CI: Schedule PR jobs based on commit specific changes

.github/workflows/build.yml

@@ -125,20 +125,29 @@ jobs:
         [ -z "$AWS_ACCESS_KEY_ID" ] || aws s3 cp mpy-cross/mpy-cross.static-raspbian s3://adafruit-circuit-python/bin/mpy-cross/mpy-cross.static-raspbian-${{ env.CP_VERSION }} --no-progress --region us-east-1
         [ -z "$AWS_ACCESS_KEY_ID" ] || aws s3 cp mpy-cross/mpy-cross.static s3://adafruit-circuit-python/bin/mpy-cross/mpy-cross.static-amd64-linux-${{ env.CP_VERSION }} --no-progress --region us-east-1
         [ -z "$AWS_ACCESS_KEY_ID" ] || aws s3 cp mpy-cross/mpy-cross.static.exe s3://adafruit-circuit-python/bin/mpy-cross/mpy-cross.static-x64-windows-${{ env.CP_VERSION }}.exe --no-progress --region us-east-1
-    - name: "Get changes"
+    - name: Get last commit with checks
+      id: get-last-commit-with-checks
       if: github.event_name == 'pull_request'
-      uses: dorny/paths-filter@v2
-      id: filter
-      with:
-        list-files: json
-        filters: |
-          changed:
-            - '**'
-    - name: "Set matrix"
+      working-directory: tools
+      env:
+        REPO: ${{ github.repository }}
+        PULL: ${{ github.event.number }}
+        GITHUB_TOKEN: ${{ github.token }}
+        EXCLUDE_COMMIT: ${{ github.event.after }}
+      run: python3 -u ci_changes_per_commit.py
+    - name: Get changes
+      id: get-changes
+      if: github.event_name == 'pull_request'
+      uses: tj-actions/changed-files@v34
+      with:
+        json: "true"
+        base_sha: ${{ steps.get-last-commit-with-checks.outputs.commit }}
+    - name: Set matrix
       id: set-matrix
       working-directory: tools
       env:
-        CHANGED_FILES: ${{ steps.filter.outputs.changed_files }}
+        CHANGED_FILES: ${{ toJSON(steps.get-changes.outputs.all_changed_and_modified_files) }}
+        LAST_FAILED_JOBS: ${{ steps.get-last-commit-with-checks.outputs.checkruns }}
       run: python3 -u ci_set_matrix.py
 
 

tools/ci_changes_per_commit.py

@@ -0,0 +1,227 @@
+#! /usr/bin/env python3
+
+# SPDX-FileCopyrightText: 2021 microDev
+#
+# SPDX-License-Identifier: MIT
+
+# GraphQL Query
+
+QUERY_COMMITS = """
+query ($owner: String!, $name: String!, $pullNumber: Int!, $commitsPerPage: Int!, $beforeCommit: String) {
+  repository(owner: $owner, name: $name) {
+    pullRequest(number: $pullNumber) {
+      commits(last: $commitsPerPage, before: $beforeCommit) {
+        totalCount
+        pageInfo {
+          startCursor
+          hasPreviousPage
+        }
+        nodes {
+          commit {
+            checkSuites(first: 3) {
+              nodes {
+                conclusion
+                workflowRun {
+                  workflow {
+                    name
+                  }
+                }
+                id
+              }
+              totalCount
+            }
+            oid
+          }
+        }
+      }
+    }
+  }
+}
+"""
+
+QUERY_CHECKRUNS = """
+query ($checkSuiteID: ID!,
+       $afterFailedRun: String, $afterIncompleteRun: String,
+       $includeFailedRuns: Boolean!, $includeIncompleteRuns: Boolean!) {
+  node(id: $checkSuiteID) {
+    ... on CheckSuite {
+      failedRuns: checkRuns(
+        first: 100
+        after: $afterFailedRun
+        filterBy: {checkType: LATEST, conclusions: [ACTION_REQUIRED, TIMED_OUT, CANCELLED, FAILURE, NEUTRAL, STARTUP_FAILURE]}
+      ) @include(if: $includeFailedRuns) {
+        nodes {
+          name
+        }
+        pageInfo {
+          endCursor
+          hasNextPage
+        }
+      }
+      incompleteRuns: checkRuns(
+        first: 100
+        after: $afterIncompleteRun
+        filterBy: {checkType: LATEST, statuses: [QUEUED, IN_PROGRESS, WAITING, PENDING, REQUESTED]}
+      ) @include(if: $includeIncompleteRuns) {
+        nodes {
+          name
+        }
+        pageInfo {
+          endCursor
+          hasNextPage
+        }
+      }
+    }
+  }
+}
+"""
+
+
+import os
+import re
+import json
+import requests
+
+
+query_variables_commits = {
+    "owner": "",
+    "name": "",
+    "pullNumber": int(os.environ["PULL"]),
+    "commitsPerPage": 20,
+    "beforeCommit": None,
+}
+
+
+query_variables_checkruns = {
+    "checkSuiteID": "",
+    "afterFailedRun": None,
+    "afterIncompleteRun": None,
+    "includeFailedRuns": True,
+    "includeIncompleteRuns": True,
+}
+
+
+headers = {"Authorization": f"Bearer {os.environ['GITHUB_TOKEN']}"}
+
+
+class Query:
+    def __init__(self, query, variables={}, headers={}):
+        self.query = query
+        self.variables = variables
+        self.headers = headers
+
+    def paginate(self, page_info, name):
+        has_page = (
+            page_info["hasNextPage"] if name.startswith("after") else page_info["hasPreviousPage"]
+        )
+        if has_page:
+            self.variables[name] = (
+                page_info["endCursor"] if name.startswith("after") else page_info["startCursor"]
+            )
+        return has_page
+
+    def fetch(self):
+        request = requests.post(
+            "https://api.github.com/graphql",
+            json={"query": self.query, "variables": self.variables},
+            headers=self.headers,
+        )
+        if request.status_code == 200:
+            return request.json()
+        else:
+            raise Exception("Query Failed: {}".format(request.status_code))
+
+
+def set_output(name, value):
+    if "GITHUB_OUTPUT" in os.environ:
+        with open(os.environ["GITHUB_OUTPUT"], "at") as f:
+            print(f"{name}={value}", file=f)
+    else:
+        print(f"Would set GitHub actions output {name} to '{value}'")
+
+
+def get_commit_and_checksuite(query_commits):
+    commits = query_commits.fetch()["data"]["repository"]["pullRequest"]["commits"]
+
+    if commits["totalCount"] > 0:
+        for commit in reversed(commits["nodes"]):
+            commit = commit["commit"]
+            commit_sha = commit["oid"]
+            if commit_sha == os.environ["EXCLUDE_COMMIT"]:
+                continue
+            checksuites = commit["checkSuites"]
+            if checksuites["totalCount"] > 0:
+                for checksuite in checksuites["nodes"]:
+                    if checksuite["workflowRun"]["workflow"]["name"] == "Build CI":
+                        return [
+                            commit_sha,
+                            checksuite["id"] if checksuite["conclusion"] != "SUCCESS" else None,
+                        ]
+        else:
+            if query_commits.paginate(commits["pageInfo"], "beforeCommit"):
+                return get_commit_and_checksuite(query_commits)
+
+    return [None, None]
+
+
+def append_runs_to_list(runs, list):
+    regex_matrix = re.compile("^build-[^ ]+")
+    regex_board = re.compile("\([^ ]+\)$")
+    for run in runs["nodes"]:
+        name = run["name"]
+        res_matrix = regex_matrix.search(name)
+        if res_matrix:
+            matrix = res_matrix.group()
+            if matrix not in list:
+                list[matrix] = []
+            list[matrix].append(regex_board.search(name).group()[1:-1])
+
+
+def get_bad_checkruns(query_checkruns, list={}):
+    checkruns = query_checkruns.fetch()["data"]["node"]
+    run_types = ["failed", "incomplete"]
+    paginate = False
+
+    for run_type in run_types:
+        run_type_camel = run_type.capitalize() + "Run"
+        run_type = run_type + "Runs"
+
+        append_runs_to_list(checkruns[run_type], list)
+
+        if query_checkruns.paginate(checkruns[run_type]["pageInfo"], "after" + run_type_camel):
+            query_checkruns.variables["include" + run_type_camel] = True
+            paginate = True
+
+    return get_bad_checkruns(query_checkruns, list) if paginate else list
+
+
+def main():
+    query_commits = Query(QUERY_COMMITS, query_variables_commits, headers)
+    query_commits.variables["owner"], query_commits.variables["name"] = os.environ["REPO"].split(
+        "/"
+    )
+
+    commit, checksuite = get_commit_and_checksuite(query_commits)
+
+    if checksuite is None:
+        if commit is None:
+            print("No checkSuites found -> Abort")
+        else:
+            set_output("commit", commit)
+        quit()
+
+    query_checkruns = Query(QUERY_CHECKRUNS, query_variables_checkruns, headers)
+    query_checkruns.variables["checkSuiteID"] = checksuite
+
+    checkruns = get_bad_checkruns(query_checkruns)
+
+    if len(checkruns) == 0:
+        print("No checkRuns found -> Abort")
+        quit()
+
+    set_output("commit", commit)
+    set_output("checkruns", json.dumps(checkruns))
+
+
+if __name__ == "__main__":
+    main()

tools/ci_set_matrix.py

@@ -25,7 +25,6 @@
 import os
 import sys
 import json
-import yaml
 import pathlib
 from concurrent.futures import ThreadPoolExecutor
 
@@ -62,14 +61,22 @@
 if len(sys.argv) > 1:
     print("Using files list on commandline")
     changed_files = sys.argv[1:]
+    last_failed_jobs = {}
 else:
     c = os.environ["CHANGED_FILES"]
     if c == "":
         print("CHANGED_FILES is in environment, but value is empty")
         changed_files = []
     else:
         print("Using files list in CHANGED_FILES")
-        changed_files = json.loads(os.environ["CHANGED_FILES"])
+        changed_files = json.loads(c)
+
+    j = os.environ["LAST_FAILED_JOBS"]
+    if j == "":
+        print("LAST_FAILED_JOBS is in environment, but value is empty")
+        last_failed_jobs = {}
+    else:
+        last_failed_jobs = json.loads(j)
 
 
 def set_output(name, value):
@@ -196,7 +203,7 @@ def get_settings(board):
     # Split boards by architecture.
     print("Building boards:")
     arch_to_boards = {"aarch": [], "arm": [], "riscv": [], "espressif": []}
-    for board in sorted(boards_to_build):
+    for board in boards_to_build:
         print(" ", board)
         port = board_to_port.get(board)
         # A board can appear due to its _deletion_ (rare)
@@ -208,10 +215,20 @@ def get_settings(board):
 
     # Set the step outputs for each architecture
     for arch in arch_to_boards:
+        # Append previous failed jobs
+        if f"build-{arch}" in last_failed_jobs:
+            failed_boards = last_failed_jobs[f"build-{arch}"]
+            for board in failed_boards:
+                if not board in arch_to_boards[arch]:
+                    arch_to_boards[arch].append(board)
+        # Set Output
         set_output(f"boards-{arch}", json.dumps(sorted(arch_to_boards[arch])))
 
 
 def set_docs_to_build(build_all):
+    if "build-doc" in last_failed_jobs:
+        build_all = True
+
     doc_match = build_all
     if not build_all:
         doc_pattern = re.compile(
@@ -224,7 +241,7 @@ def set_docs_to_build(build_all):
 
     # Set the step outputs
     print("Building docs:", doc_match)
-    set_output(f"build-doc", doc_match)
+    set_output("build-doc", doc_match)
 
 
 def check_changed_files():