synapse-sdk 1.0.0a94__tar.gz → 1.0.0a96__tar.gz

{synapse_sdk-1.0.0a94 → synapse_sdk-1.0.0a96}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0a94
+Version: 1.0.0a96
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT

{synapse_sdk-1.0.0a94 → synapse_sdk-1.0.0a96}/synapse_sdk/plugins/categories/upload/actions/upload.py

@@ -24,6 +24,20 @@ from synapse_sdk.utils.pydantic.validators import non_blank
 from synapse_sdk.utils.storage import get_pathlib
 
 
+class PathAwareJSONEncoder(json.JSONEncoder):
+    """Custom JSON encoder that handles Path-like objects."""
+
+    def default(self, obj):
+        if hasattr(obj, '__fspath__') or hasattr(obj, 'as_posix'):
+            # Handle Path-like objects (including UPath, SFTPPath, pathlib.Path, etc.)
+            return str(obj)
+        elif hasattr(obj, 'isoformat'):
+            # Handle datetime objects
+            return obj.isoformat()
+        # Let the base class handle other types
+        return super().default(obj)
+
+
 class UploadStatus(str, Enum):
     SUCCESS = 'success'
     FAILED = 'failed'
@@ -67,7 +81,8 @@ class UploadRun(Run):
             status (UploadStatus): The status of the data file.
         """
         now = datetime.now().isoformat()
-        data_file_info_str = json.dumps(data_file_info, ensure_ascii=False)
+        # Use custom JSON encoder to handle Path-like objects
+        data_file_info_str = json.dumps(data_file_info, ensure_ascii=False, cls=PathAwareJSONEncoder)
         self.log(
             'upload_data_file',
             self.DataFileLog(data_file_info=data_file_info_str, status=status.value, created=now).model_dump(),
@@ -192,8 +207,7 @@ class UploadParams(BaseModel):
         storage (int): The storage of the action.
         collection (int): The collection of the action.
         project (int | None): The project of the action.
-        use_excel_metadata (bool): Whether to use excel file for additional metadata.
-        excel_metadata_path (str | None): Path to excel file containing metadata.
+        excel_metadata_path (str | None): Path to excel file containing metadata. Defaults to 'meta.xlsx' or 'meta.xls' in the path directory.
     """
 
     name: Annotated[str, AfterValidator(non_blank)]
@@ -202,7 +216,6 @@ class UploadParams(BaseModel):
     storage: int
     collection: int
     project: int | None
-    use_excel_metadata: bool = False
     excel_metadata_path: str | None = None
 
     @field_validator('storage', mode='before')
@@ -251,7 +264,7 @@ class UploadParams(BaseModel):
     @field_validator('excel_metadata_path', mode='before')
     @classmethod
     def check_excel_metadata_path(cls, value: str, info) -> str:
-        """Validate excel metadata file exists and is secure if use_excel_metadata is True.
+        """Validate excel metadata file exists and is secure if provided.
 
         This validator performs comprehensive security checks including:
         - File existence and format validation
@@ -271,48 +284,45 @@ class UploadParams(BaseModel):
         if not value:
             return value
 
-        # Check if use_excel_metadata is True
-        data = info.data
-        if data.get('use_excel_metadata', False):
-            excel_path = Path(value)
+        excel_path = Path(value)
 
-            # Check file existence
-            if not excel_path.exists():
-                raise PydanticCustomError('file_not_found', _('Excel metadata file not found.'))
-
-            # Check file extension
-            if excel_path.suffix.lower() not in ['.xlsx', '.xls']:
-                raise PydanticCustomError('invalid_file_type', _('Excel metadata file must be .xlsx or .xls format.'))
-
-            # Security check: file size limit
-            file_size = excel_path.stat().st_size
-            excel_config = ExcelSecurityConfig()
-            if file_size > excel_config.MAX_FILE_SIZE_BYTES:
-                raise PydanticCustomError(
-                    'file_too_large',
-                    _('Excel metadata file is too large. Maximum size is {}MB.').format(excel_config.MAX_FILE_SIZE_MB),
-                )
+        # Check file existence
+        if not excel_path.exists():
+            raise PydanticCustomError('file_not_found', _('Excel metadata file not found.'))
 
-            # Basic security check: ensure file is readable and not corrupted
-            try:
-                with open(excel_path, 'rb') as f:
-                    # Read first few bytes to check if it's a valid Excel file
-                    header = f.read(8)
-                    if not header:
-                        raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be empty.'))
-
-                    # Check for valid Excel file signatures
-                    if excel_path.suffix.lower() == '.xlsx':
-                        # XLSX files start with PK (ZIP signature)
-                        if not header.startswith(b'PK'):
-                            raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be corrupted.'))
-                    elif excel_path.suffix.lower() == '.xls':
-                        # XLS files have specific OLE signatures
-                        if not (header.startswith(b'\xd0\xcf\x11\xe0') or header.startswith(b'\x09\x08')):
-                            raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be corrupted.'))
-
-            except (OSError, IOError):
-                raise PydanticCustomError('file_access_error', _('Cannot access Excel metadata file.'))
+        # Check file extension
+        if excel_path.suffix.lower() not in ['.xlsx', '.xls']:
+            raise PydanticCustomError('invalid_file_type', _('Excel metadata file must be .xlsx or .xls format.'))
+
+        # Security check: file size limit
+        file_size = excel_path.stat().st_size
+        excel_config = ExcelSecurityConfig()
+        if file_size > excel_config.MAX_FILE_SIZE_BYTES:
+            raise PydanticCustomError(
+                'file_too_large',
+                _('Excel metadata file is too large. Maximum size is {}MB.').format(excel_config.MAX_FILE_SIZE_MB),
+            )
+
+        # Basic security check: ensure file is readable and not corrupted
+        try:
+            with open(excel_path, 'rb') as f:
+                # Read first few bytes to check if it's a valid Excel file
+                header = f.read(8)
+                if not header:
+                    raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be empty.'))
+
+                # Check for valid Excel file signatures
+                if excel_path.suffix.lower() == '.xlsx':
+                    # XLSX files start with PK (ZIP signature)
+                    if not header.startswith(b'PK'):
+                        raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be corrupted.'))
+                elif excel_path.suffix.lower() == '.xls':
+                    # XLS files have specific OLE signatures
+                    if not (header.startswith(b'\xd0\xcf\x11\xe0') or header.startswith(b'\x09\x08')):
+                        raise PydanticCustomError('invalid_file', _('Excel metadata file appears to be corrupted.'))
+
+        except (OSError, IOError):
+            raise PydanticCustomError('file_access_error', _('Cannot access Excel metadata file.'))
 
         return value
 
@@ -543,6 +553,24 @@ class UploadAction(Action):
         if row_count > self.excel_config.MAX_ROWS:
             raise ExcelParsingError(f'Too many rows: {row_count} (max: {self.excel_config.MAX_ROWS})')
 
+    def _find_excel_metadata_file(self, pathlib_cwd: Path) -> Optional[Path]:
+        """Find Excel metadata file in the directory.
+
+        Checks for meta.xlsx and meta.xls in the given directory.
+
+        Args:
+            pathlib_cwd (Path): The pathlib object representing the current working directory.
+
+        Returns:
+            Optional[Path]: Path to the Excel metadata file if found, None otherwise.
+        """
+        # Check for xlsx first, then xls
+        for extension in ['.xlsx', '.xls']:
+            excel_path = pathlib_cwd / f'meta{extension}'
+            if excel_path.exists() and excel_path.is_file():
+                return excel_path
+        return None
+
     def _read_excel_metadata(self, pathlib_cwd: Path) -> Dict[str, Dict[str, Any]]:
         """Read metadata from excel file with comprehensive security validation.
 
@@ -560,13 +588,20 @@ class UploadAction(Action):
             ExcelSecurityError: If security validation fails
             ExcelParsingError: If Excel content is invalid or exceeds limits
         """
-        if not self.params.get('use_excel_metadata', False) or not self.params.get('excel_metadata_path'):
-            return {}
+        excel_path = None
 
-        excel_path = pathlib_cwd / self.params['excel_metadata_path']
-        if not excel_path.exists():
-            self.run.log_message(f'Excel metadata file not found: {excel_path}', context=Context.WARNING.value)
-            return {}
+        # Check if user provided a specific excel_metadata_path
+        if self.params.get('excel_metadata_path'):
+            excel_path = pathlib_cwd / self.params['excel_metadata_path']
+            if not excel_path.exists():
+                self.run.log_message(f'Excel metadata file not found: {excel_path}', context=Context.WARNING.value)
+                return {}
+        else:
+            # Look for default meta.xlsx or meta.xls
+            excel_path = self._find_excel_metadata_file(pathlib_cwd)
+            if not excel_path:
+                # No Excel metadata file found, return empty dict (not an error)
+                return {}
 
         try:
             # Prepare Excel file with security validation
@@ -611,28 +646,26 @@ class UploadAction(Action):
         storage = self.client.get_storage(self.params['storage'])
         pathlib_cwd = get_pathlib(storage, self.params['path'])
 
-        # Read excel metadata if configured
+        # Read excel metadata if configured or default file exists
         excel_metadata: Dict[str, Dict[str, Any]] = {}
         try:
             excel_metadata = self._read_excel_metadata(pathlib_cwd)
             if excel_metadata:
                 self.run.log_message(f'Excel metadata loaded for {len(excel_metadata)} files')
-            elif self.params.get('use_excel_metadata', False):
-                self.run.log_message('Excel metadata enabled but no entries found')
-            # Don't log anything if Excel metadata is not being used
         except ExcelSecurityError as e:
             # Security violations should stop the process entirely
             self.run.log_message(f'Excel security validation failed: {str(e)}', context=Context.ERROR.value)
             self.run.log_message('Upload aborted due to Excel security concerns.', context=Context.ERROR.value)
             return result
         except ExcelParsingError as e:
-            # Parsing errors can be non-critical if Excel metadata is optional
-            if self.params.get('use_excel_metadata', False):
+            # Parsing errors can be non-critical if user didn't explicitly provide Excel file
+            if self.params.get('excel_metadata_path'):
+                # User explicitly provided Excel file, treat as error
                 self.run.log_message(f'Excel parsing failed: {str(e)}', context=Context.ERROR.value)
                 self.run.log_message('Upload aborted due to Excel parsing failure.', context=Context.ERROR.value)
                 return result
             else:
-                # If Excel metadata is not explicitly enabled, treat as warning and continue
+                # Default Excel file found but failed to parse, treat as warning and continue
                 self.run.log_message(f'Excel parsing failed (continuing): {str(e)}', context=Context.WARNING.value)
                 excel_metadata = {}
 

{synapse_sdk-1.0.0a94 → synapse_sdk-1.0.0a96}/synapse_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0a94
+Version: 1.0.0a96
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT