Blame · Line-by-line history

`hot-files.ts`

Each line is annotated with the commit that last touched it. Click any SHA to jump to that commit and see the surrounding change.

hot-files.tsBlame160 lines · 1 contributor

Normal view Raw

74d8c4dClaude	1	/**
	2	* Hot Files analysis — git log --numstat based file churn.
	3	*
	4	* Spawns git to count how frequently each file has been modified within
	5	* a sliding time window. Returns the top 50 files ranked by churn
	6	* (total lines added + deleted), annotated with a risk level.
	7	*/
	8
	9	import { join } from "path";
	10
	11	// ─── Public types ─────────────────────────────────────────────────────────────
	12
	13	export interface HotFile {
	14	/** Repo-relative path */
	15	path: string;
	16	/** Number of commits in which this file appears */
	17	changes: number;
	18	/** Total lines added across all commits */
	19	added: number;
	20	/** Total lines deleted across all commits */
	21	deleted: number;
	22	/** added + deleted */
	23	churn: number;
	24	/** Computed risk tier */
	25	riskLevel: "high" \| "medium" \| "low";
	26	/** File extension without leading dot (e.g. "ts", "py") */
	27	ext: string;
	28	}
	29
	30	// ─── Risk heuristic ───────────────────────────────────────────────────────────
	31
	32	const HIGH_RISK_PATTERNS = [
	33	"auth",
	34	"security",
	35	"schema",
	36	"db/",
	37	"middleware",
	38	"routes/git",
	39	"crypto",
	40	];
	41
	42	const MEDIUM_RISK_PATTERNS = ["route", "api", "lib/", ".sql"];
	43
	44	function classifyRisk(filePath: string): "high" \| "medium" \| "low" {
	45	const lower = filePath.toLowerCase();
	46	if (HIGH_RISK_PATTERNS.some((p) => lower.includes(p))) return "high";
	47	if (MEDIUM_RISK_PATTERNS.some((p) => lower.includes(p))) return "medium";
	48	return "low";
	49	}
	50
	51	function extractExt(filePath: string): string {
	52	const dot = filePath.lastIndexOf(".");
	53	if (dot === -1 \|\| dot === filePath.length - 1) return "";
	54	return filePath.slice(dot + 1);
	55	}
	56
	57	// ─── Core function ────────────────────────────────────────────────────────────
	58
	59	/**
	60	* Returns the top ≤50 most-churned files in the repo within the last
	61	* `windowDays` days, ordered by churn (lines added + deleted) descending.
	62	*
	63	* The result is empty when the git repo path does not exist, git fails,
	64	* or there are no commits in the window.
	65	*/
	66	export async function getHotFiles(
	67	ownerName: string,
	68	repoName: string,
	69	windowDays: number
	70	): Promise<HotFile[]> {
	71	const repoBase = process.env.GIT_REPOS_PATH \|\| "./repos";
	72	const diskPath = join(repoBase, `${ownerName}/${repoName}.git`);
	73
	74	// ─── Spawn git ──────────────────────────────────────────────────────────
	75
	76	let raw = "";
	77	try {
	78	const proc = Bun.spawn(
	79	[
	80	"git",
	81	"--git-dir",
	82	diskPath,
	83	"log",
	84	"--numstat",
	85	`--since=${windowDays}.days.ago`,
	86	"--format=",
	87	],
	88	{ stdout: "pipe", stderr: "pipe" }
	89	);
	90	raw = await new Response(proc.stdout as ReadableStream).text();
	91	await proc.exited;
	92	} catch {
	93	return [];
	94	}
	95
	96	if (!raw.trim()) return [];
	97
	98	// ─── Parse --numstat lines ───────────────────────────────────────────────
	99	//
	100	// git --numstat emits lines of the form:
	101	// <added>\t<deleted>\t<path>
	102	//
	103	// When --format= is used the commit header lines are blank, so we only
	104	// see the numstat data lines (non-blank lines starting with a digit or "-").
	105	// Binary files show "-\t-\t<path>"; we treat those as 0/0.
	106
	107	/** Per-file aggregation keyed by file path. */
	108	const fileMap = new Map<
	109	string,
	110	{ changes: number; added: number; deleted: number }
	111	>();
	112
	113	for (const line of raw.split("\n")) {
	114	const trimmed = line.trim();
	115	if (!trimmed) continue;
	116
	117	const parts = trimmed.split("\t");
	118	if (parts.length < 3) continue;
	119
	120	const [rawAdded, rawDeleted, ...pathParts] = parts;
	121	const filePath = pathParts.join("\t"); // guard against tabs in filenames
	122	if (!filePath) continue;
	123
	124	const added = rawAdded === "-" ? 0 : parseInt(rawAdded, 10);
	125	const deleted = rawDeleted === "-" ? 0 : parseInt(rawDeleted, 10);
	126
	127	if (isNaN(added) \|\| isNaN(deleted)) continue;
	128
	129	const existing = fileMap.get(filePath);
	130	if (existing) {
	131	existing.changes += 1;
	132	existing.added += added;
	133	existing.deleted += deleted;
	134	} else {
	135	fileMap.set(filePath, { changes: 1, added, deleted });
	136	}
	137	}
	138
	139	if (fileMap.size === 0) return [];
	140
	141	// ─── Sort and cap ────────────────────────────────────────────────────────
	142
	143	const results: HotFile[] = [];
	144	for (const [path, agg] of fileMap) {
	145	const churn = agg.added + agg.deleted;
	146	results.push({
	147	path,
	148	changes: agg.changes,
	149	added: agg.added,
	150	deleted: agg.deleted,
	151	churn,
	152	riskLevel: classifyRisk(path),
	153	ext: extractExt(path),
	154	});
	155	}
	156
	157	results.sort((a, b) => b.churn - a.churn);
	158
	159	return results.slice(0, 50);
	160	}