fix: Catalog init introduces significant overhead (georgia-tech-db#1270)

Fixed georgia-tech-db#1268

Moved instrumentation after the entire execution.

<img width="1840" alt="image"
src="https://github.com/georgia-tech-db/evadb/assets/12206234/3d770689-deff-4408-bb64-9df320a95fa1">
<img width="1832" alt="image"
src="https://github.com/georgia-tech-db/evadb/assets/12206234/969ed19b-0985-4feb-ac87-045542a7b485">

evadb/executor/apply_and_merge_executor.py

@@ -43,6 +43,7 @@ def exec(self, *args, **kwargs) -> Iterator[Batch]:
         for batch in child_executor.exec(**kwargs):
             func_result = self.func_expr.evaluate(batch)
 
+<<<<<<< HEAD
 <<<<<<< HEAD
 =======
             # persist stats of function expression
@@ -55,6 +56,8 @@ def exec(self, *args, **kwargs) -> Iterator[Batch]:
                 )
 
 >>>>>>> 2dacff69 (feat: sync master staging (#1050))
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
             output = Batch.merge_column_wise([batch, func_result])
             if self.do_unnest:
                 output.unnest(func_result.columns)

evadb/executor/executor_utils.py

@@ -16,10 +16,13 @@
 import os
 from pathlib import Path
 from typing import TYPE_CHECKING, Generator, List, Union
+<<<<<<< HEAD
 
 from evadb.catalog.catalog_utils import xform_column_definitions_to_catalog_entries
 from evadb.catalog.models.utils import TableCatalogEntry
 from evadb.parser.create_statement import ColumnDefinition
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
 
 from evadb.catalog.catalog_utils import xform_column_definitions_to_catalog_entries
 from evadb.catalog.models.utils import TableCatalogEntry
@@ -73,6 +76,7 @@ def apply_project(batch: Batch, project_list: List[AbstractExpression]):
         batches = [expr.evaluate(batch) for expr in project_list]
         batch = Batch.merge_column_wise(batches)
 
+<<<<<<< HEAD
 <<<<<<< HEAD
 =======
         # persist stats of function expression
@@ -86,6 +90,8 @@ def apply_project(batch: Batch, project_list: List[AbstractExpression]):
                         func_expr._stats.prev_cost,
                     )
 >>>>>>> 2dacff69 (feat: sync master staging (#1050))
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
     return batch
 
 
@@ -95,6 +101,7 @@ def apply_predicate(batch: Batch, predicate: AbstractExpression) -> Batch:
         batch.drop_zero(outcomes)
         batch.reset_index()
 
+<<<<<<< HEAD
 <<<<<<< HEAD
 =======
         # persist stats of function expression
@@ -105,6 +112,8 @@ def apply_predicate(batch: Batch, predicate: AbstractExpression) -> Batch:
                     function_id, func_expr.function_obj.name, func_expr._stats.prev_cost
                 )
 >>>>>>> 2dacff69 (feat: sync master staging (#1050))
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
     return batch
 
 

evadb/executor/function_scan_executor.py

@@ -42,6 +42,7 @@ def exec(self, *args, **kwargs) -> Iterator[Batch]:
         if not lateral_input.empty():
             res = self.func_expr.evaluate(lateral_input)
 
+<<<<<<< HEAD
 <<<<<<< HEAD
 =======
             # persist stats of function expression
@@ -54,6 +55,8 @@ def exec(self, *args, **kwargs) -> Iterator[Batch]:
                 )
 
 >>>>>>> 2dacff69 (feat: sync master staging (#1050))
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
             if not res.empty():
                 if self.do_unnest:
                     res.unnest(res.columns)

evadb/executor/project_executor.py

@@ -19,14 +19,20 @@
 from evadb.database import EvaDBDatabase
 from evadb.executor.abstract_executor import AbstractExecutor
 <<<<<<< HEAD
+<<<<<<< HEAD
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
 from evadb.executor.executor_utils import (
     ExecutorError,
     apply_project,
     instrument_function_expression_cost,
 )
+<<<<<<< HEAD
 =======
 from evadb.executor.executor_utils import ExecutorError, apply_project
 >>>>>>> 40a10ce1 (Bump v0.3.4+ dev)
+=======
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
 from evadb.models.storage.batch import Batch
 from evadb.plan_nodes.project_plan import ProjectPlan
 
@@ -43,29 +49,43 @@ def exec(self, *args, **kwargs) -> Iterator[Batch]:
         if len(self.children) == 0:
             # Create a dummy batch with size 1
             dummy_batch = Batch(pd.DataFrame([0]))
+<<<<<<< HEAD
 <<<<<<< HEAD
             batch = apply_project(dummy_batch, self.target_list)
 =======
             batch = apply_project(dummy_batch, self.target_list, self.catalog())
 >>>>>>> 40a10ce1 (Bump v0.3.4+ dev)
+=======
+            batch = apply_project(dummy_batch, self.target_list)
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
             if not batch.empty():
                 yield batch
         # SELECT expr FROM table;
         elif len(self.children) == 1:
             child_executor = self.children[0]
             for batch in child_executor.exec(**kwargs):
+<<<<<<< HEAD
 <<<<<<< HEAD
                 batch = apply_project(batch, self.target_list)
 =======
                 batch = apply_project(batch, self.target_list, self.catalog())
 >>>>>>> 40a10ce1 (Bump v0.3.4+ dev)
+=======
+                batch = apply_project(batch, self.target_list)
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))
                 if not batch.empty():
                     yield batch
         else:
             raise ExecutorError("ProjectExecutor has more than 1 children.")
+<<<<<<< HEAD
 <<<<<<< HEAD
 
         # instrument required stats
         instrument_function_expression_cost(self.target_list, self.catalog())
 =======
 >>>>>>> 40a10ce1 (Bump v0.3.4+ dev)
+=======
+
+        # instrument required stats
+        instrument_function_expression_cost(self.target_list, self.catalog())
+>>>>>>> 842cc5f8 (fix: Catalog init introduces significant overhead  (#1270))