co-maintainer 0.4.11 → 0.4.13

package/dist/package.json

@@ -1,6 +1,6 @@
 {
     "name": "co-maintainer",
-    "version": "0.4.11",
+    "version": "0.4.13",
     "description": "Analyzes a GitHub repository and writes repository-specific contribution guidance.",
     "license": "MIT",
     "repository": {

package/dist/src/github/collect.d.ts

@@ -1,6 +1,6 @@
 import type { GitHubClient, Options } from "../types.ts";
 import type { PullRequest, Source, State } from "../knowledge/types.ts";
-export declare function collectSource(client: GitHubClient, options: Options, previous?: State, progress?: (items: PullRequest[]) => Promise<void>, codebaseProgress?: (data: {
+export declare function collectSource(client: GitHubClient, options: Options, previous?: State, progress?: (items: PullRequest[], cache: PullRequest[]) => Promise<void>, codebaseProgress?: (data: {
     tree: string[];
     treeSha: Record<string, string>;
     files: Record<string, string>;

package/dist/src/github/collect.js

@@ -248,8 +248,10 @@ async function pullRequests(client, options, previous, phase, progress) {
     if (phase)
         phase.text = "listing pull requests · page 0 · fetched 0";
     const selected = await listPullRequestPages(client, options, phase);
-    const previousByNumber = new Map((previous?.source.pullRequests ?? []).map((pr) => [pr.number, pr]));
+    const keptByNumber = new Map((previous?.source.pullRequests ?? []).map((pr) => [pr.number, pr]));
+    const cacheByNumber = new Map((previous?.source.pullRequestCache ?? []).map((pr) => [pr.number, pr]));
     const result = new Array(selected.length);
+    const cache = new Array(selected.length);
     let completed = 0;
     let save = Promise.resolve();
     if (phase) {
@@ -258,7 +260,8 @@ async function pullRequests(client, options, previous, phase, progress) {
     log("fetch", `${selected.length} pull requests · concurrency=${options.ghConcurrent}`);
     await mapPool(selected, options.ghConcurrent, async (pr, index) => {
         const number = Number(pr.number);
-        const cached = previousByNumber.get(number);
+        const full = keptByNumber.get(number);
+        const cached = full ?? cacheByNumber.get(number);
         const listedAdditions = Number(pr.additions);
         const listedDeletions = Number(pr.deletions);
         const listedStatsAvailable = Number.isFinite(listedAdditions) &&
@@ -308,10 +311,12 @@ async function pullRequests(client, options, previous, phase, progress) {
             current.changesRequested = cached.changesRequested;
         }
         const only = options.onlyRequestChangedPr;
-        const discussionUnchanged = cached?.updatedAt === current.updatedAt;
+        const discussionUnchanged = full?.updatedAt === current.updatedAt;
         const decisionKnown = typeof cached?.changesRequested === "boolean";
-        const diffUnchanged = cached?.headSha === current.headSha && Boolean(cached?.diff);
-        let droppedFromCache = only && discussionUnchanged && cached?.changesRequested === false;
+        const diffUnchanged = full?.headSha === current.headSha && Boolean(full?.diff);
+        let droppedFromCache = only &&
+            cached?.updatedAt === current.updatedAt &&
+            cached.changesRequested === false;
         let dropped = droppedFromCache;
         const loadComments = async () => {
             const comments = await client.pages(`repos/${options.repo}/issues/${number}/comments`);
@@ -383,6 +388,7 @@ async function pullRequests(client, options, previous, phase, progress) {
                 current.diff = "";
             }
         }
+        cache[index] = current;
         if (!dropped)
             result[index] = current;
         completed++;
@@ -391,29 +397,82 @@ async function pullRequests(client, options, previous, phase, progress) {
         }
         log("fetch", `PR ${completed}/${selected.length} · ${percent(completed, selected.length)} #${number} · ${discussionStatus} · ${diffStatus}`);
         if (progress) {
-            save = save.then(() => progress(result.filter((item) => !!item)));
+            save = save.then(() => progress(result.filter((item) => !!item), cache.filter((item) => !!item)));
             await save;
         }
     });
     const kept = result.filter((item) => !!item);
+    const decided = cache.filter((item) => !!item);
     if (options.onlyRequestChangedPr) {
         log("fetch", `only request-changed pr · kept ${kept.length} · dropped ${selected.length - kept.length}`);
     }
-    return kept;
+    return { kept, cache: decided };
 }
-async function commits(client, options, meta, phase) {
+function commitLimit(maxCommits) {
+    return maxCommits && maxCommits > 0 ? maxCommits : undefined;
+}
+function commitsCovered(cached, previousMax, limit) {
+    if (cached.length === 0)
+        return false;
+    const previousAll = commitLimit(previousMax) === undefined;
+    if (limit === undefined)
+        return previousAll;
+    return previousAll || cached.length >= limit;
+}
+async function commits(client, options, meta, previous, phase) {
     const branch = String(meta.default_branch ?? "main");
-    const limit = options.maxCommits && options.maxCommits > 0
-        ? options.maxCommits
-        : undefined;
-    if (phase)
-        phase.text = "commit history · page 0 · fetched 0";
-    return client.pages(`repos/${options.repo}/commits?sha=${encodeURIComponent(branch)}`, limit, (page, fetched) => {
+    const limit = commitLimit(options.maxCommits);
+    const cached = previous?.source.commits ?? [];
+    const cachedTip = String(cached[0]?.sha ?? "");
+    const endpoint = `repos/${options.repo}/commits?sha=${encodeURIComponent(branch)}`;
+    const note = (page, fetched) => {
         const message = `commit history · page ${page} · fetched ${fetched} · total unknown`;
         if (phase)
             phase.text = message;
         log("fetch", message);
-    });
+    };
+    if (phase)
+        phase.text = "commit history · page 0 · fetched 0";
+    if (!cachedTip) {
+        return client.pages(endpoint, limit, note);
+    }
+    const fresh = [];
+    let found = false;
+    for (let page = 1;; page++) {
+        const pageItems = await client.request(`${endpoint}&per_page=100&page=${page}`);
+        const items = Array.isArray(pageItems) ? pageItems : [];
+        for (const commit of items) {
+            if (String(commit.sha ?? "") === cachedTip) {
+                found = true;
+                break;
+            }
+            fresh.push(commit);
+            if (limit !== undefined && fresh.length >= limit)
+                break;
+        }
+        note(page, fresh.length);
+        if (found ||
+            items.length < 100 ||
+            (limit !== undefined && fresh.length >= limit)) {
+            break;
+        }
+    }
+    if (!found || !commitsCovered(cached, previous?.options.maxCommits, limit)) {
+        if (found)
+            return client.pages(endpoint, limit, note);
+        return limit === undefined ? fresh : fresh.slice(0, limit);
+    }
+    const selected = [...fresh];
+    for (const commit of cached) {
+        if (limit !== undefined && selected.length >= limit)
+            break;
+        selected.push(commit);
+    }
+    const reused = selected.length - fresh.length;
+    if (reused > 0) {
+        log("fetch", `commit history · reused ${reused} cached commits · ${selected.length} total`);
+    }
+    return selected;
 }
 export async function collectSource(client, options, previous, progress, codebaseProgress) {
     const phase = { text: "reading repository metadata" };
@@ -426,16 +485,17 @@ export async function collectSource(client, options, previous, progress, codebas
             : { tree: [], treeSha: {}, files: {} };
         const pullRequestData = options.includePullRequests
             ? await pullRequests(client, options, previous, phase, progress)
-            : [];
+            : { kept: [], cache: previous?.source.pullRequestCache ?? [] };
         const commitData = options.includeCommitHistory
-            ? await commits(client, options, repo, phase)
+            ? await commits(client, options, repo, previous, phase)
             : [];
         return {
             repo,
             tree: files.tree,
             treeSha: files.treeSha,
             files: files.files,
-            pullRequests: pullRequestData,
+            pullRequests: pullRequestData.kept,
+            pullRequestCache: pullRequestData.cache,
             commits: commitData,
         };
     }

package/dist/src/knowledge/types.d.ts

@@ -23,7 +23,11 @@ export type Source = {
     tree: string[];
     treeSha: Record<string, string>;
     files: Record<string, string>;
+    /** Pull requests selected for facts and the skill. */
     pullRequests: PullRequest[];
+    /** Every pull request already decided, including ones left out of the skill.
+     * The next fetch reads this so a `changesRequested: false` result is not downloaded again. */
+    pullRequestCache?: PullRequest[];
     commits: Json[];
 };
 export type Fact = {

package/dist/src/services/setup.js

@@ -15,6 +15,7 @@ import { readState, writeState } from "../store/skill_state.js";
 import { cacheSet } from "../store/cache_db.js";
 import { log, timed, withLogSink } from "../util/log.js";
 import { enqueue, registerHandler } from "./jobs.js";
+import { closeAppDb, isAppDbOpen, openAppDb } from "../store/app_db.js";
 import { getRepo, markKnowledgeBuilt } from "../store/repos.js";
 import { nowIso } from "../util/time.js";
 import { mkdir, readTextFile, remove } from "../util/runtime.js";
@@ -116,6 +117,18 @@ function addReviewLink(markdown, documents) {
     return `${markdown.trimEnd()}\n\n## Pull request review guides\n\n- Read [PR_REVIEW_GUIDE.md](PR_REVIEW_GUIDE.md).${detailedLink}\n`;
 }
 export async function runInitOrRemake(options) {
+    const openedAppDb = !isAppDbOpen();
+    if (openedAppDb)
+        await openAppDb();
+    try {
+        await initOrRemake(options);
+    }
+    finally {
+        if (openedAppDb)
+            await closeAppDb();
+    }
+}
+async function initOrRemake(options) {
     const operationStarted = performance.now();
     const aiMetrics = emptyAiMetrics();
     const previous = await readState(options.repo);
@@ -136,8 +149,9 @@ export async function runInitOrRemake(options) {
         log("ai", `${options.ai} providers configured · low=${options.lowModel} · high=${options.highModel}`);
     }
     log("fetch", `${options.repo} via ${options.auth}`);
-    const source = await timed("fetch repository data", options.logTime, () => collectSource(client, options, previous, async (pullRequests) => {
+    const source = await timed("fetch repository data", options.logTime, () => collectSource(client, options, previous, async (pullRequests, cache) => {
         checkpoint.source.pullRequests = pullRequests;
+        checkpoint.source.pullRequestCache = cache;
         checkpoint.scanDone = {
             ...checkpoint.scanDone,
             pullRequests: pullRequests.length,

package/dist/src/store/app_db.d.ts

@@ -3,8 +3,10 @@ export declare function appDbDir(): string;
 /** `CM_APP_DB` overrides the path — tests point it at a temp file. */
 export declare function appDbPath(): string;
 export declare function releaseLock(): Promise<void>;
+export declare function isAppDbOpen(): boolean;
 /** Opens (creating and migrating if needed) the single shared `app.db`
- * connection every store module reuses. Call once at `serve` startup. */
+ * connection every store module reuses. `serve` opens it for the process.
+ * `init` and `remake` open it for the run when it is not already open. */
 export declare function openAppDb(): Promise<Database>;
 export declare function getAppDb(): Database;
 export declare function closeAppDb(): Promise<void>;

package/dist/src/store/app_db.js

@@ -59,8 +59,12 @@ function migrate(database) {
     }
 }
 let db;
+export function isAppDbOpen() {
+    return db !== undefined;
+}
 /** Opens (creating and migrating if needed) the single shared `app.db`
- * connection every store module reuses. Call once at `serve` startup. */
+ * connection every store module reuses. `serve` opens it for the process.
+ * `init` and `remake` open it for the run when it is not already open. */
 export async function openAppDb() {
     if (db)
         return db;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "co-maintainer",
-  "version": "0.4.11",
+  "version": "0.4.13",
   "description": "Analyzes a GitHub repository and writes repository-specific contribution guidance.",
   "license": "MIT",
   "repository": {