oracletrace 1.0.1

Detect Python performance regressions and compare execution traces with lightweight call graph visualization

OracleTrace

Detect Python performance regressions and compare execution traces with a lightweight call graph profiler.

OracleTrace is a lightweight Python performance analysis tool designed to help developers detect performance regressions, compare execution traces, and visualize call graphs in a simple and readable way.

It is ideal for:

• Detecting performance regressions between script versions
• Comparing execution time across runs
• Visualizing function call graphs
• Lightweight profiling without heavy instrumentation
• CI performance validation

---

Why OracleTrace?

Performance regressions in Python projects are often hard to detect early.

Traditional profilers focus on deep performance analysis, but they are not optimized for quick regression comparison between two executions.

OracleTrace solves this by allowing you to:

• Run a script and generate an execution trace
• Export results to JSON
• Compare two trace files
• Identify performance differences
• Detect new or removed functions
• Measure execution time deltas

---

Key Features

Performance Regression Detection

Compare two JSON trace files and instantly see:

• Slower functions
• Faster functions
• New function calls
• Removed function calls

Execution Trace Analysis

• Total execution time per function
• Average time per call
• Call counts
• Caller → callee relationships

Call Graph Visualization

Visual tree structure of your program’s execution flow.

JSON Export

Export trace results for:

• CI performance checks
• Historical comparison
• Automation pipelines

Clean Output

Filters internal Python calls to focus only on your project code.

---

Installation

pip install oracletrace

---

Quick Example

Step 1 — Create a script

import time

def process_data():
    time.sleep(0.1)
    calculate_results()

def calculate_results():
    time.sleep(0.2)

def main():
    for _ in range(2):
        process_data()

if __name__ == "__main__":
    main()

Step 2 — Run OracleTrace

oracletrace my_app.py

Export trace to JSON

oracletrace my_app.py --json baseline.json

Compare with a new version

oracletrace my_app.py --json new.json --compare baseline.json

This allows you to detect performance regressions between two executions.

---

How It Works

OracleTrace uses Python’s built-in sys.setprofile() function to intercept:

• call
• return

It measures execution time per function and records caller-callee relationships.

By filtering functions outside your project directory, the output focuses only on relevant application code.

---

Example Output

Summary table showing top functions by total execution time and average time per call.

Call graph visualization displaying execution flow hierarchy.

Starting application...

Iteration 1:
  > Processing data...
    > Calculating results...

Iteration 2:
  > Processing data...
    > Calculating results...

Application finished.

Summary:
                         Top functions by Total Time
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┓
┃ Function                     ┃ Total Time (s) ┃ Calls ┃ Avg. Time/Call (ms) ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━┩
│ my_app.py:main               │         0.6025 │     1 │             602.510 │
│ my_app.py:process_data       │         0.6021 │     2 │             301.050 │
│ my_app.py:calculate_results  │         0.4015 │     2 │             200.750 │
└──────────────────────────────┴────────────────┴───────┴─────────────────────┘


Logic Flow:
<module>
└── my_app.py:main (1x, 0.6025s)
    └── my_app.py:process_data (2x, 0.6021s)
        └── my_app.py:calculate_results (2x, 0.4015s)

---

Use Cases

• Detect Python performance regressions in development
• Compare execution time between versions
• Lightweight alternative to heavy profilers
• CI/CD performance monitoring
• Educational demonstration of call graphs

---

Requirements

• Python >= 3.10
• rich

---

Contributing

Contributions are welcome.

If you have ideas for improving regression detection, trace comparison, or visualization features, feel free to open an issue or submit a pull request.