surshar / claude-code

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

Built by Anthropic · github.com/anthropics

claude-code / scripts / backfill-duplicate-comments.ts

Adam WolffMerge pull request #8160 from anthropics/adam/restore 272a201 · on Sep 25, 2025 History

213 lines · 6.7 KBRaw

1	#!/usr/bin/env bun
2
3	declare global {
4	var process: {
5	env: Record<string, string \| undefined>;
6	};
7	}
8
9	interface GitHubIssue {
10	number: number;
11	title: string;
12	state: string;
13	state_reason?: string;
14	user: { id: number };
15	created_at: string;
16	closed_at?: string;
17	}
18
19	interface GitHubComment {
20	id: number;
21	body: string;
22	created_at: string;
23	user: { type: string; id: number };
24	}
25
26	async function githubRequest<T>(endpoint: string, token: string, method: string = 'GET', body?: any): Promise<T> {
27	const response = await fetch(`https://api.github.com${endpoint}`, {
28	method,
29	headers: {
30	Authorization: `Bearer ${token}`,
31	Accept: "application/vnd.github.v3+json",
32	"User-Agent": "backfill-duplicate-comments-script",
33	...(body && { "Content-Type": "application/json" }),
34	},
35	...(body && { body: JSON.stringify(body) }),
36	});
37
38	if (!response.ok) {
39	throw new Error(
40	`GitHub API request failed: ${response.status} ${response.statusText}`
41	);
42	}
43
44	return response.json();
45	}
46
47	async function triggerDedupeWorkflow(
48	owner: string,
49	repo: string,
50	issueNumber: number,
51	token: string,
52	dryRun: boolean = true
53	): Promise<void> {
54	if (dryRun) {
55	console.log(`[DRY RUN] Would trigger dedupe workflow for issue #${issueNumber}`);
56	return;
57	}
58
59	await githubRequest(
60	`/repos/${owner}/${repo}/actions/workflows/claude-dedupe-issues.yml/dispatches`,
61	token,
62	'POST',
63	{
64	ref: 'main',
65	inputs: {
66	issue_number: issueNumber.toString()
67	}
68	}
69	);
70	}
71
72	async function backfillDuplicateComments(): Promise<void> {
73	console.log("[DEBUG] Starting backfill duplicate comments script");
74
75	const token = process.env.GITHUB_TOKEN;
76	if (!token) {
77	throw new Error(`GITHUB_TOKEN environment variable is required
78
79	Usage:
80	GITHUB_TOKEN=your_token bun run scripts/backfill-duplicate-comments.ts
81
82	Environment Variables:
83	GITHUB_TOKEN - GitHub personal access token with repo and actions permissions (required)
84	DRY_RUN - Set to "false" to actually trigger workflows (default: true for safety)
85	MAX_ISSUE_NUMBER - Only process issues with numbers less than this value (default: 4050)`);
86	}
87	console.log("[DEBUG] GitHub token found");
88
89	const owner = "anthropics";
90	const repo = "claude-code";
91	const dryRun = process.env.DRY_RUN !== "false";
92	const maxIssueNumber = parseInt(process.env.MAX_ISSUE_NUMBER \|\| "4050", 10);
93	const minIssueNumber = parseInt(process.env.MIN_ISSUE_NUMBER \|\| "1", 10);
94
95	console.log(`[DEBUG] Repository: ${owner}/${repo}`);
96	console.log(`[DEBUG] Dry run mode: ${dryRun}`);
97	console.log(`[DEBUG] Looking at issues between #${minIssueNumber} and #${maxIssueNumber}`);
98
99	console.log(`[DEBUG] Fetching issues between #${minIssueNumber} and #${maxIssueNumber}...`);
100	const allIssues: GitHubIssue[] = [];
101	let page = 1;
102	const perPage = 100;
103
104	while (true) {
105	const pageIssues: GitHubIssue[] = await githubRequest(
106	`/repos/${owner}/${repo}/issues?state=all&per_page=${perPage}&page=${page}&sort=created&direction=desc`,
107	token
108	);
109
110	if (pageIssues.length === 0) break;
111
112	// Filter to only include issues within the specified range
113	const filteredIssues = pageIssues.filter(issue =>
114	issue.number >= minIssueNumber && issue.number < maxIssueNumber
115	);
116	allIssues.push(...filteredIssues);
117
118	// If the oldest issue in this page is still above our minimum, we need to continue
119	// but if the oldest issue is below our minimum, we can stop
120	const oldestIssueInPage = pageIssues[pageIssues.length - 1];
121	if (oldestIssueInPage && oldestIssueInPage.number >= maxIssueNumber) {
122	console.log(`[DEBUG] Oldest issue in page #${page} is #${oldestIssueInPage.number}, continuing...`);
123	} else if (oldestIssueInPage && oldestIssueInPage.number < minIssueNumber) {
124	console.log(`[DEBUG] Oldest issue in page #${page} is #${oldestIssueInPage.number}, below minimum, stopping`);
125	break;
126	} else if (filteredIssues.length === 0 && pageIssues.length > 0) {
127	console.log(`[DEBUG] No issues in page #${page} are in range #${minIssueNumber}-#${maxIssueNumber}, continuing...`);
128	}
129
130	page++;
131
132	// Safety limit to avoid infinite loops
133	if (page > 200) {
134	console.log("[DEBUG] Reached page limit, stopping pagination");
135	break;
136	}
137	}
138
139	console.log(`[DEBUG] Found ${allIssues.length} issues between #${minIssueNumber} and #${maxIssueNumber}`);
140
141	let processedCount = 0;
142	let candidateCount = 0;
143	let triggeredCount = 0;
144
145	for (const issue of allIssues) {
146	processedCount++;
147	console.log(
148	`[DEBUG] Processing issue #${issue.number} (${processedCount}/${allIssues.length}): ${issue.title}`
149	);
150
151	console.log(`[DEBUG] Fetching comments for issue #${issue.number}...`);
152	const comments: GitHubComment[] = await githubRequest(
153	`/repos/${owner}/${repo}/issues/${issue.number}/comments`,
154	token
155	);
156	console.log(
157	`[DEBUG] Issue #${issue.number} has ${comments.length} comments`
158	);
159
160	// Look for existing duplicate detection comments (from the dedupe bot)
161	const dupeDetectionComments = comments.filter(
162	(comment) =>
163	comment.body.includes("Found") &&
164	comment.body.includes("possible duplicate") &&
165	comment.user.type === "Bot"
166	);
167
168	console.log(
169	`[DEBUG] Issue #${issue.number} has ${dupeDetectionComments.length} duplicate detection comments`
170	);
171
172	// Skip if there's already a duplicate detection comment
173	if (dupeDetectionComments.length > 0) {
174	console.log(
175	`[DEBUG] Issue #${issue.number} already has duplicate detection comment, skipping`
176	);
177	continue;
178	}
179
180	candidateCount++;
181	const issueUrl = `https://github.com/${owner}/${repo}/issues/${issue.number}`;
182
183	try {
184	console.log(
185	`[INFO] ${dryRun ? '[DRY RUN] ' : ''}Triggering dedupe workflow for issue #${issue.number}: ${issueUrl}`
186	);
187	await triggerDedupeWorkflow(owner, repo, issue.number, token, dryRun);
188
189	if (!dryRun) {
190	console.log(
191	`[SUCCESS] Successfully triggered dedupe workflow for issue #${issue.number}`
192	);
193	}
194	triggeredCount++;
195	} catch (error) {
196	console.error(
197	`[ERROR] Failed to trigger workflow for issue #${issue.number}: ${error}`
198	);
199	}
200
201	// Add a delay between workflow triggers to avoid overwhelming the system
202	await new Promise(resolve => setTimeout(resolve, 1000));
203	}
204
205	console.log(
206	`[DEBUG] Script completed. Processed ${processedCount} issues, found ${candidateCount} candidates without duplicate comments, ${dryRun ? 'would trigger' : 'triggered'} ${triggeredCount} workflows`
207	);
208	}
209
210	backfillDuplicateComments().catch(console.error);
211
212	// Make it a module
213	export {};