Spaces:
Sleeping
Sleeping
Upload folder using huggingface_hub
Browse files
app/infrastructure/database/models.py
CHANGED
|
@@ -27,7 +27,7 @@ class DocumentModel(Base):
|
|
| 27 |
upload_session_id = Column(String(100), nullable=True)
|
| 28 |
uploaded_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 29 |
indexed_at = Column(DateTime(timezone=True), nullable=True)
|
| 30 |
-
|
| 31 |
created_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 32 |
updated_at = Column(
|
| 33 |
DateTime(timezone=True), default=datetime.utcnow, onupdate=datetime.utcnow, nullable=False
|
|
@@ -50,7 +50,7 @@ class DocumentChunkModel(Base):
|
|
| 50 |
content = Column(Text, nullable=False)
|
| 51 |
token_count = Column(Integer, nullable=False)
|
| 52 |
vector_id = Column(String(100), nullable=True, index=True)
|
| 53 |
-
|
| 54 |
created_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 55 |
|
| 56 |
__table_args__ = (Index("ix_chunks_document_id_index", "document_id", "chunk_index"),)
|
|
|
|
| 27 |
upload_session_id = Column(String(100), nullable=True)
|
| 28 |
uploaded_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 29 |
indexed_at = Column(DateTime(timezone=True), nullable=True)
|
| 30 |
+
doc_metadata = Column(JSONB, default={}, nullable=False)
|
| 31 |
created_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 32 |
updated_at = Column(
|
| 33 |
DateTime(timezone=True), default=datetime.utcnow, onupdate=datetime.utcnow, nullable=False
|
|
|
|
| 50 |
content = Column(Text, nullable=False)
|
| 51 |
token_count = Column(Integer, nullable=False)
|
| 52 |
vector_id = Column(String(100), nullable=True, index=True)
|
| 53 |
+
chunk_metadata = Column(JSONB, default={}, nullable=False)
|
| 54 |
created_at = Column(DateTime(timezone=True), default=datetime.utcnow, nullable=False)
|
| 55 |
|
| 56 |
__table_args__ = (Index("ix_chunks_document_id_index", "document_id", "chunk_index"),)
|
app/infrastructure/repositories/postgres_repository.py
CHANGED
|
@@ -32,7 +32,7 @@ class PostgresDocumentRepository(IDocumentRepository):
|
|
| 32 |
upload_session_id=document.upload_session_id,
|
| 33 |
uploaded_at=document.uploaded_at,
|
| 34 |
indexed_at=document.indexed_at,
|
| 35 |
-
|
| 36 |
)
|
| 37 |
self.session.add(model)
|
| 38 |
await self.session.commit()
|
|
@@ -60,7 +60,7 @@ class PostgresDocumentRepository(IDocumentRepository):
|
|
| 60 |
model.title = document.title
|
| 61 |
model.status = document.status.value
|
| 62 |
model.indexed_at = document.indexed_at
|
| 63 |
-
model.
|
| 64 |
model.updated_at = document.updated_at
|
| 65 |
|
| 66 |
await self.session.commit()
|
|
@@ -109,7 +109,7 @@ class PostgresDocumentRepository(IDocumentRepository):
|
|
| 109 |
upload_session_id=model.upload_session_id,
|
| 110 |
uploaded_at=model.uploaded_at,
|
| 111 |
indexed_at=model.indexed_at,
|
| 112 |
-
metadata=model.
|
| 113 |
created_at=model.created_at,
|
| 114 |
updated_at=model.updated_at,
|
| 115 |
)
|
|
@@ -131,7 +131,7 @@ class PostgresChunkRepository(IChunkRepository):
|
|
| 131 |
content=chunk.content,
|
| 132 |
token_count=chunk.token_count,
|
| 133 |
vector_id=chunk.vector_id,
|
| 134 |
-
|
| 135 |
)
|
| 136 |
for chunk in chunks
|
| 137 |
]
|
|
@@ -173,6 +173,6 @@ class PostgresChunkRepository(IChunkRepository):
|
|
| 173 |
content=model.content,
|
| 174 |
token_count=model.token_count,
|
| 175 |
vector_id=model.vector_id,
|
| 176 |
-
metadata=model.
|
| 177 |
created_at=model.created_at,
|
| 178 |
)
|
|
|
|
| 32 |
upload_session_id=document.upload_session_id,
|
| 33 |
uploaded_at=document.uploaded_at,
|
| 34 |
indexed_at=document.indexed_at,
|
| 35 |
+
doc_metadata=document.metadata,
|
| 36 |
)
|
| 37 |
self.session.add(model)
|
| 38 |
await self.session.commit()
|
|
|
|
| 60 |
model.title = document.title
|
| 61 |
model.status = document.status.value
|
| 62 |
model.indexed_at = document.indexed_at
|
| 63 |
+
model.doc_metadata = document.metadata
|
| 64 |
model.updated_at = document.updated_at
|
| 65 |
|
| 66 |
await self.session.commit()
|
|
|
|
| 109 |
upload_session_id=model.upload_session_id,
|
| 110 |
uploaded_at=model.uploaded_at,
|
| 111 |
indexed_at=model.indexed_at,
|
| 112 |
+
metadata=model.doc_metadata,
|
| 113 |
created_at=model.created_at,
|
| 114 |
updated_at=model.updated_at,
|
| 115 |
)
|
|
|
|
| 131 |
content=chunk.content,
|
| 132 |
token_count=chunk.token_count,
|
| 133 |
vector_id=chunk.vector_id,
|
| 134 |
+
chunk_metadata=chunk.metadata,
|
| 135 |
)
|
| 136 |
for chunk in chunks
|
| 137 |
]
|
|
|
|
| 173 |
content=model.content,
|
| 174 |
token_count=model.token_count,
|
| 175 |
vector_id=model.vector_id,
|
| 176 |
+
metadata=model.chunk_metadata,
|
| 177 |
created_at=model.created_at,
|
| 178 |
)
|