Skip to main content

Silver

Overview

Materialize a type-converted Delta table from bronze. With keephistory: false, silver is a flat typed copy of the latest bronze. With keephistory: true, silver maintains its own Silver.his history table populated from Bronze.his; silverloadtype selects either a full reload or an incremental merge into the existing silver table. Per-column type conversions and expressions live on the same per-table YAML as bronze.

What gets generated

StageComponentOutput
BuildEasyFabric Generator (EFG)GenerateFabricObjects → Fabric lakehouse tables (silver + history)
RuntimeEasyFabric Runtime (EFR)easyfabric.load_data_silver.run, easyfabric.load_data_silver.dataframeloader

Example YAML

Dataplatform/DP/Objects/AdvWorks/orders.yaml

  Table:
Connection: adv-advworks
SourceTable: orders.csv
DataPlatformObjectname: orders
PreBronzeNotebook:
Notebook: Pull_Github
Param001: orders.csv
Param002: orders.csv
KeepHistory: true
Columns:
- SourceColumn: SalesOrderID
SourceDataType: varchar(255)
IsPrimaryKey: true
- SourceColumn: LineItem
SourceDataType: int
IsPrimaryKey: true
- SourceColumn: OrderDate
SourceDataType: datetime
- SourceColumn: CustomerID
SourceDataType: int
- SourceColumn: ProductID
SourceDataType: int
- SourceColumn: OrderQty
SourceDataType: decimal(28,5)
- SourceColumn: LineItemTotal
SourceDataType: decimal(28,5)

Schema reference

Required fields marked *. Linked types are collapsible — click to expand inline.

Fabric Object

NameTypeDescription
Connection *StringConnection to use for this object (as defined in connection)
Fields *List<FabricAttribute>Fields from the source object
SourceTable *StringName of the object in the source
Show optional fields (9)
NameTypeDescription
BronzeTableSettingsFabricTableSettingsTable Settings for Bronze
DataPlatformObjectNameStringTo override the source name that is used in the dataplatform
DescriptionStringDescription of the source
IsActivetrue/falseSet to active for generating this object (default=true)
KeepHistorytrue/falseSet to true if history is required in the silver layer (default=true). Primary key required.
LayersString[]Which layers to generate
PrefixStringSet a prefix (default = '')
SilverTableSettingsFabricTableSettingsTable Settings for Silver
SourceSchemaStringSource schema in the source (if applicable)
Fabric Attribute
NameTypeDescription
SourceColumn *StringName of the attribute in the source system
Show optional fields (14)
NameTypeDescription
ClassificationClassificationTypeSet the classification of a field (sensitive, restricted, internal, public (default = public)
DataplatformnameStringTo override the name used in the dataplatform (default is the sourcecolumn name used)
HstgTypeStringType used in HSTG. Will try to convert a value from the source to this datatype. It will also keep the original value in a separate column.
IsActivetrue/falseAttribute is active (default=true)
IsNullabletrue/falsen/a
IsPartitiontrue/falseAttribute is part of the partition column (default = false)
IsPrimaryKeytrue/falseAttribute is part of the primary key (default = false)
IsSilverOnlytrue/falseAttribute is only generated for Silver (default = false)
SilverConvertedColumnNullable<true/false>If a SilverDataType is given, there is some kind of conversion happening. For the converted value a column is created that will have the converted value. Inherits FabricConfiguration.SilverConvertedColumn when unset. (Default=true)
SilverDataTypeStringDatatype used in the Silver layer.
SilverExpressionStringThe expression that is used to fill the Silver column with the changed datatype. Example to convert a datetime from string to datetime, use: "to_timestamp(your_string_column, 'dd-MM-yyyy HH:mm:ss')". If a SilverDataType is given, there is some kind of conversion happening.
SilverInitialColumnNullable<true/false>Create a Silver column holding the original (pre-conversion) value when a Silver conversion is set. Inherits FabricConfiguration.SilverInitialColumn when unset. (Default=true)
SourceDataTypeStringDatatype of the source (use source datatype, will be converted to dataplatform types automatically)
SourceExpressionStringThe expression that is used to fill the column in the Bronze layer with a conversion. Prefer to use the SilverExpression option. Potentially obsolete.

Runtime-only fields

These fields are read by the EFR at runtime and have no EFG counterpart.

  • keephistory
  • silverloadtype
  • silverdatatype
  • silverexpression
  • issilveronly
  • isdeletedidentifier

EasyFabric Runtime

load_data_silver.run

def run(tablefile: str,
config_manager: ConfigManager = None,
overrides: LoadOverrides = None) -> str | None

Executes the silver extraction and transformation process for a table, using the configuration information provided in the ConfigManager. This involves loading, processing, and merging table data within the silver lakehouse layer.

The function ensures the proper configuration of logging, verifies layer activity and table activation status, and invokes pre-silver and post-silver workflows if defined. It uses corresponding notebooks or direct data operations for the silver loading process.

Arguments:

  • tablefile str - Path to the YAML file containing table configuration details.
  • config_manager ConfigManager - An instance of ConfigManager, pre-initialized with application configuration, connection, and lakehouse details.
  • overrides LoadOverrides - Optional per-run overrides. Only skip_presilver / skip_postsilver apply to silver; the bronze hook flags, stale-check, and force-reload fields are bronze-only and ignored here.

Returns:

  • Optional[str] - An error message containing the name of the failed table and the exception details, or None if the operation finishes successfully.

Raises:

  • Exception - If ConfigManager is not initialized before invoking this function.
  • Exception - If the bronze lakehouse configuration is not found.
  • Exception - If the stop_at_error setting is enabled and an exception occurs in processing.

load_data_silver.dataframeloader

def dataframeloader(data_frame: DataFrame,
table_config: TableConfig,
load_config: LoadConfig,
config_manager: ConfigManager = None) -> str | None

Loads data into a silver layer table in a lakehouse environment.

This function facilitates loading data from a given DataFrame into a table specified by a table configuration within a silver layer of the lakehouse architecture. It uses the provided configuration details to establish connections, manage runtime settings, and log relevant information during the operation. It validates critical configurations and raises appropriate exceptions in case of missing or invalid details.

Arguments:

  • data_frame DataFrame - The input PySpark DataFrame containing data to load.
  • table_config TableConfig - Configuration object specifying table details and related connection configurations.
  • load_config LoadConfig - Configuration object holding load-specific settings, including the layer and runtime options.
  • config_manager ConfigManager - Centralized configuration management object used to retrieve connection settings and maintain runtime parameters.
LoadConfig fields

Runtime parameter bag — construct in code and pass to the loader. All fields are optional unless flagged below.

FieldTypeDescription
_layerstrOperational layer associated with the configuration. Defaults to "not set".
dry_runboolIndicates if the process should be executed in dry-run mode. Defaults to True.
auto_null_columnboolDetermines if null values should be automatically managed for columns. Defaults to True.
load_typeLoadTypeSpecifies the type of load operation. Defaults to LoadType.FULL.
stop_at_errorboolSpecifies whether the process should stop when an error occurs. Defaults to True.
business_key_checkboolIndicates if business keys should be validated during the load. Defaults to True.
log_row_countboolDeprecated and ignored. Row counts are always logged; retained for backward compatibility. Defaults to False.
key_violation_actionstrAction to be taken when key violations occur. Defaults to "raise".
load_history_tableboolWhen true, also load the object's history table alongside the main table. Defaults to False.
model_object_nameOptional[str]Specifies the model object name used in operations. Defaults to None.

Returns:

  • str - A string indicating the result of the load operation, either as success or an error message.

Raises:

  • Exception - If load_config is None, if ConfigManager is not properly initialized, or if no silver lakehouse configuration is found.
  • Exception - If there is an issue while performing the load operation and config_manager.stop_at_error is set to True.
  • easyfabric.data.TableConfig
  • easyfabric.data.Column