Upload tracks to files

docker-compose.yaml

@@ -17,9 +17,11 @@ services:
       dockerfile: ./Dockerfile
     volumes:
       - ./src:/opt/obsAPI/src
+      - ./local/api-data:/data
     environment:
       - PORT=3000
       - MONGODB_URL=mongodb://mongo/obsTest
+      - UPLOADS_DIR=/data
     links:
       - mongo
     ports:

migrations/2020-12-12-1823-original-filename.js

@@ -0,0 +1,21 @@
+const Track = require('../src/models/Track');
+
+module.exports = {
+  async up(next) {
+    try {
+      for await (const track of Track.find()) {
+        track.originalFileName = track.slug + '.csv'
+        await track.generateOriginalFilePath();
+        await track.save()
+      }
+      next();
+    } catch(err) {
+      next(err)
+    }
+  },
+
+  async down(next) {
+    next();
+  },
+};
+

migrations/2020-12-13-2025-move-to-upload-file.js

@@ -0,0 +1,25 @@
+
+const Track = require('../src/models/Track');
+
+module.exports = {
+  async up(next) {
+    try {
+      for await (const track of Track.find()) {
+        if (!track.body) {
+          continue
+        }
+
+        await track.writeToOriginalFile(track.body)
+        delete track.body;
+        await track.save()
+      }
+      next();
+    } catch(err) {
+      next(err)
+    }
+  },
+
+  async down(next) {
+    next();
+  },
+};

package-lock.json

@@ -9308,6 +9308,14 @@
         }
       }
     },
+    "sanitize-filename": {
+      "version": "1.6.3",
+      "resolved": "https://registry.npmjs.org/sanitize-filename/-/sanitize-filename-1.6.3.tgz",
+      "integrity": "sha512-y/52Mcy7aw3gRm7IrcGDFx/bCk4AhRh2eI9luHOQM86nZsqwiRkkq2GekHXBBD+SmPidc8i2PqtYZl+pWJ8Oeg==",
+      "requires": {
+        "truncate-utf8-bytes": "^1.0.0"
+      }
+    },
     "sanitize-html": {
       "version": "1.20.1",
       "resolved": "https://registry.npmjs.org/sanitize-html/-/sanitize-html-1.20.1.tgz",
@@ -10167,6 +10175,14 @@
         "punycode": "^2.1.1"
       }
     },
+    "truncate-utf8-bytes": {
+      "version": "1.0.2",
+      "resolved": "https://registry.npmjs.org/truncate-utf8-bytes/-/truncate-utf8-bytes-1.0.2.tgz",
+      "integrity": "sha1-QFkjkJWS1W94pYGENLC3hInKXys=",
+      "requires": {
+        "utf8-byte-length": "^1.0.1"
+      }
+    },
     "tsconfig-paths": {
       "version": "3.9.0",
       "resolved": "https://registry.npmjs.org/tsconfig-paths/-/tsconfig-paths-3.9.0.tgz",
@@ -10957,6 +10973,11 @@
       "integrity": "sha1-Qw/VEKt/yVtdWRDJAteYgMIIQ2s=",
       "dev": true
     },
+    "utf8-byte-length": {
+      "version": "1.0.4",
+      "resolved": "https://registry.npmjs.org/utf8-byte-length/-/utf8-byte-length-1.0.4.tgz",
+      "integrity": "sha1-9F8VDExm7uloGGUFq5P8u4rWv2E="
+    },
     "util-deprecate": {
       "version": "1.0.2",
       "resolved": "https://registry.npmjs.org/util-deprecate/-/util-deprecate-1.0.2.tgz",

package.json

@@ -45,6 +45,7 @@
     "passport": "0.4.1",
     "passport-local": "1.0.0",
     "request": "2.88.2",
+    "sanitize-filename": "^1.6.3",
     "slug": "^3.3.5",
     "turf": "^3.0.14",
     "underscore": "^1.11.0"

src/logic/tracks.js

@@ -64,6 +64,9 @@ function replaceDollarNewlinesHack(body) {
 }
 
 function* parseTrackPoints(body, format = null) {
+  if (body instanceof Buffer) {
+    body = body.toString('utf-8')
+  }
   body = replaceDollarNewlinesHack(body);
 
   const detectedFormat = format != null ? format : detectFormat(body);

src/models/Track.js

@@ -1,35 +1,60 @@
 const mongoose = require('mongoose');
 const uniqueValidator = require('mongoose-unique-validator');
 const slug = require('slug');
+const path = require('path');
+const sanitize = require('sanitize-filename');
+const fs = require('fs')
 
 const { parseTrackPoints } = require('../logic/tracks');
 
 const TrackData = require('./TrackData');
 
+const DATA_DIR = process.env.DATA_DIR || path.resolve(__dirname, '../../data/')
+
 const schema = new mongoose.Schema(
   {
     slug: { type: String, lowercase: true, unique: true },
     title: String,
     description: String,
-    body: String,
     visible: Boolean,
     uploadedByUserAgent: String,
+    body: String, // deprecated, remove after migration has read it
     comments: [{ type: mongoose.Schema.Types.ObjectId, ref: 'Comment' }],
     author: { type: mongoose.Schema.Types.ObjectId, ref: 'User' },
     trackData: { type: mongoose.Schema.Types.ObjectId, ref: 'TrackData' },
     publicTrackData: { type: mongoose.Schema.Types.ObjectId, ref: 'TrackData' },
+    originalFileName: {
+      type: String,
+      required: true,
+      validate: {
+        validator: function (v) {
+          // Must be a sane filename, i.e. not change when being sanitized
+          return sanitize(v) === v && v.length > 0 && /.+\.csv$/i.test(v);
+        },
+        message: (props) => `${props.value} is not a valid filename`,
+      },
+    },
+    originalFilePath: String,
   },
   { timestamps: true },
 );
 
 schema.plugin(uniqueValidator, { message: 'is already taken' });
 
-schema.pre('validate', function (next) {
-  if (!this.slug) {
-    this.slugify();
-  }
+schema.pre('validate', async function (next) {
+  try {
+    if (!this.slug) {
+      this.slugify();
+    }
 
-  next();
+    if (!this.originalFilePath) {
+      await this.generateOriginalFilePath();
+    }
+
+    next();
+  } catch (err) {
+    next(err);
+  }
 });
 
 class Track extends mongoose.Model {
@@ -37,6 +62,11 @@ class Track extends mongoose.Model {
     this.slug = slug(this.title) + '-' + ((Math.random() * Math.pow(36, 6)) | 0).toString(36);
   }
 
+  async generateOriginalFilePath() {
+    await this.populate('author').execPopulate();
+    this.originalFilePath = path.join('uploads', 'originals', this.author.username, this.slug, this.originalFileName);
+  }
+
   isVisibleTo(user) {
     if (this.visible) {
       return true;
@@ -57,6 +87,24 @@ class Track extends mongoose.Model {
     return user && user._id.equals(this.author._id);
   }
 
+  async _ensureDirectoryExists() {
+    if (!this.originalFilePath) {
+      await this.generateOriginalFilePath()
+    }
+
+    const dir = path.join(DATA_DIR, path.dirname(this.originalFilePath))
+    await fs.promises.mkdir(dir, {recursive: true})
+  }
+
+  get fullOriginalFilePath() {
+    return path.join(DATA_DIR, this.originalFilePath)
+  }
+
+  async writeToOriginalFile(fileBody) {
+      await this._ensureDirectoryExists()
+      await fs.promises.writeFile(this.fullOriginalFilePath, fileBody)
+  }
+
   /**
    * Fills the trackData and publicTrackData with references to correct
    * TrackData objects.  For now, this is either the same, or publicTrackData
@@ -76,8 +124,11 @@ class Track extends mongoose.Model {
       await TrackData.findByIdAndDelete(this.publicTrackData);
     }
 
-    // parse the points from the body
-    const points = Array.from(parseTrackPoints(this.body));
+    // Parse the points from the body.
+    // TODO: Stream file contents, if possible
+    const body = await fs.promises.readFile(this.fullOriginalFilePath)
+    const points = Array.from(parseTrackPoints(body));
+
     const trackData = TrackData.createFromPoints(points);
     await trackData.save();
 
@@ -102,7 +153,12 @@ class Track extends mongoose.Model {
       updatedAt: this.updatedAt,
       visible: this.visible,
       author: this.author.toProfileJSONFor(user),
-      ...(includePrivateFields ? { uploadedByUserAgent: this.uploadedByUserAgent } : {}),
+      ...(includePrivateFields
+        ? {
+            uploadedByUserAgent: this.uploadedByUserAgent,
+            originalFileName: this.originalFileName,
+          }
+        : {}),
     };
   }
 }

src/routes/api/tracks.js

@@ -175,23 +175,28 @@ router.post(
   auth.required,
   busboy(), // parse multipart body
   wrapRoute(async (req, res) => {
-    const { body } = await getMultipartOrJsonBody(req, (body) => body.track);
+    // Read the whole file into memory. This is not optimal, instead, we should
+    // write the file data directly to the target file. However, we first have
+    // to parse the rest of the track data to know where to place the file.
+    // TODO: Stream into temporary file, then move it later.
+    const { body, fileInfo } = await getMultipartOrJsonBody(req, (body) => body.track);
 
-    const track = new Track(body);
-    track.author = req.user;
+    const {body: fileBody, visible, ...trackBody} = body
 
-    if (body.visible != null) {
-      track.visible = Boolean(body.visible);
-    } else {
-      track.visible = track.author.areTracksVisibleForAll;
-    }
+    const track = new Track({
+      ...trackBody,
+      author: req.user,
+      visible: visible == null ? req.user.areTracksVisibleForAll : Boolean(trackBody.visible)
+    })
+    track.slugify();
 
-    if (track.body) {
-      track.body = track.body.trim();
+    if (fileBody) {
       track.uploadedByUserAgent = normalizeUserAgent(req.headers['user-agent']);
+      track.originalFileName = fileInfo.body ? fileInfo.body.filename : track.slug + '.csv';
+      await track.writeToOriginalFile(fileBody)
       await track.rebuildTrackDataAndSave();
     } else {
-      await track.save();
+      await track.save()
     }
 
     // console.log(track.author);
@@ -224,32 +229,25 @@ router.put(
       return res.sendStatus(403);
     }
 
-    const { body } = await getMultipartOrJsonBody(req, (body) => body.track);
+    const { body: {body: fileBody, ...trackBody}, fileInfo } = await getMultipartOrJsonBody(req, (body) => body.track);
 
-    if (typeof body.title !== 'undefined') {
-      track.title = (body.title || '').trim() || null;
+    if (typeof trackBody.title !== 'undefined') {
+      track.title = (trackBody.title || '').trim() || null;
     }
 
-    if (typeof body.description !== 'undefined') {
-      track.description = (body.description || '').trim() || null;
+    if (typeof trackBody.description !== 'undefined') {
+      track.description = (trackBody.description || '').trim() || null;
     }
 
-    if (body.visible != null) {
-      track.visible = Boolean(body.visible);
+    if (trackBody.visible != null) {
+      track.visible = Boolean(trackBody.visible);
     }
 
-    if (typeof body.tagList !== 'undefined') {
-      track.tagList = body.tagList;
-    }
-
-    if (body.body && body.body.trim()) {
-      // delete existing
-      if (track.trackData) {
-        await TrackData.findByIdAndDelete(track.trackData);
-      }
-
-      track.body = body.body.trim();
+    if (fileBody) {
+      track.originalFileName = fileInfo.body ? fileInfo.body.filename : track.slug + '.csv';
       track.uploadedByUserAgent = normalizeUserAgent(req.headers['user-agent']);
+      await track.writeToOriginalFile(fileBody)
+
       await track.rebuildTrackDataAndSave();
     } else {
       await track.save();