Python Import Machinery & sys.modules Cache

1Concept

When `import x` is invoked, Python checks `sys.modules`. If cached, the module is returned immediately. Otherwise, Python iterates through `sys.meta_path` finders, locates the file, creates a module spec, allocates the module object, and executes its bytecode in a new namespace.

2Architecture Diagram

import my_module ---> [ Found in sys.modules? ] --Yes--> Return cached instance
                                   | No
                             Find Spec -> Load Bytecode -> Cache -> Return

3Code Example

import sys
import math

print(f"Is math cached in sys.modules: {'math' in sys.modules}")
print(f"Math module object: {sys.modules['math']}")
print(f"Total modules loaded in runtime: {len(sys.modules)}")

4Expected Output

Is math cached in sys.modules: True
Math module object: <module 'math' (built-in)>
Total modules loaded in runtime: 78

5Key Takeaways

• ✓Modules are singletons; importing the same module across 100 files executes its top-level code once.
• ✓Deleting an entry from `sys.modules` forces Python to re-import it on the next call.
• ✓Circular imports fail when module A needs an attribute from module B before B completes initialization.