.github: add shallow checkpatch to Github Actions

Let's try to run checkpatch in an open-source and fast way while
leveraging Github's very good user interface.

Run checkpatch both with and without --strict; this has been a source of
confusion in the past, see
thesofproject/linux#1988

Note this attempt does not rely on git merge-bases which has been found
to be not compatible with shallow cloning:
thesofproject/linux#2556

Signed-off-by: Marc Herbert <marc.herbert@intel.com>

.github/workflows/shallow_checkpatch.yml

@@ -0,0 +1,55 @@
+---
+# SPDX-License-Identifier: BSD-3-Clause
+# Tools that can save round-trips to github and a lot of time:
+#
+# yamllint -f parsable this.yml
+# pip3 install ruamel.yaml.cmd
+# yaml merge-expand this.yml exp.yml && diff -w -u this.yml exp.yml
+#
+# github.com also has a powerful web editor that can be used without
+# committing.
+
+name: checkpatch
+
+# yamllint disable-line rule:truthy
+on: [pull_request]
+
+jobs:
+  checkpatch:
+    runs-on: ubuntu-20.04
+    env:
+      PR_NUM: ${{github.event.number}}
+      CHK_CMD: ./scripts/checkpatch.pl --codespell --codespellfile
+          /usr/lib/python3/dist-packages/codespell_lib/data/dictionary.txt
+    steps:
+      # depth 2 so:
+      # ^1. we can show the Subject of the current target branch tip
+      # ^2. we reconnect/graft to the later fetch pull/1234/head,
+      - uses: actions/checkout@v2
+        with: {fetch-depth: 2}
+
+      - name: yamllint ourselves
+        run: yamllint .github/workflows/shallow_checkpatch.yml
+
+      - name: install codespell
+        run: sudo apt-get -y install codespell && dpkg -L codespell | grep dict
+
+      # See shallowness issue https://github.com/thesofproject/linux/issues/2556
+      - name: fetch PR commits
+        run: |
+           .github/workflows/shallowfetchPRcommits.sh \
+              ${GITHUB_REPOSITORY} "$PR_NUM"
+           # show what we got
+           git --no-pager log --oneline --graph --decorate --max-count=50
+
+      - name: checkpatch
+        run: while read -r sha ; do
+              printf '\n    -------------- \n\n';
+              ${CHK_CMD} -g $sha;
+             done < PR_SHAs.txt
+
+      - name: checkpatch --strict
+        run: while read -r sha ; do
+              printf '\n    -------------- \n\n';
+              ${CHK_CMD} --strict -g $sha;
+             done < PR_SHAs.txt

.github/workflows/shallowfetchPRcommits.sh

@@ -0,0 +1,69 @@
+#!/bin/sh
+# SPDX-License-Identifier: BSD-3-Clause
+
+set -e
+
+# This script fetches Pull Request commits missing from a shallow clone
+# and creates a PR_SHAs.txt file. This script has a limit of 500 commits but the
+# github API used has a lower limit of 250 commits.
+
+# It does not rely on git merge-bases which basically don't work with
+# shallow clones:
+# https://github.com/thesofproject/linux/issues/2556
+
+# Design goals:
+#
+# - Keep the code short and as simple as possible. No one is interested
+#   in maintaining this sort of script.
+#
+# - Fast and accurate for small Pull Requests
+#
+# - For large Pull Requests _with merges_ the only objective is to
+#   complete in a reasonable time; say less than 10 minutes. It's very
+#   unlikely will look at 250 checkpatch results and time optimizations
+#   should not make this script more complex.
+
+
+# Sample usage:
+#   $0  thesoftproject/linux  2772
+main()
+{
+    local gh_project="$1"
+    local pr_number="$2"
+
+    printf '%s: fetching PR %d for project %s\n' "$0" "$pr_number" "$gh_project"
+
+    # As of March 2021, Github's documented limit is 250 commits
+    # Let's have another cap a 500.
+    # https://docs.github.com/en/rest/reference/pulls#list-commits-on-a-pull-request
+    local pagelen PRlen=0
+    for i in 1 2 3 4 5; do
+        curl -H 'Accept: application/vnd.github.v3+json' \
+         "https://api.github.com/repos/$gh_project/pulls/$pr_number/commits?per_page=100&page=$i" \
+         > commits_"$i".json
+        pagelen=$(jq length < commits_$i.json)
+        if [ "$pagelen" -eq 0 ]; then
+            break
+        fi
+        PRlen=$((PRlen + pagelen))
+    done
+
+    printf 'Found %d commits, SHA1 list is in PR_SHAs.txt\n' "$PRlen"
+
+    # 'cut' removes double quotes
+    cat commits_?.json |
+        jq '.[] | .sha' |
+        cut -b2-41 > PR_SHAs.txt
+
+    # PRlen+1 gets us the merge base for simple, linear histories. For
+    # pull requests with merges, depth=PRLEN goes already much further
+    # than needed and +1 makes little difference. It's not clear when
+    # and for what sort of PRs git fetching individual commits would be
+    # faster so keep a single and simple fetch for now.
+
+    set -x # this command may take a while so show it
+    git fetch --depth "$((PRlen+1))" "http://github.com/$gh_project" "pull/$pr_number/head"
+
+}
+
+main "$@"