diff --git a/arkindex/dataimport/api.py b/arkindex/dataimport/api.py
index 48abf86eac582ca5045861d65aad31a67e7076d2..4349bab8fb070ac9742097e38f28b18e83fc1414 100644
--- a/arkindex/dataimport/api.py
+++ b/arkindex/dataimport/api.py
@@ -17,12 +17,12 @@ from arkindex.project.permissions import IsVerified, IsAuthenticated, IsAdminUse
from arkindex.documents.models import Corpus, Right, Element, ElementType
from arkindex.dataimport.models import \
DataImport, DataFile, DataImportFailure, Repository, Event, EventType
-from arkindex.dataimport.serializers import (
- DataImportLightSerializer, DataImportSerializer, DataImportFromFilesSerializer,
- DataImportFailureSerializer, DataFileSerializer, DataFileCreateSerializer,
- RepositorySerializer, RepositoryStartImportSerializer,
- ExternalRepositorySerializer, EventSerializer, MLToolSerializer,
-)
+from arkindex.dataimport.serializers.ml_tool import MLToolSerializer
+from arkindex.dataimport.serializers.files import DataFileSerializer, DataFileCreateSerializer
+from arkindex.dataimport.serializers.git import \
+ RepositorySerializer, RepositoryStartImportSerializer, ExternalRepositorySerializer, EventSerializer
+from arkindex.dataimport.serializers.imports import \
+ DataImportLightSerializer, DataImportSerializer, DataImportFromFilesSerializer, DataImportFailureSerializer
from arkindex.users.models import OAuthCredentials
from arkindex_common.ml_tool import MLTool
from arkindex_common.enums import DataImportMode
diff --git a/arkindex/dataimport/serializers/__init__.py b/arkindex/dataimport/serializers/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/arkindex/dataimport/serializers/files.py b/arkindex/dataimport/serializers/files.py
new file mode 100644
index 0000000000000000000000000000000000000000..056784035a135b82d6b1ab08e964a0ed45759c72
--- /dev/null
+++ b/arkindex/dataimport/serializers/files.py
@@ -0,0 +1,92 @@
+from rest_framework import serializers
+from arkindex.project.serializer_fields import EnumField
+from arkindex.project.aws import S3FileStatus
+from arkindex.dataimport.models import DataFile
+from arkindex.images.serializers import ImageSerializer
+import re
+
+
+class DataFileSerializer(serializers.ModelSerializer):
+ """
+ Serialize a single uploaded file
+ """
+
+ images = ImageSerializer(many=True, read_only=True)
+ status = EnumField(S3FileStatus)
+ s3_url = serializers.SerializerMethodField()
+
+ class Meta:
+ model = DataFile
+ fields = (
+ 'id',
+ 'name',
+ 'hash',
+ 'content_type',
+ 'size',
+ 'images',
+ 'status',
+ 's3_url',
+ )
+ read_only_fields = ('id', 'name', 'hash', 'size', 'content_type', 'images', 's3_url', )
+
+ def validate_status(self, value):
+ if value == S3FileStatus.Checked:
+ # Status has been resquested to be checked, perform validation
+ try:
+ self.instance.perform_check(raise_exc=True)
+ except (AssertionError, ValueError) as e:
+ raise serializers.ValidationError(str(e))
+ return value
+
+ def get_s3_url(self, obj):
+ if 'request' not in self.context:
+ return
+ # Only allow the S3 URL for internal users or admins
+ user = self.context['request'].user
+ if user.is_authenticated and (user.is_admin or user.is_internal):
+ return obj.s3_url
+
+
+class DataFileCreateSerializer(serializers.ModelSerializer):
+ """
+ Serialize a Datafile creation with Amazon S3 PUT uri
+ """
+
+ status = EnumField(S3FileStatus, read_only=True)
+ hash = serializers.RegexField(re.compile(r'[0-9A-Fa-f]{32}'), min_length=32, max_length=32)
+ s3_put_url = serializers.SerializerMethodField()
+
+ class Meta:
+ model = DataFile
+ fields = (
+ 'id',
+ 'name',
+ 'hash',
+ 'size',
+ 'corpus',
+ 'status',
+ 's3_url',
+ 's3_put_url',
+ )
+ read_only_fields = ('id', 'status', 's3_url', 's3_put_url')
+
+ def get_s3_put_url(self, obj):
+ if obj.status == S3FileStatus.Checked:
+ return None
+ return obj.s3_put_url
+
+ def run_validation(self, data):
+ existing_datafile = DataFile.objects.filter(hash=data['hash']).first()
+ if existing_datafile:
+ message = {
+ 'hash': ['DataFile with this hash already exists'],
+ 'id': str(existing_datafile.id),
+ 'status': existing_datafile.status.value,
+ }
+ if existing_datafile.status != S3FileStatus.Checked:
+ message['s3_put_url'] = existing_datafile.s3_put_url
+ else:
+ message['s3_url'] = existing_datafile.s3_url
+ self._errors = message
+ raise serializers.ValidationError(message)
+ return super().run_validation(data)
diff --git a/arkindex/dataimport/serializers/git.py b/arkindex/dataimport/serializers/git.py
new file mode 100644
index 0000000000000000000000000000000000000000..ad1e379bdea0e52fff6596b5930a620fab24eed2
--- /dev/null
+++ b/arkindex/dataimport/serializers/git.py
@@ -0,0 +1,119 @@
+from rest_framework import serializers
+from arkindex.project.serializer_fields import EnumField
+from arkindex.documents.models import Corpus
+from arkindex.dataimport.models import DataImport, Repository, Revision, Event, EventType
+import gitlab.v4.objects
+
+
+class RevisionSerializer(serializers.ModelSerializer):
+ """
+ Serialize a repository revision
+ """
+
+ date = serializers.DateTimeField(source='created')
+
+ class Meta:
+ model = Revision
+ fields = (
+ 'id',
+ 'date',
+ 'hash',
+ 'ref',
+ 'message',
+ 'author',
+ 'commit_url',
+ 'repo_id',
+ )
+
+
+class RepositorySerializer(serializers.ModelSerializer):
+ """
+ Serialize a repository
+ """
+
+ class Meta:
+ model = Repository
+ fields = (
+ 'id',
+ 'url',
+ 'enabled',
+ 'corpus',
+ )
+ extra_kwargs = {
+ 'id': {'read_only': True},
+ 'url': {'read_only': True},
+ 'enabled': {'read_only': True},
+ }
+
+
+class RepositoryStartImportSerializer(serializers.ModelSerializer):
+ """
+ A serializer used by the RepositoryStartImport endpoint to return a DataImport ID.
+ This serializer is required to get the OpenAPI schema generation to work.
+ """
+
+ import_id = serializers.UUIDField(source='id')
+
+ class Meta:
+ model = DataImport
+ fields = ('import_id',)
+ read_only_fields = ('import_id',)
+
+
+class ExternalRepositorySerializer(serializers.Serializer):
+ """
+ Serialize a Git repository from an external API
+ """
+
+ # Useless fields, added to prevent 500 errors when opening
+ # an API endpoint in the browser
+ id = serializers.IntegerField(min_value=0)
+ corpus = serializers.UUIDField()
+
+ def to_representation(self, obj):
+ if isinstance(obj, gitlab.v4.objects.Project):
+ return {
+ "id": obj.id,
+ "name": obj.name_with_namespace,
+ "url": obj.web_url,
+ }
+ else:
+ raise NotImplementedError
+
+ def to_internal_value(self, data):
+ """
+ Deserializing only requires a 'id' attribute
+ """
+ if not data.get('id'):
+ raise serializers.ValidationError({
+ 'id': 'This field is required.'
+ })
+ if not data.get('corpus'):
+ raise serializers.ValidationError({
+ 'corpus': 'This field is required.'
+ })
+
+ return {
+ 'id': data['id'],
+ 'corpus': Corpus.objects.writable(self.context['request'].user)
+ .get(id=data['corpus'])
+ }
+
+
+class EventSerializer(serializers.ModelSerializer):
+ """
+ Serialize a diff event for an element on a revision
+ """
+
+ type = EnumField(EventType)
+ revision = RevisionSerializer()
+ date = serializers.DateTimeField(source='created')
+
+ class Meta:
+ model = Event
+ fields = (
+ 'id',
+ 'type',
+ 'date',
+ 'revision',
+ )
diff --git a/arkindex/dataimport/serializers.py b/arkindex/dataimport/serializers/imports.py
similarity index 50%
rename from arkindex/dataimport/serializers.py
rename to arkindex/dataimport/serializers/imports.py
index fc1afdfc5962e9d5e61600f22d8af21af9459e60..20580598d5a381d2a3ddb8a150c854876ddfc8ff 100644
--- a/arkindex/dataimport/serializers.py
+++ b/arkindex/dataimport/serializers/imports.py
@@ -1,17 +1,13 @@
from rest_framework import serializers
from rest_framework.utils import model_meta
+from ponos.models import State
+from arkindex_common.enums import DataImportMode, DataImportPDFEngine
+from arkindex_common.ml_tool import MLToolType
from arkindex.project.serializer_fields import EnumField, MLToolField
-from arkindex.project.aws import S3FileStatus
-from arkindex.dataimport.models import \
- DataImport, DataImportFailure, DataFile, Repository, Revision, Event, EventType
from arkindex.documents.models import Corpus, Element, ElementType
from arkindex.documents.serializers.light import ElementLightSerializer
-from arkindex.images.serializers import ImageSerializer
-from arkindex_common.ml_tool import MLToolType
-from arkindex_common.enums import DataImportMode, DataImportPDFEngine
-from ponos.models import State
-import gitlab.v4.objects
-import re
+from arkindex.dataimport.models import DataImport, DataImportFailure, DataFile
+from arkindex.dataimport.serializers.git import RevisionSerializer
class DataImportLightSerializer(serializers.ModelSerializer):
@@ -59,27 +55,6 @@ class PDFPayloadSerializer(ImagesPayloadSerializer):
pdf_engine = EnumField(DataImportPDFEngine, default=DataImportPDFEngine.Convert)
-class RevisionSerializer(serializers.ModelSerializer):
- """
- Serialize a repository revision
- """
-
- date = serializers.DateTimeField(source='created')
-
- class Meta:
- model = Revision
- fields = (
- 'id',
- 'date',
- 'hash',
- 'ref',
- 'message',
- 'author',
- 'commit_url',
- 'repo_id',
- )
-
-
class DataImportSerializer(DataImportLightSerializer):
"""
Serialize a data importing workflow with its payload
@@ -181,92 +156,6 @@ class DataImportFromFilesSerializer(serializers.Serializer):
return data
-class DataFileSerializer(serializers.ModelSerializer):
- """
- Serialize a single uploaded file
- """
-
- images = ImageSerializer(many=True, read_only=True)
- status = EnumField(S3FileStatus)
- s3_url = serializers.SerializerMethodField()
-
- class Meta:
- model = DataFile
- fields = (
- 'id',
- 'name',
- 'hash',
- 'content_type',
- 'size',
- 'images',
- 'status',
- 's3_url',
- )
- read_only_fields = ('id', 'name', 'hash', 'size', 'content_type', 'images', 's3_url', )
-
- def validate_status(self, value):
- if value == S3FileStatus.Checked:
- # Status has been resquested to be checked, perform validation
- try:
- self.instance.perform_check(raise_exc=True)
- except (AssertionError, ValueError) as e:
- raise serializers.ValidationError(str(e))
- return value
-
- def get_s3_url(self, obj):
- if 'request' not in self.context:
- return
- # Only allow the S3 URL for internal users or admins
- user = self.context['request'].user
- if user.is_authenticated and (user.is_admin or user.is_internal):
- return obj.s3_url
-
-
-class DataFileCreateSerializer(serializers.ModelSerializer):
- """
- Serialize a Datafile creation with Amazon S3 PUT uri
- """
-
- status = EnumField(S3FileStatus, read_only=True)
- hash = serializers.RegexField(re.compile(r'[0-9A-Fa-f]{32}'), min_length=32, max_length=32)
- s3_put_url = serializers.SerializerMethodField()
-
- class Meta:
- model = DataFile
- fields = (
- 'id',
- 'name',
- 'hash',
- 'size',
- 'corpus',
- 'status',
- 's3_url',
- 's3_put_url',
- )
- read_only_fields = ('id', 'status', 's3_url', 's3_put_url')
-
- def get_s3_put_url(self, obj):
- if obj.status == S3FileStatus.Checked:
- return None
- return obj.s3_put_url
-
- def run_validation(self, data):
- existing_datafile = DataFile.objects.filter(hash=data['hash']).first()
- if existing_datafile:
- message = {
- 'hash': ['DataFile with this hash already exists'],
- 'id': str(existing_datafile.id),
- 'status': existing_datafile.status.value,
- }
- if existing_datafile.status != S3FileStatus.Checked:
- message['s3_put_url'] = existing_datafile.s3_put_url
- else:
- message['s3_url'] = existing_datafile.s3_url
- self._errors = message
- raise serializers.ValidationError(message)
- return super().run_validation(data)
-
-
class DataImportFailureSerializer(serializers.ModelSerializer):
"""
Serialize a data import error log
@@ -285,106 +174,3 @@ class DataImportFailureSerializer(serializers.ModelSerializer):
'context',
'view_url',
)
-
-
-class RepositorySerializer(serializers.ModelSerializer):
- """
- Serialize a repository
- """
-
- class Meta:
- model = Repository
- fields = (
- 'id',
- 'url',
- 'enabled',
- 'corpus',
- )
- extra_kwargs = {
- 'id': {'read_only': True},
- 'url': {'read_only': True},
- 'enabled': {'read_only': True},
- }
-
-
-class RepositoryStartImportSerializer(serializers.ModelSerializer):
- """
- A serializer used by the RepositoryStartImport endpoint to return a DataImport ID.
- This serializer is required to get the OpenAPI schema generation to work.
- """
-
- import_id = serializers.UUIDField(source='id')
-
- class Meta:
- model = DataImport
- fields = ('import_id',)
- read_only_fields = ('import_id',)
-
-
-class ExternalRepositorySerializer(serializers.Serializer):
- """
- Serialize a Git repository from an external API
- """
-
- # Useless fields, added to prevent 500 errors when opening
- # an API endpoint in the browser
- id = serializers.IntegerField(min_value=0)
- corpus = serializers.UUIDField()
-
- def to_representation(self, obj):
- if isinstance(obj, gitlab.v4.objects.Project):
- return {
- "id": obj.id,
- "name": obj.name_with_namespace,
- "url": obj.web_url,
- }
- else:
- raise NotImplementedError
-
- def to_internal_value(self, data):
- """
- Deserializing only requires a 'id' attribute
- """
- if not data.get('id'):
- raise serializers.ValidationError({
- 'id': 'This field is required.'
- })
- if not data.get('corpus'):
- raise serializers.ValidationError({
- 'corpus': 'This field is required.'
- })
-
- return {
- 'id': data['id'],
- 'corpus': Corpus.objects.writable(self.context['request'].user)
- .get(id=data['corpus'])
- }
-
-
-class EventSerializer(serializers.ModelSerializer):
- """
- Serialize a diff event for an element on a revision
- """
-
- type = EnumField(EventType)
- revision = RevisionSerializer()
- date = serializers.DateTimeField(source='created')
-
- class Meta:
- model = Event
- fields = (
- 'id',
- 'type',
- 'date',
- 'revision',
- )
-
-
-class MLToolSerializer(serializers.Serializer):
- """
- Serialize a machine learning tool for display
- """
- name = serializers.CharField()
- slug = serializers.SlugField()
- type = EnumField(MLToolType)
- version = serializers.CharField()
diff --git a/arkindex/dataimport/serializers/ml_tool.py b/arkindex/dataimport/serializers/ml_tool.py
new file mode 100644
index 0000000000000000000000000000000000000000..1f4fcb7b2ac4e029b97fd4d6224c258ebde0ece9
--- /dev/null
+++ b/arkindex/dataimport/serializers/ml_tool.py
@@ -0,0 +1,13 @@
+from rest_framework import serializers
+from arkindex_common.ml_tool import MLToolType
+from arkindex.project.serializer_fields import EnumField
+
+
+class MLToolSerializer(serializers.Serializer):
+ """
+ Serialize a machine learning tool for display
+ """
+ name = serializers.CharField()
+ slug = serializers.SlugField()
+ type = EnumField(MLToolType)
+ version = serializers.CharField()
diff --git a/arkindex/documents/serializers/elements.py b/arkindex/documents/serializers/elements.py
index f67b3a070fc419f2dd4d99935557e8a65fbbd5b6..2c3ef6ebd1d1473acb13a530d2dcfca0013566c9 100644
--- a/arkindex/documents/serializers/elements.py
+++ b/arkindex/documents/serializers/elements.py
@@ -9,7 +9,7 @@ from arkindex.images.serializers import ZoneSerializer, ImageSerializer
from arkindex.images.models import Image
from arkindex.documents.serializers.light import CorpusLightSerializer, ElementLightSerializer, EntityLightSerializer
from arkindex.documents.serializers.ml import ClassificationSerializer, TranscriptionSerializer
-from arkindex.dataimport.serializers import RevisionSerializer
+from arkindex.dataimport.serializers.git import RevisionSerializer
from arkindex.dataimport.models import EventType
from arkindex.project.serializer_fields import EnumField
from arkindex.project.polygon import Polygon