Merge pull request #15 from microsoft/refactoring_flowquery_py

Refactoring flowquery py

Author: niclasko

.husky/pre-commit

@@ -1,5 +1,23 @@
 npx lint-staged
 
+# Run ruff and mypy on Python files in flowquery-py
+if git diff --cached --name-only | grep -q "^flowquery-py/.*\.py$"; then
+    echo "Running ruff on Python files..."
+    cd flowquery-py
+    conda run -n flowquery ruff check src --select=F401,E,W,I
+    if [ $? -ne 0 ]; then
+        echo "Ruff check failed. Please fix the issues before committing."
+        exit 1
+    fi
+    echo "Running mypy on Python files..."
+    conda run -n flowquery mypy src --no-error-summary
+    if [ $? -ne 0 ]; then
+        echo "Mypy check failed. Please fix the type errors before committing."
+        exit 1
+    fi
+    cd ..
+fi
+
 # Strip outputs from notebook files in flowquery-py
 for notebook in $(git ls-files 'flowquery-py/**/*.ipynb'); do
     if [ -f "$notebook" ]; then

flowquery-py/.gitignore

@@ -73,6 +73,9 @@ venv.bak/
 .dmypy.json
 dmypy.json
 
+# ruff
+.ruff_cache/
+
 # Pyre type checker
 .pyre/
 

flowquery-py/pyproject.toml

@@ -41,6 +41,8 @@ dev = [
     "jupyter>=1.0.0",
     "ipykernel>=6.0.0",
     "nbstripout>=0.6.0",
+    "mypy>=1.0.0",
+    "ruff>=0.1.0",
 ]
 
 [build-system]
@@ -75,4 +77,45 @@ python_functions = ["test_*"]
 addopts = "-v --tb=short"
 
 [tool.pytest-asyncio]
-mode = "auto"
+mode = "auto"
+
+[tool.mypy]
+python_version = "3.10"
+strict = true
+ignore_missing_imports = true
+exclude = [
+    "tests/",
+    "__pycache__",
+    ".git",
+    "build",
+    "dist",
+]
+
+[[tool.mypy.overrides]]
+module = "src.parsing.parser"
+warn_return_any = false
+disable_error_code = ["union-attr", "arg-type"]
+
+[tool.ruff]
+target-version = "py310"
+line-length = 120
+exclude = [
+    ".git",
+    "__pycache__",
+    "build",
+    "dist",
+]
+
+[tool.ruff.lint]
+select = [
+    "F",      # Pyflakes (includes F401 unused imports)
+    "E",      # pycodestyle errors
+    "W",      # pycodestyle warnings
+    "I",      # isort
+]
+ignore = [
+    "E501",   # line too long (handled by formatter)
+]
+
+[tool.ruff.lint.isort]
+known-first-party = ["flowquery", "src"]

flowquery-py/src/__init__.py

@@ -8,17 +8,17 @@
 
 from .compute.runner import Runner
 from .io.command_line import CommandLine
-from .parsing.parser import Parser
-from .parsing.functions.function import Function
 from .parsing.functions.aggregate_function import AggregateFunction
 from .parsing.functions.async_function import AsyncFunction
-from .parsing.functions.predicate_function import PredicateFunction
-from .parsing.functions.reducer_element import ReducerElement
+from .parsing.functions.function import Function
 from .parsing.functions.function_metadata import (
+    FunctionCategory,
     FunctionDef,
     FunctionMetadata,
-    FunctionCategory,
 )
+from .parsing.functions.predicate_function import PredicateFunction
+from .parsing.functions.reducer_element import ReducerElement
+from .parsing.parser import Parser
 
 __all__ = [
     "Runner",

flowquery-py/src/compute/runner.py

@@ -9,10 +9,10 @@
 
 class Runner:
     """Executes a FlowQuery statement and retrieves the results.
-    
+
     The Runner class parses a FlowQuery statement into an AST and executes it,
     managing the execution flow from the first operation to the final return statement.
-    
+
     Example:
         runner = Runner("WITH 1 as x RETURN x")
         await runner.run()
@@ -25,24 +25,28 @@ def __init__(
         ast: Optional[ASTNode] = None
     ):
         """Creates a new Runner instance and parses the FlowQuery statement.
-        
+
         Args:
             statement: The FlowQuery statement to execute
             ast: An already-parsed AST (optional)
-            
+
         Raises:
             ValueError: If neither statement nor AST is provided
         """
         if (statement is None or statement == "") and ast is None:
             raise ValueError("Either statement or AST must be provided")
-        
-        _ast = ast if ast is not None else Parser().parse(statement)
-        self._first: Operation = _ast.first_child()
-        self._last: Operation = _ast.last_child()
+
+        _ast = ast if ast is not None else Parser().parse(statement or "")
+        first = _ast.first_child()
+        last = _ast.last_child()
+        if not isinstance(first, Operation) or not isinstance(last, Operation):
+            raise ValueError("AST must contain Operations")
+        self._first: Operation = first
+        self._last: Operation = last
 
     async def run(self) -> None:
         """Executes the parsed FlowQuery statement.
-        
+
         Raises:
             Exception: If an error occurs during execution
         """
@@ -53,7 +57,7 @@ async def run(self) -> None:
     @property
     def results(self) -> List[Dict[str, Any]]:
         """Gets the results from the executed statement.
-        
+
         Returns:
             The results from the last operation (typically a RETURN statement)
         """

flowquery-py/src/extensibility.py

@@ -4,7 +4,7 @@
 
 Example:
     from flowquery.extensibility import Function, FunctionDef
-    
+
     @FunctionDef({
         'description': "Converts a string to uppercase",
         'category': "string",
@@ -15,27 +15,27 @@ class UpperCase(Function):
         def __init__(self):
             super().__init__("uppercase")
             self._expected_parameter_count = 1
-        
+
         def value(self) -> str:
             return str(self.get_children()[0].value()).upper()
 """
 
 # Base function classes for creating custom functions
-from .parsing.functions.function import Function
 from .parsing.functions.aggregate_function import AggregateFunction
 from .parsing.functions.async_function import AsyncFunction
-from .parsing.functions.predicate_function import PredicateFunction
-from .parsing.functions.reducer_element import ReducerElement
+from .parsing.functions.function import Function
 
 # Decorator and metadata types for function registration
 from .parsing.functions.function_metadata import (
+    FunctionCategory,
     FunctionDef,
-    FunctionMetadata,
     FunctionDefOptions,
-    ParameterSchema,
+    FunctionMetadata,
     OutputSchema,
-    FunctionCategory,
+    ParameterSchema,
 )
+from .parsing.functions.predicate_function import PredicateFunction
+from .parsing.functions.reducer_element import ReducerElement
 
 __all__ = [
     "Function",

flowquery-py/src/graph/__init__.py

@@ -1,18 +1,18 @@
 """Graph module for FlowQuery."""
 
-from .node import Node
-from .relationship import Relationship
-from .pattern import Pattern
-from .patterns import Patterns
-from .pattern_expression import PatternExpression
 from .database import Database
 from .hops import Hops
+from .node import Node
 from .node_data import NodeData
 from .node_reference import NodeReference
-from .relationship_data import RelationshipData
-from .relationship_reference import RelationshipReference
+from .pattern import Pattern
+from .pattern_expression import PatternExpression
+from .patterns import Patterns
 from .physical_node import PhysicalNode
 from .physical_relationship import PhysicalRelationship
+from .relationship import Relationship
+from .relationship_data import RelationshipData
+from .relationship_reference import RelationshipReference
 
 __all__ = [
     "Node",

flowquery-py/src/graph/database.py

@@ -1,14 +1,16 @@
 """Graph database for FlowQuery."""
 
-from typing import Any, Dict, Optional, Union, TYPE_CHECKING
+from __future__ import annotations
 
-from ..parsing.ast_node import ASTNode
+from typing import Dict, Optional, Union
 
-if TYPE_CHECKING:
-    from .node import Node
-    from .relationship import Relationship
-    from .node_data import NodeData
-    from .relationship_data import RelationshipData
+from ..parsing.ast_node import ASTNode
+from .node import Node
+from .node_data import NodeData
+from .physical_node import PhysicalNode
+from .physical_relationship import PhysicalRelationship
+from .relationship import Relationship
+from .relationship_data import RelationshipData
 
 
 class Database:
@@ -18,7 +20,7 @@ class Database:
     _nodes: Dict[str, 'PhysicalNode'] = {}
     _relationships: Dict[str, 'PhysicalRelationship'] = {}
 
-    def __init__(self):
+    def __init__(self) -> None:
         pass
 
     @classmethod
@@ -29,7 +31,6 @@ def get_instance(cls) -> 'Database':
 
     def add_node(self, node: 'Node', statement: ASTNode) -> None:
         """Adds a node to the database."""
-        from .physical_node import PhysicalNode
         if node.label is None:
             raise ValueError("Node label is null")
         physical = PhysicalNode(None, node.label)
@@ -42,7 +43,6 @@ def get_node(self, node: 'Node') -> Optional['PhysicalNode']:
 
     def add_relationship(self, relationship: 'Relationship', statement: ASTNode) -> None:
         """Adds a relationship to the database."""
-        from .physical_relationship import PhysicalRelationship
         if relationship.type is None:
             raise ValueError("Relationship type is null")
         physical = PhysicalRelationship()
@@ -56,11 +56,6 @@ def get_relationship(self, relationship: 'Relationship') -> Optional['PhysicalRe
 
     async def get_data(self, element: Union['Node', 'Relationship']) -> Union['NodeData', 'RelationshipData']:
         """Gets data for a node or relationship."""
-        from .node import Node
-        from .relationship import Relationship
-        from .node_data import NodeData
-        from .relationship_data import RelationshipData
-        
         if isinstance(element, Node):
             node = self.get_node(element)
             if node is None:
@@ -75,8 +70,3 @@ async def get_data(self, element: Union['Node', 'Relationship']) -> Union['NodeD
             return RelationshipData(data)
         else:
             raise ValueError("Element is neither Node nor Relationship")
-
-
-# Import for type hints
-from .physical_node import PhysicalNode
-from .physical_relationship import PhysicalRelationship