Python Dynamic Code Execution Functions

What’s the difference between eval, exec, and compile? - Stack Overflow

Understanding the three core functions for dynamic code execution in Python:

eval() - Expression Evaluation

Purpose: Evaluates a single Python expression and returns the result

1
2
3

result = eval("2 + 3 * 4")  # Returns 14
x = 5
result = eval("x * 2")      # Returns 10

Characteristics:

• Single Expression: Only works with expressions, not statements
• Return Value: Always returns a value
• Use Cases: Mathematical calculations, simple expressions
• Limitations: Cannot handle statements like assignments or loops

exec() - Statement Execution

Purpose: Executes Python statements (does not return a value)

1
2
3
4
5

exec("x = 10; y = 20; print(x + y)")  # Prints: 30
exec("""
for i in range(3):
    print(f"Hello {i}")
""")

Characteristics:

• Multiple Statements: Can execute complex code blocks
• No Return Value: Returns None
• Use Cases: Dynamic code execution, configuration scripts
• Flexibility: Can handle any valid Python code

compile() - Code Object Creation

Purpose: Compiles source code into code objects for repeated execution

1
2
3
4
5
6

# Compile once, execute multiple times
code = compile("x * 2", "<string>", "eval")
x = 5
result1 = eval(code)  # Returns 10
x = 10
result2 = eval(code)  # Returns 20

Modes:

• ’eval’: For expressions (use with eval())
• ’exec’: For statements (use with exec())
• ‘single’: For single interactive statements

Performance Considerations

Compilation Overhead:

• eval() and exec() compile code every time
• compile() allows pre-compilation for repeated use
• Significant performance improvement for repeated execution

Example - Repeated Execution:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

import timeit

# Without compile() - slower
def slow_version():
    for i in range(1000):
        eval("2 + 3 * 4")

# With compile() - faster
code = compile("2 + 3 * 4", "<string>", "eval")
def fast_version():
    for i in range(1000):
        eval(code)

Security Considerations

Major Risks:

• Code Injection: User input can execute arbitrary code
• System Access: Malicious code can access file system, network
• Data Exposure: Can access and modify global variables

Safer Alternatives:

1
2
3
4
5
6
7
8

# Restricted globals and locals
safe_globals = {"__builtins__": {}}
safe_locals = {"x": 10, "y": 20}
result = eval("x + y", safe_globals, safe_locals)

# Use ast.literal_eval for safe data parsing
import ast
data = ast.literal_eval("{'key': 'value'}")  # Only literals

Best Practices

1. Avoid When Possible: Use alternative approaches first
2. Sanitize Input: Never execute untrusted user input
3. Restrict Scope: Use limited globals and locals dictionaries
4. Use ast.literal_eval: For parsing data structures safely
5. Pre-compile: Use compile() for repeated execution

These functions provide powerful dynamic execution capabilities but require careful consideration of security and performance implications.