From 470be08a83bd2df387121960f56a64a395688859 Mon Sep 17 00:00:00 2001
From: Kyle McCormick <kdmc@pm.me>
Date: Mon, 7 Jun 2021 22:15:03 -0400
Subject: [PATCH] build: check for __init__.py files in each directory

---
 .github/workflows/verify-dunder-init.yml | 26 +++++++
 scripts/verify-dunder-init.sh            | 99 ++++++++++++++++++++++++
 2 files changed, 125 insertions(+)
 create mode 100644 .github/workflows/verify-dunder-init.yml
 create mode 100755 scripts/verify-dunder-init.sh

diff --git a/.github/workflows/verify-dunder-init.yml b/.github/workflows/verify-dunder-init.yml
new file mode 100644
index 0000000000..aefc0f53b6
--- /dev/null
+++ b/.github/workflows/verify-dunder-init.yml
@@ -0,0 +1,26 @@
+name: CI
+
+on:
+  pull_request:
+    branches:
+      - master
+
+jobs:
+
+  verify_dunder_init:
+
+    name: Verify __init__.py Files
+    runs-on: ubuntu-20.04
+
+    steps:
+
+    - name: Check out branch
+      uses: actions/checkout@v2
+
+    - name: Ensure git is installed
+      run: |
+        sudo apt-get update && sudo apt-get install git
+
+    - name: Verify __init__.py files exist
+      run: |
+        scripts/verify-dunder-init.sh
diff --git a/scripts/verify-dunder-init.sh b/scripts/verify-dunder-init.sh
new file mode 100755
index 0000000000..5055138e3d
--- /dev/null
+++ b/scripts/verify-dunder-init.sh
@@ -0,0 +1,99 @@
+#!/usr/bin/env bash
+#
+# Recursively verify that every directory in edx-platform that contains
+# Python source code also contains an __init__.py file (aka a "dunder-init file").
+#
+# Even though the Python 3 runner does not require that an __init__.py file
+# exist in every Python directory, some of our tooling (namely, pylint and
+# import-linter) depend on it existing, and will report false-positive success
+# otherwise.
+#
+# Run from root of directory with no args.
+# Example:
+#   scripts/verify-dunder-init.sh
+#
+# Exits 0 if no errors, 1 otherwise.
+# Missing __init__.py files are printed to STDOUT.
+# Extra info is printed to STDIN.
+
+set -euo pipefail # Strict mode.
+
+
+## Directories that contain Python source code, but for which we don't care
+## whether or not there's a __init__.py file.
+exclude=''
+
+# Exclude repo root.
+exclude+='^\.$'
+
+# Exclude test data that includes Python (do NOT exclude unit test source code, though).
+exclude+='|^common/lib/capa/capa/safe_exec/tests/test_files/?.*$'
+exclude+='|^common/test/data/?.*$'
+
+# Exclude common/lib and its immediate child directories.
+# They are not Python packages, but instead are Python sub-projects (hence the setup.py
+# in each chlid directory).
+# However, we do NOT want to exclude the source directories *within* the sub-projects.
+# Example:
+#  * common/lib/xmodule -> EXCLUDE from check.
+#  * common/lib/xmodule/xmodule/modulestore -> INCLUDE in check.
+exclude+='|^common/lib$'
+exclude+='|^common/lib/(capa|safe_lxml|sandbox-packages|symmath|xmodule)$'
+
+# Docs, scripts.
+exclude+='|^docs/.*$'
+exclude+='|^lms/djangoapps/monitoring/scripts$'
+exclude+='|^scripts/?.*$'
+
+
+## Counters for directories.
+no_python=0
+excluded=0
+errored=0
+confirmed=0
+
+
+## Loop through all directories that are under version control.
+>&2 echo
+>&2 echo "============== begin list of missing files =============="
+for directory in $(git ls-files | xargs dirname | sort | uniq) ; do
+	if ! ls "$directory"/*.py &>/dev/null ; then
+		# No Python in this directory; skip it.
+		no_python=$(( no_python+1 ))
+		continue
+	fi
+	if [[ "$directory" =~ $exclude ]]; then
+		excluded=$(( excluded+1 ))
+		# Directory is specifically excluded; skip it.
+		continue
+	fi
+	if [[ -f "$directory"/__init__.py ]] ; then
+		# Directory contains __init__.py; all good!
+		confirmed=$(( confirmed+1 ))
+		continue
+	fi
+	# Error! Print missing file to STDOUT.
+	errored=$(( errored+1 ))
+	echo "$directory/__init__.py"
+done
+>&2 echo "=============== end list of missing files ==============="
+
+
+## Report results (to STDERR)
+>&2 echo
+>&2 echo "${no_python} directories do not contain Python source code."
+>&2 echo "${excluded} directories contain Python source code, but are excluded."
+>&2 echo
+>&2 echo "${confirmed} Python source directories DO contain an __init__.py file."
+>&2 echo "${errored} Python source directories do NOT contain an __init__.py file."
+>&2 echo
+
+
+## Succeed or fail.
+if ! [[ errored -eq 0 ]] ; then
+	>&2 echo "Check failed! All directories with Python source code must contain __init__.py (unless excluded)."
+	exit 1
+else
+	>&2 echo "Check passed!"
+	exit 0
+fi