Merge remote-tracking branch 'origin/master' into afo-register

ebl/common/application/schemas.py

@@ -0,0 +1,16 @@
+from marshmallow import Schema, fields, validate, post_load
+from ebl.common.domain.accession import Accession
+
+
+class AbstractMuseumNumberSchema(Schema):
+    prefix = fields.String(required=True, validate=validate.Length(min=1))
+    number = fields.String(
+        required=True, validate=(validate.Length(min=1), validate.ContainsNoneOf("."))
+    )
+    suffix = fields.String(required=True, validate=validate.ContainsNoneOf("."))
+
+
+class AccessionSchema(AbstractMuseumNumberSchema):
+    @post_load
+    def create_accession(self, data, **kwargs) -> Accession:
+        return Accession(**data)

ebl/common/domain/accession.py

@@ -0,0 +1,15 @@
+from ebl.transliteration.domain.museum_number import MuseumNumber
+import functools
+import attr
+import re
+
+
+@functools.total_ordering
+@attr.s(auto_attribs=True, frozen=True, order=False)
+class Accession(MuseumNumber):
+    @staticmethod
+    def of(source: str) -> "Accession":
+        if match := re.compile(r"(.+?)\.([^.]+)(?:\.([^.]+))?").fullmatch(source):
+            return Accession(match[1], match[2], match[3] or "")
+        else:
+            raise ValueError(f"'{source}' is not a valid accession number.")

ebl/fragmentarium/application/archaeology_schemas.py

@@ -1,25 +1,24 @@
+from ebl.common.application.schemas import AbstractMuseumNumberSchema
 from ebl.bibliography.application.reference_schema import ReferenceSchema
 from ebl.fragmentarium.application.date_schemas import (
     DateRangeSchema,
     DateWithNotesSchema,
 )
-from ebl.fragmentarium.domain.archaeology import Archaeology
-from ebl.fragmentarium.domain.findspot import BuildingType, ExcavationPlan, Findspot
+from ebl.fragmentarium.domain.archaeology import (
+    Archaeology,
+    ExcavationNumber,
+)
+from ebl.fragmentarium.domain.findspot import (
+    BuildingType,
+    ExcavationPlan,
+    Findspot,
+    ExcavationSite,
+)
 from ebl.schemas import NameEnumField
-from marshmallow import Schema, fields, post_load, validate
-from ebl.transliteration.domain.museum_number import MuseumNumber as ExcavationNumber
-from ebl.corpus.domain.provenance import Provenance as ExcavationSite
-
+from marshmallow import Schema, fields, post_load
 
-class ExcavationNumberSchema(Schema):
-    prefix = fields.String(
-        required=True, validate=(validate.Length(min=1), validate.ContainsNoneOf("."))
-    )
-    number = fields.String(
-        required=True, validate=(validate.Length(min=1), validate.ContainsNoneOf("."))
-    )
-    suffix = fields.String(required=True, validate=validate.ContainsNoneOf("."))
 
+class ExcavationNumberSchema(AbstractMuseumNumberSchema):
     @post_load
     def create_excavation_number(self, data, **kwargs) -> ExcavationNumber:
         return ExcavationNumber(**data)

ebl/fragmentarium/application/fragment_info_schema.py

@@ -4,6 +4,7 @@
     ReferenceSchema,
     ApiReferenceSchema,
 )
+from ebl.common.application.schemas import AccessionSchema
 from ebl.fragmentarium.application.fragment_schema import ScriptSchema
 from ebl.fragmentarium.application.genre_schema import GenreSchema
 from ebl.fragmentarium.domain.fragment_infos_pagination import FragmentInfosPagination
@@ -13,7 +14,7 @@
 
 class FragmentInfoSchema(Schema):
     number: fields.Field = fields.Nested(MuseumNumberSchema, required=True)
-    accession = fields.String(required=True)
+    accession = fields.Nested(AccessionSchema, allow_none=True, load_default=None)
     script = fields.Nested(ScriptSchema, required=True)
     description = fields.String(required=True)
     editor = fields.String(load_default="")

ebl/fragmentarium/application/fragment_schema.py

@@ -2,6 +2,7 @@
 from marshmallow import Schema, fields, post_dump, post_load, EXCLUDE
 
 from ebl.bibliography.application.reference_schema import ReferenceSchema
+from ebl.common.application.schemas import AccessionSchema
 from ebl.common.domain.period import Period, PeriodModifier
 from ebl.fragmentarium.application.archaeology_schemas import ArchaeologySchema
 from ebl.fragmentarium.application.genre_schema import GenreSchema
@@ -159,7 +160,7 @@ def omit_empty_numbers(self, data, **kwargs):
 
 class FragmentSchema(Schema):
     number = fields.Nested(MuseumNumberSchema, required=True, data_key="museumNumber")
-    accession = fields.String(required=True)
+    accession = fields.Nested(AccessionSchema, allow_none=True, load_default=None)
     edited_in_oracc_project = fields.String(
         required=True, data_key="editedInOraccProject"
     )

ebl/fragmentarium/domain/archaeology.py

@@ -1,9 +1,18 @@
 from typing import Optional, Sequence
 import attr
 from ebl.fragmentarium.domain.iso_date import DateWithNotes
-from ebl.transliteration.domain.museum_number import MuseumNumber as ExcavationNumber
-from ebl.corpus.domain.provenance import Provenance as ExcavationSite
-from ebl.fragmentarium.domain.findspot import Findspot
+from ebl.transliteration.domain.museum_number import MuseumNumber
+from ebl.fragmentarium.domain.findspot import Findspot, ExcavationSite
+import re
+
+
+class ExcavationNumber(MuseumNumber):
+    @staticmethod
+    def of(source: str) -> "ExcavationNumber":
+        if match := re.compile(r"(.+?)\.([^.]+)(?:\.([^.]+))?").fullmatch(source):
+            return ExcavationNumber(match[1], match[2], match[3] or "")
+        else:
+            raise ValueError(f"'{source}' is not a valid excavation number.")
 
 
 @attr.s(auto_attribs=True, frozen=True)

ebl/fragmentarium/domain/findspot.py

@@ -3,7 +3,10 @@
 from enum import Enum, auto
 from ebl.bibliography.domain.reference import Reference
 from ebl.fragmentarium.domain.iso_date import DateRange
-from ebl.corpus.domain.provenance import Provenance as ExcavationSite
+from ebl.corpus.domain.provenance import Provenance
+
+
+ExcavationSite = Provenance
 
 
 class BuildingType(Enum):

ebl/fragmentarium/domain/fragment.py

@@ -5,6 +5,7 @@
 import pydash
 
 from ebl.bibliography.domain.reference import Reference
+from ebl.common.domain.accession import Accession
 from ebl.common.domain.period import Period, PeriodModifier
 from ebl.common.domain.scopes import Scope
 from ebl.fragmentarium.application.matches.create_line_to_vec import create_line_to_vec
@@ -113,7 +114,7 @@ class ExternalNumbers:
 @attr.s(auto_attribs=True, frozen=True)
 class Fragment:
     number: MuseumNumber
-    accession: str = ""
+    accession: Optional[Accession] = None
     edited_in_oracc_project: str = ""
     publication: str = ""
     description: str = ""

ebl/fragmentarium/domain/fragment_info.py

@@ -3,6 +3,7 @@
 import attr
 
 from ebl.bibliography.domain.reference import Reference
+from ebl.common.domain.accession import Accession
 from ebl.fragmentarium.domain.fragment import Fragment, Genre, Script
 from ebl.fragmentarium.domain.record import RecordEntry, RecordType
 from ebl.transliteration.domain.museum_number import MuseumNumber
@@ -12,7 +13,7 @@
 @attr.s(frozen=True, auto_attribs=True)
 class FragmentInfo:
     number: MuseumNumber
-    accession: str
+    accession: Optional[Accession]
     script: Script
     description: str
     matching_lines: Optional[Text]

ebl/fragmentarium/infrastructure/mongo_fragment_repository.py

@@ -36,7 +36,7 @@
 from ebl.transliteration.application.museum_number_schema import MuseumNumberSchema
 from ebl.transliteration.domain.museum_number import MuseumNumber
 from ebl.transliteration.infrastructure.collections import FRAGMENTS_COLLECTION
-from ebl.transliteration.infrastructure.queries import museum_number_is
+from ebl.transliteration.infrastructure.queries import query_number_is
 
 RETRIEVE_ALL_LIMIT = 1000
 
@@ -171,7 +171,7 @@ def query_by_museum_number(
     ):
         data = self._fragments.aggregate(
             [
-                {"$match": museum_number_is(number)},
+                {"$match": query_number_is(number)},
                 *(
                     self._omit_text_lines()
                     if exclude_lines
@@ -206,7 +206,7 @@ def fetch_date(self, number: MuseumNumber) -> Optional[Date]:
     def fetch_scopes(self, number: MuseumNumber) -> List[Scope]:
         fragment = next(
             self._fragments.find_many(
-                museum_number_is(number), projection={"authorizedScopes": True}
+                query_number_is(number), projection={"authorizedScopes": True}
             ),
             {},
         )

ebl/fragmentarium/infrastructure/queries.py

@@ -1,5 +1,7 @@
 from typing import List, Sequence
+from ebl.common.domain.accession import Accession
 from ebl.common.domain.scopes import Scope
+from ebl.fragmentarium.domain.archaeology import ExcavationNumber
 
 from ebl.fragmentarium.domain.fragment import Fragment
 from ebl.fragmentarium.domain.record import RecordType
@@ -9,7 +11,7 @@
     FRAGMENTS_COLLECTION,
     FINDSPOTS_COLLECTION,
 )
-from ebl.transliteration.infrastructure.queries import museum_number_is
+from ebl.transliteration.infrastructure.queries import query_number_is
 
 HAS_TRANSLITERATION: dict = {"text.lines.type": {"$exists": True}}
 NUMBER_OF_LATEST_TRANSLITERATIONS: int = 50
@@ -18,15 +20,17 @@
 
 
 def fragment_is(fragment: Fragment) -> dict:
-    return museum_number_is(fragment.number)
+    return query_number_is(fragment.number)
 
 
 def number_is(number: str) -> dict:
-    or_ = [{"externalNumbers.cdliNumber": number}, {"accession": number}]
-    try:
-        or_.append(museum_number_is(MuseumNumber.of(number)))
-    except ValueError:
-        pass
+    or_ = [{"externalNumbers.cdliNumber": number}]
+
+    for number_class in [MuseumNumber, Accession, ExcavationNumber]:
+        try:
+            or_.append(query_number_is(number_class.of(number)))
+        except ValueError:
+            pass
     return {"$or": or_}
 
 

ebl/tests/common/test_accession.py

@@ -0,0 +1,24 @@
+import pytest
+from ebl.common.application.schemas import AccessionSchema
+from ebl.common.domain.accession import Accession
+
+
+ACCESSION = Accession("A", "38")
+ACCESSION_DTO = {"prefix": "A", "number": "38", "suffix": ""}
+
+
+def test_of():
+    assert Accession.of("A.38") == ACCESSION
+
+
+def test_of_invalid():
+    with pytest.raises(ValueError, match="'invalid.' is not a valid accession number."):
+        Accession.of("invalid.")
+
+
+def test_serialize():
+    assert AccessionSchema().dump(ACCESSION) == ACCESSION_DTO
+
+
+def test_deserialize():
+    assert AccessionSchema().load(ACCESSION_DTO) == ACCESSION

ebl/tests/factories/archaeology.py

@@ -1,9 +1,8 @@
-from ebl.fragmentarium.domain.archaeology import Archaeology
+from ebl.fragmentarium.domain.archaeology import Archaeology, ExcavationNumber
 from ebl.fragmentarium.domain.iso_date import DateRange, DateWithNotes
 from ebl.fragmentarium.domain.findspot import BuildingType, ExcavationPlan, Findspot
 from ebl.tests.factories.bibliography import ReferenceFactory
 from ebl.tests.factories.collections import TupleFactory
-from ebl.transliteration.domain.museum_number import MuseumNumber as ExcavationNumber
 from ebl.corpus.domain.provenance import Provenance as ExcavationSite
 import factory.fuzzy
 

ebl/tests/factories/fragment.py

@@ -2,6 +2,7 @@
 
 import factory.fuzzy
 import random
+from ebl.common.domain.accession import Accession
 from ebl.common.domain.period import Period, PeriodModifier
 from ebl.common.domain.project import ResearchProject
 
@@ -181,7 +182,7 @@ class Meta:
 
     number = factory.Sequence(lambda n: MuseumNumber("X", str(n)))
     edited_in_oracc_project = factory.Sequence(lambda n: f"editedInOracc-{n}")
-    accession = factory.Sequence(lambda n: f"accession-{n}")
+    accession = factory.Sequence(lambda n: Accession("A", str(n)))
     museum = factory.Faker("word")
     collection = factory.Faker("word")
     publication = factory.Faker("sentence")

ebl/tests/fragmentarium/test_dtos.py

@@ -1,6 +1,7 @@
 import attr
 import pydash
 import pytest
+from ebl.common.application.schemas import AccessionSchema
 
 from ebl.errors import DataError
 from ebl.fragmentarium.application.archaeology_schemas import ArchaeologySchema
@@ -44,7 +45,7 @@ def expected_dto(lemmatized_fragment, has_photo):
     return pydash.omit_by(
         {
             "museumNumber": attr.asdict(lemmatized_fragment.number),
-            "accession": lemmatized_fragment.accession,
+            "accession": AccessionSchema().dump(lemmatized_fragment.accession),
             "editedInOraccProject": lemmatized_fragment.edited_in_oracc_project,
             "publication": lemmatized_fragment.publication,
             "description": lemmatized_fragment.description,
@@ -131,7 +132,7 @@ def test_create_fragment_info_dto():
     is_transliteration = record_entry.type == RecordType.TRANSLITERATION
     assert ApiFragmentInfoSchema().dump(info) == {
         "number": str(info.number),
-        "accession": info.accession,
+        "accession": AccessionSchema().dump(info.accession),
         "script": ScriptSchema().dump(info.script),
         "description": info.description,
         "matchingLines": TextSchema().dump(text),

ebl/tests/fragmentarium/test_fragment_archaeology_route.py

@@ -4,16 +4,19 @@
 import falcon
 import pytest
 from ebl.fragmentarium.application.archaeology_schemas import ArchaeologySchema
-from ebl.fragmentarium.domain.archaeology import Archaeology
+from ebl.fragmentarium.domain.archaeology import (
+    Archaeology,
+    ExcavationNumber,
+)
+from ebl.fragmentarium.domain.findspot import ExcavationSite
 from ebl.fragmentarium.domain.fragment import Fragment
 
 from ebl.fragmentarium.web.dtos import create_response_dto
 from ebl.tests.factories.archaeology import DateWithNotesFactory
 from ebl.tests.factories.fragment import FragmentFactory
-from ebl.transliteration.domain.museum_number import MuseumNumber
-from ebl.corpus.domain.provenance import Provenance as ExcavationSite
 
-ARCHAEOLOGY = Archaeology(MuseumNumber("F", "1"), ExcavationSite.KALHU)
+
+ARCHAEOLOGY = Archaeology(ExcavationNumber("F", "1"), ExcavationSite.KALHU)
 ARCHAEOLOGIES = [
     ARCHAEOLOGY,
     attr.evolve(ARCHAEOLOGY, site=None),

ebl/tests/fragmentarium/test_fragment_schema.py

@@ -61,3 +61,16 @@ def test_scope_deserialization():
         "authorizedScopes": SERIALIZED_SCOPES,
     }
     assert FragmentSchema().load(data).authorized_scopes == SCOPES
+
+
+def test_empty_accession_serialization():
+    fragment = FragmentFactory.build(accession=None)
+    assert "accession" not in FragmentSchema().dump(fragment)
+
+
+def test_empty_accession_deserialization():
+    data = {
+        **FragmentSchema().dump(FragmentFactory.build()),
+        "accession": None,
+    }
+    assert FragmentSchema().load(data).accession is None

ebl/tests/fragmentarium/test_fragments_search_route.py

@@ -54,7 +54,8 @@ def query_item_of(
     [
         lambda fragment: str(fragment.number),
         lambda fragment: fragment.cdli_number,
-        lambda fragment: fragment.accession,
+        lambda fragment: str(fragment.accession),
+        lambda fragment: str(fragment.archaeology.excavation_number),
     ],
 )
 def test_query_fragmentarium_number(get_number, client, fragmentarium):

ebl/transliteration/application/museum_number_schema.py

@@ -1,15 +1,9 @@
-from marshmallow import Schema, fields, post_load, validate
-
+from marshmallow import post_load
 from ebl.transliteration.domain.museum_number import MuseumNumber
+from ebl.common.application.schemas import AbstractMuseumNumberSchema
 
 
-class MuseumNumberSchema(Schema):
-    prefix = fields.String(required=True, validate=validate.Length(min=1))
-    number = fields.String(
-        required=True, validate=(validate.Length(min=1), validate.ContainsNoneOf("."))
-    )
-    suffix = fields.String(required=True, validate=validate.ContainsNoneOf("."))
-
+class MuseumNumberSchema(AbstractMuseumNumberSchema):
     @post_load
     def create_museum_number(self, data, **kwargs) -> MuseumNumber:
         return MuseumNumber(**data)