py-flow-mapper 0.1.0b3.dev0__tar.gz → 0.1.0b5.dev0__tar.gz

{py_flow_mapper-0.1.0b3.dev0/src/py_flow_mapper.egg-info → py_flow_mapper-0.1.0b5.dev0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: py-flow-mapper
-Version: 0.1.0b3.dev0
+Version: 0.1.0b5.dev0
 Summary: Python project analyzer and visualization tool
 Home-page: https://github.com/ArunKoundinya/py-flow-mapper
 Author: Arun Koundinya Parasa

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0}/setup.py

@@ -5,7 +5,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
 
 setup(
     name="py-flow-mapper",
-    version="0.1.0b3.dev",
+    version="0.1.0b5.dev",
     author="Arun Koundinya Parasa",
     author_email="parasa.arunkoundinya@gmail.com",
     description="Python project analyzer and visualization tool",

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0}/src/py_flow_mapper/__init__.py

@@ -2,7 +2,7 @@
 PyFlowMapper - A Python project analyzer and visualization tool.
 """
 
-__version__ = "0.1.0b3.dev"
+__version__ = "0.1.0b5.dev"
 __author__ = "Arun Koundinya Parasa"
 __description__ = "Analyze Python projects and generate dependency graphs"
 __github__ = "https://github.com/ArunKoundinya/py-flow-mapper"

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0}/src/py_flow_mapper/analyzer.py

@@ -107,8 +107,24 @@ class ProjectAnalyzer:
         """Find all Python files in the project, excluding virtual environments and other unwanted directories."""
         python_files = []
         
-        # Prefixes of directories to exclude
-        exclude_prefixes = ('venv', '.venv', 'env', '.env', '__pycache__', '.git', 'node_modules')
+        # Prefixes of directories to exclude        
+        exclude_prefixes = (
+            "venv", ".venv", "env", ".env", "virtualenv", ".virtualenv",
+            "__pycache__", ".mypy_cache", ".pytest_cache", ".ruff_cache",
+            ".coverage", "htmlcov", ".hypothesis",
+            ".git", ".hg", ".svn",
+            "build", "dist", ".eggs", "egg-info", "*.egg-info",
+            ".tox", ".nox", ".pdm-build",
+            "site-packages", "node_modules", "bower_components",
+            "docs", "doc", "site", "_site", "mkdocs",
+            ".ipynb_checkpoints", "notebooks", "examples", "experiments", "scratch",
+            "data", "datasets", "models", "checkpoints", "outputs", "results",
+            "logs", "tmp", "temp", "cache",
+            ".idea", ".vscode", ".DS_Store", "__MACOSX",
+            ".github", ".gitlab", ".circleci",
+            "tests", "test", "testing",
+            )
+
         
         for root, dirs, files in os.walk(self.base_path):
             # Filter out directories that start with any of the exclude prefixes
@@ -505,24 +521,21 @@ class DataFlowAnalyzer(ast.NodeVisitor):
         
     def visit_Assign(self, node):
         """Track assignments of function return values."""
-        # Get the variable name being assigned to
-        if isinstance(node.targets[0], ast.Name):
+        if node.targets and isinstance(node.targets[0], ast.Name):
             var_name = node.targets[0].id
-            
-            # Check if the value is a function call
+
             if isinstance(node.value, ast.Call):
                 call_info = self._extract_call_info(node.value)
                 if call_info:
-                    self.calls.append(call_info['func_name'])
-                    if var_name not in self.return_assignments:
-                        self.return_assignments[var_name] = []
-                    self.return_assignments[var_name].append(call_info['func_name'])
-            
-            # Check if the value is a Name (could be a variable holding a return value)
+                    func_name = call_info["func_name"]
+
+                    if func_name not in self.calls:
+                        self.calls.append(func_name)
+                    self.return_assignments.setdefault(var_name, []).append(func_name)
+
             elif isinstance(node.value, ast.Name):
-                # Track variable assignments for data flow
                 pass
-        
+
         self.generic_visit(node)
     
     def visit_Call(self, node):

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0}/src/py_flow_mapper/mermaid_generator.py

@@ -71,23 +71,16 @@ class MermaidGenerator:
                 return ""
             return ",".join(sorted(set(cleaned)))
         
-        current_module_ctx = ""
 
-        def module_import_mapping() -> dict:
-            return (modules.get(current_module_ctx, {}) or {}).get("import_mapping", {}) or {}
+        def module_import_mapping(mod: str) -> dict:
+            return (modules.get(mod, {}) or {}).get("import_mapping", {}) or {}
 
-        def external_root_name(call_name: str) -> str:
+        def external_root_name(call_name: str, current_module: str) -> str:
             if not call_name:
                 return ""
-
             root = call_name.split(".")[0]
-
-            # module-specific import mapping (from analyzer)
-            imp_map = module_import_mapping() or {}
-
-            # merge common aliases (module imports override defaults)
+            imp_map = module_import_mapping(current_module)
             merged_map = {**COMMON_ALIAS_MAP, **imp_map}
-
             return merged_map.get(root, root)
 
         # Keep the graph clean: ignore common builtins + attribute-noise
@@ -99,42 +92,35 @@ class MermaidGenerator:
             "Path",
         }
 
-        def keep_external(call_name: str) -> bool:
+        def keep_external(call_name: str, current_module: str) -> bool:
             if not call_name:
                 return False
 
             base = short_label(call_name)
 
-            # forced external always wins
-            root = external_root_name(call_name)
+            root = external_root_name(call_name, current_module)
             if root and root in self.force_external:
                 return True
             if base in self.force_external:
                 return True
 
-            # ✅ NEW: module-specific import mapping
-            imp_map = module_import_mapping() or {}
+            imp_map = module_import_mapping(current_module)
             if base in imp_map:
                 mapped = imp_map[base]
                 mapped_module = ".".join(mapped.split(".")[:-1])
-
                 if mapped_module in modules:
                     return False
-
                 return True
 
-            # internal class => NOT external
             if base in internal_classes:
                 return False
 
             if base in NOISY_EXTERNAL:
                 return False
 
-            # dotted calls like obj.method: too noisy at this level
             if "." in call_name:
                 return False
 
-            # keep CamelCase (tool-ish classes / constructors)
             return is_camel_case(base)
 
         def resolve_internal(call: str, current_module: str) -> str:
@@ -174,7 +160,7 @@ class MermaidGenerator:
 
                 info = function_map.get(fn_key, {})
                 current_module = info.get("module", "") or ""
-                current_module_ctx = current_module
+                #current_module_ctx = current_module
                 for c in (info.get("calls") or []):
                     target = self._find_function_full_name(c, current_module)
                     if target and target in function_map:
@@ -210,15 +196,15 @@ class MermaidGenerator:
 
         for _, info in function_map.items():
             current_module = info.get("module", "") or ""
-            current_module_ctx = current_module
+            #current_module_ctx = current_module
 
             # prefer call_arguments keys
             call_args = info.get("call_arguments", {}) or {}
             for callee in call_args.keys():
                 if resolve_internal(callee, current_module):
                     continue
-                if keep_external(callee):
-                    external_nodes.add(external_root_name(callee) or callee)
+                if keep_external(callee, current_module):
+                    external_nodes.add(external_root_name(callee, current_module) or callee)
 
             # scan raw calls for Done + missed externals
             for callee in (info.get("calls") or []):
@@ -227,8 +213,8 @@ class MermaidGenerator:
 
                 if resolve_internal(callee, current_module):
                     continue
-                if keep_external(callee):
-                    external_nodes.add(external_root_name(callee) or callee)
+                if keep_external(callee, current_module):
+                    external_nodes.add(external_root_name(callee, current_module) or callee)
 
         if external_nodes or uses_done:
             lines.append("    subgraph External [External]")
@@ -247,7 +233,7 @@ class MermaidGenerator:
 
             info = function_map[caller]
             current_module = info.get("module", "") or ""
-            current_module_ctx = current_module
+            #current_module_ctx = current_module
             src = nid(caller)
 
             call_args = info.get("call_arguments", {}) or {}
@@ -274,8 +260,8 @@ class MermaidGenerator:
                     else:
                         lines.append(f"    {src} --> {nid(target_internal)}")
                 else:
-                    if keep_external(callee):
-                        ext = external_root_name(callee) or callee
+                    if keep_external(callee, current_module):
+                        ext = external_root_name(callee, current_module) or callee
                         edge_key = (src, nid("ext:" + callee), label)
                         if edge_key in seen:
                             continue
@@ -294,7 +280,7 @@ class MermaidGenerator:
             info = function_map[fn]
             dst = nid(fn)
             current_module = info.get("module", "") or ""
-            current_module_ctx = current_module
+            #current_module_ctx = current_module
 
             return_assignments = info.get("return_assignments", {}) or {}
             for var_name, producers in return_assignments.items():
@@ -303,20 +289,20 @@ class MermaidGenerator:
                     if internal_p:
                         lines.append(f"    {nid(internal_p)} -.->|{var_name}| {dst}")
                     else:
-                        if keep_external(p):
-                            ext = external_root_name(p) or p
+                        if keep_external(p, current_module):
+                            ext = external_root_name(p,current_module) or p
                             lines.append(f"    {nid('ext:' + ext)} -.->|{var_name}| {dst}")
 
         # ---------- pipeline heuristic: external tool A -> external tool B ----------
         for _, info in function_map.items():
             current_module = info.get("module", "") or ""
-            current_module_ctx = current_module
+            #current_module_ctx = current_module
             calls = info.get("calls") or []
             call_args = info.get("call_arguments") or {}
 
             meaningful = []
             for c in calls:
-                if c and keep_external(c) and not resolve_internal(c, current_module):
+                if c and keep_external(c,current_module) and not resolve_internal(c, current_module):
                     if c not in meaningful:
                         meaningful.append(c)
 
@@ -325,7 +311,7 @@ class MermaidGenerator:
 
             fileish_callee = None
             for callee, args in call_args.items():
-                if keep_external(callee) and any(is_fileish_arg(a) for a in (args or []) if isinstance(a, str)):
+                if keep_external(callee, current_module) and any(is_fileish_arg(a) for a in (args or []) if isinstance(a, str)):
                     fileish_callee = callee
                     break
 
@@ -338,8 +324,8 @@ class MermaidGenerator:
 
             vars_passed = call_args.get(fileish_callee, []) or []
             label = normalize_vars(vars_passed) or "value"
-            prod_ext = external_root_name(producer) or producer
-            file_ext = external_root_name(fileish_callee) or fileish_callee
+            prod_ext = external_root_name(producer,current_module) or producer
+            file_ext = external_root_name(fileish_callee,current_module) or fileish_callee
             lines.append(f"    {nid('ext:' + prod_ext)} -->|{label}| {nid('ext:' + file_ext)}")
 
         # ---------- Done edge ----------

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0}/src/py_flow_mapper/utils.py

@@ -120,17 +120,36 @@ def load_json(file_path: Path) -> Any:
 def get_project_structure(base_path: Path, exclude_dirs: List[str] = None) -> Dict[str, Any]:
     """Get the project directory structure."""
     if exclude_dirs is None:
-        exclude_dirs = ['venv', '.venv', '__pycache__', '.git', '.idea', '.vscode']
-    
-    structure = {}
-    
-    for item in sorted(base_path.iterdir()):
-        if item.name in exclude_dirs:
+        exclude_dirs = [
+            "venv", ".venv", "env", ".env", "virtualenv",
+            "__pycache__", ".mypy_cache", ".pytest_cache", ".ruff_cache",
+            ".coverage", "htmlcov",
+            ".git", ".hg", ".svn",
+            ".idea", ".vscode", ".DS_Store", "__MACOSX",
+            "build", "dist", ".eggs", ".tox", ".nox",
+            "node_modules", "site-packages",
+            "docs", "doc", "notebooks", ".ipynb_checkpoints",
+            "models", "outputs", "results",
+            "logs", "tmp", "temp",
+        ]
+
+    exclude_set = set(exclude_dirs)
+    structure: Dict[str, Any] = {}
+
+    for item in sorted(base_path.iterdir(), key=lambda p: p.name):
+        name = item.name
+
+        # Unified exclusion logic (works for both files and dirs)
+        if (
+            name in exclude_set
+            or name.endswith(".egg-info")                          # e.g., mypkg.egg-info
+            or any(name.startswith(excl) for excl in exclude_set)  # e.g., venv311, build_temp
+        ):
             continue
-        
+
         if item.is_dir():
-            structure[item.name] = get_project_structure(item, exclude_dirs)
+            structure[name] = get_project_structure(item, exclude_dirs)
         else:
-            structure[item.name] = 'file'
-    
-    return structure
+            structure[name] = "file"
+
+    return structure

{py_flow_mapper-0.1.0b3.dev0 → py_flow_mapper-0.1.0b5.dev0/src/py_flow_mapper.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: py-flow-mapper
-Version: 0.1.0b3.dev0
+Version: 0.1.0b5.dev0
 Summary: Python project analyzer and visualization tool
 Home-page: https://github.com/ArunKoundinya/py-flow-mapper
 Author: Arun Koundinya Parasa