DH-4703 Set table-desription status (#183)

* DH-4703 Set table-desription status

* DH-4703 Added script to update table_descriptions rows

dataherald/api/fastapi.py

@@ -6,14 +6,15 @@
 from bson import json_util
 from fastapi import BackgroundTasks, HTTPException
 from overrides import override
+from sqlalchemy import MetaData, inspect
 
 from dataherald.api import API
 from dataherald.api.types import Query
 from dataherald.config import System
 from dataherald.context_store import ContextStore
 from dataherald.db import DB
 from dataherald.db_scanner import Scanner
-from dataherald.db_scanner.models.types import TableSchemaDetail
+from dataherald.db_scanner.models.types import TableDescriptionStatus, TableSchemaDetail
 from dataherald.db_scanner.repository.base import DBScannerRepository
 from dataherald.eval import Evaluator
 from dataherald.repositories.base import NLQueryResponseRepository
@@ -221,10 +222,36 @@ def list_table_descriptions(
         self, db_connection_id: str | None = None, table_name: str | None = None
     ) -> list[TableSchemaDetail]:
         scanner_repository = DBScannerRepository(self.storage)
-        return scanner_repository.find_by(
+        table_descriptions = scanner_repository.find_by(
             {"db_connection_id": db_connection_id, "table_name": table_name}
         )
 
+        if db_connection_id:
+            db_connection_repository = DatabaseConnectionRepository(self.storage)
+            db_connection = db_connection_repository.find_by_id(db_connection_id)
+            database = SQLDatabase.get_sql_engine(db_connection)
+            inspector = inspect(database.engine)
+            meta = MetaData(bind=database.engine)
+            MetaData.reflect(meta, views=True)
+            all_tables = inspector.get_table_names() + inspector.get_view_names()
+
+            for table_description in table_descriptions:
+                if table_description.table_name not in all_tables:
+                    table_description.status = TableDescriptionStatus.DEPRECATED.value
+                else:
+                    all_tables.remove(table_description.table_name)
+            for table in all_tables:
+                table_descriptions.append(
+                    TableSchemaDetail(
+                        table_name=table,
+                        status=TableDescriptionStatus.NOT_SYNCHRONIZED.value,
+                        db_connection_id=db_connection_id,
+                        columns=[],
+                    )
+                )
+
+        return table_descriptions
+
     @override
     def add_golden_records(
         self, golden_records: List[GoldenRecordRequest]

dataherald/db_scanner/models/types.py

@@ -1,4 +1,5 @@
 from datetime import datetime
+from enum import Enum
 from typing import Any
 
 from pydantic import BaseModel
@@ -19,13 +20,22 @@ class ColumnDetail(BaseModel):
     foreign_key: ForeignKeyDetail | None
 
 
+class TableDescriptionStatus(Enum):
+    NOT_SYNCHRONIZED = "NOT_SYNCHRONIZED"
+    SYNCHRONIZING = "SYNCHRONIZING"
+    DEPRECATED = "DEPRECATED"
+    SYNCHRONIZED = "SYNCHRONIZED"
+    FAILED = "FAILED"
+
+
 class TableSchemaDetail(BaseModel):
     id: Any
     db_connection_id: str
     table_name: str
     description: str | None
     table_schema: str | None
-    columns: list[ColumnDetail]
+    columns: list[ColumnDetail] = []
     examples: list = []
     last_schema_sync: datetime | None
-    status: str = "synchrinozed"
+    status: str = TableDescriptionStatus.SYNCHRONIZED.value
+    error_message: str | None

dataherald/db_scanner/sqlalchemy.py

@@ -8,7 +8,11 @@
 from sqlalchemy.sql import func
 
 from dataherald.db_scanner import Scanner
-from dataherald.db_scanner.models.types import ColumnDetail, TableSchemaDetail
+from dataherald.db_scanner.models.types import (
+    ColumnDetail,
+    TableDescriptionStatus,
+    TableSchemaDetail,
+)
 from dataherald.db_scanner.repository.base import DBScannerRepository
 from dataherald.sql_database.base import SQLDatabase
 
@@ -146,7 +150,7 @@ def scan_single_table(
                 meta=meta, db_engine=db_engine, table=table, rows_number=3
             ),
             last_schema_sync=datetime.now(),
-            status="syncronized",
+            status="SYNCHRONIZED",
         )
 
         repository.save_table_info(object)
@@ -171,13 +175,32 @@ def scan(
             ]
         if len(tables) == 0:
             raise ValueError("No table found")
-        result = []
+
+        # persist tables to be scanned
         for table in tables:
-            obj = self.scan_single_table(
-                meta=meta,
-                table=table,
-                db_engine=db_engine,
-                db_connection_id=db_connection_id,
-                repository=repository,
+            repository.save_table_info(
+                TableSchemaDetail(
+                    db_connection_id=db_connection_id,
+                    table_name=table,
+                    status=TableDescriptionStatus.SYNCHRONIZING.value,
+                )
             )
-            result.append(obj)
+
+        for table in tables:
+            try:
+                self.scan_single_table(
+                    meta=meta,
+                    table=table,
+                    db_engine=db_engine,
+                    db_connection_id=db_connection_id,
+                    repository=repository,
+                )
+            except Exception as e:
+                repository.save_table_info(
+                    TableSchemaDetail(
+                        db_connection_id=db_connection_id,
+                        table_name=table,
+                        status=TableDescriptionStatus.FAILED.value,
+                        error_message=f"{e}",
+                    )
+                )

dataherald/scripts/migrate_v002_to_v003.py

@@ -0,0 +1,17 @@
+import dataherald.config
+from dataherald.config import System
+from dataherald.db import DB
+
+if __name__ == "__main__":
+    settings = dataherald.config.Settings()
+    system = System(settings)
+    system.start()
+    storage = system.instance(DB)
+    # Update table_descriptions status
+    collection_rows = storage.find_all("table_descriptions")
+    for collection_row in collection_rows:
+        collection_row["status"] = "SYNCHRONIZED"
+        # update object
+        storage.update_or_create(
+            "table_descriptions", {"_id": collection_row["_id"]}, collection_row
+        )

dataherald/sql_generator/dataherald_sqlagent.py

@@ -30,7 +30,7 @@
 
 from dataherald.context_store import ContextStore
 from dataherald.db import DB
-from dataherald.db_scanner.models.types import TableSchemaDetail
+from dataherald.db_scanner.models.types import TableDescriptionStatus, TableSchemaDetail
 from dataherald.db_scanner.repository.base import DBScannerRepository
 from dataherald.sql_database.base import SQLDatabase, SQLInjectionError
 from dataherald.sql_database.models.types import (
@@ -581,7 +581,8 @@ def generate_response(
         )
         repository = DBScannerRepository(storage)
         db_scan = repository.get_all_tables_by_db(
-            db_connection_id=database_connection.id
+            db_connection_id=database_connection.id,
+            status=TableDescriptionStatus.SYNCHRONIZED.value,
         )
         if not db_scan:
             raise ValueError("No scanned tables found for database")

docs/api.list_table_description.rst

@@ -3,7 +3,7 @@
 List table descriptions
 =======================
 
-Once you have scanned a db connection you can list the table descriptions by requesting this endpoint.
+This endpoint returns the database connection tables and includes a status field that indicates whether the tables have been scanned or not.
 
 Request this ``GET`` endpoint::
 
@@ -15,7 +15,7 @@ Request this ``GET`` endpoint::
    :header: "Name", "Type", "Description"
    :widths: 20, 20, 60
 
-   "db_connection_id", "string", "Filter by connection id, ``Optional``"
+   "db_connection_id", "string", "Filter by connection id, ``Optional``. By configuring this field, it establishes a connection with the database to fetch table names and subsequently merges this data with the pre-existing rows in our MongoDB."
    "table_name", "string", "Filter by table name, ``Optional``"
 
 **Responses**
@@ -31,6 +31,8 @@ HTTP 200 code response
         "table_name": "string",
         "description": "string",
         "table_schema": "string",
+        "status": "NOT_SYNCHRONIZED | SYNCHRONIZING | DEPRECATED | SYNCHRONIZED | FAILED"
+        "error_message": "string",
         "columns": [
           {
             "name": "string",
@@ -51,6 +53,19 @@ HTTP 200 code response
       }
     ]
 
+.. csv-table::
+   :header: "Name", "Type", "Description"
+   :widths: 20, 20, 60
+
+   "status", "string", "It can be one of the next options:
+    - `NOT_SYNCHRONIZED` if the table has not been scanned
+    - `SYNCHRONIZING` while the sync schema process is running
+    - `DEPRECATED` if there is a row in our `table-descriptions` collection that is no longer in the database, probably because the table/view was deleted or renamed
+    - `SYNCHRONIZED` when we have scanned the table
+    - `FAILED` if anything failed during the sync schema process, and the `error_message` field stores the error."
+   "error_message", "string", "This field is set only if the async schema process fails"
+
+
 **Request example**
 
 .. code-block:: rst