repo-dive 0.4.0 → 0.4.1

package/README.md

@@ -68,8 +68,8 @@ export default defineConfig({
       ["alice@work.example", "alice@personal.example"],
       // Rich form: a display name, a profile link and an explicit kind.
       {
-        emails: ["bob@work.example", "12345+bob@users.noreply.github.com"],
         displayName: "Bob",
+        emails: ["bob@work.example", "12345+bob@users.noreply.github.com"],
         url: "https://github.com/bob",
       },
     ],

package/dist/cli.js

@@ -52170,7 +52170,7 @@ var runWith = (command, config) => {
 };
 var package_default = {
 	name: "repo-dive",
-	version: "0.4.0",
+	version: "0.4.1",
 	description: "Dive into a git repository's history: per-commit snapshots, an indexed metrics catalog and an interactive dashboard",
 	keywords: [
 		"git",
@@ -52364,6 +52364,15 @@ var recordAt = (value, key) => asRecord$1(asRecord$1(value)?.[key]) ?? {};
 var arrayAt = (value, key) => asArray(asRecord$1(value)?.[key]);
 //#endregion
 //#region src/lib/collectors/types.ts
+/**
+* Whether a collector describes the *state of the tree* at a commit rather than
+* facts about the commit itself. Such snapshots are only meaningful on the
+* first-parent chain: a commit that lives on a side branch — or that arrived
+* with a foreign history absorbed by an unrelated-histories merge — carries a
+* tree that was never the repository's state, so charting it puts a cliff into
+* every timeline.
+*/
+var describesTreeState = (collector) => collector.strategy !== "log";
 /** File extension used as a category key, e.g. ".ts"; files without one map to "(none)". */
 var extensionOf = (filePath) => {
 	const basename = filePath.split("/").at(-1) ?? "";
@@ -53546,12 +53555,57 @@ var resolveCollectors = (names) => {
 	return resolved;
 };
 //#endregion
+//#region src/lib/sampling.ts
+var parseSamplingPolicy = (input) => {
+	if (input === "all" || input === "weekly" || input === "monthly" || input === "quarterly") return input;
+	const everyNthMatch = /^every-nth:(\d+)$/.exec(input);
+	if (everyNthMatch?.[1]) {
+		const everyNth = Number(everyNthMatch[1]);
+		if (everyNth >= 1) return { everyNth };
+	}
+	return /* @__PURE__ */ new Error(`Unknown sampling policy: ${input}. Expected all, weekly, monthly, quarterly or every-nth:<n>.`);
+};
+/** How a policy is spelled in CLI output (`collectors`, `scan`, `status`). */
+var samplingLabel = (policy) => typeof policy === "object" ? `every-nth:${policy.everyNth}` : policy;
+var isoWeekOf = (date) => {
+	const utc = new Date(Date.UTC(date.getUTCFullYear(), date.getUTCMonth(), date.getUTCDate()));
+	const dayOfWeek = utc.getUTCDay() || 7;
+	utc.setUTCDate(utc.getUTCDate() + 4 - dayOfWeek);
+	const yearStart = Date.UTC(utc.getUTCFullYear(), 0, 1);
+	const week = Math.ceil(((utc.getTime() - yearStart) / 864e5 + 1) / 7);
+	return `${utc.getUTCFullYear()}-W${String(week).padStart(2, "0")}`;
+};
+var bucketOf = (policy, authorDate) => {
+	const date = new Date(authorDate);
+	if (policy === "weekly") return isoWeekOf(date);
+	const month = date.getUTCMonth();
+	return policy === "monthly" ? `${date.getUTCFullYear()}-${String(month + 1).padStart(2, "0")}` : `${date.getUTCFullYear()}-Q${Math.floor(month / 3) + 1}`;
+};
+/**
+* Picks the sampled subset of commits for a policy. `commits` must be ordered
+* newest first (as `git log` emits them); period policies keep the newest
+* commit of each period, so HEAD is always included.
+*/
+var sampleCommits = (commits, policy) => {
+	if (policy === "all") return [...commits];
+	if (typeof policy === "object") return commits.filter((_, index) => index % policy.everyNth === 0);
+	const seenBuckets = /* @__PURE__ */ new Set();
+	const sampled = [];
+	for (const commit of commits) {
+		const bucket = bucketOf(policy, commit.authorDate);
+		if (!seenBuckets.has(bucket)) {
+			seenBuckets.add(bucket);
+			sampled.push(commit);
+		}
+	}
+	return sampled;
+};
+//#endregion
 //#region src/commands/collectors.ts
-var samplingLabel$1 = (policy) => typeof policy === "object" ? `every-nth:${policy.everyNth}` : policy;
 var collectorsCommand = make$3("collectors").pipe(withDescription("List available collectors, their versions, strategies and default sampling"), withHandler(() => gen(function* () {
 	for (const collector of builtInCollectors) yield* log([
 		`${collector.name} (v${collector.version})`,
-		`  strategy: ${collector.strategy}, sampling: ${samplingLabel$1(collector.defaultSampling)}`,
+		`  strategy: ${collector.strategy}, sampling: ${samplingLabel(collector.defaultSampling)}`,
 		`  ${collector.description}`
 	].join("\n"));
 })));
@@ -53817,6 +53871,7 @@ var runIndex = ({ repoPath }) => gen(function* () {
 	const commitsPath = path.join(catalogPath, "commits");
 	const registry = new Map(builtInCollectors.map((collector) => [collector.name, collector]));
 	const gitCommits = yield* listCommits(repoRoot);
+	const firstParentShas = yield* listFirstParentShas(repoRoot);
 	const catalogShas = new Set(yield* tryPromise({
 		try: async () => {
 			try {
@@ -53830,10 +53885,12 @@ var runIndex = ({ repoPath }) => gen(function* () {
 	const orderedCommits = gitCommits.toReversed().filter((commit) => catalogShas.has(commit.hash));
 	if (orderedCommits.length === 0) return yield* fail$3(/* @__PURE__ */ new Error(`No collected commits found in ${commitsPath} — run \`repo-dive scan\` first.`));
 	let unknownCollectorDirs = 0;
+	let offMainlineSnapshots = 0;
 	const commitFacts = [];
 	yield* forEach$1(orderedCommits, (commit) => tryPromise({
 		try: async () => {
 			const commitDir = path.join(commitsPath, commit.hash);
+			const onMainline = firstParentShas.has(commit.hash);
 			const factsByCollector = /* @__PURE__ */ new Map();
 			for (const collectorName of await readdir(commitDir)) {
 				const collector = registry.get(collectorName);
@@ -53841,6 +53898,10 @@ var runIndex = ({ repoPath }) => gen(function* () {
 					unknownCollectorDirs += 1;
 					continue;
 				}
+				if (!onMainline && describesTreeState(collector)) {
+					offMainlineSnapshots += 1;
+					continue;
+				}
 				const raw = JSON.parse(await readFile(path.join(commitDir, collectorName, "output.json"), "utf8"));
 				factsByCollector.set(collectorName, collector.normalize(raw));
 			}
@@ -53882,7 +53943,8 @@ var runIndex = ({ repoPath }) => gen(function* () {
 		`Indexed ${commitFacts.length} commits into ${factCount} facts.`,
 		`Cube: ${dbPath}`,
 		`Dashboard data: ${dashboardPath}`,
-		...unknownCollectorDirs > 0 ? [`Skipped ${unknownCollectorDirs} outputs from unknown collectors (see \`gc --stale\`).`] : []
+		...unknownCollectorDirs > 0 ? [`Skipped ${unknownCollectorDirs} outputs from unknown collectors (see \`gc --stale\`).`] : [],
+		...offMainlineSnapshots > 0 ? [`Skipped ${offMainlineSnapshots} tree snapshots taken off HEAD's first-parent chain.`] : []
 	].join("\n"));
 });
 //#endregion
@@ -54035,50 +54097,6 @@ var loadConfig = (repoRoot) => gen(function* () {
 	});
 });
 //#endregion
-//#region src/lib/sampling.ts
-var parseSamplingPolicy = (input) => {
-	if (input === "all" || input === "weekly" || input === "monthly" || input === "quarterly") return input;
-	const everyNthMatch = /^every-nth:(\d+)$/.exec(input);
-	if (everyNthMatch?.[1]) {
-		const everyNth = Number(everyNthMatch[1]);
-		if (everyNth >= 1) return { everyNth };
-	}
-	return /* @__PURE__ */ new Error(`Unknown sampling policy: ${input}. Expected all, weekly, monthly, quarterly or every-nth:<n>.`);
-};
-var isoWeekOf = (date) => {
-	const utc = new Date(Date.UTC(date.getUTCFullYear(), date.getUTCMonth(), date.getUTCDate()));
-	const dayOfWeek = utc.getUTCDay() || 7;
-	utc.setUTCDate(utc.getUTCDate() + 4 - dayOfWeek);
-	const yearStart = Date.UTC(utc.getUTCFullYear(), 0, 1);
-	const week = Math.ceil(((utc.getTime() - yearStart) / 864e5 + 1) / 7);
-	return `${utc.getUTCFullYear()}-W${String(week).padStart(2, "0")}`;
-};
-var bucketOf = (policy, authorDate) => {
-	const date = new Date(authorDate);
-	if (policy === "weekly") return isoWeekOf(date);
-	const month = date.getUTCMonth();
-	return policy === "monthly" ? `${date.getUTCFullYear()}-${String(month + 1).padStart(2, "0")}` : `${date.getUTCFullYear()}-Q${Math.floor(month / 3) + 1}`;
-};
-/**
-* Picks the sampled subset of commits for a policy. `commits` must be ordered
-* newest first (as `git log` emits them); period policies keep the newest
-* commit of each period, so HEAD is always included.
-*/
-var sampleCommits = (commits, policy) => {
-	if (policy === "all") return [...commits];
-	if (typeof policy === "object") return commits.filter((_, index) => index % policy.everyNth === 0);
-	const seenBuckets = /* @__PURE__ */ new Set();
-	const sampled = [];
-	for (const commit of commits) {
-		const bucket = bucketOf(policy, commit.authorDate);
-		if (!seenBuckets.has(bucket)) {
-			seenBuckets.add(bucket);
-			sampled.push(commit);
-		}
-	}
-	return sampled;
-};
-//#endregion
 //#region src/lib/worktree.ts
 var toError$4 = (error) => error instanceof Error ? error : new Error(String(error));
 /**
@@ -54157,6 +54175,18 @@ var listCommits = (repoRoot) => runGit([
 	"log",
 	`--format=${gitLogFormat}`
 ]).pipe(catch_$2((error) => error instanceof GitCommandError && error.stderr.includes("does not have any commits yet") ? succeed$3("") : fail$3(error)), map$7(parseGitLog));
+/**
+* Shas on HEAD's first-parent chain, i.e. the states the repository actually
+* passed through. See {@link describesTreeState} for why snapshot collectors
+* are restricted to them.
+*/
+var listFirstParentShas = (repoRoot) => runGit([
+	"-C",
+	repoRoot,
+	"log",
+	"--first-parent",
+	"--format=%H"
+]).pipe(catch_$2((error) => error instanceof GitCommandError && error.stderr.includes("does not have any commits yet") ? succeed$3("") : fail$3(error)), map$7((stdout) => new Set(stdout.split("\n").filter(Boolean))));
 var summarizeCommits = (commits) => {
 	const authorEmails = new Set(commits.map((commit) => commit.authorEmail));
 	const dates = commits.map((commit) => commit.authorDate).filter(Boolean).toSorted();
@@ -54220,7 +54250,6 @@ var collectCommit = ({ catalog, sha, collectors, cacheKeyOf, force, failures })
 		skipped
 	};
 });
-var samplingLabel = (policy) => typeof policy === "object" ? `every-nth:${policy.everyNth}` : policy;
 var runScan = ({ repoPath, collectorNames, maxCommits, sample, force = false }) => gen(function* () {
 	const collectors = resolveCollectors(collectorNames);
 	if (collectors instanceof Error) return yield* fail$3(collectors);
@@ -54238,12 +54267,14 @@ var runScan = ({ repoPath, collectorNames, maxCommits, sample, force = false })
 	const config = yield* loadConfig(repoRoot);
 	const cacheKeys = new Map(collectors.map((collector) => [collector.name, collectorCacheKey(collector, config)]));
 	const cacheKeyOf = (collector) => cacheKeys.get(collector.name) ?? collectorCacheKey(collector, config);
+	const firstParentShas = yield* listFirstParentShas(repoRoot);
 	const plans = collectors.map((collector) => {
 		const policy = sampleOverride ?? collector.defaultSampling;
+		const candidates = describesTreeState(collector) ? selected.filter((commit) => firstParentShas.has(commit.hash)) : selected;
 		return {
 			collector,
 			policy,
-			shas: new Set(sampleCommits(selected, policy).map((commit) => commit.hash))
+			shas: new Set(sampleCommits(candidates, policy).map((commit) => commit.hash))
 		};
 	});
 	yield* log(`Plan: ${plans.map((plan) => `${plan.collector.name} → ${plan.shas.size} commits (${samplingLabel(plan.policy)})`).join(", ")}`);
@@ -59199,6 +59230,7 @@ var exists = (filePath) => promise(() => access(filePath).then(() => true, () =>
 var runStatus = ({ repoPath }) => gen(function* () {
 	const repoRoot = yield* resolveRepoRoot(repoPath);
 	const commits = yield* listCommits(repoRoot);
+	const firstParentShas = yield* listFirstParentShas(repoRoot);
 	const config = yield* loadConfig(repoRoot);
 	const catalogPath = path.join(repoRoot, catalogDirName);
 	if (!(yield* exists(path.join(catalogPath, "catalog.json")))) {
@@ -59220,15 +59252,16 @@ var runStatus = ({ repoPath }) => gen(function* () {
 		`Catalog: ${catalogPath}`
 	];
 	for (const collector of builtInCollectors) {
+		const target = sampleCommits(describesTreeState(collector) ? commits.filter((commit) => firstParentShas.has(commit.hash)) : commits, collector.defaultSampling);
 		let collected = 0;
 		const cacheKey = collectorCacheKey(collector, config);
-		yield* forEach$1(commits, (commit) => isCollected(catalog, commit.hash, collector, cacheKey).pipe(map$7((done) => {
+		yield* forEach$1(target, (commit) => isCollected(catalog, commit.hash, collector, cacheKey).pipe(map$7((done) => {
 			if (done) collected += 1;
 		})), {
 			concurrency: 16,
 			discard: true
 		});
-		lines.push(`  ${collector.name}: ${collected}/${commits.length} commits collected`);
+		lines.push(collector.defaultSampling === "all" ? `  ${collector.name}: ${collected}/${target.length} commits collected` : `  ${collector.name}: ${collected}/${target.length} commits collected (${samplingLabel(collector.defaultSampling)} sample of ${commits.length})`);
 	}
 	yield* log(lines.join("\n"));
 });

package/dist/config.d.ts

@@ -15,8 +15,8 @@
  *       ["alice@work.example", "alice@personal.example"],
  *       // Rich form: a display name, a profile link and a kind.
  *       {
- *         emails: ["bob@work.example", "bob@personal.example"],
  *         displayName: "Bob",
+ *         emails: ["bob@work.example", "bob@personal.example"],
  *         url: "https://github.com/bob",
  *       },
  *     ],

package/dist/config.js

@@ -15,8 +15,8 @@
  *       ["alice@work.example", "alice@personal.example"],
  *       // Rich form: a display name, a profile link and a kind.
  *       {
- *         emails: ["bob@work.example", "bob@personal.example"],
  *         displayName: "Bob",
+ *         emails: ["bob@work.example", "bob@personal.example"],
  *         url: "https://github.com/bob",
  *       },
  *     ],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "repo-dive",
-  "version": "0.4.0",
+  "version": "0.4.1",
   "description": "Dive into a git repository's history: per-commit snapshots, an indexed metrics catalog and an interactive dashboard",
   "keywords": [
     "git",