Remove OpenTelemetry tracing infrastructure (#662)

This commit is contained in:
Thomas Durieux
2026-04-15 04:39:08 +02:00
committed by GitHub
parent c6d6806d7a
commit 655ae92c4c
19 changed files with 848 additions and 1366 deletions
+44 -61
View File
@@ -4,7 +4,6 @@ import { OctokitResponse } from "@octokit/types";
import storage from "../storage";
import GitHubBase, { GitHubBaseData } from "./GitHubBase";
import { trace } from "@opentelemetry/api";
import { FILE_TYPE } from "../storage/Storage";
import { octokit } from "../GitHubUtils";
import AnonymousError from "../AnonymousError";
@@ -27,47 +26,39 @@ export default class GitHubDownload extends GitHubBase {
}
async download(progress?: (status: string) => void) {
const span = trace.getTracer("ano-file").startSpan("GHDownload.download");
span.setAttribute("repoId", this.data.repoId);
let response: OctokitResponse<unknown, number>;
try {
let response: OctokitResponse<unknown, number>;
try {
response = await this.getZipUrl();
} catch (error) {
span.recordException(error as Error);
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status || 404,
object: this.data,
cause: error as Error,
});
}
await storage.mk(this.data.repoId);
try {
const downloadStream = got.stream(response.url);
downloadStream.addListener(
"downloadProgress",
(p: { transferred?: number }) => {
if (progress && p.transferred) {
progress("Repository download: " + humanFileSize(p.transferred));
}
response = await this.getZipUrl();
} catch (error) {
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status || 404,
object: this.data,
cause: error as Error,
});
}
await storage.mk(this.data.repoId);
try {
const downloadStream = got.stream(response.url);
downloadStream.addListener(
"downloadProgress",
(p: { transferred?: number }) => {
if (progress && p.transferred) {
progress("Repository download: " + humanFileSize(p.transferred));
}
);
await storage.extractZip(
this.data.repoId,
"",
downloadStream,
this.type
);
} catch (error) {
span.recordException(error as Error);
throw new AnonymousError("unable_to_download", {
httpStatus: 500,
cause: error as Error,
object: this.data,
});
}
} finally {
span.end();
}
);
await storage.extractZip(
this.data.repoId,
"",
downloadStream,
this.type
);
} catch (error) {
throw new AnonymousError("unable_to_download", {
httpStatus: 500,
cause: error as Error,
object: this.data,
});
}
}
@@ -75,29 +66,21 @@ export default class GitHubDownload extends GitHubBase {
file: AnonymizedFile,
progress?: (status: string) => void
): Promise<Readable> {
const span = trace
.getTracer("ano-file")
.startSpan("GHDownload.getFileContent");
span.setAttribute("repoId", this.data.repoId);
try {
const exists = await storage.exists(this.data.repoId, file.filePath);
if (exists === FILE_TYPE.FILE) {
return storage.read(this.data.repoId, file.filePath);
} else if (exists === FILE_TYPE.FOLDER) {
throw new AnonymousError("folder_not_supported", {
httpStatus: 400,
object: file,
});
}
// will throw an error if the file is not in the repository
await file.originalPath();
// the cache is not ready, we need to download the repository
await this.download(progress);
const exists = await storage.exists(this.data.repoId, file.filePath);
if (exists === FILE_TYPE.FILE) {
return storage.read(this.data.repoId, file.filePath);
} finally {
span.end();
} else if (exists === FILE_TYPE.FOLDER) {
throw new AnonymousError("folder_not_supported", {
httpStatus: 400,
object: file,
});
}
// will throw an error if the file is not in the repository
await file.originalPath();
// the cache is not ready, we need to download the repository
await this.download(progress);
return storage.read(this.data.repoId, file.filePath);
}
async getFiles(progress?: (status: string) => void) {
+153 -189
View File
@@ -1,7 +1,6 @@
import { Branch } from "../types";
import * as gh from "parse-github-url";
import { RestEndpointMethodTypes } from "@octokit/rest";
import { trace } from "@opentelemetry/api";
import AnonymousError from "../AnonymousError";
import { isConnected } from "../../server/database";
@@ -55,81 +54,64 @@ export class GitHubRepository {
accessToken: string;
}
) {
const span = trace
.getTracer("ano-file")
.startSpan("GHRepository.getCommitInfo");
span.setAttribute("owner", this.owner);
span.setAttribute("repo", this.repo);
try {
const oct = octokit(opt.accessToken);
const commit = await oct.repos.getCommit({
owner: this.owner,
repo: this.repo,
ref: sha,
});
return commit.data;
} finally {
span.end();
}
const oct = octokit(opt.accessToken);
const commit = await oct.repos.getCommit({
owner: this.owner,
repo: this.repo,
ref: sha,
});
return commit.data;
}
async branches(opt: {
accessToken: string;
force?: boolean;
}): Promise<Branch[]> {
const span = trace.getTracer("ano-file").startSpan("GHRepository.branches");
span.setAttribute("owner", this.owner);
span.setAttribute("repo", this.repo);
try {
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) {
span.recordException(error as Error);
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status,
cause: error as Error,
object: this,
});
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 } }
);
}
} else if (isConnected) {
const q = await RepositoryModel.findOne({ externalId: this.id }).select(
"branches"
);
this._data.branches = q?.branches;
} catch (error) {
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status,
cause: error as Error,
object: this,
});
}
return this._data.branches || [];
} finally {
span.end();
} 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: {
@@ -137,60 +119,52 @@ export class GitHubRepository {
force?: boolean;
accessToken: string;
}): Promise<string | undefined> {
const span = trace.getTracer("ano-file").startSpan("GHRepository.readme");
span.setAttribute("owner", this.owner);
span.setAttribute("repo", this.repo);
try {
if (!opt.branch) opt.branch = this._data.defaultBranch || "master";
if (!opt.branch) opt.branch = this._data.defaultBranch || "master";
const model = await RepositoryModel.findOne({
externalId: this.id,
}).select("branches");
const model = await RepositoryModel.findOne({
externalId: this.id,
}).select("branches");
if (!model) {
throw new AnonymousError("repo_not_found", { httpStatus: 404 });
}
if (!model) {
throw new AnonymousError("repo_not_found", { httpStatus: 404 });
}
this._data.branches = await this.branches(opt);
model.branches = this._data.branches;
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) {
span.recordException(error as Error);
throw new AnonymousError("readme_not_available", {
httpStatus: 404,
cause: error as Error,
object: this,
});
}
}
if (!selected) {
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,
});
}
return selected.readme;
} finally {
span.end();
}
if (!selected) {
throw new AnonymousError("readme_not_available", {
httpStatus: 404,
object: this,
});
}
return selected.readme;
}
public get owner(): string {
@@ -235,60 +209,44 @@ export async function getRepositoryFromGitHub(opt: {
accessToken: string;
force?: boolean;
}) {
const span = trace
.getTracer("ano-file")
.startSpan("GHRepository.getRepositoryFromGitHub");
span.setAttribute("owner", opt.owner);
span.setAttribute("repo", opt.repo);
try {
if (opt.repo.indexOf(".git") > -1) {
opt.repo = opt.repo.replace(".git", "");
}
let dbModel = null;
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"];
try {
r = (
await oct.repos.get({
owner: opt.owner,
repo: opt.repo,
if (opt.repo.indexOf(".git") > -1) {
opt.repo = opt.repo.replace(".git", "");
}
let dbModel = null;
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,
})
).data;
} catch (error) {
span.recordException(error as 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,
});
}
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status,
: null;
}
if (dbModel && !opt.force) {
return new GitHubRepository(dbModel);
}
const oct = octokit(opt.accessToken);
let r: RestEndpointMethodTypes["repos"]["get"]["response"]["data"];
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,
@@ -296,32 +254,38 @@ export async function getRepositoryFromGitHub(opt: {
cause: error as Error,
});
}
if (!r)
throw new AnonymousError("repo_not_found", {
httpStatus: 404,
object: {
owner: opt.owner,
repo: opt.repo,
},
});
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 (model.hasPage) {
const ghPageRes = await oct.repos.getPages({
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status,
object: {
owner: opt.owner,
repo: opt.repo,
});
model.pageSource = ghPageRes.data.source;
}
if (isConnected) {
await model.save();
}
return new GitHubRepository(model);
} finally {
span.end();
},
cause: error as Error,
});
}
if (!r)
throw new AnonymousError("repo_not_found", {
httpStatus: 404,
object: {
owner: opt.owner,
repo: opt.repo,
},
});
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 (model.hasPage) {
const ghPageRes = await oct.repos.getPages({
owner: opt.owner,
repo: opt.repo,
});
model.pageSource = ghPageRes.data.source;
}
if (isConnected) {
await model.save();
}
return new GitHubRepository(model);
}
+90 -142
View File
@@ -7,7 +7,6 @@ import { basename, dirname } from "path";
import * as stream from "stream";
import AnonymousError from "../AnonymousError";
import { trace } from "@opentelemetry/api";
import { FILE_TYPE } from "../storage/Storage";
import { octokit } from "../GitHubUtils";
import FileModel from "../model/files/files.model";
@@ -21,8 +20,6 @@ export default class GitHubStream extends GitHubBase {
}
downloadFile(token: string, sha: string) {
const span = trace.getTracer("ano-file").startSpan("GHStream.downloadFile");
span.setAttribute("sha", sha);
const oct = octokit(token);
try {
const { url } = oct.rest.git.getBlob.endpoint({
@@ -40,14 +37,11 @@ export default class GitHubStream extends GitHubBase {
});
} catch (error) {
console.error(error);
// span.recordException(error as Error);
throw new AnonymousError("repo_not_accessible", {
httpStatus: 404,
object: this.data,
cause: error as Error,
});
} finally {
span.end();
}
}
@@ -56,12 +50,6 @@ export default class GitHubStream extends GitHubBase {
repoId: string,
fileSha: () => Promise<string> | string
) {
const span = trace
.getTracer("ano-file")
.startSpan("GHStream.getFileContent");
span.setAttribute("repoId", repoId);
span.setAttribute("file", filePath);
const fileInfo = await storage.exists(repoId, filePath);
if (fileInfo == FILE_TYPE.FILE) {
return storage.read(repoId, filePath);
@@ -76,10 +64,6 @@ export default class GitHubStream extends GitHubBase {
await fileSha()
);
content.on("close", () => {
span.end();
});
// duplicate the stream to write it to the storage
const stream1 = content.pipe(new stream.PassThrough());
const stream2 = content.pipe(new stream.PassThrough());
@@ -99,45 +83,30 @@ export default class GitHubStream extends GitHubBase {
}
async getFileContent(file: AnonymizedFile): Promise<stream.Readable> {
const span = trace
.getTracer("ano-file")
.startSpan("GHStream.getFileContent");
span.setAttribute("repoId", file.repository.repoId);
span.setAttribute("file", file.anonymizedPath);
try {
try {
void file.filePath;
} catch (_) {
// compute the original path if ambiguous
await file.originalPath();
}
return this.getFileContentCache(
file.filePath,
file.repository.repoId,
async () => {
const fileSha = await file.sha();
if (!fileSha) {
throw new AnonymousError("file_not_accessible", {
httpStatus: 404,
object: file,
});
}
return fileSha;
}
);
} finally {
span.end();
void file.filePath;
} catch (_) {
// compute the original path if ambiguous
await file.originalPath();
}
return this.getFileContentCache(
file.filePath,
file.repository.repoId,
async () => {
const fileSha = await file.sha();
if (!fileSha) {
throw new AnonymousError("file_not_accessible", {
httpStatus: 404,
object: file,
});
}
return fileSha;
}
);
}
async getFiles(progress?: (status: string) => void) {
const span = trace.getTracer("ano-file").startSpan("GHStream.getFiles");
span.setAttribute("repoId", this.data.repoId);
try {
return this.getTruncatedTree(this.data.commit, progress);
} finally {
span.end();
}
return this.getTruncatedTree(this.data.commit, progress);
}
private async getGHTree(
@@ -145,25 +114,19 @@ export default class GitHubStream extends GitHubBase {
count = { request: 0, file: 0 },
opt = { recursive: true, callback: () => {} }
) {
const span = trace.getTracer("ano-file").startSpan("GHStream.getGHTree");
span.setAttribute("sha", sha);
try {
const oct = octokit(await this.data.getToken());
const ghRes = await oct.git.getTree({
owner: this.data.organization,
repo: this.data.repoName,
tree_sha: sha,
recursive: opt.recursive === true ? "1" : undefined,
});
count.request++;
count.file += ghRes.data.tree.length;
if (opt.callback) {
opt.callback();
}
return ghRes.data;
} finally {
span.end();
const oct = octokit(await this.data.getToken());
const ghRes = await oct.git.getTree({
owner: this.data.organization,
repo: this.data.repoName,
tree_sha: sha,
recursive: opt.recursive === true ? "1" : undefined,
});
count.request++;
count.file += ghRes.data.tree.length;
if (opt.callback) {
opt.callback();
}
return ghRes.data;
}
private async getTruncatedTree(
@@ -175,65 +138,56 @@ export default class GitHubStream extends GitHubBase {
request: 0,
file: 0,
};
const span = trace
.getTracer("ano-file")
.startSpan("GHStream.getTruncatedTree");
span.setAttribute("sha", sha);
span.setAttribute("parentPath", parentPath);
const output: IFile[] = [];
let data = null;
try {
let data = null;
try {
data = await this.getGHTree(sha, count, {
recursive: false,
callback: () => {
if (progress) {
progress("List file: " + count.file);
}
},
});
output.push(...this.tree2Tree(data.tree, parentPath));
} catch (error) {
console.log(error);
if ((error as any).status == 409 || (error as any).status == 404) {
// empty repo
data = { tree: [] };
} else {
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status || 404,
object: this.data,
cause: error as Error,
});
}
}
const promises: ReturnType<GitHubStream["getGHTree"]>[] = [];
const parentPaths: string[] = [];
for (const file of data.tree) {
if (file.type == "tree" && file.path && file.sha) {
const elementPath = path.join(parentPath, file.path);
parentPaths.push(elementPath);
promises.push(
this.getGHTree(file.sha, count, {
recursive: true,
callback: () => {
if (progress) {
progress("List file: " + count.file);
}
},
})
);
}
}
(await Promise.all(promises)).forEach((data, i) => {
if (data.truncated) {
// TODO: the tree is truncated
}
output.push(...this.tree2Tree(data.tree, parentPaths[i]));
data = await this.getGHTree(sha, count, {
recursive: false,
callback: () => {
if (progress) {
progress("List file: " + count.file);
}
},
});
return output;
} finally {
span.end();
output.push(...this.tree2Tree(data.tree, parentPath));
} catch (error) {
console.log(error);
if ((error as any).status == 409 || (error as any).status == 404) {
// empty repo
data = { tree: [] };
} else {
throw new AnonymousError("repo_not_found", {
httpStatus: (error as any).status || 404,
object: this.data,
cause: error as Error,
});
}
}
const promises: ReturnType<GitHubStream["getGHTree"]>[] = [];
const parentPaths: string[] = [];
for (const file of data.tree) {
if (file.type == "tree" && file.path && file.sha) {
const elementPath = path.join(parentPath, file.path);
parentPaths.push(elementPath);
promises.push(
this.getGHTree(file.sha, count, {
recursive: true,
callback: () => {
if (progress) {
progress("List file: " + count.file);
}
},
})
);
}
}
(await Promise.all(promises)).forEach((data, i) => {
if (data.truncated) {
// TODO: the tree is truncated
}
output.push(...this.tree2Tree(data.tree, parentPaths[i]));
});
return output;
}
private tree2Tree(
@@ -247,25 +201,19 @@ export default class GitHubStream extends GitHubBase {
}[],
parentPath: string = ""
) {
const span = trace.getTracer("ano-file").startSpan("GHStream.tree2Tree");
span.setAttribute("parentPath", parentPath);
try {
return tree.map((elem) => {
const fullPath = path.join(parentPath, elem.path || "");
let pathFile = dirname(fullPath);
if (pathFile === ".") {
pathFile = "";
}
return new FileModel({
name: basename(fullPath),
path: pathFile,
repoId: this.data.repoId,
size: elem.size,
sha: elem.sha,
});
return tree.map((elem) => {
const fullPath = path.join(parentPath, elem.path || "");
let pathFile = dirname(fullPath);
if (pathFile === ".") {
pathFile = "";
}
return new FileModel({
name: basename(fullPath),
path: pathFile,
repoId: this.data.repoId,
size: elem.size,
sha: elem.sha,
});
} finally {
span.end();
}
});
}
}