Debugging & Profiling

1# ============================================================
2# 1. pdb / breakpoint() — Interactive Debugging
3# ============================================================
4import sys
5from pathlib import Path
6
7
8def calculate_discount(items, membership_level):
9    """
10    Production function with a subtle bug to debug.
11    Items: list of dicts with 'name', 'price', 'quantity'.
12    """
13    subtotal = sum(
14        item["price"] * item["quantity"] for item in items
15    )
16
17    # Drop into debugger to inspect state:
18    # breakpoint()  # Uncomment to debug interactively
19
20    discount_rates = {
21        "bronze": 0.05,
22        "silver": 0.10,
23        "gold": 0.15,
24        "platinum": 0.20,
25    }
26
27    rate = discount_rates.get(membership_level, 0)
28    discount = subtotal * rate
29    total = subtotal - discount
30
31    return {
32        "subtotal": round(subtotal, 2),
33        "discount": round(discount, 2),
34        "total": round(total, 2),
35        "rate": rate,
36    }
37
38
39# ============================================================
40# 2. Conditional breakpoints and post-mortem debugging
41# ============================================================
42def find_anomalies(data):
43    """Process data with conditional debugging."""
44    results = []
45    for i, value in enumerate(data):
46        processed = value ** 0.5 if value >= 0 else None
47
48        # Conditional breakpoint — only pause on suspicious values
49        # if processed is not None and processed > 100:
50        #     breakpoint()
51
52        results.append({"index": i, "original": value, "processed": processed})
53    return results
54
55
56def debug_with_post_mortem():
57    """
58    Post-mortem debugging: inspect state AFTER a crash.
59    Run with: python -m pdb script.py
60    When it crashes, pdb drops you into the frame where the
61    exception occurred.
62    """
63    data = [100, 200, -1, 400, 0]
64    try:
65        results = [1 / x for x in data]
66    except ZeroDivisionError:
67        import pdb
68        # pdb.post_mortem()  # Uncomment to debug at crash site
69        print("ZeroDivisionError caught — would enter post-mortem debugger")
70
71
72# ============================================================
73# 3. Structured Logging (production debugging)
74# ============================================================
75import logging
76
77# Configure logging with structured format
78logging.basicConfig(
79    level=logging.DEBUG,
80    format="%(asctime)s [%(levelname)s] %(name)s:%(funcName)s:%(lineno)d — %(message)s",
81    datefmt="%Y-%m-%d %H:%M:%S",
82)
83logger = logging.getLogger(__name__)
84
85
86def process_order(order_id, items):
87    """Production code with proper logging levels."""
88    logger.info("Processing order %s with %d items", order_id, len(items))
89
90    for item in items:
91        logger.debug(
92            "Item: %s, price=%.2f, qty=%d",
93            item["name"],
94            item["price"],
95            item["quantity"],
96        )
97
98    try:
99        result = calculate_discount(items, "gold")
100        logger.info(
101            "Order %s total: $%.2f (discount: $%.2f)",
102            order_id,
103            result["total"],
104            result["discount"],
105        )
106        return result
107    except Exception as e:
108        logger.exception("Failed to process order %s", order_id)
109        raise
110
111
112# ============================================================
113# 4. Custom exception hooks and traceback formatting
114# ============================================================
115import traceback
116
117
118def robust_processor(data_batch):
119    """Collect errors without stopping the entire batch."""
120    results = []
121    errors = []
122
123    for i, item in enumerate(data_batch):
124        try:
125            processed = 100 / item["value"]
126            results.append({"index": i, "result": processed})
127        except (ZeroDivisionError, KeyError, TypeError) as e:
128            error_info = {
129                "index": i,
130                "item": item,
131                "error": str(e),
132                "traceback": traceback.format_exc(),
133            }
134            errors.append(error_info)
135            logger.warning("Error at index %d: %s", i, e)
136
137    if errors:
138        logger.warning(
139            "Batch completed with %d errors out of %d items",
140            len(errors),
141            len(data_batch),
142        )
143
144    return results, errors
145
146
147# ============================================================
148# 5. cProfile — Function-level profiling
149# ============================================================
150import cProfile
151import io
152import pstats
153
154
155def fibonacci(n):
156    """Deliberately unoptimized for profiling demonstration."""
157    if n <= 1:
158        return n
159    return fibonacci(n - 1) + fibonacci(n - 2)
160
161
162def profile_fibonacci():
163    """Profile with cProfile and display sorted results."""
164    profiler = cProfile.Profile()
165    profiler.enable()
166
167    result = fibonacci(30)
168
169    profiler.disable()
170
171    # Capture profile output
172    stream = io.StringIO()
173    stats = pstats.Stats(profiler, stream=stream)
174    stats.sort_stats("cumulative")
175    stats.print_stats(10)  # top 10 functions
176    print(stream.getvalue())
177    print(f"Result: {result}")
178
179
180# Alternative: profile from command line
181# python -m cProfile -s cumulative my_script.py
182# python -m cProfile -o profile_output.prof my_script.py
183
184
185# ============================================================
186# 6. Timing utilities for targeted profiling
187# ============================================================
188import time
189from functools import wraps
190from contextlib import contextmanager
191
192
193def timer(func):
194    """Decorator to measure function execution time."""
195    @wraps(func)
196    def wrapper(*args, **kwargs):
197        start = time.perf_counter()
198        result = func(*args, **kwargs)
199        elapsed = time.perf_counter() - start
200        logger.info("%s executed in %.4f seconds", func.__name__, elapsed)
201        return result
202    return wrapper
203
204
205@contextmanager
206def timed_block(label="block"):
207    """Context manager for timing arbitrary code blocks."""
208    start = time.perf_counter()
209    yield
210    elapsed = time.perf_counter() - start
211    logger.info("%s completed in %.4f seconds", label, elapsed)
212
213
214@timer
215def sort_large_list():
216    """Example function to profile."""
217    import random
218    data = [random.randint(0, 1_000_000) for _ in range(500_000)]
219    return sorted(data)
220
221
222def demo_timed_block():
223    """Demonstrate context manager timing."""
224    with timed_block("list comprehension"):
225        squares = [x ** 2 for x in range(1_000_000)]
226
227    with timed_block("generator sum"):
228        total = sum(x ** 2 for x in range(1_000_000))
229
230
231# ============================================================
232# 7. tracemalloc — Memory profiling (stdlib)
233# ============================================================
234import tracemalloc
235
236
237def memory_profile_demo():
238    """Track memory allocations to find leaks."""
239    tracemalloc.start()
240
241    # Allocate some memory
242    data = [list(range(1000)) for _ in range(1000)]
243
244    snapshot = tracemalloc.take_snapshot()
245    top_stats = snapshot.statistics("lineno")
246
247    print("\nTop 5 memory allocations:")
248    for stat in top_stats[:5]:
249        print(f"  {stat}")
250
251    current, peak = tracemalloc.get_traced_memory()
252    print(f"\nCurrent memory: {current / 1024:.1f} KB")
253    print(f"Peak memory: {peak / 1024:.1f} KB")
254
255    tracemalloc.stop()
256
257
258# ============================================================
259# Usage
260# ============================================================
261if __name__ == "__main__":
262    # Debugging demo
263    items = [
264        {"name": "Widget", "price": 25.99, "quantity": 3},
265        {"name": "Gadget", "price": 49.99, "quantity": 1},
266        {"name": "Doohickey", "price": 12.50, "quantity": 5},
267    ]
268    process_order("ORD-001", items)
269
270    # Profiling demo
271    profile_fibonacci()
272
273    # Timing demo
274    sort_large_list()
275    demo_timed_block()
276
277    # Memory demo
278    memory_profile_demo()