Add repetitions substitution

backend/app/documents/query.py

@@ -1,7 +1,7 @@
 from datetime import datetime
 from typing import Iterable
 
-from sqlalchemy import case, func, select, Row
+from sqlalchemy import Row, case, func, select
 from sqlalchemy.orm import Session
 
 from app.base.exceptions import BaseQueryException
@@ -96,7 +96,7 @@ def get_document_records_paged(
         repetitions_subquery = (
             select(
                 DocumentRecord.source,
-                func.count(DocumentRecord.id).label('repetitions_count')
+                func.count(DocumentRecord.id).label("repetitions_count"),
             )
             .filter(DocumentRecord.document_id == doc.id)
             .group_by(DocumentRecord.source)
@@ -106,12 +106,14 @@ def get_document_records_paged(
         return self.__db.execute(
             select(
                 DocumentRecord,
-                func.coalesce(repetitions_subquery.c.repetitions_count, 0).label('repetitions_count')
+                func.coalesce(repetitions_subquery.c.repetitions_count, 0).label(
+                    "repetitions_count"
+                ),
             )
             .filter(DocumentRecord.document_id == doc.id)
             .outerjoin(
                 repetitions_subquery,
-                DocumentRecord.source == repetitions_subquery.c.source
+                DocumentRecord.source == repetitions_subquery.c.source,
             )
             .order_by(DocumentRecord.id)
             .offset(page_records * page)
@@ -131,19 +133,39 @@ def update_record(
             raise NotFoundDocumentRecordExc()
 
         record.target = data.target
+
         if data.approved is not None:
             record.approved = data.approved
 
-            if data.approved is True:
-                bound_tm = None
-                for memory in record.document.memory_associations:
-                    if memory.mode == TmMode.write:
-                        bound_tm = memory.tm_id
-
-                if bound_tm:
-                    TranslationMemoryQuery(self.__db).add_or_update_record(
-                        bound_tm, record.source, record.target
+        # If update_repetitions is True, find all records with the same source
+        if data.update_repetitions:
+            repeated_records = (
+                self.__db.execute(
+                    select(DocumentRecord).filter(
+                        DocumentRecord.document_id == record.document_id,
+                        DocumentRecord.source == record.source,
                     )
+                )
+                .scalars()
+                .all()
+            )
+
+            # Update all repeated records
+            for repeated_record in repeated_records:
+                repeated_record.target = data.target
+                if data.approved is not None:
+                    repeated_record.approved = data.approved
+
+        if data.approved is True:
+            bound_tm = None
+            for memory in record.document.memory_associations:
+                if memory.mode == TmMode.write:
+                    bound_tm = memory.tm_id
+
+            if bound_tm:
+                TranslationMemoryQuery(self.__db).add_or_update_record(
+                    bound_tm, record.source, record.target
+                )
 
         self.__db.commit()
         return record

backend/app/documents/schema.py

@@ -36,6 +36,7 @@ class DocumentRecordUpdateResponse(Identified):
 class DocumentRecordUpdate(BaseModel):
     target: str
     approved: Optional[bool]
+    update_repetitions: bool
 
 
 class DocumentProcessingSettings(BaseModel):

backend/app/routers/document.py

@@ -152,8 +152,13 @@ def update_doc_record(
     db: Annotated[Session, Depends(get_db)],
 ) -> doc_schema.DocumentRecordUpdateResponse:
     try:
-        record = GenericDocsQuery(db).update_record(record_id, record)
-        return record
+        updated_record = GenericDocsQuery(db).update_record(record_id, record)
+        return doc_schema.DocumentRecordUpdateResponse(
+            id=updated_record.id,
+            source=updated_record.source,
+            target=updated_record.target,
+            approved=updated_record.approved,
+        )
     except NotFoundDocumentRecordExc as e:
         raise HTTPException(
             status_code=status.HTTP_404_NOT_FOUND, detail="Record not found"

backend/tests/routers/test_routes_doc_records.py

@@ -125,8 +125,8 @@ def test_doc_records_returns_404_for_nonexistent_document(
 @pytest.mark.parametrize(
     "arguments",
     [
-        {"target": "Updated", "approved": None},
-        {"target": "Updated", "approved": True},
+        {"target": "Updated", "approved": None, "update_repetitions": False},
+        {"target": "Updated", "approved": True, "update_repetitions": False},
     ],
 )
 def test_can_update_doc_record(
@@ -201,7 +201,12 @@ def test_record_approving_creates_memory(
         s.commit()
 
     response = user_logged_client.put(
-        "/document/record/1", json={"target": "Updated", "approved": True}
+        "/document/record/1",
+        json={
+            "target": "Updated",
+            "approved": True,
+            "update_repetitions": False,
+        },
     )
     assert response.status_code == 200, response.text
 
@@ -251,7 +256,12 @@ def test_record_approving_updates_memory(
         s.commit()
 
     response = user_logged_client.put(
-        "/document/record/1", json={"target": "Updated", "approved": True}
+        "/document/record/1",
+        json={
+            "target": "Updated",
+            "approved": True,
+            "update_repetitions": False,
+        },
     )
     assert response.status_code == 200, response.text
 
@@ -274,7 +284,12 @@ def test_returns_404_for_nonexistent_doc_when_updating_record(
     user_logged_client: TestClient,
 ):
     response = user_logged_client.put(
-        "/document/record/3", json={"target": "Updated", "approved": None}
+        "/document/record/3",
+        json={
+            "target": "Updated",
+            "approved": None,
+            "update_repetitions": False,
+        },
     )
     assert response.status_code == 404
 
@@ -298,3 +313,90 @@ def test_returns_404_for_nonexistent_record(
         "/document/1/record/3", json={"target": "Updated"}
     )
     assert response.status_code == 404
+
+
+def test_can_update_doc_record_with_repetitions(
+    user_logged_client: TestClient, session: Session
+):
+    """Test updating all records with the same source text"""
+    with session as s:
+        records = [
+            DocumentRecord(source="Hello World", target="Привет Мир"),
+            DocumentRecord(source="Hello World", target="Здравствуйте Мир"),
+            DocumentRecord(source="Goodbye", target="Пока"),
+        ]
+        s.add(
+            Document(
+                name="test_doc.txt",
+                type=DocumentType.txt,
+                records=records,
+                processing_status="pending",
+                created_by=1,
+            )
+        )
+        s.commit()
+
+    # Update record 1 with repetition update enabled
+    response = user_logged_client.put(
+        "/document/record/1",
+        json={"target": "Updated Hello", "approved": True, "update_repetitions": True},
+    )
+    assert response.status_code == 200
+
+    with session as s:
+        # Check that all records with "Hello World" source were updated
+        record1 = s.query(DocumentRecord).filter(DocumentRecord.id == 1).one()
+        record2 = s.query(DocumentRecord).filter(DocumentRecord.id == 2).one()
+        record3 = s.query(DocumentRecord).filter(DocumentRecord.id == 3).one()
+
+        assert record1.target == "Updated Hello"
+        assert record1.approved is True
+
+        assert record2.target == "Updated Hello"  # updated
+        assert record2.approved is True
+
+        assert record3.target == "Пока"  # unchanged (different source)
+        assert record3.approved is False
+
+
+def test_update_repetitions_default_behavior(
+    user_logged_client: TestClient, session: Session
+):
+    """Test that update_repetitions defaults to False when not specified"""
+    with session as s:
+        records = [
+            DocumentRecord(source="Hello World", target="Привет Мир"),
+            DocumentRecord(source="Hello World", target="Здравствуйте Мир"),
+        ]
+        s.add(
+            Document(
+                name="test_doc.txt",
+                type=DocumentType.txt,
+                records=records,
+                processing_status="pending",
+                created_by=1,
+            )
+        )
+        s.commit()
+
+    # Update without specifying update_repetitions (should default to False)
+    response = user_logged_client.put(
+        "/document/record/1",
+        json={
+            "target": "Updated Hello",
+            "approved": True,
+            "update_repetitions": False,
+        },
+    )
+    assert response.status_code == 200
+
+    with session as s:
+        # Check that only record 1 was updated
+        record1 = s.query(DocumentRecord).filter(DocumentRecord.id == 1).one()
+        record2 = s.query(DocumentRecord).filter(DocumentRecord.id == 2).one()
+
+        assert record1.target == "Updated Hello"
+        assert record1.approved is True
+
+        assert record2.target == "Здравствуйте Мир"  # unchanged
+        assert record2.approved is False

backend/tests/routers/test_routes_documents.py

@@ -917,7 +917,9 @@ def test_setting_glossaries_returns_404_for_non_existing_glossaries(
     assert response.status_code == 404
 
 
-def test_get_doc_records_with_repetitions(user_logged_client: TestClient, session: Session):
+def test_get_doc_records_with_repetitions(
+    user_logged_client: TestClient, session: Session
+):
     """Test that document records endpoint returns repetition counts"""
     with session as s:
         records = [
@@ -947,7 +949,9 @@ def test_get_doc_records_with_repetitions(user_logged_client: TestClient, sessio
 
     # Check that repetition counts are correct
     # "Hello World" appears 3 times, others appear once
-    record_counts = {record["source"]: record["repetitions_count"] for record in response_json}
+    record_counts = {
+        record["source"]: record["repetitions_count"] for record in response_json
+    }
     assert record_counts["Hello World"] == 3
     assert record_counts["Goodbye"] == 1
     assert record_counts["Test"] == 1

backend/tests/test_worker.py

@@ -273,7 +273,7 @@ def test_process_task_uses_correct_tm_ids(session: Session):
                 create_doc(name="small.xliff", type_=DocumentType.xliff),
                 create_xliff_doc(file_data),
                 create_task(),
-                DocMemoryAssociation(doc_id=1, tm_id=2, mode='read')
+                DocMemoryAssociation(doc_id=1, tm_id=2, mode="read"),
             ]
         )
         s.commit()
@@ -318,8 +318,8 @@ def test_process_task_uses_tm_mode(mode: str, trans_result: str, session: Sessio
                 create_doc(name="small.xliff", type_=DocumentType.xliff),
                 create_xliff_doc(file_data),
                 create_task(usage=TranslationMemoryUsage(mode)),
-                DocMemoryAssociation(doc_id=1, tm_id=1, mode='read'),
-                DocMemoryAssociation(doc_id=1, tm_id=2, mode='read')
+                DocMemoryAssociation(doc_id=1, tm_id=1, mode="read"),
+                DocMemoryAssociation(doc_id=1, tm_id=2, mode="read"),
             ]
         )
         s.commit()

backend/worker.py

@@ -106,7 +106,9 @@ def process_document(
     )
 
     start_time = time.time()
-    translate_indices = substitute_segments(settings, session, segments, tm_ids, glossary_ids)
+    translate_indices = substitute_segments(
+        settings, session, segments, tm_ids, glossary_ids
+    )
     logging.info(
         "Segments substitution time: %.2f seconds, speed: %.2f segment/second, segments: %d/%d",
         time.time() - start_time,

frontend/mocks/documentMocks.ts

@@ -7,12 +7,38 @@ import {
 } from '../src/client/services/DocumentService'
 import {DocumentStatus} from '../src/client/schemas/DocumentStatus'
 
+const segments = [
+  {
+    id: 10000,
+    approved: false,
+    source: 'Adventure Hooks',
+    target: 'Зацепки приключения',
+    repetitions_count: 2,
+  },
+  {
+    id: 10001,
+    approved: true,
+    source: 'Adventure Hooks',
+    target: 'Зацепки приключения',
+    repetitions_count: 2,
+  },
+  {
+    id: 10002,
+    approved: false,
+    source:
+      'The moment the Cynidiceans pried the horn from the monolith, their city was doomed.',
+    target:
+      'В тот момент, когда кинидийцы извлекли рог из монолита, их город был обречен.',
+    repetitions_count: 1,
+  },
+]
+
 const docs = [
   {
     id: 1,
     created_by: 12,
-    records_count: 3,
-    approved_records_count: 1,
+    records_count: segments.length,
+    approved_records_count: segments.filter(({approved}) => approved).length,
     name: 'Some document',
     status: 'done' as DocumentStatus,
     type: 'XLIFF',
@@ -36,31 +62,9 @@ export const documentMocks = [
     ({params}) => {
       const doc = docs.find((doc) => doc.id === Number(params.id))
       if (doc !== undefined) {
-        return HttpResponse.json<AwaitedReturnType<typeof getDocRecords>>([
-          {
-            id: 1,
-            approved: false,
-            source: 'Adventure Hooks',
-            target: 'Зацепки приключения',
-            repetitions_count: 2,
-          },
-          {
-            id: 2,
-            approved: true,
-            source: 'Adventure Hooks',
-            target: 'Зацепки приключения',
-            repetitions_count: 2,
-          },
-          {
-            id: 3,
-            approved: false,
-            source:
-              'The moment the Cynidiceans pried the horn from the monolith, their city was doomed.',
-            target:
-              'В тот момент, когда кинидийцы извлекли рог из монолита, их город был обречен.',
-            repetitions_count: 1,
-          },
-        ])
+        return HttpResponse.json<AwaitedReturnType<typeof getDocRecords>>(
+          segments
+        )
       } else {
         new HttpResponse(null, {status: 404})
       }

frontend/src/client/schemas/DocumentRecordUpdate.ts

@@ -3,4 +3,5 @@
 export interface DocumentRecordUpdate {
   target: string
   approved: boolean | null
+  update_repetitions: boolean
 }

frontend/src/client/schemas/DocumentRecordUpdateResponse.ts

@@ -0,0 +1,8 @@
+// This file is autogenerated, do not edit directly.
+
+export interface DocumentRecordUpdateResponse {
+  id: number
+  source: string
+  target: string
+  approved: boolean
+}