feat(files): implement file service with vector search

.env.example

@@ -26,3 +26,11 @@ LLM_CONTEXT_MESSAGES_LIMIT=20
 # Celery
 CELERY_BROKER_URL=redis://localhost:6379/0
 CELERY_RESULT_BACKEND=redis://localhost:6379/1
+
+# S3
+S3_ENDPOINT_URL=http://localhost:9000
+S3_ACCESS_KEY_ID=minioadmin
+S3_SECRET_ACCESS_KEY=minioadmin
+S3_REGION=us-east-1
+S3_BUCKET_NAME=documents
+S3_PRESIGNED_URL_EXPIRE_SECONDS=3600

.github/workflows/ci.yml

@@ -23,7 +23,7 @@ jobs:
 
     services:
       postgres:
-        image: postgres:16
+        image: pgvector/pgvector:pg16
         env:
           POSTGRES_DB: ai_notes
           POSTGRES_USER: postgres
@@ -63,6 +63,11 @@ jobs:
       CELERY_BROKER_URL: redis://localhost:6379/0
       CELERY_RESULT_BACKEND: redis://localhost:6379/0
 
+      S3_ENDPOINT_URL: http://localhost:9000
+      S3_ACCESS_KEY_ID: test
+      S3_SECRET_ACCESS_KEY: test
+      S3_BUCKET_NAME: documents
+
     steps:
       - uses: actions/checkout@v4
 

README.md

@@ -96,6 +96,12 @@ Required variables:
 * `LLM_CONTEXT_MESSAGES_LIMIT` - number of recent messages sent as context
 * `CELERY_BROKER_URL` - Redis URL for the Celery broker
 * `CELERY_RESULT_BACKEND` - Redis URL for the Celery result backend
+* `S3_ENDPOINT_URL` - S3 endpoint URL
+* `S3_ACCESS_KEY_ID` - S3 access key ID
+* `S3_SECRET_ACCESS_KEY` - S3 secret access key
+* `S3_REGION` - S3 region name, default `us-east-1`
+* `S3_BUCKET_NAME` - bucket used to store documents, default `documents`
+* `S3_PRESIGNED_URL_EXPIRE_SECONDS` - presigned document URL lifetime in seconds
 
 The database connection URL is composed automatically from the `POSTGRES_*` values.
 
@@ -165,6 +171,176 @@ task lint
 task check
 ```
 
+## 🗄 Database schema
+
+Entity-relationship diagram for the database models:
+
+```mermaid
+erDiagram
+    users {
+        uuid id PK
+        string email UK
+        string username "null"
+        string hashed_password
+        bool is_active
+        bool is_superuser
+        datetime created_at
+        datetime updated_at
+    }
+
+    notes {
+        uuid id PK
+        uuid user_id FK
+        string title
+        text content
+        string_array tags
+        enum source "model_source"
+        string model_name "null"
+        jsonb model_metadata
+        datetime created_at
+        datetime updated_at
+        datetime deleted_at "null"
+    }
+
+    chat_sessions {
+        uuid id PK
+        uuid user_id FK
+        string title
+        enum generation_status "chat_session_generation_status"
+        uuid generation_id "null"
+        datetime generation_started_at "null"
+        datetime created_at
+        datetime updated_at
+        datetime deleted_at "null"
+    }
+
+    messages {
+        uuid id PK
+        uuid session_id FK
+        text content
+        enum role "message_role"
+        string provider "null"
+        string model_name "null"
+        int prompt_tokens "null"
+        int completion_tokens "null"
+        int total_tokens "null"
+        datetime created_at
+        datetime updated_at
+        datetime deleted_at "null"
+    }
+
+    chat_memories {
+        uuid id PK
+        uuid session_id FK,UK
+        text summary
+        jsonb facts
+        bool is_summarizing
+        uuid last_summarized_message_id FK "null"
+        datetime created_at
+        datetime updated_at
+    }
+
+    generation_jobs {
+        uuid id PK
+        uuid user_id FK
+        uuid session_id FK
+        enum status "generation_job_status"
+        text input_message
+        uuid output_message_id FK "null"
+        text error "null"
+        datetime started_at "null"
+        datetime finished_at "null"
+        datetime created_at
+        datetime updated_at
+    }
+
+    documents {
+        uuid id PK
+        uuid user_id FK
+        uuid session_id FK
+        string filename
+        string content_type
+        int file_size
+        string checksum_sha256
+        string storage_bucket
+        string storage_object_name
+        enum status "document_status"
+        text error_message "null"
+        datetime created_at
+        datetime updated_at
+        datetime deleted_at "null"
+    }
+
+    document_chunks {
+        uuid id PK
+        uuid user_id FK
+        uuid session_id FK
+        uuid document_id FK
+        int chunk_index
+        text content
+        string content_hash
+        vector embedding
+        string embedding_model
+        int token_count "null"
+        datetime created_at
+        datetime updated_at
+        datetime deleted_at "null"
+    }
+
+    rag_queries {
+        uuid id PK
+        uuid user_id FK
+        uuid session_id FK
+        text question
+        text answer "null"
+        string provider "null"
+        string model "null"
+        int prompt_tokens "null"
+        int completion_tokens "null"
+        int total_tokens "null"
+        int top_k
+        enum status "rag_query_status"
+        datetime finished_at "null"
+        text error_message "null"
+        datetime created_at
+        datetime updated_at
+    }
+
+    rag_query_sources {
+        uuid id PK
+        uuid rag_query_id FK
+        uuid document_id FK
+        uuid chunk_id FK
+        float score
+        int rank
+        text content_preview
+        datetime created_at
+        datetime updated_at
+    }
+
+    users ||--o{ notes : owns
+    users ||--o{ chat_sessions : owns
+    users ||--o{ generation_jobs : owns
+    users ||--o{ documents : owns
+    users ||--o{ document_chunks : owns
+    users ||--o{ rag_queries : owns
+
+    chat_sessions ||--o{ messages : contains
+    chat_sessions ||--o| chat_memories : has
+    chat_sessions ||--o{ generation_jobs : contains
+    chat_sessions ||--o{ documents : contains
+    chat_sessions ||--o{ document_chunks : contains
+    chat_sessions ||--o{ rag_queries : contains
+
+    messages ||--o| generation_jobs : "output of"
+    messages ||--o| chat_memories : "last summarized"
+
+    documents ||--o{ document_chunks : "split into"
+    documents ||--o{ rag_query_sources : "referenced by"
+    document_chunks ||--o{ rag_query_sources : "referenced by"
+    rag_queries ||--o{ rag_query_sources : "retrieved"
+```
+
 ## 🛠 Database migrations
 
 * Create a new Alembic revision:

alembic/versions/239f767ca4fd_add_document_chunk_table.py

@@ -0,0 +1,32 @@
+"""'Add document_chunk table'
+
+Revision ID: 239f767ca4fd
+Revises: 692926ebb7e2
+Create Date: 2026-06-23 05:44:02.407327
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = '239f767ca4fd'
+down_revision: Union[str, Sequence[str], None] = '692926ebb7e2'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Upgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    pass
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    """Downgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    pass
+    # ### end Alembic commands ###

alembic/versions/48add2a2bda8_add_document_processing_jobs_table.py

@@ -0,0 +1,45 @@
+"""'Add document_processing_jobs table'
+
+Revision ID: 48add2a2bda8
+Revises: e495a670b858
+Create Date: 2026-06-23 07:33:08.783932
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = '48add2a2bda8'
+down_revision: Union[str, Sequence[str], None] = 'e495a670b858'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Upgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.create_table('document_processing_jobs',
+    sa.Column('id', sa.Uuid(), nullable=False),
+    sa.Column('document_id', sa.Uuid(), nullable=False),
+    sa.Column('status', sa.Enum('queued', 'running', 'completed', 'failed', name='document_processing_job_status'), nullable=False),
+    sa.Column('started_at', sa.DateTime(timezone=True), nullable=True),
+    sa.Column('finished_at', sa.DateTime(timezone=True), nullable=True),
+    sa.Column('error', sa.Text(), nullable=True),
+    sa.Column('created_at', sa.DateTime(timezone=True), nullable=False),
+    sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False),
+    sa.ForeignKeyConstraint(['document_id'], ['documents.id'], ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id')
+    )
+    op.create_index(op.f('ix_document_processing_jobs_document_id'), 'document_processing_jobs', ['document_id'], unique=False)
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    """Downgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_index(op.f('ix_document_processing_jobs_document_id'), table_name='document_processing_jobs')
+    op.drop_table('document_processing_jobs')
+    # ### end Alembic commands ###

alembic/versions/5c4081c49d7e_remove_default_none.py

@@ -0,0 +1,32 @@
+"""'Remove default=None'
+
+Revision ID: 5c4081c49d7e
+Revises: 239f767ca4fd
+Create Date: 2026-06-23 05:45:13.909308
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = '5c4081c49d7e'
+down_revision: Union[str, Sequence[str], None] = '239f767ca4fd'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Upgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    pass
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    """Downgrade schema."""
+    # ### commands auto generated by Alembic - please adjust! ###
+    pass
+    # ### end Alembic commands ###