importfiler

Profile your imports, find what is dragging your module load time.

importfiler builds on top of the output generated by the interpreter -X importtime traces. It is just a different output targeted more at a specific use-case. Finding what modules impact the longest on a specific one you are trying to optimize.

If you need a more detailed result, check out py-spy or any other runtime profiler.

Getting dependencies import time

Given an import, shows the the time spent importing direct dependencies. The import time does include the time spent on importing other modules.

This helps you finding the most time consuming imports in your library/app.

Examples:

$ python -m importfiler requests
modules:            time(%)
------------------- --------
requests            100 %
urllib3             61 %
charset_normalizer  11 %
http.cookiejar      10 %
idna                2 %
certifi             2 %
http.cookies        1 %
zipfile             1 %

Getting the time in milliseconds:

$ python -m importfiler --unit=ms requests
modules:            time(ms)
------------------- ---------
requests            48 ms
urllib3             29 ms
charset_normalizer  5 ms
http.cookiejar      4 ms
idna                1 ms
certifi             1 ms

Getting all_modules import time

Given an import, shows the the time spent loading all its dependencies (including transitive). The import time does not include the time spent on importing other modules.

This helps you finding the most time consuming modules across your chain of dependencies.

Examples:

$ python -m importfiler --mode="all_modules" requests
modules:             time (%)             submodules
-------------------- -------------------- --------------------
urllib3              17 %                 urllib3.util.url, urllib3.packages.six, urllib3.exceptions, urllib3.poolmanager, ...
email                7 %                  email._policybase, email.message, email.header, email.feedparser, ...
charset_normalizer   6 %                  charset_normalizer.constant, charset_normalizer.models, charset_normalizer.cd, ...
requests             6 %                  requests.utils, requests.models, requests.exceptions, requests.packages, ...
http                 6 %                  http.cookiejar, http.client, http.cookies
urllib               4 %                  urllib.request, urllib.parse, urllib.error, urllib.response
typing               3 %
ssl                  2 %
importlib            2 %                  importlib.resources._adapters, importlib.resources.abc, importlib.resources._common, ...
re                   2 %                  re._parser, re._compiler, re._constants, re._casefix

Getting the time in milliseconds:

$ python -m importfiler --mode="all_modules" --unit=ms requests
modules:             time (ms)            submodules
-------------------- -------------------- --------------------
urllib3              18 ms                urllib3.util.url, urllib3.packages.six, urllib3.exceptions, urllib3.connectionpool, ...
charset_normalizer   8 ms                 charset_normalizer.md__mypyc, charset_normalizer.constant, charset_normalizer.models, ...
requests             7 ms                 requests.utils, requests.exceptions, requests.models, requests.packages, ...
email                3 ms                 email._policybase, email.header, email.feedparser, email.utils, ...
http                 3 ms                 http.cookiejar, http.cookies, http.client
idna                 2 ms                 idna.core, idna.idnadata, idna.package_data, idna.intranges
urllib               1 ms                 urllib.request, urllib.parse, urllib.error, urllib.response
importlib            1 ms                 importlib.resources._adapters, importlib.resources.abc, importlib.resources._common, ...
typing               1 ms
ssl                  1 ms
re                   1 ms                 re._parser, re._compiler, re._constants, re._casefix