Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 1 | #!/usr/bin/env python3 |
| 2 | |
| 3 | # Copyright (c) 2022 Intel Corp. |
| 4 | # SPDX-License-Identifier: Apache-2.0 |
| 5 | |
| 6 | import argparse |
| 7 | import sys |
| 8 | import os |
| 9 | import time |
| 10 | import datetime |
| 11 | from github import Github, GithubException |
Anas Nashif | 6027152 | 2022-07-18 19:37:31 -0400 | [diff] [blame] | 12 | from github.GithubException import UnknownObjectException |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 13 | from collections import defaultdict |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 14 | from west.manifest import Manifest |
| 15 | from west.manifest import ManifestProject |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 16 | |
| 17 | TOP_DIR = os.path.join(os.path.dirname(__file__)) |
| 18 | sys.path.insert(0, os.path.join(TOP_DIR, "scripts")) |
| 19 | from get_maintainer import Maintainers |
| 20 | |
| 21 | def log(s): |
| 22 | if args.verbose > 0: |
| 23 | print(s, file=sys.stdout) |
| 24 | |
| 25 | def parse_args(): |
| 26 | global args |
| 27 | parser = argparse.ArgumentParser( |
| 28 | description=__doc__, |
Jamie McCrae | ec70444 | 2023-01-04 16:08:36 +0000 | [diff] [blame] | 29 | formatter_class=argparse.RawDescriptionHelpFormatter, allow_abbrev=False) |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 30 | |
| 31 | parser.add_argument("-M", "--maintainer-file", required=False, default="MAINTAINERS.yml", |
| 32 | help="Maintainer file to be used.") |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 33 | |
| 34 | group = parser.add_mutually_exclusive_group() |
| 35 | group.add_argument("-P", "--pull_request", required=False, default=None, type=int, |
| 36 | help="Operate on one pull-request only.") |
Fabio Baltieri | b6cbcba | 2023-08-22 17:04:31 +0000 | [diff] [blame] | 37 | group.add_argument("-I", "--issue", required=False, default=None, type=int, |
| 38 | help="Operate on one issue only.") |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 39 | group.add_argument("-s", "--since", required=False, |
| 40 | help="Process pull-requests since date.") |
| 41 | group.add_argument("-m", "--modules", action="store_true", |
| 42 | help="Process pull-requests from modules.") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 43 | |
| 44 | parser.add_argument("-y", "--dry-run", action="store_true", default=False, |
| 45 | help="Dry run only.") |
| 46 | |
| 47 | parser.add_argument("-o", "--org", default="zephyrproject-rtos", |
| 48 | help="Github organisation") |
| 49 | |
| 50 | parser.add_argument("-r", "--repo", default="zephyr", |
| 51 | help="Github repository") |
| 52 | |
| 53 | parser.add_argument("-v", "--verbose", action="count", default=0, |
| 54 | help="Verbose Output") |
| 55 | |
| 56 | args = parser.parse_args() |
| 57 | |
| 58 | def process_pr(gh, maintainer_file, number): |
| 59 | |
| 60 | gh_repo = gh.get_repo(f"{args.org}/{args.repo}") |
| 61 | pr = gh_repo.get_pull(number) |
| 62 | |
| 63 | log(f"working on https://github.com/{args.org}/{args.repo}/pull/{pr.number} : {pr.title}") |
| 64 | |
| 65 | labels = set() |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 66 | area_counter = defaultdict(int) |
| 67 | maint = defaultdict(int) |
| 68 | |
| 69 | num_files = 0 |
| 70 | all_areas = set() |
| 71 | fn = list(pr.get_files()) |
| 72 | if len(fn) > 500: |
| 73 | log(f"Too many files changed ({len(fn)}), skipping....") |
| 74 | return |
| 75 | for f in pr.get_files(): |
| 76 | num_files += 1 |
| 77 | log(f"file: {f.filename}") |
| 78 | areas = maintainer_file.path2areas(f.filename) |
| 79 | |
| 80 | if areas: |
| 81 | all_areas.update(areas) |
| 82 | for a in areas: |
| 83 | area_counter[a.name] += 1 |
| 84 | labels.update(a.labels) |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 85 | for p in a.maintainers: |
| 86 | maint[p] += 1 |
| 87 | |
| 88 | ac = dict(sorted(area_counter.items(), key=lambda item: item[1], reverse=True)) |
| 89 | log(f"Area matches: {ac}") |
| 90 | log(f"labels: {labels}") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 91 | |
Stephanos Ioannidis | faf4208 | 2022-10-20 21:51:03 +0900 | [diff] [blame] | 92 | # Create a list of collaborators ordered by the area match |
| 93 | collab = list() |
| 94 | for a in ac: |
| 95 | collab += maintainer_file.areas[a].maintainers |
| 96 | collab += maintainer_file.areas[a].collaborators |
| 97 | collab = list(dict.fromkeys(collab)) |
| 98 | log(f"collab: {collab}") |
| 99 | |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 100 | sm = dict(sorted(maint.items(), key=lambda item: item[1], reverse=True)) |
| 101 | |
| 102 | log(f"Submitted by: {pr.user.login}") |
| 103 | log(f"candidate maintainers: {sm}") |
| 104 | |
Fabio Baltieri | d06450b | 2022-10-03 14:51:40 +0000 | [diff] [blame] | 105 | maintainer = "None" |
| 106 | maintainers = list(sm.keys()) |
| 107 | |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 108 | prop = 0 |
Fabio Baltieri | d06450b | 2022-10-03 14:51:40 +0000 | [diff] [blame] | 109 | if maintainers: |
| 110 | maintainer = maintainers[0] |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 111 | |
| 112 | if len(ac) > 1 and list(ac.values())[0] == list(ac.values())[1]: |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 113 | for aa in ac: |
| 114 | if 'Documentation' in aa: |
| 115 | log("++ With multiple areas of same weight including docs, take something else other than Documentation as the maintainer") |
| 116 | for a in all_areas: |
Fabio Baltieri | d06450b | 2022-10-03 14:51:40 +0000 | [diff] [blame] | 117 | if (a.name == aa and |
| 118 | a.maintainers and a.maintainers[0] == maintainer and |
| 119 | len(maintainers) > 1): |
| 120 | maintainer = maintainers[1] |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 121 | elif 'Platform' in aa: |
Fabio Baltieri | 279ab43 | 2022-10-03 15:00:10 +0000 | [diff] [blame] | 122 | log("++ Platform takes precedence over subsystem...") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 123 | log(f"Set maintainer of area {aa}") |
| 124 | for a in all_areas: |
| 125 | if a.name == aa: |
| 126 | if a.maintainers: |
| 127 | maintainer = a.maintainers[0] |
| 128 | break |
| 129 | |
| 130 | |
| 131 | # if the submitter is the same as the maintainer, check if we have |
| 132 | # multiple maintainers |
| 133 | if pr.user.login == maintainer: |
| 134 | log("Submitter is same as Assignee, trying to find another assignee...") |
| 135 | aff = list(ac.keys())[0] |
| 136 | for a in all_areas: |
| 137 | if a.name == aff: |
| 138 | if len(a.maintainers) > 1: |
| 139 | maintainer = a.maintainers[1] |
| 140 | else: |
| 141 | log(f"This area has only one maintainer, keeping assignee as {maintainer}") |
| 142 | |
| 143 | prop = (maint[maintainer] / num_files) * 100 |
| 144 | if prop < 20: |
| 145 | maintainer = "None" |
Fabio Baltieri | d06450b | 2022-10-03 14:51:40 +0000 | [diff] [blame] | 146 | |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 147 | log(f"Picked maintainer: {maintainer} ({prop:.2f}% ownership)") |
| 148 | log("+++++++++++++++++++++++++") |
| 149 | |
| 150 | # Set labels |
Fabio Baltieri | 16d723e | 2023-01-26 17:03:13 +0000 | [diff] [blame] | 151 | if labels: |
| 152 | if len(labels) < 10: |
| 153 | for l in labels: |
| 154 | log(f"adding label {l}...") |
| 155 | if not args.dry_run: |
| 156 | pr.add_to_labels(l) |
| 157 | else: |
| 158 | log(f"Too many labels to be applied") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 159 | |
| 160 | if collab: |
| 161 | reviewers = [] |
| 162 | existing_reviewers = set() |
| 163 | |
| 164 | revs = pr.get_reviews() |
| 165 | for review in revs: |
| 166 | existing_reviewers.add(review.user) |
| 167 | |
| 168 | rl = pr.get_review_requests() |
| 169 | page = 0 |
| 170 | for r in rl: |
| 171 | existing_reviewers |= set(r.get_page(page)) |
| 172 | page += 1 |
| 173 | |
| 174 | for c in collab: |
Anas Nashif | 6027152 | 2022-07-18 19:37:31 -0400 | [diff] [blame] | 175 | try: |
| 176 | u = gh.get_user(c) |
| 177 | if pr.user != u and gh_repo.has_in_collaborators(u): |
| 178 | if u not in existing_reviewers: |
| 179 | reviewers.append(c) |
| 180 | except UnknownObjectException as e: |
| 181 | log(f"Can't get user '{c}', account does not exist anymore? ({e})") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 182 | |
Stephanos Ioannidis | faf4208 | 2022-10-20 21:51:03 +0900 | [diff] [blame] | 183 | if len(existing_reviewers) < 15: |
| 184 | reviewer_vacancy = 15 - len(existing_reviewers) |
| 185 | reviewers = reviewers[:reviewer_vacancy] |
| 186 | |
| 187 | if reviewers: |
| 188 | try: |
| 189 | log(f"adding reviewers {reviewers}...") |
| 190 | if not args.dry_run: |
| 191 | pr.create_review_request(reviewers=reviewers) |
| 192 | except GithubException: |
| 193 | log("cant add reviewer") |
| 194 | else: |
| 195 | log("not adding reviewers because the existing reviewer count is greater than or " |
| 196 | "equal to 15") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 197 | |
| 198 | ms = [] |
| 199 | # assignees |
Anas Nashif | d63c2c4 | 2022-06-16 11:25:52 -0400 | [diff] [blame] | 200 | if maintainer != 'None' and not pr.assignee: |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 201 | try: |
| 202 | u = gh.get_user(maintainer) |
| 203 | ms.append(u) |
| 204 | except GithubException: |
| 205 | log(f"Error: Unknown user") |
| 206 | |
| 207 | for mm in ms: |
| 208 | log(f"Adding assignee {mm}...") |
| 209 | if not args.dry_run: |
| 210 | pr.add_to_assignees(mm) |
Anas Nashif | d63c2c4 | 2022-06-16 11:25:52 -0400 | [diff] [blame] | 211 | else: |
| 212 | log("not setting assignee") |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 213 | |
| 214 | time.sleep(1) |
| 215 | |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 216 | |
Fabio Baltieri | b6cbcba | 2023-08-22 17:04:31 +0000 | [diff] [blame] | 217 | def process_issue(gh, maintainer_file, number): |
| 218 | gh_repo = gh.get_repo(f"{args.org}/{args.repo}") |
| 219 | issue = gh_repo.get_issue(number) |
| 220 | |
| 221 | log(f"Working on {issue.url}: {issue.title}") |
| 222 | |
| 223 | if issue.assignees: |
| 224 | print(f"Already assigned {issue.assignees}, bailing out") |
| 225 | return |
| 226 | |
| 227 | label_to_maintainer = defaultdict(set) |
| 228 | for _, area in maintainer_file.areas.items(): |
| 229 | if not area.labels: |
| 230 | continue |
| 231 | |
| 232 | labels = set() |
| 233 | for label in area.labels: |
| 234 | labels.add(label.lower()) |
| 235 | labels = tuple(sorted(labels)) |
| 236 | |
| 237 | for maintainer in area.maintainers: |
| 238 | label_to_maintainer[labels].add(maintainer) |
| 239 | |
| 240 | # Add extra entries for areas with multiple labels so they match with just |
| 241 | # one label if it's specific enough. |
| 242 | for areas, maintainers in dict(label_to_maintainer).items(): |
| 243 | for area in areas: |
| 244 | if tuple([area]) not in label_to_maintainer: |
| 245 | label_to_maintainer[tuple([area])] = maintainers |
| 246 | |
| 247 | issue_labels = set() |
| 248 | for label in issue.labels: |
| 249 | label_name = label.name.lower() |
| 250 | if tuple([label_name]) not in label_to_maintainer: |
| 251 | print(f"Ignoring label: {label}") |
| 252 | continue |
| 253 | issue_labels.add(label_name) |
| 254 | issue_labels = tuple(sorted(issue_labels)) |
| 255 | |
| 256 | print(f"Using labels: {issue_labels}") |
| 257 | |
| 258 | if issue_labels not in label_to_maintainer: |
| 259 | print(f"no match for the label set, not assigning") |
| 260 | return |
| 261 | |
| 262 | for maintainer in label_to_maintainer[issue_labels]: |
| 263 | log(f"Adding {maintainer} to {issue.html_url}") |
| 264 | if not args.dry_run: |
| 265 | issue.add_to_assignees(maintainer) |
| 266 | |
| 267 | |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 268 | def process_modules(gh, maintainers_file): |
| 269 | manifest = Manifest.from_file() |
| 270 | |
| 271 | repos = {} |
| 272 | for project in manifest.get_projects([]): |
| 273 | if not manifest.is_active(project): |
| 274 | continue |
| 275 | |
| 276 | if isinstance(project, ManifestProject): |
| 277 | continue |
| 278 | |
| 279 | area = f"West project: {project.name}" |
| 280 | if area not in maintainers_file.areas: |
| 281 | log(f"No area for: {area}") |
| 282 | continue |
| 283 | |
| 284 | maintainers = maintainers_file.areas[area].maintainers |
| 285 | if not maintainers: |
| 286 | log(f"No maintainers for: {area}") |
| 287 | continue |
| 288 | |
| 289 | log(f"Found {area}, maintainers={maintainers}") |
| 290 | repo_name = f"{args.org}/{project.name}" |
| 291 | repos[repo_name] = maintainers_file.areas[area] |
| 292 | |
| 293 | query = f"is:open is:pr no:assignee" |
| 294 | for repo in repos: |
| 295 | query += f" repo:{repo}" |
| 296 | |
| 297 | issues = gh.search_issues(query=query) |
| 298 | for issue in issues: |
| 299 | pull = issue.as_pull_request() |
| 300 | |
| 301 | if pull.draft: |
| 302 | continue |
| 303 | |
| 304 | if pull.assignees: |
| 305 | log(f"ERROR: {pull.html_url} should have no assignees, found {pull.assignees}") |
| 306 | continue |
| 307 | |
| 308 | repo_name = f"{args.org}/{issue.repository.name}" |
| 309 | area = repos[repo_name] |
| 310 | |
| 311 | for maintainer in area.maintainers: |
| 312 | log(f"Adding {maintainer} to {pull.html_url}") |
| 313 | if not args.dry_run: |
| 314 | pull.add_to_assignees(maintainer) |
| 315 | |
| 316 | |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 317 | def main(): |
| 318 | parse_args() |
| 319 | |
| 320 | token = os.environ.get('GITHUB_TOKEN', None) |
| 321 | if not token: |
| 322 | sys.exit('Github token not set in environment, please set the ' |
| 323 | 'GITHUB_TOKEN environment variable and retry.') |
| 324 | |
| 325 | gh = Github(token) |
| 326 | maintainer_file = Maintainers(args.maintainer_file) |
| 327 | |
| 328 | if args.pull_request: |
| 329 | process_pr(gh, maintainer_file, args.pull_request) |
Fabio Baltieri | 5e78660 | 2023-08-25 13:49:27 +0000 | [diff] [blame] | 330 | elif args.issue: |
Fabio Baltieri | b6cbcba | 2023-08-22 17:04:31 +0000 | [diff] [blame] | 331 | process_issue(gh, maintainer_file, args.issue) |
Fabio Baltieri | 9a1f4ab | 2023-08-15 14:31:32 +0000 | [diff] [blame] | 332 | elif args.modules: |
| 333 | process_modules(gh, maintainer_file) |
Anas Nashif | 2048316 | 2022-02-25 18:37:47 -0500 | [diff] [blame] | 334 | else: |
| 335 | if args.since: |
| 336 | since = args.since |
| 337 | else: |
| 338 | today = datetime.date.today() |
| 339 | since = today - datetime.timedelta(days=1) |
| 340 | |
| 341 | common_prs = f'repo:{args.org}/{args.repo} is:open is:pr base:main -is:draft no:assignee created:>{since}' |
| 342 | pulls = gh.search_issues(query=f'{common_prs}') |
| 343 | |
| 344 | for issue in pulls: |
| 345 | process_pr(gh, maintainer_file, issue.number) |
| 346 | |
| 347 | |
| 348 | if __name__ == "__main__": |
| 349 | main() |