "BitwiseOr": "|"}
+        return f"{format_expression(root.left)} {_map.get(root.value, root.value).upper()} {format_expression(root.right)}"
+    if node_type == NodeType.UNARY_OPERATOR:
+        _map = {"IsNull": "%s IS NULL", "IsNotNull": "%s IS NOT NULL"}
+        return _map.get(root.value, root.value + "(%s)").replace(
+            "%s", format_expression(root.centre)
+        )
+    if node_type == NodeType.NOT:
+        return f"NOT {format_expression(root.centre)}"
+    if node_type in (NodeType.AND, NodeType.OR, NodeType.XOR):
+        _map = {
+            NodeType.AND: "AND",
+            NodeType.OR: "OR",
+            NodeType.XOR: "XOR",
+        }  # type:ignore
+        return f"{format_expression(root.left)} {_map[node_type]} {format_expression(root.right)}"
+
+    return str(root.value)

tests/misc/test_connection.py

@@ -45,7 +45,6 @@ def test_execute():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_connection()
-    test_execute()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_connection_arrow.py

@@ -47,8 +47,6 @@ def test_direct_as_arrow_no_limit():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_as_arrow_no_limit()
-    test_as_arrow_with_limit()
-    test_direct_as_arrow_no_limit()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_connection_errors.py

@@ -52,8 +52,6 @@ def test_fetching():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_connection_warnings()
-    test_connection_parameter_mismatch()
-    test_fetching()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_connection_pandas.py

@@ -34,7 +34,6 @@ def test_as_pandas_with_limit():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_as_pandas_no_limit()
-    test_as_pandas_with_limit()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_connection_polars.py

@@ -30,7 +30,6 @@ def test_as_polars_with_limit():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_as_polars_no_limit()
-    test_as_polars_with_limit()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_connector_prefixes.py

@@ -45,6 +45,6 @@ def test_connector_prefixes():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_connector_prefixes()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_cursor.py

@@ -119,12 +119,6 @@ def test_execute_error():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_execute()
-    test_rowcount()
-    test_shape()
-    test_fetchone()
-    test_fetchmany()
-    test_fetchall()
-    test_execute_error()
-
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_date_trunc.py

@@ -65,11 +65,6 @@ def test_truncate_to_week():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_truncate_to_day()
-    test_truncate_to_hour()
-    test_truncate_to_minute()
-    test_truncate_to_second()
-    test_truncate_to_week()
-    test_truncate_to_year()
-
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_documentation.py

@@ -186,18 +186,6 @@ def get_user_permissions(user_roles):
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_documentation_connect_example()
-    test_readme_1()  # execute-a-simple-query-in-python
-    test_readme_2()  # execute-sql-on-a-pandas-dataframe
-    test_readme_3()  # query-data-on-local-disk
-    test_readme_4()  # query-data-on-gcs
-    test_readme_5()  # query-data-in-sqlite
-    test_get_started()
-    test_pandas_integration_input()
-    test_pandas_integration_output()
-    test_polars_integration_input()
-    test_polars_integration_output()
-    test_permissions_example()
-    test_role_based_permissions()
-
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_lruk.py

@@ -143,7 +143,6 @@ def test_lru_cache_eviction():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_lruk()
-    test_lru2()
-    test_lru_cache_eviction()
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_node.py

@@ -33,5 +33,6 @@ def test_node():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_node()
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_node_formatter.py

@@ -0,0 +1,24 @@
+import os
+import sys
+
+sys.path.insert(1, os.path.join(sys.path[0], "../.."))
+
+from orso.types import OrsoTypes
+
+from opteryx.managers.expression import format_expression
+from opteryx.managers.expression import NodeType
+from opteryx.models.node import Node
+
+
+def test_format_nodes():
+    left_node = Node(NodeType.LITERAL, type=OrsoTypes.INTEGER, value=1)
+    right_node = Node(NodeType.LITERAL, type=OrsoTypes.DOUBLE, value=1.1)
+
+    print(format_expression(left_node))
+    print(format_expression(right_node))
+
+
+if __name__ == "__main__":  # pragma: no cover
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_pandas_integration.py

@@ -42,7 +42,6 @@ def test_documentation():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_pandas()
-    test_documentation()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_polars_integration.py

@@ -26,6 +26,6 @@ def test_polars():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_polars()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_pyarrow_ops_join.py

@@ -29,6 +29,6 @@ def test_hash_join_consistency():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_hash_join_consistency()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_relation_schema.py

@@ -33,7 +33,6 @@ def test_dictionary_column():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_constant_column()
-    test_dictionary_column()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()

tests/misc/test_formatter.py → tests/misc/test_sql_formatter.py

@@ -15,5 +15,7 @@ def test_format_sql():
     ), (str(formatted_sql.encode()) + "\n" + formatted_sql)
 
 
-if __name__ == "__main__":
-    test_format_sql()
+if __name__ == "__main__":  # pragma: no cover
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_string_functions.py

@@ -54,7 +54,6 @@ def test_random_string():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_slice_left()
-    test_slice_right()
-    test_random_string()
-    print("✅ okay")
+    from tests.tools import run_tests
+
+    run_tests()

tests/misc/test_suggestions.py

@@ -20,6 +20,6 @@ def test_hint_hints():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    test_hint_hints()
+    from tests.tools import run_tests
 
-    print("✅ okay")
+    run_tests()