Metadata extraction from image files (#22)

* Tested locally metadata extraction with diff types of files

* Added Dockerfile for frontend

* Handled metadata nulls and checked for proper metadata updation

* Fixed linter issues

api/internal/models/mediaitem.go

@@ -25,11 +25,11 @@ type (
 	}
 
 	Photo struct {
-		CameraMake      *string  `json:"cameraMake"`
-		CameraModel     *string  `json:"cameraModel"`
-		FocalLength     *int     `json:"focalLength"`
-		ApertureFNumber *float64 `json:"apertureFNumber"`
-		IsoEquivalent   *int     `json:"isoEquivalent"`
-		ExposureTime    *string  `json:"exposureTime"`
+		CameraMake      *string `json:"cameraMake"`
+		CameraModel     *string `json:"cameraModel"`
+		FocalLength     *string `json:"focalLength"`
+		ApertureFNumber *string `json:"apertureFNumber"`
+		IsoEquivalent   *int    `json:"isoEquivalent"`
+		ExposureTime    *string `json:"exposureTime"`
 	}
 )

api/schema.graphql

@@ -8,7 +8,7 @@ type MediaItem {
   mimeType: String!
   fileName: String!
   fileSize: Int!
-  mediaMetadata: MediaMetaData!
+  mediaMetadata: MediaMetaData
   createdAt: Time!
   updatedAt: Time!
 }
@@ -23,8 +23,8 @@ type MediaMetaData {
 type Photo  {
   cameraMake: String
   cameraModel: String
-  focalLength: Int
-  apertureFNumber: Float
+  focalLength: String
+  apertureFNumber: String
   isoEquivalent: Int
   exposureTime: String
 }

frontend/Dockerfile

@@ -0,0 +1,10 @@
+FROM node:14-alpine as build-step
+RUN mkdir /app
+WORKDIR /app
+COPY package.json /app
+RUN npm install
+COPY . /app
+RUN npm run build
+
+FROM nginx:1.21.3-alpine
+COPY --from=build-step /app/build /usr/share/nginx/html

ml/.pylintrc

@@ -4,4 +4,5 @@ ignore=CVS, .svn, .git
 [MESSAGES CONTROL]
 disable=missing-module-docstring,invalid-name
 
-indent-string='  '
+indent-string='  '
+max-line-length=160

ml/pipeline/component.py

@@ -1,19 +1,31 @@
 """Component"""
 import json
+import urllib.request
+from bson.objectid import ObjectId
 
 
 class Component:
   """Component"""
-  def __init__(self, name, queue):
+  def __init__(self, name, queue, db):
     self.name = name
     self.queue = queue
+    self.db = db
 
   def callback(self, _, __, ___, body):
     """RabbitMQ Callback"""
     data = json.loads(body)
     print(f'[{self.name}]: {data}')
     self.process(data['id'], data['imageUrl'])
 
+  def download(self, oid, image_url): # pylint: disable=no-self-use
+    """Downloads the file for processing"""
+    urllib.request.urlretrieve(image_url, f'image-{oid}')
+    print(f'downloaded file image-{oid}')
+
+  def update(self, oid, data):
+    """Updates database with the pipeline result"""
+    self.db['mediaitems'].update_one({'_id': ObjectId(oid)}, {'$set': data})
+
   def process(self, oid, image_url):
     """Component Process"""
     raise NotImplementedError

ml/pipeline/metadata.py

@@ -1,12 +1,30 @@
 """Metadata"""
+import exifread
 from .component import Component
 
 
 class Metadata(Component):
   """Metadata Component"""
-  def __init__(self):
-    super().__init__('metadata', 'pipeline.metadata')
+  def __init__(self, db):
+    super().__init__('metadata', 'pipeline.metadata', db)
 
   def process(self, oid, image_url):
-    print(oid)
-    print(image_url)
+    self.download(oid, image_url)
+    with open(f'image-{oid}', 'rb') as f:
+      tags = exifread.process_file(f)
+      if len(tags.keys()) > 0:
+        self.update(oid, {
+          'mediaMetadata': {
+            'creationTime': str(tags['EXIF DateTimeOriginal']) if 'EXIF DateTimeOriginal' in tags else None,
+            'width': tags['EXIF ExifImageLength'].values[0] if 'EXIF ExifImageLength' in tags else None,
+            'height': tags['EXIF ExifImageWidth'].values[0] if 'EXIF ExifImageWidth' in tags else None,
+            'photo': {
+              'cameraMake': str(tags['Image Make']) if 'Image Make' in tags else None,
+              'cameraModel': str(tags['Image Model']) if 'Image Model' in tags else None,
+              'focalLength': str(tags['EXIF FocalLength'].values[0].decimal()) if 'EXIF FocalLength' in tags else None,
+              'apertureFNumber': str(tags['EXIF FNumber'].values[0].decimal()) if 'EXIF FNumber' in tags else None,
+              'isoEquivalent': tags['EXIF ISOSpeedRatings'].values[0] if 'EXIF ISOSpeedRatings' in tags else None,
+              'exposureTime': str(tags['EXIF ExposureTime']) if 'EXIF ExposureTime' in tags else None,
+            },
+          }
+        })

ml/pipeline/people.py

@@ -4,8 +4,8 @@
 
 class People(Component):
   """People Component"""
-  def __init__(self):
-    super().__init__('people', 'pipeline.people')
+  def __init__(self, db):
+    super().__init__('people', 'pipeline.people', db)
 
   def process(self, oid, image_url):
     print(oid)

ml/pipeline/places.py

@@ -4,8 +4,8 @@
 
 class Places(Component):
   """Places Component"""
-  def __init__(self):
-    super().__init__('places', 'pipeline.places')
+  def __init__(self, db):
+    super().__init__('places', 'pipeline.places', db)
 
   def process(self, oid, image_url):
     print(oid)

ml/pipeline/things.py

@@ -4,8 +4,8 @@
 
 class Things(Component):
   """Things Component"""
-  def __init__(self):
-    super().__init__('things', 'pipeline.things')
+  def __init__(self, db):
+    super().__init__('things', 'pipeline.things', db)
 
   def process(self, oid, image_url):
     print(oid)

ml/requirements.txt

@@ -1 +1,3 @@
-pika==1.2.0
+pika==1.2.0
+ExifRead==2.3.2
+pymongo==3.11.3

ml/worker.py

@@ -1,9 +1,12 @@
 """ML Worker"""
 import pika
+from pymongo import MongoClient
 from pipeline import Metadata, People, Places, Things
 
+client = MongoClient('mongodb://root:root@database:5010/iris?authSource=admin')
+db = client['iris']
 
-pipeline = [Metadata(), People(), Places(), Things()]
+pipeline = [Metadata(db), People(db), Places(db), Things(db)]
 
 def start_consumers():
   """Init rabbitmq connection and start consumers"""