Merge pull request #47 from IBM/support-slim-jsons-from-codeanalyzer

Version 0.2.0—support slim jsons from codeanalyzer

Author: rofrano

cldk/analysis/java/treesitter/javasitter.py

@@ -1,3 +1,4 @@
+from ipdb import set_trace
 from itertools import groupby
 from typing import List, Set, Dict
 from tree_sitter import Language, Node, Parser, Query

cldk/models/java/models.py

@@ -1,13 +1,13 @@
-import re
 import json
-from ipdb import set_trace
-from contextvars import ContextVar
+import re
 from typing import Dict, List, Optional
-from .constants_namespace import ConstantsNamespace
 from pydantic import BaseModel, field_validator, model_validator
+from contextvars import ContextVar
+from .constants_namespace import ConstantsNamespace
 
 constants = ConstantsNamespace()
 context_concrete_class = ContextVar("context_concrete_class")  # context var to store class concreteness
+_CALLABLES_LOOKUP_TABLE = dict()
 
 
 class JField(BaseModel):
@@ -340,12 +340,13 @@ class JGraphEdges(BaseModel):
     @field_validator("source", "target", mode="before")
     @classmethod
     def validate_source(cls, value) -> JMethodDetail:
-        callable_dict = json.loads(value)
-        j_callable = JCallable(**json.loads(callable_dict["callable"]))  # parse the value which is a quoted string
-        class_name = callable_dict["class_interface_declarations"]
+        file_path, type_declaration, callable_declaration = value["file_path"], value["type_declaration"], value["callable_declaration"]
+        j_callable = _CALLABLES_LOOKUP_TABLE.get((file_path, type_declaration, callable_declaration), None)
+        if j_callable is None:
+            raise ValueError(f"Callable not found in lookup table: {file_path}, {type_declaration}, {callable_declaration}")
+        class_name = type_declaration
         method_decl = j_callable.declaration
-        mc = JMethodDetail(method_declaration=method_decl, klass=class_name, method=j_callable)
-        return mc
+        return JMethodDetail(method_declaration=method_decl, klass=class_name, method=j_callable)
 
     def __hash__(self):
         return hash(tuple(self))
@@ -365,3 +366,14 @@ class JApplication(BaseModel):
 
     symbol_table: Dict[str, JCompilationUnit]
     system_dependency_graph: List[JGraphEdges] = None
+
+    @field_validator("symbol_table", mode="after")
+    @classmethod
+    def validate_source(cls, symbol_table):
+        from ipdb import set_trace
+
+        # Populate the lookup table for callables
+        for file_path, j_compulation_unit in symbol_table.items():
+            for type_declaration, jtype in j_compulation_unit.type_declarations.items():
+                for callable_declaration, j_callable in jtype.callable_declarations.items():
+                    _CALLABLES_LOOKUP_TABLE[(file_path, type_declaration, callable_declaration)] = j_callable

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "cldk"
-version = "0.1.1"
+version = "0.2.0"
 description = "codellm-devkit: A python library for seamless integration with LLMs."
 authors = ["Rahul Krishna <i.m.ralk@gmail.com>", "Rangeet Pan <rangeet.pan@ibm.com>", "Saurabh Sinhas <sinhas@us.ibm.com>",
             "Raju Pavuluri <pavuluri@us.ibm.com>"]
@@ -64,5 +64,5 @@ line-length = 180
 
 [tool.cldk.testing]
 sample-application = "tests/resources/java/application/"
-sample-application-analysis-json = "tests/resources/java/analysis_db"
+sample-application-analysis-json = "tests/resources/java/analysis_json/slim"
 codeanalyzer-jar-path = "tests/resources/java/codeanalyzer/build/libs/"

tests/conftest.py

@@ -17,43 +17,43 @@ def analysis_json_fixture():
     return config["tool"]["cldk"]["testing"]["sample-application-analysis-json"]
 
 
-@pytest.fixture(scope="session", autouse=True)
-def test_fixture():
-    """
-    Returns the path to the test data directory.
-
-    Yields:
-        Path : The path to the test data directory.
-    """
-    # ----------------------------------[ SETUP ]----------------------------------
-    # Path to your pyproject.toml
-    pyproject_path = Path(__file__).parent.parent / "pyproject.toml"
-
-    # Load the configuration
-    config = toml.load(pyproject_path)
-
-    # Access the test data path
-    test_data_path = config["tool"]["cldk"]["testing"]["sample-application"]
-
-    if not Path(test_data_path).exists():
-        Path(test_data_path).mkdir(parents=True)
-    url = "https://github.com/OpenLiberty/sample.daytrader8/archive/refs/tags/v1.2.zip"
-    filename = Path(test_data_path).absolute() / "v1.2.zip"
-    urlretrieve(url, filename)
-
-    # Extract the zip file to the test data path
-    with zipfile.ZipFile(filename, "r") as zip_ref:
-        zip_ref.extractall(test_data_path)
-
-    # Remove the zip file
-    filename.unlink()
-    # --------------------------------------------------------------------------------
-    # Daytrader8 sample application path
-    yield Path(test_data_path) / "sample.daytrader8-1.2"
-
-    # -----------------------------------[ TEARDOWN ]----------------------------------
-    # Remove the daytrader8 sample application that was downloaded for testing
-    for directory in Path(test_data_path).iterdir():
-        if directory.exists() and directory.is_dir():
-            shutil.rmtree(directory)
-    # ---------------------------------------------------------------------------------
+# @pytest.fixture(scope="session", autouse=True)
+# def test_fixture():
+#     """
+#     Returns the path to the test data directory.
+
+#     Yields:
+#         Path : The path to the test data directory.
+#     """
+#     # ----------------------------------[ SETUP ]----------------------------------
+#     # Path to your pyproject.toml
+#     pyproject_path = Path(__file__).parent.parent / "pyproject.toml"
+
+#     # Load the configuration
+#     config = toml.load(pyproject_path)
+
+#     # Access the test data path
+#     test_data_path = config["tool"]["cldk"]["testing"]["sample-application"]
+
+#     if not Path(test_data_path).exists():
+#         Path(test_data_path).mkdir(parents=True)
+#     url = "https://github.com/OpenLiberty/sample.daytrader8/archive/refs/tags/v1.2.zip"
+#     filename = Path(test_data_path).absolute() / "v1.2.zip"
+#     urlretrieve(url, filename)
+
+#     # Extract the zip file to the test data path
+#     with zipfile.ZipFile(filename, "r") as zip_ref:
+#         zip_ref.extractall(test_data_path)
+
+#     # Remove the zip file
+#     filename.unlink()
+#     # --------------------------------------------------------------------------------
+#     # Daytrader8 sample application path
+#     yield Path(test_data_path) / "sample.daytrader8-1.2"
+
+#     # -----------------------------------[ TEARDOWN ]----------------------------------
+#     # Remove the daytrader8 sample application that was downloaded for testing
+#     for directory in Path(test_data_path).iterdir():
+#         if directory.exists() and directory.is_dir():
+#             shutil.rmtree(directory)
+#     # ---------------------------------------------------------------------------------

tests/models/java/test_java_models.py

@@ -1,11 +1,12 @@
+from ipdb import set_trace
 from typing import List, Tuple
 from cldk import CLDK
 
 
 def test_get_class_call_graph(analysis_json_fixture):
     # Initialize the CLDK object with the project directory, language, and analysis_backend.
     cldk = CLDK(language="java")
-
     analysis = cldk.analysis(
         project_path=analysis_json_fixture, analysis_backend="codeanalyzer", analysis_json_path=analysis_json_fixture, eager=False, analysis_level="call-graph"
     )
+    assert analysis.get_call_graph_json() is not None

tests/resources/java/analysis_json/.gitignore

@@ -0,0 +1 @@
+!*.json