github-label-issues-to-triage.py

 1#!/usr/bin/env python3
 2"""Add `state:needs triage` label to open GitHub issues of types Bug and Crash
 3if they're missing area, priority, or frequency labels. Don't touch issues
 4with an assignee or another `state:` label.
 5
 6Requires `requests` library and a GitHub access token with "Issues (write)"
 7permission passed as an environment variable. Was used as a quick-and-dirty
 8one-off-bulk-operation script to surface older untriaged issues in the `zed`
 9repository. Leaving it here for reference only; there's no error handling or
10guardrails, you've been warned.
11"""
12
13import itertools
14import logging
15import os
16
17import requests
18
19
20logging.basicConfig(level=logging.INFO)
21log = logging.getLogger(__name__)
22
23GITHUB_API_BASE_URL = "https://api.github.com"
24REPO_OWNER = "zed-industries"
25REPO_NAME = "zed"
26GITHUB_TOKEN = os.getenv("GITHUB_TOKEN")
27HEADERS = {
28    "Authorization": f"token {GITHUB_TOKEN}",
29    "Accept": "application/vnd.github+json"
30}
31REQUIRED_LABELS_PREFIXES = ["area:", "priority:", "frequency:"]
32NEEDS_TRIAGE_LABEL = "state:needs triage"
33
34
35def get_open_issues(repo, issue_type):
36    """Get open issues of certain type(s) via GitHub's REST API."""
37    issues = []
38    issues_url = f"{GITHUB_API_BASE_URL}/repos/{REPO_OWNER}/{repo}/issues"
39
40    log.info("Start fetching open issues from the GitHub API.")
41    params = {
42        "state": "open",
43        "type": issue_type,
44        "page": 1,
45        "per_page": 100, # worked fine despite the docs saying 30
46    }
47    while True:
48        response = requests.get(issues_url, headers=HEADERS, params=params)
49        response.raise_for_status()
50        issues.extend(response.json())
51        log.info(f"Fetched the next page, total issues so far: {len(issues)}.")
52
53        # is there a next page?
54        link_header = response.headers.get('Link', '')
55        if 'rel="next"' not in link_header:
56            break
57        params['page'] += 1
58
59    log.info("Done fetching issues.")
60    return issues
61
62
63def is_untriaged(issue):
64    issue_labels = [label['name'] for label in issue['labels']]
65    # don't want to overwrite existing state labels
66    no_state_label = not any(label.startswith('state:') for label in issue_labels)
67    # we want at least one label for each of the required prefixes
68    has_all_required_labels = all(
69        any(label.startswith(prefix) for label in issue_labels)
70        for prefix in REQUIRED_LABELS_PREFIXES
71    )
72    # let's also assume if we managed to assign an issue it's triaged enough
73    no_assignee = not issue['assignee']
74    return no_state_label and no_assignee and not has_all_required_labels
75
76
77def label_issues(issues, label):
78    for issue in issues:
79        log.debug(f"Processing issue {issue['number']}.")
80        api_url_add_label = f"{GITHUB_API_BASE_URL}/repos/{REPO_OWNER}/{REPO_NAME}/issues/{issue['number']}/labels"
81        add_response = requests.post(
82            api_url_add_label, headers=HEADERS, json={"labels": [label]}
83        )
84        add_response.raise_for_status()
85        log.info(f"Added label '{label}' to issue {issue['title']}.")
86
87
88if __name__ == "__main__":
89    open_bugs = get_open_issues(REPO_NAME, "Bug")
90    open_crashes = get_open_issues(REPO_NAME, "Crash")
91    untriaged_issues = filter(
92        is_untriaged, itertools.chain(open_bugs, open_crashes))
93    label_issues(untriaged_issues, label=NEEDS_TRIAGE_LABEL)