KBMBF-452: #37 Show ratio of oer-content

KBMBF-502: Improving frontend
SD_WLO-238: Showing OER

Cleanup and planning for refactoring

src/app/api/analytics/analytics.py

@@ -36,15 +36,13 @@ class Config(ElasticConfig):
         pass
 
 
-COUNT_STATISTICS_TYPE = dict[str, int]
+CountStatistics = dict[str, int]
 
 
 class StatsResponse(ResponseModel):
     derived_at: datetime
-    stats: dict[str, dict[str, COUNT_STATISTICS_TYPE]]
-    oer_ratio: int = Field(
-        default=0, ge=0, le=100, description="Overall ratio of OER content"
-    )
+    stats: dict[str, dict[str, CountStatistics]]
+    oer_ratio: int = Field(ge=0, le=100, description="Overall ratio of OER content")
 
 
 ValidationStatsT = TypeVar("ValidationStatsT")

src/app/api/analytics/background_task.py

@@ -1,5 +1,6 @@
 import asyncio
 import os
+import uuid
 from datetime import datetime
 
 from fastapi import APIRouter
@@ -117,7 +118,7 @@ def run():
     )
 
     all_collections = [
-        Row(id=value["value"], title=key)
+        Row(id=uuid.UUID(value["value"]), title=key)
         for key, value in COLLECTION_NAME_TO_ID.items()
     ]
     print("Tree ready to iterate. Length: ", len(all_collections))

src/app/api/analytics/models.py

@@ -5,6 +5,6 @@
 
 # TODO: Rename, as used for materials in background_task, as well
 class Collection(BaseModel):
-    id: str
+    id: str  # TODO: Refactored to UUID
     doc: dict
     derived_at: datetime

src/app/api/analytics/stats.py

@@ -9,8 +9,8 @@
 from glom import merge
 
 from app.api.analytics.analytics import (
-    COUNT_STATISTICS_TYPE,
     CollectionValidationStats,
+    CountStatistics,
     MaterialValidationStats,
     StatsNotFoundException,
     StatsResponse,
@@ -142,7 +142,7 @@ def nodes(data: list[CollectionNode]) -> list:
 
 def query_material_types(
     node_id: uuid.UUID, oer_only: bool
-) -> dict[str, COUNT_STATISTICS_TYPE]:
+) -> dict[str, CountStatistics]:
     """
     get collections with parent id equal to node_id
 
@@ -162,7 +162,7 @@ def query_material_types(
     # TODO: Refactor with filter and dict comprehension
     for collection in collections:
         for count in counts:
-            if str(collection.id) == str(count.noderef_id):
+            if collection.id == str(count.noderef_id):
                 stats.update(
                     {str(collection.id): {"total": count.total, **count.counts}}
                 )
@@ -179,7 +179,7 @@ def filtered_collections(collections: list[Collection], node_id: uuid.UUID):
 
 async def query_search_statistics(
     node_id: uuid.UUID,
-) -> dict[str, COUNT_STATISTICS_TYPE]:
+) -> dict[str, CountStatistics]:
     for stats in global_store.search:
         if str(node_id) == str(stats.node_id):
             return {

src/app/api/analytics/storage.py

@@ -1,6 +1,8 @@
 import uuid
 from dataclasses import dataclass
 
+from app.api.analytics.analytics import CountStatistics
+
 _COLLECTIONS = "collections"
 _MATERIALS = "materials"
 _COLLECTION_COUNT = "counts"
@@ -17,20 +19,20 @@
 }  # TODO: Refactor me ASAP
 
 
-@dataclass
+@dataclass(frozen=True)
 class SearchStoreCollection:
     node_id: uuid.UUID
-    missing_materials: dict[str, int]
+    missing_materials: CountStatistics
 
 
-@dataclass
+@dataclass(frozen=True)
 class SearchStore:
     __slots__ = "node_id", "collections"
     node_id: uuid.UUID
     collections: dict[uuid.UUID, SearchStoreCollection]
 
 
-@dataclass
+@dataclass(frozen=True)
 class Store:
     __slots__ = "search"
     search: list[SearchStore]

src/app/api/collections/missing_materials.py

@@ -184,14 +184,14 @@ def missing_attributes_search(
             qmatch(**{"collections.path": node_id}),
             qmatch(**{"collections.nodeRef.id": node_id}),
         ],
-        "filter": type_filter[
-            ResourceType.MATERIAL
-        ].copy(),  # copy otherwise appending the query causes mutation
+        "filter": [
+            *type_filter[
+                ResourceType.MATERIAL
+            ].copy(),  # copy otherwise appending the query causes mutation
+            Q("bool", **{"must_not": [{"term": {"aspects": "ccm:io_childobject"}}]}),
+            Q({"term": {"content.mimetype.keyword": "text/plain"}}),
+        ],
     }
-    query["filter"].append(
-        Q("bool", **{"must_not": [{"term": {"aspects": "ccm:io_childobject"}}]})
-    )
-    query["filter"].append(Q({"term": {"content.mimetype.keyword": "text/plain"}}))
     if missing_attribute == LearningMaterialAttribute.LICENSES.path:
         query["filter"].append(query_missing_material_license().to_dict())
     else:

src/app/api/score/models.py

@@ -4,75 +4,70 @@
 
 
 class MissingCollectionProperties(BaseModel):
-    total: int = Field(default=0, ge=0, description="Number of entries")
+    total: int = Field(ge=0, description="Number of entries")
     short_description: float = Field(
-        default=0.0,
         ge=0.0,
         le=1.0,
         description="Ratio of entries without short description",
     )
     short_title: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without short title"
+        ge=0.0, le=1.0, description="Ratio of entries without short title"
     )
     missing_edu_context: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without edu context"
+        ge=0.0, le=1.0, description="Ratio of entries without edu context"
     )
     missing_description: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without description"
+        ge=0.0, le=1.0, description="Ratio of entries without description"
     )
     few_keywords: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries with few keywords"
+        ge=0.0, le=1.0, description="Ratio of entries with few keywords"
     )
     missing_keywords: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without keywords"
+        ge=0.0, le=1.0, description="Ratio of entries without keywords"
     )
     missing_title: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without title"
+        ge=0.0, le=1.0, description="Ratio of entries without title"
     )
 
 
 class MissingMaterialProperties(BaseModel):
     total: int = Field(default=0, ge=0, description="Number of entries")
     missing_title: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without title"
+        ge=0.0, le=1.0, description="Ratio of entries without title"
     )
     missing_material_type: float = Field(
-        default=0.0,
         ge=0.0,
         le=1.0,
         description="Ratio of entries missing material type",
     )
     missing_subjects: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries missing subjects"
+        ge=0.0, le=1.0, description="Ratio of entries missing subjects"
     )
     missing_url: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without url"
+        ge=0.0, le=1.0, description="Ratio of entries without url"
     )
     missing_license: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries with missing license"
+        ge=0.0, le=1.0, description="Ratio of entries with missing license"
     )
     missing_publisher: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without publisher"
+        ge=0.0, le=1.0, description="Ratio of entries without publisher"
     )
     missing_description: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without description"
+        ge=0.0, le=1.0, description="Ratio of entries without description"
     )
     missing_intended_end_user_role: float = Field(
-        default=0.0,
         ge=0.0,
         le=1.0,
         description="Ratio of entries without intended end user role",
     )
     missing_edu_context: float = Field(
-        default=0.0, ge=0.0, le=1.0, description="Ratio of entries without edu context"
+        ge=0.0, le=1.0, description="Ratio of entries without edu context"
     )
 
 
 class ScoreOutput(BaseModel):
-    score: int = Field(default=0, ge=0, le=100, description="Overall score")
-    oer_ratio: int = Field(
-        default=0, ge=0, le=100, description="Overall ratio of OER content"
-    )
+    score: int = Field(ge=0, le=100, description="Overall score")
+    oer_ratio: int = Field(ge=0, le=100, description="Overall ratio of OER content")
     collections: MissingCollectionProperties = Field(
         description="Score for specific collection properties"
     )

src/app/core/constants.py

@@ -1,4 +1,4 @@
-COLLECTION_NAME_TO_ID = {
+COLLECTION_NAME_TO_ID: dict[str, dict[str, str]] = {
     "Physik": {"value": "94f22c9b-0d3a-4c1c-8987-4c8e83f3a92e"},
     "Mathematik": {"value": "bd8be6d5-0fbe-4534-a4b3-773154ba6abc"},
     "Biologie": {"value": "15fce411-54d9-467f-8f35-61ea374a298d"},
@@ -25,6 +25,6 @@
     "Franzoesisch": {"value": "86b990ef-0955-45ad-bdae-ec2623cf0e1a"},
     "Musik": {"value": "2eda0065-f69b-46c8-ae09-d258c8226a5e"},
     "Philosophie": {"value": "9d364fd0-4374-40b4-a153-3c722b9cda35"},
-}
+}  # TODO: Refactor by removing value and refactoring the API constructor respectively
 OPEN_API_VERSION = "2.1.0"
 COLLECTION_ROOT_ID = "5e40e372-735c-4b17-bbf7-e827a5702b57"