feat(annotation): poc optimization of zoom level management with the new osekit [APLOSE-371]

.gitlab-ci.yml

@@ -8,7 +8,6 @@
 #         * OSMOSE_DB_USER : DB user (ex: osmose)
 #         * OSMOSE_DB_PWD : DB password
 #         * OSMOSE_DB_BASE : DB Aplose dedicated database name (ex: osmose)
-#         * OSMOSE_PROXY_URL : Proxy host:port if needed (ex: proxy-wiz.osmose.fr:3128)
 #     * Exposed port: 8000
 #     * Folder dedicated to "datawork" mount: /opt/datawork
 #     * Folder dedicated to server static files mount: /opt/staticfiles

backend/api/models/data/spectrogram.py

@@ -1,18 +1,23 @@
 """Spectrogram model"""
 import csv
 from datetime import datetime, timedelta
+from os import path
 from os.path import join
-from pathlib import Path
+from pathlib import Path, PureWindowsPath
 
 from django.conf import settings
 from django.db import models
 from django.db.models import Q, F, QuerySet
+from django.utils import timezone
 from metadatax.data.models import FileFormat
 from osekit.config import TIMESTAMP_FORMAT_EXPORTED_FILES_LOCALIZED
-
-# from osekit.core_api.spectro_data import SpectroData
-# from osekit.core_api.spectro_dataset import SpectroDataset
-from backend.utils.osekit_replace import SpectroDataset, SpectroData
+from osekit.core_api.audio_data import AudioData
+from osekit.core_api.audio_file import AudioFile
+from osekit.core_api.spectro_data import SpectroData
+from osekit.core_api.spectro_dataset import SpectroDataset
+from pandas import Timestamp
+from scipy.signal import ShortTimeFFT
+from scipy.signal.windows import hamming
 
 from .__abstract_file import AbstractFile
 from .__abstract_time_segment import TimeSegment
@@ -157,6 +162,110 @@ def __str__(self):
 
     analysis = models.ManyToManyField(SpectrogramAnalysis, related_name="spectrograms")
 
+    def get_audio_path(self, analysis: SpectrogramAnalysis) -> str:
+        if analysis.dataset.legacy:
+            folders = PureWindowsPath(analysis.path).as_posix().split("/")
+            folders.pop()
+            return path.join(
+                analysis.dataset.path.split(
+                    settings.DATASET_EXPORT_PATH.stem + "/"
+                ).pop(),
+                PureWindowsPath(settings.DATASET_FILES_FOLDER),
+                PureWindowsPath(folders.pop()),
+                PureWindowsPath(f"{self.filename}.wav"),
+            )
+        else:
+            spectro_data: SpectroData = self.get_spectro_data_for(analysis)
+            audio_files = list(spectro_data.audio_data.files)
+            if len(audio_files) != 1:
+                return None
+
+            audio_file = audio_files[0]
+            if audio_file.begin != (
+                self.start if audio_file.begin.tz else timezone.make_naive(self.start)
+            ):
+                return None
+            if audio_file.end < (
+                self.end if audio_file.end.tz else timezone.make_naive(self.end)
+            ):
+                return None
+
+            audio_path = str(audio_file.path)
+            return (
+                audio_path.split(str(settings.DATASET_EXPORT_PATH)).pop().lstrip("\\")
+            )
+
+    def get_base_spectro_path(self, analysis: SpectrogramAnalysis) -> str:
+        if analysis.dataset.legacy:
+            return path.join(
+                PureWindowsPath(
+                    analysis.dataset.path.split(
+                        settings.DATASET_EXPORT_PATH.stem + "/"
+                    ).pop()
+                ),
+                PureWindowsPath(analysis.path),
+                PureWindowsPath("image"),
+                PureWindowsPath(f"{self.filename}.{self.format.name}"),
+            )
+        else:
+            spectro_dataset: SpectroDataset = analysis.get_osekit_spectro_dataset()
+            spectro_dataset_path = str(spectro_dataset.folder).split(
+                str(settings.DATASET_EXPORT_PATH)
+            )[1]
+            return path.join(
+                PureWindowsPath(spectro_dataset_path),
+                PureWindowsPath("spectrogram"),  # TODO: avoid static path parts!!!
+                PureWindowsPath(f"{self.filename}.{self.format.name}"),
+            ).lstrip("\\")
+
     def get_spectro_data_for(self, analysis: SpectrogramAnalysis) -> SpectroData:
-        spectro_dataset: SpectroDataset = analysis.get_osekit_spectro_dataset()
-        return [d for d in spectro_dataset.data if d.name == self.filename].pop()
+        if analysis.legacy:
+            audio_path = Path(
+                join(
+                    settings.VOLUMES_ROOT,
+                    settings.DATASET_EXPORT_PATH,
+                    self.get_audio_path(analysis),
+                )
+            )
+            audio_file = AudioFile(
+                path=audio_path,
+                begin=Timestamp.fromisoformat(self.start.isoformat()),
+            )
+            audio_data = AudioData.from_files([audio_file])
+            overlap = analysis.fft.overlap or 0.95
+            hop = round(analysis.fft.window_size * (1 - overlap))
+            return SpectroData.from_audio_data(
+                data=audio_data,
+                fft=ShortTimeFFT(
+                    win=hamming(analysis.fft.window_size),
+                    hop=hop,
+                    fs=analysis.fft.sampling_frequency,
+                    scale_to="magnitude",
+                ),
+                colormap="viridis",  # This is the default value
+            )
+        else:
+            spectro_dataset: SpectroDataset = analysis.get_osekit_spectro_dataset()
+            return [d for d in spectro_dataset.data if d.name == self.filename].pop()
+
+    def get_audio_data_for(self, analysis: SpectrogramAnalysis) -> AudioData:
+        if analysis.legacy:
+            audio_path = Path(
+                join(
+                    settings.VOLUMES_ROOT,
+                    settings.DATASET_EXPORT_PATH,
+                    self.get_audio_path(analysis),
+                )
+            )
+            audio_file = AudioFile(
+                path=audio_path,
+                begin=Timestamp.fromisoformat(self.start.isoformat()),
+            )
+            return AudioData.from_files([audio_file])
+        else:
+            spectro_dataset: SpectroDataset = analysis.get_osekit_spectro_dataset()
+            return (
+                [d for d in spectro_dataset.data if d.name == self.filename]
+                .pop()
+                .audio_data
+            )

backend/api/models/data/spectrogram_analysis.py

@@ -14,8 +14,8 @@
 
 from backend.aplose.models import User
 
-# from osekit.core_api.spectro_dataset import SpectroDataset
-from backend.utils.osekit_replace import SpectroDataset
+from osekit.core_api.spectro_dataset import SpectroDataset
+
 from .__abstract_analysis import AbstractAnalysis
 from .colormap import Colormap
 from .dataset import Dataset

backend/api/schema/nodes/annotation_spectrogram.py

@@ -5,16 +5,11 @@
 import graphene
 import graphene_django_optimizer
 from django.conf import settings
-from django.utils import timezone
 from django_extension.schema.errors import NotFoundError
 from django_extension.schema.fields import AuthenticatedPaginationConnectionField
 from django_extension.schema.types import ExtendedNode
 from graphql import GraphQLResolveInfo
 
-# from osekit.core_api.spectro_data import SpectroData
-# from osekit.core_api.spectro_dataset import SpectroDataset
-from backend.utils.osekit_replace import SpectroDataset, SpectroData
-
 from backend.api.models import (
     Spectrogram,
     AnnotationCampaign,
@@ -102,77 +97,21 @@ def resolve_is_assigned(
     @graphene_django_optimizer.resolver_hints()
     def resolve_audio_path(self: Spectrogram, info, analysis_id: int):
         analysis: SpectrogramAnalysis = self.analysis.get(id=analysis_id)
-
-        audio_path: str
-        if analysis.dataset.legacy:
-            folders = PureWindowsPath(analysis.path).as_posix().split("/")
-            folders.pop()
-            audio_path = path.join(
-                analysis.dataset.path.split(
-                    settings.DATASET_EXPORT_PATH.stem + "/"
-                ).pop(),
-                PureWindowsPath(settings.DATASET_FILES_FOLDER),
-                PureWindowsPath(folders.pop()),
-                PureWindowsPath(f"{self.filename}.wav"),
-            )
-        else:
-            spectro_data: SpectroData = self.get_spectro_data_for(analysis)
-            audio_files = list(spectro_data.audio_data.files)
-            if len(audio_files) != 1:
-                return None
-
-            audio_file = audio_files[0]
-            if audio_file.begin != (
-                self.start if audio_file.begin.tz else timezone.make_naive(self.start)
-            ):
-                return None
-            if audio_file.end < (
-                self.end if audio_file.end.tz else timezone.make_naive(self.end)
-            ):
-                return None
-
-            audio_path = str(audio_file.path)
-            audio_path = (
-                audio_path.split(str(settings.DATASET_EXPORT_PATH)).pop().lstrip("\\")
-            )
         return path.join(
             PureWindowsPath(settings.STATIC_URL),
             PureWindowsPath(settings.DATASET_EXPORT_PATH),
-            PureWindowsPath(audio_path),
+            PureWindowsPath(self.get_audio_path(analysis)),
         )
 
     path = graphene.String(analysis_id=graphene.ID(required=True), required=True)
 
     @graphene_django_optimizer.resolver_hints()
     def resolve_path(self: Spectrogram, info, analysis_id: int):
         analysis: SpectrogramAnalysis = self.analysis.get(id=analysis_id)
-
-        spectrogram_path: str
-        if analysis.dataset.legacy:
-            spectrogram_path = path.join(
-                PureWindowsPath(
-                    analysis.dataset.path.split(
-                        settings.DATASET_EXPORT_PATH.stem + "/"
-                    ).pop()
-                ),
-                PureWindowsPath(analysis.path),
-                PureWindowsPath("image"),
-                PureWindowsPath(f"{self.filename}.{self.format.name}"),
-            )
-        else:
-            spectro_dataset: SpectroDataset = analysis.get_osekit_spectro_dataset()
-            spectro_dataset_path = str(spectro_dataset.folder).split(
-                str(settings.DATASET_EXPORT_PATH)
-            )[1]
-            spectrogram_path = path.join(
-                PureWindowsPath(spectro_dataset_path),
-                PureWindowsPath("spectrogram"),  # TODO: avoid static path parts!!!
-                PureWindowsPath(f"{self.filename}.{self.format.name}"),
-            ).lstrip("\\")
         return path.join(
             PureWindowsPath(settings.STATIC_URL),
             PureWindowsPath(settings.DATASET_EXPORT_PATH),
-            PureWindowsPath(spectrogram_path),
+            PureWindowsPath(self.get_base_spectro_path(analysis)),
         )
 
     task = graphene.Field(

backend/api/urls.py

@@ -5,8 +5,12 @@
     AnnotationViewSet,
     DownloadViewSet,
 )
+from backend.api.views.zoom import ZoomViewSet
 
 # API urls are meant to be used by our React frontend
 api_router = routers.DefaultRouter()
 api_router.register(r"annotation", AnnotationViewSet, basename="annotation")
 api_router.register("download", DownloadViewSet, basename="download")
+
+
+api_router.register("data", ZoomViewSet, basename="data-zoom")