back cli: add user dump and list commands

This commit is contained in:
Eric Doughty-Papassideris
2025-02-28 00:52:19 +01:00
committed by Anton Shepilov
parent eec23b230b
commit 8efcb2546c
6 changed files with 439 additions and 0 deletions
@@ -0,0 +1,116 @@
import yargs from "yargs";
import runWithPlatform from "../../lib/run-with-platform";
import gr from "../../../services/global-resolver";
import { messageQueueLogger, platformLogger } from "../../../core/platform/framework";
import {
buildUserDeletionRepositories,
descendDriveItemsDepthFirstRandomOrder,
loadRawVersionsOfItemForDeletion,
type TUserDeletionRepos,
} from "../../../core/platform/services/admin/utils";
import type User from "../../../services/user/entities/user";
import { EntityByKind } from "../../utils/print-entities";
let globalArgv: {
verbose: boolean;
user_id: string;
json_output: boolean;
with_storage_paths: boolean;
user?: User; // Not an argument per say, hydrated in handler
};
const lengthOfLongestEntityKind = Object.keys(EntityByKind).reduce(
(acc, kind) => Math.max(acc, kind.length),
0,
);
const entityToHeaderString = <K extends keyof typeof EntityByKind>(
kind: K & string,
entity: Parameters<(typeof EntityByKind)[K]["headerOf"]>[0],
depth: number = 0,
): string => {
if (globalArgv.json_output) return kind + " " + JSON.stringify(entity); // Can't really print out json directly, valid json seems to get prettified by pino
// ts doesn't seem to infer entityHeaderPrinter[kind] correctly even with help
// eslint-disable-next-line @typescript-eslint/no-explicit-any
return `${kind.padEnd(lengthOfLongestEntityKind)} ${entity.id} ${new Array(depth + 1).join(
" ",
)}${EntityByKind[kind].headerOf(entity as any)}`;
};
async function dumpAccount(repos: TUserDeletionRepos): Promise<number> {
console.log(entityToHeaderString("user", globalArgv.user));
await descendDriveItemsDepthFirstRandomOrder(
repos,
"user_" + globalArgv.user.id,
async (item, _children, parents) => {
const depth = parents.length + 1;
if (!item.is_directory) {
const versionsAssets = await loadRawVersionsOfItemForDeletion(
repos,
item,
globalArgv.with_storage_paths,
);
for (const { version, file, paths } of versionsAssets) {
const outputPath = (path: string) =>
console.log(entityToHeaderString("s3", { id: file.id, path }, depth + 2));
if (globalArgv.json_output) {
for (const path of paths || []) outputPath(path);
} else if (paths) {
paths.sort((a, b) => a.localeCompare(b));
if (paths.length > 0) outputPath(paths[0]);
if (paths.length == 3) outputPath(paths[1]);
else if (paths.length > 3) outputPath(`... skipping ${paths.length - 2} paths ...`);
if (paths.length > 1) outputPath(paths[paths.length - 1]);
}
file && console.log(entityToHeaderString("file", file, depth + 2));
version && console.log(entityToHeaderString("version", version, depth + 1));
}
}
console.log(entityToHeaderString("item", item, depth));
},
);
return 0;
}
export default {
command: "dump <user_id>",
describe: "Output a list of entities related to a user",
builder: {
user_id: {
demandOption: true,
type: "string",
describe:
"id of the user, or a search string for the email.\n(if the id doesn't match, will print the matching users and exit)",
},
json_output: {
type: "boolean",
alias: "j",
describe: "Output in lines of '[kind] [json_object]'.",
},
with_storage_paths: {
type: "boolean",
alias: "p",
describe: "If set, also query the storage to include all relevant paths in the output",
},
},
handler: async (argv: typeof globalArgv) => {
globalArgv = argv;
if (!argv.verbose) {
platformLogger.level = "warn";
messageQueueLogger.level = "warn";
}
await runWithPlatform("dump_account", async ({ spinner: _spinner, platform: _platform }) => {
const repos = await buildUserDeletionRepositories(gr.database, gr.platformServices.search);
argv.user = await repos.user.findOne({ id: argv.user_id });
if (!argv.user) {
console.error(`Error, unknown user ${JSON.stringify(argv.user_id)}`);
return 1;
}
await dumpAccount(repos);
return 0;
});
},
} as yargs.CommandModule<object, typeof globalArgv>;
@@ -0,0 +1,69 @@
import yargs from "yargs";
import runWithPlatform from "../../lib/run-with-platform";
import { messageQueueLogger, platformLogger } from "../../../core/platform/framework";
import { buildUserDeletionRepositories } from "../../../core/platform/services/admin/utils";
import type User from "../../../services/user/entities/user";
import { EntityByKind } from "../../utils/print-entities";
let globalArgv: {
verbose: boolean;
email_like: string;
json_output: boolean;
user?: User; // Not an argument per say, hydrated in handler
};
const entityToHeaderString = <K extends keyof typeof EntityByKind>(
kind: K & string,
entity: Parameters<(typeof EntityByKind)[K]["headerOf"]>[0],
): string => {
if (globalArgv.json_output) return kind + " " + JSON.stringify(entity); // Can't really print out json directly, valid json seems to get prettified by pino
// ts doesn't seem to infer entityHeaderPrinter[kind] correctly even with help
// eslint-disable-next-line @typescript-eslint/no-explicit-any
return `${entity.id} ${EntityByKind[kind].headerOf(entity as any)}`;
};
export default {
command: "list",
describe: "list users",
builder: {
email_like: {
alias: "e",
type: "string",
describe:
"id of the user, or a search string for the email.\n(if the id doesn't match, will print the matching users and exit)",
},
json_output: {
type: "boolean",
alias: "j",
describe: "Output in lines of '[kind] [json_object]'.",
},
},
handler: async (argv: typeof globalArgv) => {
globalArgv = argv;
if (!argv.verbose) {
platformLogger.level = "warn";
messageQueueLogger.level = "warn";
}
await runWithPlatform("dump_account", async ({ spinner: _spinner, platform: _platform }) => {
const repos = await buildUserDeletionRepositories();
console.log(
(
await repos.user.find(
{},
{
...(argv.email_like ? { $like: [["email_canonical", argv.email_like]] } : {}),
sort: { email_canonical: "asc" },
},
)
)
.getEntities()
.map(user => entityToHeaderString("user", user))
.join("\n"),
);
return 0;
});
},
} as yargs.CommandModule<object, typeof globalArgv>;
@@ -35,6 +35,8 @@ export default async function runWithPlatform(
spinner.fail(err.stack || err);
process.exitCode = 1;
}
// Spinner seems to interrupt still buffered output otherwise
await new Promise(resolve => setTimeout(resolve, 200));
spinner.start("Platform: shutting down...");
await platform.stop();
spinner.succeed("Platform: shutdown");
@@ -0,0 +1,92 @@
import type { EntityTarget } from "../../core/platform/services/search/api";
import gr from "../../services/global-resolver";
import User, { TYPE as UserTYPE } from "../../services/user/entities/user";
import { DriveFile, TYPE as DriveFileTYPE } from "../../services/documents/entities/drive-file";
import { File } from "../../services/files/entities/file";
import {
FileVersion,
TYPE as FileVersionTYPE,
} from "../../services/documents/entities/file-version";
import CompanyUser, { TYPE as CompanyUserTYPE } from "../../services/user/entities/company_user";
import {
MissedDriveFile,
TYPE as MissedDriveFileTYPE,
} from "../../services/documents/entities/missed-drive-file";
import Company, { TYPE as CompanyTYPE } from "../../services/user/entities/company";
const FileTYPE = "files";
/** Fake entity to represent S3 path entries as similar resources */
interface S3Object {
/** id of the File */
id: string;
/** path in bucket of the file */
path: string;
}
const epochToISO = epoch => (epoch ? new Date(epoch).toISOString() : "<null date>");
export const EntityKindToString = {
user: (entity: User): string => `🤺 ${entity.email_canonical}`,
// extUser: (entity: ExternalUser): string => `🤺 ${entity.user_id} <-> 🌍 ${entity.external_id}`,
compUser: (entity: CompanyUser): string => `🤺 ${entity.user_id} <-> 🏦 ${entity.group_id}`,
// session: (entity: Session): string => `🤺 ${entity.sub} <-> 🏦 ${entity.company_id}`,
company: (entity: Company): string => `🏦 ${entity.name || entity.displayName}`,
item: (entity: DriveFile): string =>
`${entity.is_in_trash ? "🗑️ " : ""}${entity.is_directory ? "📁" : "📄"}${
entity.scope != "personal" ? "🏢 " : ""
} ${entity.name}`,
version: (entity: FileVersion): string =>
`🕰️ ${epochToISO(entity.date_added)} - ${entity.file_size}`,
file: (entity: File): string => `💿 ${epochToISO(entity.updated_at)}`,
s3: (entity: S3Object): string => `☁️ ${entity.path}`,
missed: (entity: MissedDriveFile): string =>
`😭 ${entity.is_in_trash ? "🗑️ " : ""} ${entity.name}`,
};
const makeEntityKind = <T>(
kind: keyof typeof EntityKindToString,
db?: { entity: EntityTarget<T>; type: string },
) => {
const headerToString = EntityKindToString[kind];
return {
kind,
headerOf: headerToString,
db: db
? {
getRepository: () => gr.database.getRepository<T>(db.type, db.entity),
...db,
}
: undefined,
};
};
export type TEntityNames = keyof typeof Entities;
export const Entities = {
User: makeEntityKind<User>("user", { entity: User, type: UserTYPE }),
DriveFile: makeEntityKind<DriveFile>("item", { entity: DriveFile, type: DriveFileTYPE }),
File: makeEntityKind<File>("file", { entity: File, type: FileTYPE }),
FileVersion: makeEntityKind<FileVersion>("version", {
entity: FileVersion,
type: FileVersionTYPE,
}),
// ExternalUser: makeEntityKind<ExternalUser>("extUser", {
// entity: ExternalUser,
// type: ExternalUserTYPE,
// }),
CompanyUser: makeEntityKind<CompanyUser>("compUser", {
entity: CompanyUser,
type: CompanyUserTYPE,
}),
MissedDriveFile: makeEntityKind<MissedDriveFile>("missed", {
entity: MissedDriveFile,
type: MissedDriveFileTYPE,
}),
Company: makeEntityKind<Company>("company", { entity: Company, type: CompanyTYPE }),
// Session: makeEntityKind<Session>("session", { entity: Session, type: SessionTYPE }),
$S3: makeEntityKind<S3Object>("s3"),
};
export const EntityByKind = Object.fromEntries(
Object.entries(Entities).map(([_, def]) => [def.kind, def]),
);
@@ -0,0 +1,150 @@
import { getLogger } from "../../framework";
import type { DatabaseServiceAPI } from "../database/api";
import type { SearchServiceAPI } from "../search/api";
import gr from "../../../../services/global-resolver";
import type Repository from "../database/services/orm/repository/repository";
import User, { TYPE as UserTYPE } from "../../../../services/user/entities/user";
import {
DriveFile,
TYPE as DriveFileTYPE,
} from "../../../../services/documents/entities/drive-file";
import { File } from "../../../../services/files/entities/file";
import {
FileVersion,
TYPE as FileVersionTYPE,
} from "../../../../services/documents/entities/file-version";
import ExternalUser, {
TYPE as ExternalUserTYPE,
} from "../../../../services/user/entities/external_user";
import CompanyUser, {
TYPE as CompanyUserType,
} from "../../../../services/user/entities/company_user";
import {
MissedDriveFile,
TYPE as MissedDriveFileTYPE,
} from "../../../../services/documents/entities/missed-drive-file";
import Session from "../../../../services/console/entities/session";
import { fisherYattesShuffleInPlace } from "../../../../utils/arrays";
import { getFilePath } from "../../../../services/files/services";
import Company, { TYPE as CompanyType } from "../../../../services/user/entities/company";
const FileTYPE = "files";
export const adminLogger = getLogger("Admin");
/** Run the `map` function on sets of `batchSize` in `list`, sequentially between batches */
async function runInBatches<T, R>(
batchSize: number,
list: readonly T[],
map: (items: T[]) => Promise<R>,
): Promise<R[]> {
const listCopy = [...list];
let batch;
const result = [] as R[];
while ((batch = listCopy.splice(0, batchSize)).length) {
console.error("batch", batch);
result.push(await map(batch));
}
return result;
}
/** Call {@see runInBatches} expecting a boolean result from `map`, return `true` only if empty or all batches returned true */
export async function runInBatchesAreAllTrue<T>(
batchSize: number,
list: T[],
map: (items: T[]) => Promise<boolean[]>,
): Promise<boolean> {
return (
await runInBatches(batchSize, list, async results => (await map(results)).every(x => !!x))
).every(x => !!x);
}
export type TUserDeletionRepos = Awaited<ReturnType<typeof buildUserDeletionRepositories>>;
/**
* Create all repositories required for deleting a user
* @deprecated Do not use this outside of this file, it is exported exclusively for e2e tests
*/
export async function buildUserDeletionRepositories(
db: DatabaseServiceAPI = gr.database,
search: SearchServiceAPI = gr.platformServices.search,
) {
return {
driveFile: await db.getRepository<DriveFile>(DriveFileTYPE, DriveFile),
file: await db.getRepository<File>(FileTYPE, File),
fileVersion: await db.getRepository<FileVersion>(FileVersionTYPE, FileVersion),
missedDriveFile: await db.getRepository<MissedDriveFile>(MissedDriveFileTYPE, MissedDriveFile),
user: await db.getRepository<User>(UserTYPE, User),
companyUser: await db.getRepository<CompanyUser>(CompanyUserType, CompanyUser),
externalUser: await db.getRepository<ExternalUser>(ExternalUserTYPE, ExternalUser),
company: await db.getRepository<Company>(CompanyType, Company),
// This one is not typical because it's only valid for remote account type:
session: gr.services.console.getSessionRepo() as Repository<Session> | null,
//TODO: checkout what to do with session and user_online (seen in prod)
search: {
driveFile: await search.getRepository<DriveFile>(DriveFileTYPE, DriveFile),
user: await search.getRepository<User>(UserTYPE, User),
},
};
}
/** Return versions, sorted by oldest first, and related DB and S3 entries for a given DriveFile */
export async function loadRawVersionsOfItemForDeletion(
repos: TUserDeletionRepos,
item: DriveFile,
loadStoragePaths = false,
): Promise<
{
version: FileVersion;
file?: File;
paths?: string[];
}[]
> {
return Promise.all(
(await repos.fileVersion.find({ drive_item_id: item.id }, { sort: { date_added: "asc" } }))
.getEntities()
.map(async version => {
const file = await repos.file.findOne({ id: version.file_metadata.external_id });
if (!loadStoragePaths) return { version, file };
const paths = file
? await gr.platformServices.storage.enumeratePathsForFile(getFilePath(file))
: [];
return { version, file, paths };
}),
);
}
/**
* Recursively descends through drive items in a depth-first manner with a random order at each level.
*
* @warn No security or filtering is done (e.g., no rights are checked, items in trash are included, etc.).
*
* @param {string} parent_id - The ID of the parent drive item (or virtual root) to start the descent from.
* @param {(item: DriveFile, children: undefined | T[], parents: DriveFile[]) => Promise<T>} map - The mapping function to process each drive item.
* - `item`: The current drive item being processed.
* - `children`: The results of processing the children of the current drive item, or `undefined` if the item is not a directory.
* - `parents`: An array of parent drive items leading to the current item.
* @returns {Promise<T[]>} A promise that resolves to an array of results produced by the `map` function.
*/
export async function descendDriveItemsDepthFirstRandomOrder<T>(
repos: TUserDeletionRepos,
parent_id: string,
map: (item: DriveFile, children: undefined | T[], parents: readonly DriveFile[]) => Promise<T>,
) {
async function descend(parent_id: string, parents: readonly DriveFile[]): Promise<T[]> {
const items = (await repos.driveFile.find({ parent_id })).getEntities();
fisherYattesShuffleInPlace(items);
const result = new Array(items.length);
for (const [index, item] of items.entries()) {
const children = item.is_directory
? await descend(item.id, parents.concat([item]))
: undefined;
result[index] = await map(item, children, parents);
}
return result;
}
return descend(parent_id, []);
}
+10
View File
@@ -0,0 +1,10 @@
/** Shuffle an array in place, returns its parameter for convenience */
export function fisherYattesShuffleInPlace<T>(list: T[]): T[] {
let index = list.length;
while (index) {
const randomIndex = Math.floor(Math.random() * index);
index--;
[list[index], list[randomIndex]] = [list[randomIndex], list[index]];
}
return list;
}