philiprehberger-import-profiler

Show how long each Python import takes during startup.

Installation

pip install philiprehberger-import-profiler

Usage

Basic Profiling

from philiprehberger_import_profiler import profile_imports

report = profile_imports("my_package")

# Slowest imports
for entry in report.slowest(10):
    print(f"{entry.name}: {entry.duration_ms:.1f}ms")

# Tree view
report.print_tree(threshold_ms=5.0)
# └── my_package (124.5ms)
#     ├── requests (45.2ms)
#     │   └── urllib3 (22.1ms)
#     └── numpy (62.0ms)

# Summary
print(f"Total: {report.total_ms:.1f}ms, Modules: {report.module_count}")

Self Time and Export

report = profile_imports("my_package")

# Self time excludes children's duration
for entry in report.slowest(5):
    print(f"{entry.name}: total={entry.duration_ms:.1f}ms self={entry.self_ms:.1f}ms")

# Export as list of dicts
data = report.to_dict()
# [{"name": "requests", "duration_ms": 45.2, "self_ms": 23.1, "parent": "my_package"}, ...]

Summary and JSON export

Use summary() for a compact dict you can log or pretty-print, and to_json() for the full entry list as a JSON string.

report = profile_imports("my_package")

print(report.summary(slowest=3))
# {"total_ms": 124.5, "module_count": 42, "slowest": [
#     {"name": "numpy", "duration_ms": 62.0},
#     {"name": "requests", "duration_ms": 45.2},
#     {"name": "urllib3", "duration_ms": 22.1},
# ]}

# JSON string (compact when indent=None)
print(report.to_json(indent=None))

Sorting and filtering reports

report = profile_imports("my_package")

# Top N slowest entries (descending by duration_ms)
for entry in report.slowest(3):
    print(f"{entry.name}: {entry.duration_ms:.1f}ms")

# Filter to entries whose module name starts with a prefix
numpy_only = report.filter("numpy")
print(f"numpy.* modules: {numpy_only.module_count}, total={numpy_only.total_ms:.1f}ms")

# filter returns a new report; the original is unchanged
assert report.module_count >= numpy_only.module_count

API

Development

pip install -e .
python -m pytest tests/ -v

Support

If you find this project useful:

⭐ Star the repo

🐛 Report issues

💡 Suggest features

❤️ Sponsor development

🌐 All Open Source Projects

💻 GitHub Profile

🔗 LinkedIn Profile

License

MIT