Pre-compile all python code and remove original .py files. (#476)

* Remove urllib3 dist-info files as well.

* Pre-compile all python code and remove original .py files.

* Compress at optimization level 2.

* Add PYTHONNODEBUGRANGES=1 to compile step.

* Add comment.

* Use different commandline option to support py3.8

* Update size limits.

* Leave __init__.py contrib files.

* Remove text like files.

* oops

Author: purple4reina

Dockerfile

@@ -11,23 +11,32 @@ WORKDIR /build
 COPY . .
 RUN pip install . -t ./python/lib/$runtime/site-packages
 
-# Remove *.pyc files
-RUN find ./python/lib/$runtime/site-packages -name \*.pyc -delete
-
 # Remove botocore (40MB) to reduce package size. aws-xray-sdk
 # installs it, while it's already provided by the Lambda Runtime.
 RUN rm -rf ./python/lib/$runtime/site-packages/botocore*
 RUN rm -rf ./python/lib/$runtime/site-packages/setuptools
 RUN rm -rf ./python/lib/$runtime/site-packages/jsonschema/tests
 RUN find . -name 'libddwaf.so' -delete
-RUN rm -rf ./python/lib/$runtime/site-packages/urllib3
+RUN rm -rf ./python/lib/$runtime/site-packages/urllib3*
 RUN rm ./python/lib/$runtime/site-packages/ddtrace/appsec/_iast/_taint_tracking/*.so
 RUN rm ./python/lib/$runtime/site-packages/ddtrace/appsec/_iast/_stacktrace*.so
 RUN rm ./python/lib/$runtime/site-packages/ddtrace/internal/datadog/profiling/libdd_wrapper.so
 RUN rm ./python/lib/$runtime/site-packages/ddtrace/internal/datadog/profiling/ddup/_ddup.*.so
 RUN rm ./python/lib/$runtime/site-packages/ddtrace/internal/datadog/profiling/stack_v2/_stack_v2.*.so
 RUN find . -name "*.dist-info" -type d | xargs rm -rf
 
+# Precompile all .pyc files and remove .py files. This speeds up load time.
+# Compile with optimization level 2 (-OO) and PYTHONNODEBUGRANGES=1 to redtce
+# size of .pyc files.
+# See https://docs.python.org/3/tutorial/modules.html#compiled-python-files
+# https://docs.python.org/3.11/using/cmdline.html#cmdoption-O
+# https://docs.python.org/3/using/cmdline.html#envvar-PYTHONNODEBUGRANGES
+RUN PYTHONNODEBUGRANGES=1 python -OO -m compileall -b ./python/lib/$runtime/site-packages
+# remove all .py files except ddtrace/contrib/*/__init__.py which are necessary
+# for ddtrace.patch to discover instrumationation packages.
+RUN find ./python/lib/$runtime/site-packages -name \*.py | grep -v ddtrace/contrib | xargs rm -rf
+RUN find ./python/lib/$runtime/site-packages/ddtrace/contrib -name \*.py | grep -v __init__ | xargs rm -rf
+RUN find ./python/lib/$runtime/site-packages -name __pycache__ -type d -exec rm -r {} \+
 
 FROM scratch
 COPY --from=builder /build/python /

scripts/check_layer_size.sh

@@ -7,10 +7,9 @@
 
 # Compares layer size to threshold, and fails if below that threshold
 
-# 7 mb size limit
 set -e
-MAX_LAYER_COMPRESSED_SIZE_KB=$(expr 4 \* 1024)
-MAX_LAYER_UNCOMPRESSED_SIZE_KB=$(expr 13 \* 1024)
+MAX_LAYER_COMPRESSED_SIZE_KB=$(expr 5 \* 1024)
+MAX_LAYER_UNCOMPRESSED_SIZE_KB=$(expr 12 \* 1024)
 
 
 LAYER_FILES_PREFIX="datadog_lambda_py"