Validate wheel content (#33)

Summary:
This PR checks the wheel content and asserts that:

- all of `libtorchcodec[4567].so` are present in the wheel
- libtorchcodec.so is not in the wheel
- none of libavfilter.so, libavcodec.so etc. are present in the wheel
- none of the doc, test or benchmarks folders are present in the wheel

This adds more reassurance that we won't be publishing broken wheel and thus preserve the user's experience.

Pull Request resolved: #33

Reviewed By: ahmadsharif1

Differential Revision: D58646828

Pulled By: NicolasHug

fbshipit-source-id: 05df551cf455cc519d27619726bae169a52028cc

.github/scripts/helpers.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+
+_list_wheel_files() {
+    unzip -l "$1" | awk '{print $4}'
+}
+
+# $1 = path to wheel
+# $2 = pattern to grep for in wheel files
+# If files matching $2 are found in the wheel, the function errors.
+assert_not_in_wheel() {
+    wheel_files=$(_list_wheel_files "$1")
+    if grep -q "$2" <<< "$wheel_files"
+    then
+        echo "Found files in $1 that start with $2. Exiting!!"
+        exit 1
+    fi
+}
+
+# See assert_not_in_wheel
+assert_in_wheel() {
+    wheel_files=$(_list_wheel_files "$1")
+    if ! grep -q "$2" <<< "$wheel_files"
+    then
+        echo "Did not find files in $1 that start with $2. Exiting!!"
+        exit 1
+    fi
+}

.github/workflows/wheel.yaml

@@ -40,15 +40,27 @@ jobs:
           # Just for sanity, make sure FFmpeg isn't installed or needed for buidling.
           .github/scripts/assert_ffmpeg_not_installed.sh
 
-          # TODO: Need to make sure the wheel only contains torchcodec and not
-          # the other folders (tests, docs, etc.)
           python -m pip install build
           BUILD_AGAINST_ALL_FFMPEG_FROM_S3=1 python -m build . -vvv --no-isolation
-      - name: Show the .so files in the wheel
+      - name: Validate wheel content
         run: |
-          # This should show libtorchcodec4.so, libtorchcodec5.so, etc.
+          source .github/scripts/helpers.sh
           wheel_path=`find dist -type f -name "*.whl"`
-          unzip -l $wheel_path | grep .so
+          echo "Wheel content:"
+          unzip -l $wheel_path
+
+          for ffmpeg_major_version in 4 5 6 7; do
+            assert_in_wheel $wheel_path torchcodec/libtorchcodec${ffmpeg_major_version}.so
+          done
+          assert_not_in_wheel $wheel_path libtorchcodec.so
+
+          for ffmpeg_so in libavcodec.so libavfilter.so libavformat.so libavutil.so libavdevice.so ; do
+            assert_not_in_wheel $wheel_path $ffmpeg_so
+          done
+
+          assert_not_in_wheel $wheel_path "^test"
+          assert_not_in_wheel $wheel_path "^doc"
+          assert_not_in_wheel $wheel_path "^benchmarks"
       - uses: actions/upload-artifact@v4
         with:
           name: sdist-and-wheel-linux_x86_${{ matrix.python-version }}