Initial WhaleTown V2 backend

2026-07-20 02:00:52 +08:00
commit c995891c1f
265 changed files with 75689 additions and 0 deletions
--- a/src/business/rankings/rankings.service.ts
+++ b/src/business/rankings/rankings.service.ts
@@ -0,0 +1,436 @@
+import { BadGatewayException, Injectable, Logger, OnModuleInit } from '@nestjs/common';
+import { Cron } from '@nestjs/schedule';
+import axios from 'axios';
+import {
+  DatawhaleCommitStats,
+  DatawhaleHonorRankingPayload,
+  DatawhaleMemberRow,
+  DatawhaleWeeklyCommitsPayload,
+  RankingCategory,
+  RankingCategoryId,
+  RankingEntry,
+  RankingUser,
+} from './rankings.types';
+
+const DATAWHALE_MEMBERS_URL = 'https://mv.datawhale.cc/data/members.json';
+const DATAWHALE_WEEKLY_COMMITS_URL = 'https://mv.datawhale.cc/data/commits_weekly.json';
+const DATAWHALE_ASSET_BASE_URL = 'https://mv.datawhale.cc/';
+const DEFAULT_CATEGORY: RankingCategoryId = 'weekly_commits';
+const DEFAULT_LIMIT = 10;
+
+const CATEGORIES: RankingCategory[] = [
+  {
+    id: 'weekly_commits',
+    label: '一周卷王',
+    title: '一周卷王',
+    description: '近 7 天 commit 数 + 连续性、多仓库、质量奖励',
+    icon: '🔥',
+  },
+  {
+    id: 'night_owl',
+    label: '夜猫榜',
+    title: '夜猫榜',
+    description: '深夜提交数量与深夜活跃比例排行',
+    icon: '🌙',
+  },
+  {
+    id: 'popularity',
+    label: '人气王',
+    title: '人气王',
+    description: 'Followers 与组织仓库 Stars 的综合影响力',
+    icon: '👑',
+  },
+  {
+    id: 'productive',
+    label: '多产榜',
+    title: '多产榜',
+    description: '参与 Datawhale 组织仓库数量排行',
+    icon: '🏆',
+  },
+  {
+    id: 'social',
+    label: '社交达人',
+    title: '社交达人',
+    description: 'GitHub Following 数量排行',
+    icon: '💬',
+  },
+  {
+    id: 'rising',
+    label: '新星榜',
+    title: '新星榜',
+    description: '按仓库数量归一后的潜力新星排行',
+    icon: '🌠',
+  },
+  {
+    id: 'comprehensive',
+    label: '综合实力',
+    title: '综合实力',
+    description: 'Stars、Followers、仓库数、社交和贡献数综合评分',
+    icon: '🌟',
+  },
+];
+
+@Injectable()
+export class RankingsService implements OnModuleInit {
+  private readonly logger = new Logger(RankingsService.name);
+  private members: DatawhaleMemberRow[] = [];
+  private weeklyCommits: DatawhaleWeeklyCommitsPayload | null = null;
+  private syncedAt: Date | null = null;
+  private syncing: Promise<void> | null = null;
+
+  async onModuleInit(): Promise<void> {
+    this.syncNow().catch(error => {
+      this.logger.warn(`Datawhale荣誉榜启动同步失败：${this.errorMessage(error)}`);
+    });
+  }
+
+  @Cron('15 3 * * *')
+  async syncDaily(): Promise<void> {
+    await this.syncNow();
+  }
+
+  async getDatawhaleHonorRanking(
+    category: RankingCategoryId = DEFAULT_CATEGORY,
+    limit: number = DEFAULT_LIMIT,
+    refresh = false,
+  ): Promise<DatawhaleHonorRankingPayload> {
+    if (refresh || this.members.length === 0) {
+      await this.syncNow();
+    }
+    return this.getCachedPayload(category, limit);
+  }
+
+  async syncNow(): Promise<DatawhaleHonorRankingPayload> {
+    if (this.syncing) {
+      await this.syncing;
+      return this.getCachedPayload(DEFAULT_CATEGORY, DEFAULT_LIMIT);
+    }
+
+    this.syncing = this.fetchAndReplace();
+    try {
+      await this.syncing;
+    } finally {
+      this.syncing = null;
+    }
+    return this.getCachedPayload(DEFAULT_CATEGORY, DEFAULT_LIMIT);
+  }
+
+  private getCachedPayload(
+    category: RankingCategoryId,
+    limit: number,
+  ): DatawhaleHonorRankingPayload {
+    const normalizedCategory = this.normalizeCategory(category);
+    const normalizedLimit = this.normalizeLimit(limit);
+    const entries = this.buildEntries(normalizedCategory, normalizedLimit);
+    return {
+      source: 'datawhale-members-visualization',
+      activeCategory: normalizedCategory,
+      categories: CATEGORIES,
+      topRankers: entries.slice(0, 3),
+      rankers: entries.slice(3, normalizedLimit),
+      myRank: {
+        rank: null,
+        score: 0,
+        reward: 0,
+      },
+      total: entries.length,
+      syncedAt: this.syncedAt ? this.syncedAt.toISOString() : null,
+      sourceUpdatedAt: this.weeklyCommits?.update_time ?? null,
+    };
+  }
+
+  private async fetchAndReplace(): Promise<void> {
+    try {
+      const [membersResponse, commitsResponse] = await Promise.all([
+        axios.get(DATAWHALE_MEMBERS_URL, { timeout: 15000 }),
+        axios.get(DATAWHALE_WEEKLY_COMMITS_URL, { timeout: 15000 }),
+      ]);
+
+      if (!Array.isArray(membersResponse.data)) {
+        throw new BadGatewayException('Datawhale成员接口返回格式异常');
+      }
+      const commitsPayload = commitsResponse.data as DatawhaleWeeklyCommitsPayload;
+      if (!commitsPayload || typeof commitsPayload !== 'object' || !commitsPayload.user_commits) {
+        throw new BadGatewayException('Datawhale周贡献接口返回格式异常');
+      }
+
+      this.members = membersResponse.data as DatawhaleMemberRow[];
+      this.weeklyCommits = commitsPayload;
+      this.syncedAt = new Date();
+      this.logger.log(`Datawhale荣誉榜同步完成，共 ${this.members.length} 位成员`);
+    } catch (error) {
+      if (this.members.length > 0) {
+        this.logger.warn(`Datawhale荣誉榜同步失败，继续使用缓存：${this.errorMessage(error)}`);
+        return;
+      }
+      throw error;
+    }
+  }
+
+  private buildEntries(category: RankingCategoryId, limit: number): RankingEntry[] {
+    const users = category === 'weekly_commits' || category === 'night_owl'
+      ? this.buildCommitDrivenUsers(category)
+      : this.members.map(member => this.toRankingUser(member, category));
+
+    return users
+      .filter(user => user.score > 0)
+      .sort((a, b) => b.score - a.score)
+      .slice(0, limit)
+      .map((user, index) => ({
+        ...user,
+        rank: index + 1,
+      }));
+  }
+
+  private buildCommitDrivenUsers(category: RankingCategoryId): RankingUser[] {
+    const memberById = new Map(
+      this.members
+        .map(member => [this.cleanString(member.id), member] as const)
+        .filter(([id]) => id.length > 0),
+    );
+    const userCommits = this.weeklyCommits?.user_commits ?? {};
+
+    return Object.entries(userCommits).map(([rawId, commits]) => {
+      const id = this.cleanString(rawId);
+      const member = memberById.get(id) ?? { id };
+      return this.toRankingUser(member, category, commits);
+    });
+  }
+
+  private toRankingUser(
+    member: DatawhaleMemberRow,
+    category: RankingCategoryId,
+    commitsOverride?: DatawhaleCommitStats,
+  ): RankingUser {
+    const id = this.cleanString(member.id);
+    const commits = commitsOverride ?? this.weeklyCommits?.user_commits?.[id] ?? {};
+    const score = this.scoreMember(member, commits, category);
+    const domains = this.domainList(member);
+    return {
+      id,
+      name: this.displayName(member),
+      avatarText: this.avatarText(member),
+      avatarUrl: this.avatarUrl(member),
+      githubUrl: this.githubUrl(member),
+      domain: domains[0] ?? '',
+      domains,
+      location: this.cleanString(member.location),
+      score,
+      scoreLabel: this.scoreLabel(member, commits, category),
+      contrib: this.numberValue(member.org_total_contributions) || this.numberValue(commits.total_commits),
+      answers: this.numberValue(commits.repo_count) || this.numberValue(member.org_repos_count),
+      likes: this.numberValue(member.org_total_stars),
+      reward: this.rewardForScore(score),
+    };
+  }
+
+  private scoreMember(
+    member: DatawhaleMemberRow,
+    commits: DatawhaleCommitStats,
+    category: RankingCategoryId,
+  ): number {
+    switch (category) {
+      case 'weekly_commits':
+        return this.weeklyCommitScore(commits);
+      case 'night_owl':
+        return this.nightOwlScore(commits);
+      case 'popularity':
+        return Math.round(
+          this.numberValue(member.followers ?? member.followers_count) * 0.6 +
+          this.numberValue(member.org_total_stars) * 0.4,
+        );
+      case 'productive':
+        return this.numberValue(member.org_repos_count);
+      case 'social':
+        return this.numberValue(member.following);
+      case 'rising': {
+        const repoCount = Math.max(this.numberValue(member.org_repos_count), 1);
+        const bonus = repoCount < 5 ? 1.5 : 1.0;
+        return Math.round((this.numberValue(member.followers ?? member.followers_count) + this.numberValue(member.org_total_stars)) / repoCount * bonus);
+      }
+      case 'comprehensive':
+        return Math.round(
+          this.numberValue(member.org_total_stars) * 0.3 +
+          this.numberValue(member.followers ?? member.followers_count) * 0.25 +
+          this.numberValue(member.org_repos_count) * 0.2 +
+          this.numberValue(member.following) * 0.15 +
+          this.numberValue(member.org_total_contributions) * 0.1,
+        );
+      default:
+        return 0;
+    }
+  }
+
+  private weeklyCommitScore(commits: DatawhaleCommitStats): number {
+    let score = this.numberValue(commits.total_commits);
+    const activeDays = this.numberValue(commits.active_days);
+    if (activeDays >= 7) {
+      score += 10;
+    } else if (activeDays >= 5) {
+      score += 5;
+    } else if (activeDays >= 3) {
+      score += 2;
+    }
+
+    const repoCount = this.numberValue(commits.repo_count);
+    if (repoCount >= 5) {
+      score += 5;
+    } else if (repoCount >= 3) {
+      score += 3;
+    } else if (repoCount >= 2) {
+      score += 1;
+    }
+
+    const avgCommitsPerDay = this.numberValue(commits.avg_commits_per_day);
+    if (avgCommitsPerDay >= 5) {
+      score += 8;
+    } else if (avgCommitsPerDay >= 3) {
+      score += 5;
+    } else if (avgCommitsPerDay >= 2) {
+      score += 2;
+    }
+    return Math.round(score);
+  }
+
+  private nightOwlScore(commits: DatawhaleCommitStats): number {
+    let score = this.numberValue(commits.night_owl_commits) * 2;
+    const percentage = this.numberValue(commits.night_owl_percentage);
+    if (percentage >= 50) {
+      score += 10;
+    } else if (percentage >= 30) {
+      score += 5;
+    } else if (percentage >= 20) {
+      score += 2;
+    }
+    const activeDays = this.numberValue(commits.active_days);
+    if (activeDays >= 5) {
+      score += 8;
+    } else if (activeDays >= 3) {
+      score += 4;
+    }
+    const repoCount = this.numberValue(commits.repo_count);
+    if (repoCount >= 3) {
+      score += 3;
+    } else if (repoCount >= 2) {
+      score += 1;
+    }
+    return Math.round(score);
+  }
+
+  private scoreLabel(
+    member: DatawhaleMemberRow,
+    commits: DatawhaleCommitStats,
+    category: RankingCategoryId,
+  ): string {
+    switch (category) {
+      case 'weekly_commits':
+        return `${this.numberValue(commits.total_commits)} commits`;
+      case 'night_owl':
+        return `${this.numberValue(commits.night_owl_commits)} 深夜`;
+      case 'popularity':
+        return `${this.numberValue(member.followers ?? member.followers_count)} followers`;
+      case 'productive':
+        return `${this.numberValue(member.org_repos_count)} 仓库`;
+      case 'social':
+        return `${this.numberValue(member.following)} following`;
+      case 'rising':
+        return '活跃度';
+      case 'comprehensive':
+        return '综合分';
+      default:
+        return '分数';
+    }
+  }
+
+  private domainList(member: DatawhaleMemberRow): string[] {
+    const primaryDomain = this.cleanString(member.primary_domain);
+    const domains = this.cleanString(member.domain)
+      .split(';')
+      .map(domain => this.cleanString(domain))
+      .filter(Boolean);
+    const result = primaryDomain ? [primaryDomain, ...domains] : domains;
+    return [...new Set(result)].slice(0, 3);
+  }
+
+  private displayName(member: DatawhaleMemberRow): string {
+    const name = this.cleanString(member.name);
+    if (name && !['null', 'undefined', 'none'].includes(name.toLowerCase())) {
+      return name;
+    }
+    return this.cleanString(member.id) || '未知用户';
+  }
+
+  private avatarText(member: DatawhaleMemberRow): string {
+    const name = this.displayName(member);
+    return name.length > 0 ? name.slice(0, 1).toUpperCase() : '鲸';
+  }
+
+  private avatarUrl(member: DatawhaleMemberRow): string {
+    const avatar = this.cleanString(member.avatar);
+    if (!avatar) {
+      return '';
+    }
+    if (avatar.startsWith('http://') || avatar.startsWith('https://')) {
+      return avatar;
+    }
+    if (avatar.startsWith('/')) {
+      return `${DATAWHALE_ASSET_BASE_URL.replace(/\/$/, '')}${avatar}`;
+    }
+    return `${DATAWHALE_ASSET_BASE_URL}${avatar}`;
+  }
+
+  private githubUrl(member: DatawhaleMemberRow): string {
+    const github = this.cleanString(member.github);
+    if (github.startsWith('http')) {
+      return github;
+    }
+    const id = this.cleanString(member.id);
+    return id ? `https://github.com/${id}` : '';
+  }
+
+  private normalizeCategory(category: RankingCategoryId): RankingCategoryId {
+    return CATEGORIES.some(item => item.id === category) ? category : DEFAULT_CATEGORY;
+  }
+
+  private normalizeLimit(limit: number): number {
+    if (!Number.isFinite(limit)) {
+      return DEFAULT_LIMIT;
+    }
+    return Math.min(Math.max(Math.floor(limit), 3), DEFAULT_LIMIT);
+  }
+
+  private rewardForScore(score: number): number {
+    if (score >= 1000) {
+      return 80;
+    }
+    if (score >= 500) {
+      return 60;
+    }
+    if (score >= 120) {
+      return 50;
+    }
+    if (score >= 50) {
+      return 40;
+    }
+    if (score >= 20) {
+      return 30;
+    }
+    return 20;
+  }
+
+  private numberValue(value: unknown): number {
+    const parsed = Number(value || 0);
+    return Number.isFinite(parsed) ? parsed : 0;
+  }
+
+  private cleanString(value: unknown): string {
+    return String(value ?? '').trim();
+  }
+
+  private errorMessage(error: unknown): string {
+    if (error instanceof Error) {
+      return error.message;
+    }
+    return String(error);
+  }
+}