anonymous_github/src/core/source/GitHubRepository.ts

import { Branch } from "../types";
import * as gh from "parse-github-url";
import { RestEndpointMethodTypes } from "@octokit/rest";

import AnonymousError from "../AnonymousError";
import { isConnected } from "../../server/database";
import { octokit } from "../GitHubUtils";
import { IRepositoryDocument } from "../model/repositories/repositories.types";
import RepositoryModel from "../model/repositories/repositories.model";

export class GitHubRepository {
  private _data: Partial<{
    [P in keyof IRepositoryDocument]: IRepositoryDocument[P];
  }>;
  constructor(
    data: Partial<{ [P in keyof IRepositoryDocument]: IRepositoryDocument[P] }>
  ) {
    this._data = data;
  }

  toJSON() {
    return {
      id: this.model._id,
      externalId: this._data.externalId,
      repo: this.repo,
      owner: this.owner,
      hasPage: this._data.hasPage,
      pageSource: this._data.pageSource,
      fullName: this.fullName,
      defaultBranch: this._data.defaultBranch,
      size: this.size,
    };
  }

  get model() {
    return this._data;
  }

  public get fullName(): string | undefined {
    return this._data.name;
  }

  public get id(): string | undefined {
    return this._data.externalId;
  }

  public get size(): number | undefined {
    return this._data.size;
  }

  async getCommitInfo(
    sha: string,
    opt: {
      accessToken: string;
    }
  ) {
    const oct = octokit(opt.accessToken);
    try {
      const commit = await oct.repos.getCommit({
        owner: this.owner,
        repo: this.repo,
        ref: sha,
      });
      return commit.data;
    } catch (error) {
      const status = (error as { status?: number }).status;
      if (status === 404) {
        // Distinguish: does the repo itself still exist?
        let repoExists: boolean;
        try {
          await oct.repos.get({ owner: this.owner, repo: this.repo });
          repoExists = true;
        } catch {
          repoExists = false;
        }
        throw new AnonymousError(
          repoExists ? "commit_not_found" : "repo_not_found",
          {
            httpStatus: 404,
            cause: error as Error,
            object: this,
          }
        );
      }
      throw error;
    }
  }

  async branches(opt: {
    accessToken: string;
    force?: boolean;
  }): Promise<Branch[]> {
    if (
      !this._data.branches ||
      this._data.branches.length == 0 ||
      opt?.force === true
    ) {
      // get the list of repo from github
      const oct = octokit(opt.accessToken);
      try {
        const branches = (
          await oct.paginate("GET /repos/{owner}/{repo}/branches", {
            owner: this.owner,
            repo: this.repo,
            per_page: 100,
          })
        ).map((b) => {
          return {
            name: b.name,
            commit: b.commit.sha,
            readme: this._data.branches?.filter(
              (f: Branch) => f.name == b.name
            )[0]?.readme,
          } as Branch;
        });
        this._data.branches = branches;
        if (isConnected) {
          await RepositoryModel.updateOne(
            { externalId: this.id },
            { $set: { branches } }
          );
        }
      } catch (error) {
        throw new AnonymousError("repo_not_found", {
          httpStatus: (error as { status?: number }).status,
          cause: error as Error,
          object: this,
        });
      }
    } else if (isConnected) {
      const q = await RepositoryModel.findOne({ externalId: this.id }).select(
        "branches"
      );
      this._data.branches = q?.branches;
    }

    return this._data.branches || [];
  }

  async readme(opt: {
    branch?: string;
    force?: boolean;
    accessToken: string;
  }): Promise<string | undefined> {
    if (!opt.branch) opt.branch = this._data.defaultBranch || "master";

    const model = await RepositoryModel.findOne({
      externalId: this.id,
    }).select("branches");

    if (!model) {
      throw new AnonymousError("repo_not_found", { httpStatus: 404 });
    }

    this._data.branches = await this.branches(opt);
    model.branches = this._data.branches;

    const selected = model.branches.filter((f) => f.name == opt.branch)[0];
    if (selected && (!selected.readme || opt?.force === true)) {
      // get the list of repo from github
      const oct = octokit(opt.accessToken);
      try {
        const ghRes = await oct.repos.getReadme({
          owner: this.owner,
          repo: this.repo,
          ref: selected?.commit,
        });
        const readme = Buffer.from(
          ghRes.data.content,
          ghRes.data.encoding as BufferEncoding
        ).toString("utf-8");
        selected.readme = readme;
        await model.save();
      } catch (error) {
        throw new AnonymousError("readme_not_available", {
          httpStatus: 404,
          cause: error as Error,
          object: this,
        });
      }
    }

    if (!selected) {
      throw new AnonymousError("readme_not_available", {
        httpStatus: 404,
        object: this,
      });
    }

    return selected.readme;
  }

  public get owner(): string {
    if (!this.fullName) {
      throw new AnonymousError("invalid_repo", {
        httpStatus: 400,
        object: this,
      });
    }
    const repo = gh(this.fullName);
    if (!repo) {
      throw new AnonymousError("invalid_repo", {
        httpStatus: 400,
        object: this,
      });
    }
    return repo.owner || this.fullName;
  }

  public get repo(): string {
    if (!this.fullName) {
      throw new AnonymousError("invalid_repo", {
        httpStatus: 400,
        object: this,
      });
    }
    const repo = gh(this.fullName);
    if (!repo) {
      throw new AnonymousError("invalid_repo", {
        httpStatus: 400,
        object: this,
      });
    }
    return repo.name || this.fullName;
  }
}

export async function getRepositoryFromGitHub(opt: {
  owner: string;
  repo: string;
  repositoryID?: string;
  accessToken: string;
  force?: boolean;
}) {
  if (opt.repo.endsWith(".git")) {
    opt.repo = opt.repo.slice(0, -4);
  }
  let dbModel;
  if (opt.repositoryID) {
    dbModel = isConnected
      ? await RepositoryModel.findById(opt.repositoryID)
      : null;
    opt.owner = dbModel?.name?.split("/")[0] || opt.owner;
    opt.repo = dbModel?.name?.split("/")[1] || opt.repo;
  } else {
    dbModel = isConnected
      ? await RepositoryModel.findOne({
          name: opt.owner + "/" + opt.repo,
        })
      : null;
  }
  if (dbModel && !opt.force) {
    return new GitHubRepository(dbModel);
  }
  const oct = octokit(opt.accessToken);
  let r: RestEndpointMethodTypes["repos"]["get"]["response"]["data"];
  // Recover the numeric GitHub repository id from `externalId` (stored as
  // "gh_<id>") so we can re-fetch by id if the owner/repo we cached has
  // since been renamed on GitHub — see #409.
  const ghIdFromExternal =
    typeof dbModel?.externalId === "string" &&
    dbModel.externalId.startsWith("gh_")
      ? dbModel.externalId.slice(3)
      : null;
  try {
    r = (
      await oct.repos.get({
        owner: opt.owner,
        repo: opt.repo,
      })
    ).data;
  } catch (error) {
    if (
      error instanceof Error &&
      error.message.includes(
        "organization has enabled OAuth App access restrictions"
      )
    ) {
      throw new AnonymousError("repo_access_limited", {
        httpStatus: 403,
        object: {
          owner: opt.owner,
          repo: opt.repo,
        },
        cause: error as Error,
      });
    }
    // If the name 404s but we know the GitHub repo id, the repo was
    // probably renamed. Look it up by id and continue with the new name.
    const status = (error as { status?: number }).status;
    if (status === 404 && ghIdFromExternal) {
      try {
        r = (
          await oct.request("GET /repositories/{id}", {
            id: ghIdFromExternal,
          })
        ).data as RestEndpointMethodTypes["repos"]["get"]["response"]["data"];
      } catch (idError) {
        throw new AnonymousError("repo_not_found", {
          httpStatus: (idError as { status?: number }).status || 404,
          object: { owner: opt.owner, repo: opt.repo },
          cause: idError as Error,
        });
      }
    } else {
      throw new AnonymousError("repo_not_found", {
        httpStatus: status,
        object: {
          owner: opt.owner,
          repo: opt.repo,
        },
        cause: error as Error,
      });
    }
  }
  if (!r)
    throw new AnonymousError("repo_not_found", {
      httpStatus: 404,
      object: {
        owner: opt.owner,
        repo: opt.repo,
      },
    });
  let pageSource:
    | RestEndpointMethodTypes["repos"]["getPages"]["response"]["data"]["source"]
    | undefined;
  if (r.has_pages) {
    const ghPageRes = await oct.repos.getPages({
      owner: opt.owner,
      repo: opt.repo,
    });
    pageSource = ghPageRes.data.source;
  }

  if (!isConnected) {
    const model = dbModel || new RepositoryModel({ externalId: "gh_" + r.id });
    model.name = r.full_name;
    model.url = r.html_url;
    model.size = r.size;
    model.defaultBranch = r.default_branch;
    model.hasPage = r.has_pages;
    if (pageSource) model.pageSource = pageSource;
    return new GitHubRepository(model);
  }

  // Atomic upsert keyed on externalId so concurrent requests for the same
  // GitHub repo can't both insert and trip the unique-index race (E11000).
  const update: Record<string, unknown> = {
    $set: {
      name: r.full_name,
      url: r.html_url,
      size: r.size,
      defaultBranch: r.default_branch,
      hasPage: r.has_pages,
      ...(pageSource ? { pageSource } : {}),
    },
    $setOnInsert: { externalId: "gh_" + r.id },
  };

  let model: IRepositoryDocument | null;
  try {
    model = await RepositoryModel.findOneAndUpdate(
      { externalId: "gh_" + r.id },
      update,
      { upsert: true, new: true, setDefaultsOnInsert: true }
    );
  } catch (error) {
    // Mongo can still raise E11000 on a concurrent upsert insert; retry
    // once with a plain find — the winning insert is now visible.
    if ((error as { code?: number }).code === 11000) {
      model = await RepositoryModel.findOne({ externalId: "gh_" + r.id });
    } else {
      throw error;
    }
  }
  if (!model) {
    throw new AnonymousError("repo_not_found", {
      httpStatus: 404,
      object: { owner: opt.owner, repo: opt.repo },
    });
  }
  return new GitHubRepository(model);
}