awslabs.s3-tables-mcp-server 0.0.13__py3-none-any.whl → 0.0.15__py3-none-any.whl

awslabs/s3_tables_mcp_server/__init__.py

@@ -15,4 +15,4 @@
 # This file is part of the awslabs namespace.
 # It is intentionally minimal to support PEP 420 namespace packages.
 
-__version__ = '0.0.13'
+__version__ = '0.0.15'

awslabs/s3_tables_mcp_server/engines/pyiceberg.py

@@ -14,19 +14,118 @@
 
 """Engine for interacting with Iceberg tables using pyiceberg and daft (read-only)."""
 
-import io
-import json
 import pyarrow as pa
-import pyarrow.json as pj
 from ..utils import pyiceberg_load_catalog
 from daft import Catalog as DaftCatalog
 from daft.session import Session
+from datetime import datetime
 from pydantic import BaseModel
 
 # pyiceberg and daft imports
 from typing import Any, Dict, Optional
 
 
+def convert_temporal_fields(rows: list[dict], arrow_schema: pa.Schema) -> list[dict]:
+    """Convert string temporal fields to appropriate datetime objects based on Arrow schema.
+
+    Args:
+        rows: List of row dictionaries with string temporal values
+        arrow_schema: PyArrow schema defining field types
+
+    Returns:
+        List of row dictionaries with converted temporal values
+    """
+    converted_rows = []
+
+    for row in rows:
+        converted_row = {}
+        for field_name, value in row.items():
+            # Early skip for non-string values
+            if not isinstance(value, str):
+                converted_row[field_name] = value
+                continue
+
+            # Get the field type from schema
+            field = arrow_schema.field(field_name)
+            field_type = field.type
+
+            # Date32 or Date64 - calendar date without timezone or time
+            if pa.types.is_date(field_type):
+                # Format: "2025-03-14"
+                converted_row[field_name] = datetime.strptime(value, '%Y-%m-%d').date()
+
+            # Time64 - time of day, microsecond precision, without date or timezone
+            elif pa.types.is_time(field_type):
+                # Format: "17:10:34.123456" or "17:10:34"
+                fmt = '%H:%M:%S.%f' if '.' in value else '%H:%M:%S'
+                converted_row[field_name] = datetime.strptime(value, fmt).time()
+
+            # Timestamp without timezone
+            elif pa.types.is_timestamp(field_type) and field_type.tz is None:
+                # Format: "2025-03-14 17:10:34.123456" or "2025-03-14T17:10:34.123456"
+                value_normalized = value.replace('T', ' ')
+                if '.' in value_normalized:
+                    # Truncate nanoseconds to microseconds if needed
+                    parts = value_normalized.split('.')
+                    if len(parts[1]) > 6:
+                        value_normalized = f'{parts[0]}.{parts[1][:6]}'
+                    fmt = '%Y-%m-%d %H:%M:%S.%f'
+                else:
+                    fmt = '%Y-%m-%d %H:%M:%S'
+                converted_row[field_name] = datetime.strptime(value_normalized, fmt)
+
+            # Timestamp with timezone (stored in UTC)
+            elif pa.types.is_timestamp(field_type) and field_type.tz is not None:
+                # Format: "2025-03-14 17:10:34.123456-07" or "2025-03-14T17:10:34.123456+00:00"
+                value_normalized = value.replace('T', ' ')
+                from datetime import timezone
+
+                # Truncate nanoseconds to microseconds if present
+                if '.' in value_normalized:
+                    # Split on timezone indicator (+ or -)
+                    # Find the last occurrence of + or - which should be the timezone
+                    tz_idx = max(value_normalized.rfind('+'), value_normalized.rfind('-'))
+                    if tz_idx > 10:  # Make sure it's not the date separator
+                        timestamp_part = value_normalized[:tz_idx]
+                        tz_part = value_normalized[tz_idx:]
+
+                        # Truncate fractional seconds to 6 digits
+                        if '.' in timestamp_part:
+                            parts = timestamp_part.split('.')
+                            if len(parts[1]) > 6:
+                                timestamp_part = f'{parts[0]}.{parts[1][:6]}'
+
+                        value_normalized = timestamp_part + tz_part
+
+                # Try different timezone formats
+                for fmt in [
+                    '%Y-%m-%d %H:%M:%S.%f%z',
+                    '%Y-%m-%d %H:%M:%S%z',
+                    '%Y-%m-%d %H:%M:%S.%f',
+                    '%Y-%m-%d %H:%M:%S',
+                ]:
+                    try:
+                        dt = datetime.strptime(value_normalized, fmt)
+                        if dt.tzinfo is None:
+                            dt = dt.replace(tzinfo=timezone.utc)
+                        converted_row[field_name] = dt.astimezone(timezone.utc)
+                        break
+                    except ValueError:
+                        continue
+                else:
+                    raise ValueError(
+                        f'Could not parse timestamp with timezone: {value} for field {field_name}'
+                    )
+
+            else:
+                # Not a temporal field, keep as is
+                converted_row[field_name] = value
+
+        converted_rows.append(converted_row)
+
+    return converted_rows
+
+
 class PyIcebergConfig(BaseModel):
     """Configuration for PyIceberg/Daft connection."""
 
@@ -107,7 +206,7 @@ class PyIcebergEngine:
             return False
 
     def append_rows(self, table_name: str, rows: list[dict]) -> None:
-        """Append rows to an Iceberg table using pyiceberg with JSON encoding.
+        """Append rows to an Iceberg table using pyiceberg.
 
         Args:
             table_name: The name of the table (e.g., 'namespace.tablename' or just 'tablename' if namespace is set)
@@ -127,28 +226,23 @@ class PyIcebergEngine:
 
             # Load the Iceberg table
             table = self._catalog.load_table(full_table_name)
-            # Encode rows as JSON (line-delimited format)
-            json_lines = []
-            for row in rows:
-                json_lines.append(json.dumps(row))
-            json_data = '\n'.join(json_lines)
 
-            # Create a file-like object from the JSON data
-            json_buffer = io.BytesIO(json_data.encode('utf-8'))
+            # Convert Iceberg schema to Arrow schema to ensure types/order match
+            arrow_schema = table.schema().as_arrow()
 
-            # Read JSON data into PyArrow Table using pyarrow.json.read_json
-            # This enforces the Iceberg schema and validates the data
+            # Convert temporal fields from strings to datetime objects
+            converted_rows = convert_temporal_fields(rows, arrow_schema)
+
+            # Create PyArrow table directly from pylist with schema validation
             try:
-                new_data_table = pj.read_json(
-                    json_buffer, read_options=pj.ReadOptions(use_threads=True)
-                )
+                pa_table = pa.Table.from_pylist(converted_rows, schema=arrow_schema)
             except pa.ArrowInvalid as e:
                 raise ValueError(
                     f'Schema mismatch detected: {e}. Please ensure your data matches the table schema.'
                 )
 
-            # Append the new data to the Iceberg table
-            table.append(new_data_table)
+            # Append the PyArrow table to the Iceberg table
+            table.append(pa_table)
 
         except Exception as e:
             raise Exception(f'Error appending rows: {str(e)}')

awslabs/s3_tables_mcp_server/file_processor/csv.py

@@ -34,7 +34,7 @@ async def import_csv_to_table(
     rest_sigv4_enabled: str = 'true',
     preserve_case: bool = False,
 ):
-    """Import a CSV file into an S3 table using PyArrow."""
+    """Import a CSV file into an existing S3 table using PyArrow."""
     return await import_file_to_table(
         warehouse=warehouse,
         region=region,

awslabs/s3_tables_mcp_server/file_processor/utils.py

@@ -20,6 +20,7 @@ particularly focusing on column name conversion and schema transformation.
 
 import os
 import pyarrow as pa
+import pyarrow.compute as pc
 from ..utils import get_s3_client, pyiceberg_load_catalog
 from io import BytesIO
 from pydantic.alias_generators import to_snake
@@ -64,6 +65,44 @@ def convert_column_names_to_snake_case(schema: pa.Schema) -> pa.Schema:
     return pa.schema(new_fields, metadata=schema.metadata)
 
 
+def convert_temporal_fields_in_table(
+    pyarrow_table: pa.Table, target_schema: pa.Schema
+) -> pa.Table:
+    """Convert string temporal fields in PyArrow table to appropriate temporal types.
+
+    Args:
+        pyarrow_table: PyArrow table with string temporal values
+        target_schema: Target schema with temporal field types
+
+    Returns:
+        PyArrow table with converted temporal columns
+    """
+    # Use PyArrow's cast which can handle ISO 8601 formatted strings
+    # This is simpler and more robust than strptime for mixed formats
+    try:
+        # Try direct cast - PyArrow can parse ISO 8601 strings automatically
+        converted_table = pyarrow_table.cast(target_schema, safe=False)
+        return converted_table
+    except pa.ArrowInvalid:
+        # If direct cast fails, fall back to column-by-column conversion
+        arrays = []
+        for i, field in enumerate(target_schema):
+            col_name = field.name
+            col_data = pyarrow_table.column(col_name)
+            field_type = field.type
+
+            # Try to cast the column to the target type
+            try:
+                col_data = pc.cast(col_data, field_type, safe=False)
+            except pa.ArrowInvalid:
+                # If cast fails, keep original data
+                pass
+
+            arrays.append(col_data)
+
+        return pa.Table.from_arrays(arrays, schema=target_schema)
+
+
 async def import_file_to_table(
     warehouse: str,
     region: str,
@@ -117,29 +156,30 @@ async def import_file_to_table(
                     'error': f'Column name conversion failed: {str(conv_err)}',
                 }
 
-        table_created = False
         try:
             # Try to load existing table
             table = catalog.load_table(f'{namespace}.{table_name}')
+            # Convert temporal fields to match existing table schema
+            target_schema = table.schema().as_arrow()
+            pyarrow_table = convert_temporal_fields_in_table(pyarrow_table, target_schema)
         except NoSuchTableError:
-            # Table doesn't exist, create it using the schema
-            try:
-                table = catalog.create_table(
-                    identifier=f'{namespace}.{table_name}',
-                    schema=pyarrow_schema,
-                )
-                table_created = True
-            except Exception as create_error:
-                return {
-                    'status': 'error',
-                    'error': f'Failed to create table: {str(create_error)}',
-                }
+            # Table doesn't exist - return error with schema information
+            # Build column information from the source file schema
+            columns_info = []
+            for field in pyarrow_schema:
+                columns_info.append({'name': field.name, 'type': str(field.type)})
+
+            return {
+                'status': 'error',
+                'error': f'Table {namespace}.{table_name} does not exist. Please create the table first before importing data.',
+                'columns': columns_info,
+            }
 
         # Append data to Iceberg table
         table.append(pyarrow_table)
 
         # Build message with warnings if applicable
-        message = f'Successfully imported {pyarrow_table.num_rows} rows{" and created new table" if table_created else ""}'
+        message = f'Successfully imported {pyarrow_table.num_rows} rows'
         if columns_converted:
             message += '. WARNING: Column names were converted to snake_case format. To preserve the original case, set preserve_case to True.'
 
@@ -148,7 +188,6 @@ async def import_file_to_table(
             'message': message,
             'rows_processed': pyarrow_table.num_rows,
             'file_processed': os.path.basename(key),
-            'table_created': table_created,
             'table_uuid': table.metadata.table_uuid,
             'columns': pyarrow_schema.names,
         }

awslabs/s3_tables_mcp_server/server.py

@@ -307,45 +307,115 @@ async def create_table(
     """Create a new S3 table in an S3 table bucket.
 
     Creates a new S3 table associated with the given S3 namespace in an S3 table bucket.
-    The S3 table can be configured with specific format and metadata settings. Metadata contains the schema of the table. Use double type for decimals.
+    The S3 table can be configured with specific format and metadata settings. Metadata contains the schema of the table.
     Do not use the metadata parameter if the schema is unclear.
 
+    Supported Iceberg Primitive Types:
+    - boolean: True or false
+    - int: 32-bit signed integers (can promote to long)
+    - long: 64-bit signed integers
+    - float: 32-bit IEEE 754 floating point (can promote to double)
+    - double: 64-bit IEEE 754 floating point
+    - decimal(P,S): Fixed-point decimal with precision P and scale S (precision must be 38 or less)
+    - date: Calendar date without timezone or time
+    - time: Time of day, microsecond precision, without date or timezone
+    - timestamp: Timestamp, microsecond precision, without timezone (represents date and time regardless of zone)
+    - timestamptz: Timestamp, microsecond precision, with timezone (stored as UTC)
+    - string: Arbitrary-length character sequences (UTF-8 encoded)
+
+    Note: Binary field types (binary, fixed, uuid) are not supported.
+
     Example of S3 table metadata:
     {
         "metadata": {
             "iceberg": {
                 "schema": {
                     "type": "struct",
-                    "fields": [{
+                    "fields": [
+                        {
                             "id": 1,
-                            "name": "customer_id",
+                            "name": "id",
                             "type": "long",
                             "required": true
                         },
                         {
                             "id": 2,
-                            "name": "customer_name",
-                            "type": "string",
-                            "required": true
+                            "name": "bool_field",
+                            "type": "boolean",
+                            "required": false
                         },
                         {
                             "id": 3,
-                            "name": "customer_balance",
+                            "name": "int_field",
+                            "type": "int",
+                            "required": false
+                        },
+                        {
+                            "id": 4,
+                            "name": "long_field",
+                            "type": "long",
+                            "required": false
+                        },
+                        {
+                            "id": 5,
+                            "name": "float_field",
+                            "type": "float",
+                            "required": false
+                        },
+                        {
+                            "id": 6,
+                            "name": "double_field",
                             "type": "double",
                             "required": false
+                        },
+                        {
+                            "id": 7,
+                            "name": "decimal_field",
+                            "type": "decimal(10,2)",
+                            "required": false
+                        },
+                        {
+                            "id": 8,
+                            "name": "date_field",
+                            "type": "date",
+                            "required": false
+                        },
+                        {
+                            "id": 9,
+                            "name": "time_field",
+                            "type": "time",
+                            "required": false
+                        },
+                        {
+                            "id": 10,
+                            "name": "timestamp_field",
+                            "type": "timestamp",
+                            "required": false
+                        },
+                        {
+                            "id": 11,
+                            "name": "timestamptz_field",
+                            "type": "timestamptz",
+                            "required": false
+                        },
+                        {
+                            "id": 12,
+                            "name": "string_field",
+                            "type": "string",
+                            "required": false
                         }
                     ]
                 },
                 "partition-spec": [
                     {
-                        "source-id": 1,
+                        "source-id": 8,
                         "field-id": 1000,
                         "transform": "month",
-                        "name": "sale_date_month"
+                        "name": "date_field_month"
                     }
                 ],
                 "table-properties": {
-                    "description": "Customer information table with customer_id for joining with transactions"
+                    "description": "Example table demonstrating supported Iceberg primitive types"
                 }
             }
         }
@@ -353,7 +423,6 @@ async def create_table(
 
     Permissions:
     You must have the s3tables:CreateTable permission to use this operation.
-    If using metadata parameter, you must have the s3tables:PutTableData permission.
     """
     from awslabs.s3_tables_mcp_server.models import OpenTableFormat, TableMetadata
 
@@ -665,11 +734,10 @@ async def import_parquet_to_table(
         bool, Field(..., description='Preserve case of column names')
     ] = False,
 ) -> dict:
-    """Import data from a Parquet file into an S3 table.
+    """Import data from a Parquet file into an existing S3 table.
 
-    This tool reads data from a Parquet file stored in S3 and imports it into an S3 table.
-    If the table doesn't exist, it will be created with a schema inferred from the Parquet file.
-    If the table exists, the Parquet file schema must be compatible with the table's schema.
+    This tool reads data from a Parquet file stored in S3 and imports it into an existing S3 table.
+    The table must already exist. The Parquet file schema must be compatible with the table's schema.
     The tool will validate the schema before attempting to import the data.
     If preserve_case is True, the column names will not be converted to snake_case. Otherwise, the column names will be converted to snake_case.
 
@@ -677,6 +745,7 @@ async def import_parquet_to_table(
         - URL is not a valid S3 URL
         - File is not a Parquet file
         - File cannot be accessed
+        - Table does not exist
         - Parquet schema is incompatible with existing table schema
         - Any other error occurs
 
@@ -685,7 +754,6 @@ async def import_parquet_to_table(
         - message: Success message with row count
         - rows_processed: Number of rows imported
         - file_processed: Name of the processed file
-        - table_created: True if a new table was created
 
     Example input values:
         warehouse: 'arn:aws:s3tables:<Region>:<accountID>:bucket/<bucketname>'
@@ -704,7 +772,6 @@ async def import_parquet_to_table(
     - s3:GetObject permission for the Parquet file
     - s3tables:GetTable and s3tables:GetTables permissions to access table information
     - s3tables:PutTableData permission to write to the table
-    - s3tables:CreateTable permission (if table doesn't exist)
     """
     if uri is None:
         uri = _default_uri_for_region(region)

{awslabs_s3_tables_mcp_server-0.0.13.dist-info → awslabs_s3_tables_mcp_server-0.0.15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: awslabs.s3-tables-mcp-server
-Version: 0.0.13
+Version: 0.0.15
 Summary: An AWS Labs Model Context Protocol (MCP) server for awslabs.s3-tables-mcp-server
 Project-URL: homepage, https://awslabs.github.io/mcp/
 Project-URL: docs, https://awslabs.github.io/mcp/servers/s3-tables-mcp-server/
@@ -25,9 +25,9 @@ Requires-Dist: boto3==1.40.8
 Requires-Dist: daft==0.5.8
 Requires-Dist: loguru==0.7.3
 Requires-Dist: mcp[cli]==1.11.0
-Requires-Dist: pyarrow==20.0.0
+Requires-Dist: pyarrow==22.0.0
 Requires-Dist: pydantic==2.9.2
-Requires-Dist: pyiceberg==0.9.1
+Requires-Dist: pyiceberg==0.10.0
 Requires-Dist: sqlparse==0.5.3
 Description-Content-Type: text/markdown
 

{awslabs_s3_tables_mcp_server-0.0.13.dist-info → awslabs_s3_tables_mcp_server-0.0.15.dist-info}/RECORD

@@ -1,24 +1,24 @@
 awslabs/__init__.py,sha256=BHp8_uaBohApyLlmVWvYVe5bSrH59FvLJ5cNTigMV_8,644
-awslabs/s3_tables_mcp_server/__init__.py,sha256=QeZWLpBaBNyp4RSHkvo54jorNB64Sjc0DbXfQ6ZYp3g,755
+awslabs/s3_tables_mcp_server/__init__.py,sha256=msY9peBnKjuzjURfUobE4kUverHTk69z9sQNh_cvdbk,755
 awslabs/s3_tables_mcp_server/constants.py,sha256=qCWY9A9PAQXdVz-anO26zbQ72Dp79nGM7xeLR062a_o,4971
 awslabs/s3_tables_mcp_server/database.py,sha256=YorxcSx-9typfQ5W_LzwNPZkP47u__QSLJlp0fBsZLg,3851
 awslabs/s3_tables_mcp_server/models.py,sha256=zWTFJLBhIZRLEgOCTyNcGvbItxqYbFJKH6se1EzXDjY,8097
 awslabs/s3_tables_mcp_server/namespaces.py,sha256=KZqxJiEnlpxkqvbfygezbr0szwyDP2O0J6osyiPUzwg,2071
 awslabs/s3_tables_mcp_server/resources.py,sha256=PXZo0sTVn34tXJ4mlw_OS90p12SNoLZs4Re0gV815wk,8281
 awslabs/s3_tables_mcp_server/s3_operations.py,sha256=-XS57fGBPryJ8ytfE-poNEEgnhWx2QSWyqpKiYusRuA,1645
-awslabs/s3_tables_mcp_server/server.py,sha256=dP3dv-4o-Dq4mDU7jt1n7Rl9eCo7TVvKwth76b06C3o,32630
+awslabs/s3_tables_mcp_server/server.py,sha256=ER-dJ0-ynrODs3YCGVkbjsNPOOI2AvRAOJALdQ_z7Z0,35301
 awslabs/s3_tables_mcp_server/table_buckets.py,sha256=JHmpB_P9h0Hz5Uis25_GPTD1G-mIODVwjaswwIGyCS4,4471
 awslabs/s3_tables_mcp_server/tables.py,sha256=ITnRDHHrtRWLsRhff4TP4B7gGT_jRXy994oxK3x10a4,10143
 awslabs/s3_tables_mcp_server/utils.py,sha256=SReyS3KsdikI9ycL5RsvtVI7MiRnA1W9bTiXGKf1lHc,4517
 awslabs/s3_tables_mcp_server/engines/__init__.py,sha256=O4wlFva3THWmjfaXfJAwi29mxJSKIhM0jcebVfd3S5U,615
-awslabs/s3_tables_mcp_server/engines/pyiceberg.py,sha256=wzkySQZgx7L8Fn2Oqk8Yz4V-hQZDUempd8q0IwvCY_4,5784
+awslabs/s3_tables_mcp_server/engines/pyiceberg.py,sha256=VS0eBEyL_BDP00htF6nrOcLVbf0nZgc7iGAKCEIvpVw,10017
 awslabs/s3_tables_mcp_server/file_processor/__init__.py,sha256=8PeggFRY3ZKBdxcFPEqSSHkSJBZ57eOs-z0fqkMHn9E,978
-awslabs/s3_tables_mcp_server/file_processor/csv.py,sha256=1JfFFKbnK1qvDvpOKWoXIv9zca8MRlPXk2Phd7-GgIU,1650
+awslabs/s3_tables_mcp_server/file_processor/csv.py,sha256=XHp_OwMblqbkd051Ctro4yIZnsre4Ea6A5pVvK_UqKc,1659
 awslabs/s3_tables_mcp_server/file_processor/parquet.py,sha256=TA9HmVPAnwOc3hv3zsL7J7sRd8gMINGFp5_r1eX7uUE,1458
-awslabs/s3_tables_mcp_server/file_processor/utils.py,sha256=MkNE0c1tlR_bs5ZeCTqAuRCLQWOc3mJqnJxATbGwoEc,5613
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/METADATA,sha256=Vr4lcaFTJqOfJbWKGKWzqK41fHF8vqX37ZefOfP4spg,12096
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/entry_points.txt,sha256=WRA45Bi2dVY5hskxkka_e7BAGRqG1KiW3ImTBnHSyLs,90
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/licenses/LICENSE,sha256=CeipvOyAZxBGUsFoaFqwkx54aPnIKEtm9a5u2uXxEws,10142
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/licenses/NOTICE,sha256=jm-1A_8i-wl7KYs2Ynj2A29vXWJCMKLHmGfy4P_B51Y,96
-awslabs_s3_tables_mcp_server-0.0.13.dist-info/RECORD,,
+awslabs/s3_tables_mcp_server/file_processor/utils.py,sha256=QJP-ddetCIqChPC6vz2_sz2RlEsSz6WJqHmSO4fOCrc,7167
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/METADATA,sha256=1ZygPK1VrYl0J9yp-Ah2yvY3TlQ8SolEX6tJdZZE8-4,12097
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/entry_points.txt,sha256=WRA45Bi2dVY5hskxkka_e7BAGRqG1KiW3ImTBnHSyLs,90
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/licenses/LICENSE,sha256=CeipvOyAZxBGUsFoaFqwkx54aPnIKEtm9a5u2uXxEws,10142
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/licenses/NOTICE,sha256=jm-1A_8i-wl7KYs2Ynj2A29vXWJCMKLHmGfy4P_B51Y,96
+awslabs_s3_tables_mcp_server-0.0.15.dist-info/RECORD,,