Migrate annotations (old)

ebl/app.py

@@ -23,6 +23,9 @@
 from ebl.ebl_ai_client import EblAiClient
 from ebl.files.infrastructure.grid_fs_file_repository import GridFsFileRepository
 from ebl.files.web.bootstrap import create_files_route
+from ebl.fragmentarium.infrastructure.cropped_sign_images_repository import (
+    MongoCroppedSignImagesRepository,
+)
 from ebl.fragmentarium.infrastructure.mongo_annotations_repository import (
     MongoAnnotationsRepository,
 )
@@ -67,6 +70,7 @@ def create_context():
     return Context(
         ebl_ai_client=ebl_ai_client,
         auth_backend=auth_backend,
+        cropped_sign_images_repository=MongoCroppedSignImagesRepository(database),
         word_repository=MongoWordRepository(database),
         sign_repository=MongoSignRepository(database),
         public_file_repository=GridFsFileRepository(database, "fs"),

ebl/context.py

@@ -15,6 +15,9 @@
 from ebl.fragmentarium.application.transliteration_update_factory import (
     TransliterationUpdateFactory,
 )
+from ebl.fragmentarium.infrastructure.cropped_sign_images_repository import (
+    MongoCroppedSignImagesRepository,
+)
 from ebl.lemmatization.application.suggestion_finder import LemmaRepository
 from ebl.transliteration.application.parallel_line_injector import ParallelLineInjector
 from ebl.transliteration.application.sign_repository import SignRepository
@@ -27,6 +30,7 @@
 class Context:
     ebl_ai_client: EblAiClient
     auth_backend: AuthBackend
+    cropped_sign_images_repository: MongoCroppedSignImagesRepository
     word_repository: WordRepository
     sign_repository: SignRepository
     public_file_repository: FileRepository

ebl/fragmentarium/application/annotations_image_extractor.py

@@ -1,137 +0,0 @@
-import base64
-from functools import singledispatchmethod
-import io
-from typing import Sequence, NewType
-
-import attr
-from PIL import Image
-
-from ebl.files.application.file_repository import FileRepository
-from ebl.fragmentarium.application.annotations_repository import AnnotationsRepository
-from ebl.fragmentarium.application.fragment_repository import FragmentRepository
-from ebl.fragmentarium.domain.annotation import BoundingBox, Annotation
-from ebl.transliteration.domain.museum_number import MuseumNumber
-from ebl.transliteration.domain.line_label import LineLabel
-from ebl.transliteration.domain.line_number import (
-    AbstractLineNumber,
-    LineNumber,
-    LineNumberRange,
-)
-
-Base64 = NewType("Base64", str)
-
-
-@attr.attrs(auto_attribs=True, frozen=True)
-class CroppedAnnotation:
-    image: Base64
-    fragment_number: MuseumNumber
-    script: str
-    label: str
-
-
-class AnnotationImageExtractor:
-    def __init__(
-        self,
-        fragment_repository: FragmentRepository,
-        annotations_repository: AnnotationsRepository,
-        photos_repository: FileRepository,
-    ):
-        self._fragments_repository = fragment_repository
-        self._annotations_repository = annotations_repository
-        self._photos_repository = photos_repository
-
-    def _format_label(self, label: LineLabel) -> str:
-        line_number = label.line_number
-        column = label.column
-        surface = label.surface
-        object = label.object
-        line_atf = line_number.atf if line_number else ""
-        column_abbr = column.abbreviation if column else ""
-        surface_abbr = surface.abbreviation if surface else ""
-        object_abbr = object.abbreviation if object else ""
-        return " ".join(
-            filter(
-                bool,
-                [column_abbr, surface_abbr, object_abbr, line_atf.replace(".", "")],
-            )
-        )
-
-    def _cropped_image_from_annotation(
-        self, annotation: Annotation, fragment_number: MuseumNumber
-    ) -> Base64:
-        fragment_image = self._photos_repository.query_by_file_name(
-            f"{fragment_number}.jpg"
-        )
-        image_bytes = fragment_image.read()
-        image = Image.open(io.BytesIO(image_bytes), mode="r")
-        bounding_box = BoundingBox.from_annotations(
-            image.size[0], image.size[1], [annotation]
-        )[0]
-        area = (
-            bounding_box.top_left_x,
-            bounding_box.top_left_y,
-            bounding_box.top_left_x + bounding_box.width,
-            bounding_box.top_left_y + bounding_box.height,
-        )
-        cropped_image = image.crop(area)
-        buf = io.BytesIO()
-        cropped_image.save(buf, format="PNG")
-        return Base64(base64.b64encode(buf.getvalue()).decode("utf-8"))
-
-    @singledispatchmethod
-    def _is_matching_number(self, line_number: AbstractLineNumber, number: int) -> bool:
-        raise ValueError("No default for overloading")
-
-    @_is_matching_number.register(LineNumber)
-    def _(self, line_number: LineNumber, number: int):
-        return number == line_number.number
-
-    @_is_matching_number.register(LineNumberRange)
-    def _(self, line_number: LineNumberRange, number: int):
-        return line_number.start.number <= number <= line_number.end.number
-
-    def _cropped_image_from_annotations(
-        self, fragment_number: MuseumNumber, annotations: Sequence[Annotation]
-    ) -> Sequence[CroppedAnnotation]:
-        cropped_annotations = []
-        for annotation in annotations:
-            fragment = self._fragments_repository.query_by_museum_number(
-                fragment_number
-            )
-            script = fragment.script
-            labels = fragment.text.labels
-            label = next(
-                (
-                    label
-                    for label in labels
-                    if self._is_matching_number(
-                        label.line_number, annotation.data.path[0]
-                    )
-                ),
-                None,
-            )
-
-            cropped_image = self._cropped_image_from_annotation(
-                annotation, fragment_number
-            )
-            cropped_annotations.append(
-                CroppedAnnotation(
-                    cropped_image,
-                    fragment_number,
-                    script,
-                    self._format_label(label) if label else "",
-                )
-            )
-        return cropped_annotations
-
-    def cropped_images_from_sign(self, sign: str) -> Sequence[CroppedAnnotation]:
-        annotations = self._annotations_repository.find_by_sign(sign)
-        cropped_annotations = []
-        for single_annotation in annotations:
-            fragment_number = single_annotation.fragment_number
-            cropped_annotations.extend(
-                self._cropped_image_from_annotations(
-                    fragment_number, single_annotation.annotations
-                )
-            )
-        return cropped_annotations

ebl/fragmentarium/application/annotations_schema.py

@@ -1,5 +1,6 @@
-from marshmallow import Schema, fields, post_load, EXCLUDE
-
+from marshmallow import Schema, fields, post_load, post_dump, EXCLUDE
+import pydash
+from ebl.fragmentarium.application.cropped_sign_image import CroppedSignSchema
 from ebl.fragmentarium.domain.annotation import (
     Geometry,
     AnnotationData,
@@ -40,18 +41,22 @@ class Meta:
 
     geometry = fields.Nested(GeometrySchema(), required=True)
     data = fields.Nested(AnnotationDataSchema(), required=True)
+    cropped_sign = fields.Nested(
+        CroppedSignSchema(), load_default=None, data_key="croppedSign"
+    )
 
     @post_load
     def make_annotation(self, data, **kwargs):
         return Annotation(**data)
 
+    @post_dump
+    def filter_none(self, data, **kwargs):
+        return pydash.omit_by(data, pydash.is_none)
 
-class AnnotationsSchema(Schema):
-    class Meta:
-        unknown = EXCLUDE
 
+class AnnotationsSchema(Schema):
     fragment_number = fields.String(required=True, data_key="fragmentNumber")
-    annotations = fields.Nested(AnnotationSchema, many=True, required=True)
+    annotations = fields.List(fields.Nested(AnnotationSchema(), required=True))
 
     @post_load
     def make_annotation(self, data, **kwargs):

ebl/fragmentarium/application/annotations_service.py

@@ -1,10 +1,29 @@
+from io import BytesIO
+from typing import Tuple, Sequence
+
+import attr
+from PIL import Image
+
 from ebl.changelog import Changelog
 from ebl.ebl_ai_client import EblAiClient
-import attr
 from ebl.files.application.file_repository import FileRepository
 from ebl.fragmentarium.application.annotations_repository import AnnotationsRepository
 from ebl.fragmentarium.application.annotations_schema import AnnotationsSchema
-from ebl.fragmentarium.domain.annotation import Annotations
+from ebl.fragmentarium.application.cropped_sign_image import CroppedSign
+from ebl.fragmentarium.application.cropped_sign_images_repository import (
+    CroppedSignImage,
+    CroppedSignImagesRepository,
+)
+from ebl.fragmentarium.application.fragment_repository import FragmentRepository
+from ebl.fragmentarium.domain.annotation import (
+    Annotations,
+    AnnotationValueType,
+    Annotation,
+    BoundingBox,
+    AnnotationValueType,
+)
+
+from ebl.transliteration.domain.line_label import LineLabel
 from ebl.transliteration.domain.museum_number import MuseumNumber
 from ebl.users.domain.user import User
 
@@ -15,6 +34,9 @@ class AnnotationsService:
     _annotations_repository: AnnotationsRepository
     _photo_repository: FileRepository
     _changelog: Changelog
+    _fragments_repository: FragmentRepository
+    _photos_repository: FileRepository
+    _cropped_sign_images_repository: CroppedSignImagesRepository
 
     def generate_annotations(
         self, number: MuseumNumber, threshold: float = 0.3
@@ -27,17 +49,94 @@ def generate_annotations(
     def find(self, number: MuseumNumber) -> Annotations:
         return self._annotations_repository.query_by_museum_number(number)
 
+    def _label_by_line_number(
+        self, line_number_to_match: int, labels: Sequence[LineLabel]
+    ) -> str:
+        matching_label = None
+        for label in labels:
+            label_line_number = label.line_number
+            if label_line_number and label_line_number.is_matching_number(
+                line_number_to_match
+            ):
+                matching_label = label
+        return matching_label.formatted_label if matching_label else ""
+
+    def _cropped_image_from_annotations_helper(
+        self,
+        annotations: Annotations,
+        image: Image.Image,
+        script: str,
+        labels: Sequence[LineLabel],
+    ) -> Tuple[Annotations, Sequence[CroppedSignImage]]:
+        cropped_sign_images = []
+        updated_cropped_annotations = []
+
+        for annotation in annotations.annotations:
+            label = (
+                self._label_by_line_number(annotation.data.path[0], labels)
+                if annotation.data.type != AnnotationValueType.BLANK
+                else ""
+            )
+            cropped_image = annotation.crop_image(image)
+            cropped_sign_image = CroppedSignImage.create(cropped_image)
+            cropped_sign_images.append(cropped_sign_image)
+
+            updated_cropped_annotation = attr.evolve(
+                annotation,
+                cropped_sign=CroppedSign(
+                    cropped_sign_image.image_id,
+                    script,
+                    label,
+                ),
+            )
+            updated_cropped_annotations.append(updated_cropped_annotation)
+        return (
+            attr.evolve(annotations, annotations=updated_cropped_annotations),
+            cropped_sign_images,
+        )
+
+    def _cropped_image_from_annotations(
+        self, annotations: Annotations
+    ) -> Tuple[Annotations, Sequence[CroppedSignImage]]:
+        fragment = self._fragments_repository.query_by_museum_number(
+            annotations.fragment_number
+        )
+        fragment_image = self._photos_repository.query_by_file_name(
+            f"{annotations.fragment_number}.jpg"
+        )
+        image_bytes = fragment_image.read()
+        image = Image.open(BytesIO(image_bytes), mode="r")
+        return self._cropped_image_from_annotations_helper(
+            annotations, image, fragment.script, fragment.text.labels
+        )
+
     def update(self, annotations: Annotations, user: User) -> Annotations:
         old_annotations = self._annotations_repository.query_by_museum_number(
             annotations.fragment_number
         )
         _id = str(annotations.fragment_number)
         schema = AnnotationsSchema()
+        (
+            annotations_with_image_ids,
+            cropped_sign_images,
+        ) = self._cropped_image_from_annotations(annotations)
+
+        self._annotations_repository.create_or_update(annotations_with_image_ids)
+        self._cropped_sign_images_repository.create_many(cropped_sign_images)
+
         self._changelog.create(
             "annotations",
             user.profile,
             {"_id": _id, **schema.dump(old_annotations)},
-            {"_id": _id, **schema.dump(annotations)},
+            {"_id": _id, **schema.dump(annotations_with_image_ids)},
         )
-        self._annotations_repository.create_or_update(annotations)
-        return annotations
+        return annotations_with_image_ids
+
+    def migrate(self, annotations: Annotations) -> Annotations:
+        (
+            annotations_with_image_ids,
+            cropped_sign_images,
+        ) = self._cropped_image_from_annotations(annotations)
+        self._annotations_repository.create_or_update(annotations_with_image_ids)
+        self._cropped_sign_images_repository.create_many(cropped_sign_images)
+        return annotations_with_image_ids

ebl/fragmentarium/application/cropped_annotations_service.py

@@ -0,0 +1,38 @@
+from typing import Sequence
+
+from ebl.fragmentarium.application.annotations_repository import AnnotationsRepository
+from ebl.fragmentarium.application.cropped_sign_image import CroppedAnnotation
+from ebl.fragmentarium.application.cropped_sign_images_repository import (
+    CroppedSignImagesRepository,
+)
+
+
+class CroppedAnnotationService:
+    def __init__(
+        self,
+        annotations_repository: AnnotationsRepository,
+        cropped_sign_images_repository: CroppedSignImagesRepository,
+    ):
+        self._annotations_repository = annotations_repository
+        self._cropped_sign_image_repository = cropped_sign_images_repository
+
+    def find_annotations_by_sign(self, sign: str) -> Sequence[CroppedAnnotation]:
+        annotations = self._annotations_repository.find_by_sign(sign)
+        cropped_image_annotations = []
+        for annotation in annotations:
+            for annotation_elem in annotation.annotations:
+                cropped_sign = annotation_elem.cropped_sign
+                if cropped_sign:
+                    cropped_sign_image = (
+                        self._cropped_sign_image_repository.query_by_id(
+                            cropped_sign.image_id
+                        )
+                    )
+                    cropped_image_annotations.append(
+                        CroppedAnnotation.from_cropped_sign(
+                            annotation.fragment_number,
+                            cropped_sign_image.image,
+                            cropped_sign,
+                        )
+                    )
+        return cropped_image_annotations