PronounsPage/server/cleanupImages.ts

import './setup.ts';

import { S3 } from '@aws-sdk/client-s3';
import type { ListObjectsV2Output, ObjectIdentifier, _Object } from '@aws-sdk/client-s3';

import { awsConfig, awsParams } from './aws.ts';
import dbConnection from './db.ts';

import type { Profile } from '~/src/profile.ts';

const FRESH_CUTOFF = 6 * 60 * 60 * 1000; // 6 hours

const execute = process.env.EXECUTE === '1';

console.log(execute ? 'WILL REMOVE FILES!' : 'Dry run');

const validateIds = (type: string, data: Record<string, boolean>, expectedMinimumSize: number = 0): void => {
    console.log(`${type}: ${Object.keys(data).length}`);

    if (Object.keys(data).length <= expectedMinimumSize) {
        throw new Error('List of ids has unexpectedly few items!');
    }

    const invalidKeys = [];
    for (const key of Object.keys(data)) {
        if (!key.match(/^[A-Z0-9]{26}$/)) {
            invalidKeys.push(key);
        }
    }

    if (invalidKeys.length) {
        console.warn(`${invalidKeys.length} invalid ${type} keys: ${invalidKeys.join(', ')}`);
        if (invalidKeys.length > 10) {
            throw new Error('Too many invalid keys');
        }
    }
};

async function cleanup(): Promise<void> {
    console.log('--- Fetching ids expected to stay ---');
    const db = await dbConnection();

    const avatars: Record<string, true> = {};
    for (const row of await db.all(`
        SELECT avatarSource
        FROM users
        WHERE avatarSource LIKE 'https://%/images/%'
    `)) {
        avatars[row.avatarSource.match('https://[^/]+/images/(.*)-(?:thumb|avatar).png')[1]] = true;
    }

    const flags: Record<string, true> = {};
    for (const row of await db.all(`
        SELECT customFlags
        FROM profiles
        WHERE customFlags != '[]'
    `)) {
        for (const { value } of JSON.parse(row.customFlags)) {
            flags[value] = true;
        }
    }

    const sources: Record<string, true> = {};
    for (const row of await db.all(`
        SELECT images
        FROM sources
        WHERE images is not null AND images != '' AND deleted = 0
    `)) {
        for (const key of row.images.split(',')) {
            sources[key] = true;
        }
    }

    const terms: Record<string, true> = {};
    for (const row of await db.all(`
        SELECT images
        FROM terms
        WHERE images is not null AND images != ''
    `)) {
        for (const key of row.images.split(',')) {
            terms[key] = true;
        }
    }

    const cards: Record<string, true> = {};
    for (const row of await db.all<Pick<Profile, 'card'>[]>(`
        SELECT card
        FROM profiles
        WHERE card is not null AND card != ''
    `)) {
        cards[row.card!.match(
            /(https:\/\/pronouns-page.s3.eu-west-1.amazonaws.com\/card\/[^/]+\/.+-([^-]+)\.png)/,
        )![1]] = true;
    }
    for (const row of await db.all(`
        SELECT cardDark
        FROM profiles
        WHERE cardDark is not null AND cardDark != ''
    `)) {
        const m = row.cardDark.match(
            /https:\/\/pronouns-page.s3.eu-west-1.amazonaws.com\/card\/[^/]+\/.+-([^-]+)-dark\.png/,
        );
        if (!m) {
            console.error(row.cardDark);
            continue;
        }
        cards[m[1]] = true;
    }

    validateIds('Avatars', avatars, 500_000);
    validateIds('Flags', flags, 500_000);
    validateIds('Sources', sources, 200);
    validateIds('Terms', terms, 4_000);
    validateIds('Cards', cards, 50_000);

    console.log('--- Cleaning up S3: Images ---');
    const s3 = new S3(awsConfig);
    let overall = 0;
    let fresh = 0;
    let removed = 0;
    let removedSize = 0;
    let awsRequests = 0;

    const chunkSize = 1000;
    let continuationToken: string | undefined = undefined;
    while (true) {
        console.log('Making a request');
        awsRequests++;
        const objects: ListObjectsV2Output = await s3.listObjectsV2({
            Prefix: 'images/',
            MaxKeys: chunkSize,
            ContinuationToken: continuationToken,
            ...awsParams,
        });
        if (!objects.Contents) {
            break;
        }
        continuationToken = objects.NextContinuationToken;

        const toRemove: ObjectIdentifier[] = [];
        const remove = async (object: _Object, reason: string): Promise<void> => {
            console.log(`REMOVING: ${object.Key} (${reason})`);
            toRemove.push({ Key: object.Key! });
            removed += 1;
            removedSize += object.Size!;
        };

        for (const object of objects.Contents) {
            overall++;

            if (object.LastModified!.getTime() > new Date().getTime() - FRESH_CUTOFF) {
                fresh++;
                continue;
            }

            const [, id, size] = object.Key!.match('images/(.*)-(.*).png')!;

            if (avatars[id]) {
                if (size !== 'thumb' && size !== 'avatar') {
                    await remove(object, 'avatar');
                }
            } else if (flags[id]) {
                if (size !== 'flag') {
                    await remove(object, 'flag');
                }
            } else if (sources[id]) {
                if (size !== 'big' && size !== 'thumb') {
                    await remove(object, 'source');
                }
            } else if (terms[id]) {
                if (size !== 'flag') {
                    await remove(object, 'term');
                }
            } else {
                await remove(object, 'not used');
            }
        }

        if (execute && toRemove.length) {
            console.log('--- Removal request ---');
            awsRequests++;
            await s3.deleteObjects({
                Delete: {
                    Objects: toRemove,
                },
                ...awsParams,
            });
        }

        if (objects.Contents.length < chunkSize) {
            break;
        }
    }

    console.log('--- Cleaning up S3: Cards ---');
    continuationToken = undefined;
    while (true) {
        console.log('Making a request');
        awsRequests++;
        const objects: ListObjectsV2Output = await s3.listObjectsV2({
            Prefix: 'card/',
            MaxKeys: chunkSize,
            ContinuationToken: continuationToken,
            ...awsParams,
        });
        if (!objects.Contents) {
            break;
        }
        continuationToken = objects.NextContinuationToken;

        const toRemove = [];

        for (const object of objects.Contents) {
            overall++;

            if (object.LastModified!.getTime() > new Date().getTime() - FRESH_CUTOFF) {
                fresh++;
                continue;
            }

            const id = object.Key!.endsWith('-dark.png')
                ? object.Key!.match(/card\/[^/]+\/.+-([^-]+)-dark\.png/)![1]
                : object.Key!.match(/card\/[^/]+\/.+-([^-]+)\.png/)![1];

            if (!cards[id]) {
                console.log(`REMOVING: ${object.Key}`);
                toRemove.push({ Key: object.Key! });
                removed += 1;
                removedSize += object.Size!;
            }
        }

        if (execute && toRemove.length) {
            console.log('--- Removal request ---');
            awsRequests++;
            await s3.deleteObjects({
                Delete: {
                    Objects: toRemove,
                },
                ...awsParams,
            });
        }

        if (objects.Contents.length < chunkSize) {
            break;
        }
    }

    console.log('--- Summary ---');
    console.log(`Overall: ${overall}`);
    console.log(`Fresh: ${fresh}`);
    console.log(`Removed: ${removed}`);
    console.log(`Removed size: ${Math.round(removedSize / 1024 / 1024)} MB`);
    console.log(`AWS requests: ${awsRequests}`);
}

cleanup();