fix: cap contributor graph file path loading

src/db/repositories.ts

@@ -1,4 +1,4 @@
 import { and, desc, eq, gte, inArray, not, or, sql, type SQL } from "drizzle-orm";
 import { getDb } from "./client";
 import {
   advisories,
@@ -129,6 +129,7 @@
   ProductUsageSurface,
   ProductUsageSurfaceActivationFunnel,
   ProductUsageSurfaceRetention,
+  PullRequestFilePathRecord,
   PullRequestFileRecord,
   PullRequestDetailSyncStateRecord,
   PullRequestRecord,
@@ -2657,6 +2658,28 @@
   return rows.map(toPullRequestFileRecord);
 }
 
+export async function listRepoPullRequestFilePaths(
+  env: Env,
+  fullName: string,
+  options: { pullNumbers?: number[] | undefined; limit?: number | undefined } = {},
+): Promise<PullRequestFilePathRecord[]> {
+  const db = getDb(env.DB);
+  const pullNumbers = [...new Set(options.pullNumbers ?? [])].filter((number) => Number.isInteger(number) && number > 0);
+  if (options.pullNumbers && pullNumbers.length === 0) return [];
+  const where = pullNumbers.length > 0
+    ? and(eq(pullRequestFiles.repoFullName, fullName), inArray(pullRequestFiles.pullNumber, pullNumbers))
+    : eq(pullRequestFiles.repoFullName, fullName);
+  return db
+    .select({
+      repoFullName: pullRequestFiles.repoFullName,
+      pullNumber: pullRequestFiles.pullNumber,
+      path: pullRequestFiles.path,
+    })
+    .from(pullRequestFiles)
+    .where(where)
+    .limit(Math.max(0, Math.min(options.limit ?? 500, 500)));
+}
+
 export async function listRepoPullRequestFiles(env: Env, fullName: string): Promise<PullRequestFileRecord[]> {
   const db = getDb(env.DB);
   const rows = await db.select().from(pullRequestFiles).where(eq(pullRequestFiles.repoFullName, fullName)).limit(2000);

src/queue/processors.ts

@@ -1,4 +1,4 @@
 import {
   countOpenIssues,
   countOpenPullRequests,
   getAgentCommandAnswer,
@@ -30,7 +30,7 @@
   listRecentMergedPullRequests,
   updatePullRequestSlopAssessment,
   listRepoLabels,
-  listRepoPullRequestFiles,
+  listRepoPullRequestFilePaths,
   listRepoSyncStates,
   listRepoSyncSegments,
   listRepositories,
@@ -140,6 +140,7 @@
   detectGittensorContributor,
   PR_PANEL_RETRIGGER_MARKER,
   unionScopedOverlapClusters,
+  type ContributorProfile,
 } from "../signals/engine";
 import { buildIssueSlopAssessment, buildSlopAssessment, type SlopBand } from "../signals/slop";
 import { runGittensoryAiSlopAdvisory } from "../services/ai-slop";
@@ -148,7 +149,7 @@
 import { buildFocusManifestGuidance, resolveEffectiveSettings } from "../signals/focus-manifest";
 import type { LocalBranchAnalysisInput } from "../signals/local-branch";
 import { runGittensoryAiReview } from "../services/ai-review";
-import type { AdvisoryFinding, ContributorEvidenceRecord, DetectedNotificationEvent, GitHubWebhookPayload, JobMessage, JsonValue, PullRequestRecord, RepositorySettings } from "../types";
+import type { AdvisoryFinding, ContributorEvidenceRecord, ContributorRepoStatRecord, DetectedNotificationEvent, GitHubWebhookPayload, IssueRecord, JobMessage, JsonValue, PullRequestFilePathRecord, PullRequestRecord, RepositoryRecord, RepositorySettings } from "../types";
 import { sha256Hex } from "../utils/crypto";
 import { errorMessage, nowIso } from "../utils/json";
 
@@ -545,6 +546,41 @@
   return [...new Set([...pullRequests, ...issues].flatMap((record) => (record.authorLogin ? [record.authorLogin] : [])))].slice(0, 200);
 }
 
+const CONTRIBUTOR_EVIDENCE_MAX_PR_FILE_PATHS = 2000;
+const CONTRIBUTOR_EVIDENCE_PR_FILE_PATHS_PER_REPO = 200;
+
+async function loadContributorPullRequestFilePaths(
+  env: Env,
+  args: {
+    login: string;
+    profile: ContributorProfile;
+    pullRequests: PullRequestRecord[];
+    issues: IssueRecord[];
+    repoStats: ContributorRepoStatRecord[];
+    repositories: RepositoryRecord[];
+  },
+): Promise<PullRequestFilePathRecord[]> {
+  const pullNumbersByRepo = new Map<string, Set<number>>();
+  for (const pr of args.pullRequests) {
+    if (pr.authorLogin?.toLowerCase() !== args.login.toLowerCase()) continue;
+    const key = pr.repoFullName.toLowerCase();
+    const current = pullNumbersByRepo.get(key) ?? new Set<number>();
+    current.add(pr.number);
+    pullNumbersByRepo.set(key, current);
+  }
+  const files: PullRequestFilePathRecord[] = [];
+  for (const repoFullName of evidenceGraphTouchedRepoFullNames(args)) {
+    if (files.length >= CONTRIBUTOR_EVIDENCE_MAX_PR_FILE_PATHS) break;
+    const remaining = CONTRIBUTOR_EVIDENCE_MAX_PR_FILE_PATHS - files.length;
+    const repoFiles = await listRepoPullRequestFilePaths(env, repoFullName, {
+      pullNumbers: [...(pullNumbersByRepo.get(repoFullName.toLowerCase()) ?? [])],
+      limit: Math.min(CONTRIBUTOR_EVIDENCE_PR_FILE_PATHS_PER_REPO, remaining),
+    });
+    files.push(...repoFiles);
+  }
+  return files;
+}
+
 async function buildContributorEvidence(env: Env, login?: string): Promise<void> {
   const [allPullRequests, allIssues, repositories, syncStates, allBounties, snapshot] = await Promise.all([
     listAllPullRequests(env),
@@ -569,18 +605,14 @@
     ]);
     const repoStats = authoritativeContributorRepoStats(gittensorSnapshot, cachedRepoStats);
     const profile = buildContributorProfile(contributorLogin, github, contributorPullRequests, contributorIssues, repoStats, gittensorSnapshot);
-    const pullRequestFiles = (
-      await Promise.all(
-        evidenceGraphTouchedRepoFullNames({
-          login: contributorLogin,
-          profile,
-          pullRequests: contributorPullRequests,
-          issues: contributorIssues,
-          repoStats,
-          repositories,
-        }).map((repoFullName) => listRepoPullRequestFiles(env, repoFullName)),
-      )
-    ).flat();
+    const pullRequestFiles = await loadContributorPullRequestFilePaths(env, {
+      login: contributorLogin,
+      profile,
+      pullRequests: contributorPullRequests,
+      issues: contributorIssues,
+      repoStats,
+      repositories,
+    });
     const fit = buildContributorFit(profile, repositories, allIssues, allPullRequests, syncStates, repoStats, allBounties, issueQualityByRepo);
     const scoringProfile = buildContributorScoringProfile({ login: contributorLogin, fit, scoringSnapshot: snapshot });
     const outcomeHistory = buildContributorOutcomeHistory({ login: contributorLogin, profile, repositories, pullRequests: allPullRequests, issues: allIssues, repoStats, cachedRepoStats });

src/services/contributor-evidence-graph.ts

@@ -1,9 +1,9 @@
 import type { GittensorContributorSnapshot } from "../gittensor/api";
 import type { ContributorOutcomeHistory, ContributorProfile, RoleContext } from "../signals/engine";
 import type {
   ContributorRepoStatRecord,
   IssueRecord,
-  PullRequestFileRecord,
+  PullRequestFilePathRecord,
   PullRequestRecord,
   RepositoryRecord,
   RepoSyncStateRecord,
@@ -153,7 +153,7 @@
   issues?: IssueRecord[] | undefined;
   repoStats?: ContributorRepoStatRecord[] | undefined;
   syncStates?: RepoSyncStateRecord[] | undefined;
-  pullRequestFiles?: PullRequestFileRecord[] | undefined;
+  pullRequestFiles?: PullRequestFilePathRecord[] | undefined;
   gittensorSnapshot?: GittensorContributorSnapshot | null | undefined;
 };
 
@@ -374,7 +374,7 @@
     );
 }
 
-function buildPathEdges(login: string, contributorPullRequests: PullRequestRecord[], files: PullRequestFileRecord[], generatedAt: string): ContributorEvidenceGraphPath[] {
+function buildPathEdges(login: string, contributorPullRequests: PullRequestRecord[], files: PullRequestFilePathRecord[], generatedAt: string): ContributorEvidenceGraphPath[] {
   const prByKey = new Map(contributorPullRequests.filter((pr) => sameLogin(pr.authorLogin, login)).map((pr) => [`${pr.repoFullName.toLowerCase()}#${pr.number}`, pr]));
   const buckets = new Map<string, PathBucket>();
   for (const file of files) {

src/services/decision-pack.ts

@@ -1,4 +1,4 @@
 import { loadRepoFocusManifests } from "../signals/focus-manifest-loader";
 import type { FocusManifest, FocusManifestIssueDiscoveryPolicy, FocusManifestLinkedIssuePolicy, FocusManifestSource } from "../signals/focus-manifest";
 import { isFocusManifestPublicSafe } from "../signals/focus-manifest";
 import {
@@ -10,7 +10,7 @@
   listContributorPullRequests,
   listContributorRepoStats,
   listLatestRepoGithubTotalsSnapshots,
-  listRepoPullRequestFiles,
+  listRepoPullRequestFilePaths,
   listRepositories,
   listRepoSyncSegments,
   listRepoSyncStates,
@@ -57,7 +57,7 @@
   ContributorRepoStatRecord,
   IssueRecord,
   JsonValue,
-  PullRequestFileRecord,
+  PullRequestFilePathRecord,
   PullRequestRecord,
   RepositoryRecord,
   RepoGithubTotalsSnapshotRecord,
@@ -71,6 +71,8 @@
 export const CONTRIBUTOR_DECISION_PACK_SIGNAL = "contributor-decision-pack";
 export const DECISION_PACK_MAX_AGE_MS = 6 * 60 * 60 * 1000;
 const DEFAULT_OSS_EMISSION_SHARE = DEFAULT_SCORING_CONSTANTS.OSS_EMISSION_SHARE ?? 0.9;
+const DECISION_PACK_MAX_PR_FILE_PATHS = 2000;
+const DECISION_PACK_PR_FILE_PATHS_PER_REPO = 200;
 
 function resolveOssEmissionShare(constants: Record<string, number> | undefined): number {
   const value = constants?.OSS_EMISSION_SHARE;
@@ -79,6 +81,38 @@
 export const DECISION_PACK_REBUILD_DEBOUNCE_MS = 15 * 1000;
 const pendingDecisionPackRebuilds = new Map<string, Promise<boolean>>();
 
+async function loadContributorPullRequestFilePaths(
+  env: Env,
+  args: {
+    login: string;
+    profile: ContributorProfile;
+    pullRequests: PullRequestRecord[];
+    issues: IssueRecord[];
+    repoStats: ContributorRepoStatRecord[];
+    repositories: RepositoryRecord[];
+  },
+): Promise<PullRequestFilePathRecord[]> {
+  const pullNumbersByRepo = new Map<string, Set<number>>();
+  for (const pr of args.pullRequests) {
+    if (pr.authorLogin?.toLowerCase() !== args.login.toLowerCase()) continue;
+    const key = pr.repoFullName.toLowerCase();
+    const current = pullNumbersByRepo.get(key) ?? new Set<number>();
+    current.add(pr.number);
+    pullNumbersByRepo.set(key, current);
+  }
+  const files: PullRequestFilePathRecord[] = [];
+  for (const repoFullName of evidenceGraphTouchedRepoFullNames(args)) {
+    if (files.length >= DECISION_PACK_MAX_PR_FILE_PATHS) break;
+    const remaining = DECISION_PACK_MAX_PR_FILE_PATHS - files.length;
+    const repoFiles = await listRepoPullRequestFilePaths(env, repoFullName, {
+      pullNumbers: [...(pullNumbersByRepo.get(repoFullName.toLowerCase()) ?? [])],
+      limit: Math.min(DECISION_PACK_PR_FILE_PATHS_PER_REPO, remaining),
+    });
+    files.push(...repoFiles);
+  }
+  return files;
+}
+
 export type DecisionRecommendation = "pursue" | "cleanup_first" | "maintainer_lane" | "avoid_for_now" | "watch";
 export type DecisionActionKind = "cleanup_existing_prs" | "land_existing_prs" | "open_new_direct_pr" | "file_issue_discovery" | "maintainer_lane_improve_repo" | "maintainer_cut_readiness";
 export type DecisionPackFreshness = "fresh" | "stale" | "rebuilding" | "missing";
@@ -442,18 +476,14 @@
     repositories.filter((repo) => repo.isRegistered).map((repo) => repo.fullName),
   );
   const profile = buildContributorProfile(login, github, contributorPullRequests, contributorIssues, repoStats, gittensorSnapshot);
-  const pullRequestFiles = (
-    await Promise.all(
-      evidenceGraphTouchedRepoFullNames({
-        login,
-        profile,
-        pullRequests: contributorPullRequests,
-        issues: contributorIssues,
-        repoStats,
-        repositories,
-      }).map((repoFullName) => listRepoPullRequestFiles(env, repoFullName)),
-    )
-  ).flat();
+  const pullRequestFiles = await loadContributorPullRequestFilePaths(env, {
+    login,
+    profile,
+    pullRequests: contributorPullRequests,
+    issues: contributorIssues,
+    repoStats,
+    repositories,
+  });
   const outcomeHistory = buildContributorOutcomeHistory({
     login,
     profile,
@@ -547,7 +577,7 @@
   contributorPullRequests: Parameters<typeof buildRoleContext>[0]["pullRequests"];
   contributorIssues: Parameters<typeof buildRoleContext>[0]["issues"];
   repoStats?: ContributorRepoStatRecord[] | undefined;
-  pullRequestFiles?: PullRequestFileRecord[] | undefined;
+  pullRequestFiles?: PullRequestFilePathRecord[] | undefined;
   gittensorSnapshot?: Awaited<ReturnType<typeof fetchGittensorContributorSnapshot>> | undefined;
   issueQualityByRepo?: Map<string, IssueQualityReport> | undefined;
   openPrMonitor: ContributorOpenPrMonitor;

src/types.ts

@@ -1,4 +1,4 @@
 export type JsonPrimitive = string | number | boolean | null;
 export type JsonValue = JsonPrimitive | JsonValue[] | { [key: string]: JsonValue };
 
 export type JobMessage =
@@ -703,6 +703,8 @@
   payload: Record<string, JsonValue>;
 };
 
+export type PullRequestFilePathRecord = Pick<PullRequestFileRecord, "repoFullName" | "pullNumber" | "path">;
+
 export type PullRequestReviewRecord = {
   id: string;
   repoFullName: string;