Parse device identifiers and create UserDevice entries in database

api/migrations/versions/f7b21148126a_add_user_device.py

@@ -0,0 +1,41 @@
+"""add user_device
+
+Revision ID: f7b21148126a
+Revises: a9627f63fbed
+Create Date: 2022-09-15 17:48:06.764342
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = "f7b21148126a"
+down_revision = "a049e5eb24dd"
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    op.create_table(
+        "user_device",
+        sa.Column("id", sa.Integer, autoincrement=True, primary_key=True),
+        sa.Column("user_id", sa.Integer, sa.ForeignKey("user.id", ondelete="CASCADE")),
+        sa.Column("identifier", sa.String, nullable=False),
+        sa.Column("display_name", sa.String, nullable=True),
+        sa.Index("user_id_identifier", "user_id", "identifier", unique=True),
+    )
+    op.add_column(
+        "track",
+        sa.Column(
+            "user_device_id",
+            sa.Integer,
+            sa.ForeignKey("user_device.id", ondelete="RESTRICT"),
+            nullable=True,
+        ),
+    )
+
+
+def downgrade():
+    op.drop_column("track", "user_device_id")
+    op.drop_table("user_device")

api/obs/api/db.py

@@ -221,6 +221,12 @@ class Track(Base):
         Integer, ForeignKey("user.id", ondelete="CASCADE"), nullable=False
     )
 
+    user_device_id = Column(
+        Integer,
+        ForeignKey("user_device.id", ondelete="RESTRICT"),
+        nullable=True,
+    )
+
     # Statistics... maybe we'll drop some of this if we can easily compute them from SQL
     recorded_at = Column(DateTime)
     recorded_until = Column(DateTime)
@@ -409,6 +415,28 @@ class User(Base):
         self.username = new_name
 
 
+class UserDevice(Base):
+    __tablename__ = "user_device"
+    id = Column(Integer, autoincrement=True, primary_key=True)
+    user_id = Column(Integer, ForeignKey("user.id", ondelete="CASCADE"))
+    identifier = Column(String, nullable=False)
+    display_name = Column(String, nullable=True)
+
+    __table_args__ = (
+        Index("user_id_identifier", "user_id", "identifier", unique=True),
+    )
+
+    def to_dict(self, for_user_id=None):
+        if for_user_id != self.user_id:
+            return {}
+
+        return {
+            "id": self.id,
+            "identifier": self.identifier,
+            "displayName": self.display_name,
+        }
+
+
 class Comment(Base):
     __tablename__ = "comment"
     id = Column(Integer, autoincrement=True, primary_key=True)
@@ -468,6 +496,14 @@ Track.overtaking_events = relationship(
     passive_deletes=True,
 )
 
+Track.user_device = relationship("UserDevice", back_populates="tracks")
+UserDevice.tracks = relationship(
+    "Track",
+    order_by=Track.created_at,
+    back_populates="user_device",
+    passive_deletes=False,
+)
+
 
 # 0..4 Night, 4..10 Morning, 10..14 Noon, 14..18 Afternoon, 18..22 Evening, 22..00 Night
 # Two hour intervals

api/obs/api/process.py

@@ -8,7 +8,7 @@ import pytz
 from os.path import join
 from datetime import datetime
 
-from sqlalchemy import delete, select
+from sqlalchemy import delete, select, and_
 from sqlalchemy.orm import joinedload
 
 from obs.face.importer import ImportMeasurementsCsv
@@ -27,7 +27,7 @@ from obs.face.filter import (
 
 from obs.face.osm import DataSource, DatabaseTileSource, OverpassTileSource
 
-from obs.api.db import OvertakingEvent, RoadUsage, Track, make_session
+from obs.api.db import OvertakingEvent, RoadUsage, Track, UserDevice, make_session
 from obs.api.app import app
 
 log = logging.getLogger(__name__)
@@ -144,10 +144,11 @@ async def process_track(session, track, data_source):
         os.makedirs(output_dir, exist_ok=True)
 
         log.info("Annotating and filtering CSV file")
-        imported_data, statistics = ImportMeasurementsCsv().read(
+        imported_data, statistics, track_metadata = ImportMeasurementsCsv().read(
             original_file_path,
             user_id="dummy",  # TODO: user username or id or nothing?
             dataset_id=Track.slug,  # TODO: use track id or slug or nothing?
+            return_metadata=True,
         )
 
         annotator = AnnotateMeasurements(
@@ -217,6 +218,36 @@ async def process_track(session, track, data_source):
         await clear_track_data(session, track)
         await session.commit()
 
+        device_identifier = track_metadata.get("DeviceId")
+        if device_identifier:
+            if isinstance(device_identifier, list):
+                device_identifier = device_identifier[0]
+
+            log.info("Finding or creating device %s", device_identifier)
+            user_device = (
+                await session.execute(
+                    select(UserDevice).where(
+                        and_(
+                            UserDevice.user_id == track.author_id,
+                            UserDevice.identifier == device_identifier,
+                        )
+                    )
+                )
+            ).scalar()
+
+            log.debug("user_device is %s", user_device)
+
+            if not user_device:
+                user_device = UserDevice(
+                    user_id=track.author_id, identifier=device_identifier
+                )
+                log.debug("Create new device for this user")
+                session.add(user_device)
+
+            track.user_device = user_device
+        else:
+            log.info("No DeviceId in track metadata.")
+
         log.info("Import events into database...")
         await import_overtaking_events(session, track, overtaking_events)
 

api/scripts

@@ -1 +1 @@
-Subproject commit 8e9395fd3cd0f1e83b4413546bc2d3cb0c726738
+Subproject commit bbc6feca08aee9ea4f4263bb7c07e199d9c989ee