img conversion

src/scraper.ts

@@ -1,6 +1,29 @@
-import axios from 'axios';
+import path from 'path'
 
-export default function scraper(discordAuth) {
+import axios from 'axios'
+import mime from 'mime-types'
+
+import type { PrismaClient, Attachment } from '@prisma/client';
+import type { ImageAnnotatorClient } from '@google-cloud/vision';
+import type Minio from 'minio'
+
+import sharp from 'sharp';
+
+const IMAGE_TYPES = [
+    'image/png',
+    'image/jpeg',
+    'image/webp',
+    'image/gif',
+]
+
+const VIDEO_TYPES = [
+    'video/mp4',
+    'video/quicktime'
+]
+
+const STORAGE_BUCKET = process.env.STORAGE_BUCKET
+
+export default function scraper(discordAuth, vision: ImageAnnotatorClient, minio: Minio.Client) {
     const headers = {
         'Authorization': discordAuth,
     }
@@ -25,12 +48,93 @@ export default function scraper(discordAuth) {
         return res.data
     }
 
+    async function detectText(attachment: Attachment) {
+        // bail out of anything not an image
+        if (!IMAGE_TYPES.includes(attachment.contentType)) return {}
+
+        const annotateRequest = {
+            image: {
+                source: {
+                    imageUri: attachment.url
+                }
+            },
+            features: [{ type: 'TEXT_DETECTION' }],
+        }
+        const res = await vision.annotateImage(annotateRequest)
+        // comes back as an array
+        return res[0] || {}
+    }
+
+    async function downloadAttachment(a: Attachment) {
+        const download = await axios.get(a.url, { responseType: 'stream' })
+        const isImage = IMAGE_TYPES.includes(a.contentType)
+
+        const objectName = () => {
+            if (isImage)
+                return `${a.memeId}.webp`
+
+            return `${a.memeId}.${mime.extension(a.contentType)}`
+        }
+
+        const stream = async () => {
+            if (isImage) {
+                const converter = sharp().webp()
+                download.data.pipe(converter)
+                return converter
+            }
+            return download.data
+        }
+
+        return minio.putObject(STORAGE_BUCKET, objectName(), await stream())
+    }
+
+    function filterAttachments(messages, guild, channel) {
+        const acceptedContentTypes = [
+            ...IMAGE_TYPES,
+            ...VIDEO_TYPES,
+        ]
+
+        return Promise.all(messages
+            .filter(m => m.attachments)
+            // bundle message and attachment together
+            .map(m => m.attachments.map(a => [a, m]))
+            .flat()
+            .filter(([a, m]) => acceptedContentTypes.includes(a.content_type))
+            .map(async ([a, m]) => {
+                const attachment: Attachment = {
+                    id: BigInt(a.id),
+                    msgId: BigInt(m.id),
+                    channelId: BigInt(m.channel_id),
+                    url: a.url,
+                    contentType: a.content_type,
+                    message: m,
+
+                    // create this ahead of time to associate with our copy of the media
+                    memeId: crypto.randomUUID(),
+
+                    // @ts-ignore it's fine
+                    vision: {},
+                    guild,
+                    channel,
+
+                    createdAt: undefined,
+                    updatedAt: undefined
+                }
+
+                const visionData = await detectText(attachment)
+                return { ...attachment, vision: visionData }
+            }))
+    }
+
     async function getChannelAttachments(channelId) {
         try {
             const channel = await fetchChannel(channelId)
             const guild = await fetchGuildDetails(channel.guild_id)
             const messages = await fetchChannelMessages(channelId)
-            return { messages, channel, guild }
+
+            return filterAttachments(messages, guild, channel);
+
+            // return { messages, channel, guild }
         } catch (e) {
             console.error('scrape error', {
                 err: e.toJSON(),
@@ -41,6 +145,7 @@ export default function scraper(discordAuth) {
 
     return {
         getChannelAttachments,
+        downloadAttachment,
     }
 }