From 3cd7ce96101ddc94144e4715900c64c1f2e5e217 Mon Sep 17 00:00:00 2001
From: jctanner <tanner.jc@gmail.com>
Date: Mon, 9 May 2022 19:27:19 -0400
Subject: [PATCH] Use the PR diff for TDD checks (#1257)

Use the PR diff url to get the list of changed files instead of making a checkout. The master branch on forked repos isn't always kept up to date and gives false positives on files changed when compared to HEAD of the PR branch.

No-Issue

Signed-off-by: James Tanner <tanner.jc@gmail.com>
---
 .github/workflows/ci_tdd.yml |  6 +++---
 dev/common/tdd.py            | 28 +++++++++++++++++++++++-----
 2 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/ci_tdd.yml b/.github/workflows/ci_tdd.yml
index 508fefc4d1..78c2882283 100644
--- a/.github/workflows/ci_tdd.yml
+++ b/.github/workflows/ci_tdd.yml
@@ -16,14 +16,14 @@ jobs:
         with:
           python-version: "3.8"
 
-      - name: make the checkout in a sane way.
-        run: git clone https://github.com/${{ github.event.pull_request.user.login }}/galaxy_ng; cd galaxy_ng ; git checkout "${{ github.head_ref }}"
+      - uses: actions/checkout@v3
 
       - name: Ensure changes were made to tests
         env:
           PY_COLORS: '1'
+          GITHUB_PR_NUMBER: ${{github.event.number}}
           GITHUB_BASE_REF: ${{ github.base_ref }}
           GITHUB_BRANCH: ${{ github.head_ref }}
           GITHUB_USER: ${{ github.event.pull_request.user.login }}
           GITHUB_CONTEXT: ${{ toJson(github) }}
-        run: cd galaxy_ng; python dev/common/tdd.py
+        run: python dev/common/tdd.py
diff --git a/dev/common/tdd.py b/dev/common/tdd.py
index dc779c5734..2a5ddf2e93 100644
--- a/dev/common/tdd.py
+++ b/dev/common/tdd.py
@@ -20,11 +20,29 @@ def get_current_branch():
 
 
 def get_changed_files(pr_branch, target_branch="master"):
-    cmd = f'git diff --name-only {pr_branch}..{target_branch}'
-    pid = subprocess.run(cmd, shell=True, stdout=subprocess.PIPE, check=True)
-    filenames = pid.stdout.decode('utf-8')
-    filenames = filenames.split('\n')
-    filenames = [x.strip() for x in filenames if x.strip()]
+    if os.environ.get('GITHUB_PR_NUMBER'):
+        url = "https://patch-diff.githubusercontent.com/raw/ansible/galaxy_ng"
+        url += f"/pull/{os.environ['GITHUB_PR_NUMBER']}.diff"
+        print(f'GET diff from {url}')
+        cmd = f'curl -s -o /tmp/pr.diff {url}'
+        pid = subprocess.run(cmd, shell=True)
+        assert pid.returncode == 0
+        with open('/tmp/pr.diff', 'r') as f:
+            raw = f.read()
+        filenames = raw.split('\n')
+        filenames = [x for x in filenames if x.startswith('---') or x.startswith('+++')]
+        filenames = [x.split(None, 1)[1] for x in filenames]
+        filenames = [x for x in filenames if not x.startswith('/dev/null')]
+        filenames = [x.lstrip('a/') for x in filenames]
+        filenames = [x.lstrip('b/') for x in filenames]
+        filenames = sorted(set(filenames))
+    else:
+        # diffs the local checkout ...
+        cmd = f'git diff --name-only "{pr_branch}".."{target_branch}"'
+        pid = subprocess.run(cmd, shell=True, stdout=subprocess.PIPE, check=True)
+        filenames = pid.stdout.decode('utf-8')
+        filenames = filenames.split('\n')
+        filenames = [x.strip() for x in filenames if x.strip()]
     return filenames