"""
return pckg.parameter_declaration(PARA_ROW,
name='Row',
data_type=pckg.DT_ROW_INDEX,
index=index)
VIZUAL_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_VIZUAL,
commands=[
pckg.command_declaration(
identifier=VIZUAL_DEL_COL,
name='Delete Column',
parameters=[pckg.para_dataset(0),
pckg.para_column(1)],
format=[
pckg.constant_format('DELETE'),
pckg.constant_format('COLUMN'),
pckg.variable_format(pckg.PARA_COLUMN),
pckg.constant_format('FROM'),
pckg.variable_format(pckg.PARA_DATASET)
]),
pckg.command_declaration(
identifier=VIZUAL_DEL_ROW,
name='Delete Row',
parameters=[pckg.para_dataset(0),
para_row_id(1)],
format=[
pckg.constant_format('DELETE'),
pckg.constant_format('ROW'),
pckg.variable_format(PARA_ROW),
pckg.constant_format('FROM'),
"""
return pckg.parameter_declaration(PARA_ROW,
name='Row',
data_type=pckg.DT_ROW_INDEX,
index=index)
VIZUAL_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_VIZUAL,
commands=[
pckg.command_declaration(
identifier=VIZUAL_DEL_COL,
name='Delete Column',
parameters=[pckg.para_dataset(0),
pckg.para_column(1)],
format=[
pckg.constant_format('DELETE'),
pckg.constant_format('COLUMN'),
pckg.variable_format(pckg.PARA_COLUMN),
pckg.constant_format('FROM'),
pckg.variable_format(pckg.PARA_DATASET)
]),
pckg.command_declaration(
identifier=VIZUAL_DEL_ROW,
name='Delete Row',
parameters=[pckg.para_dataset(0),
para_row_index(1)],
format=[
pckg.constant_format('DELETE'),
pckg.constant_format('ROW'),
pckg.variable_format(PARA_ROW),
pckg.constant_format('FROM'),
required=False,
hidden=True
)
MIMIR_LENSES = pckg.package_declaration(
identifier=PACKAGE_MIMIR,
commands=[
pckg.command_declaration(
identifier=MIMIR_DOMAIN,
name='Domain Lens',
parameters=[
pckg.para_dataset(0),
pckg.para_column(1),
para_materialize_input(2)
],
format=[
pckg.constant_format('DOMAIN'),
pckg.constant_format('FOR'),
pckg.variable_format(pckg.PARA_COLUMN),
pckg.constant_format('IN'),
pckg.variable_format(pckg.PARA_DATASET)
]
),
pckg.command_declaration(
identifier=MIMIR_GEOCODE,
name='Geocode Lens',
parameters=[
pckg.para_dataset(0),
pckg.parameter_declaration(
identifier=PARA_HOUSE_NUMBER,
name='House Nr.',
pckg.command_declaration(
identifier=PLOT_SIMPLE_CHART,
name='Simple Chart',
parameters=[
pckg.para_dataset(0),
pckg.parameter_declaration(identifier=pckg.PARA_NAME,
name='Chart Name',
data_type=pckg.DT_STRING,
index=1,
required=False),
pckg.parameter_declaration(identifier=PARA_SERIES,
name='Data Series',
data_type=pckg.DT_LIST,
index=2),
pckg.parameter_declaration(identifier=PARA_SERIES_COLUMN,
name='Column',
data_type=pckg.DT_COLUMN_ID,
index=3,
parent=PARA_SERIES),
pckg.parameter_declaration(identifier=PARA_SERIES_RANGE,
name='Range',
data_type=pckg.DT_STRING,
index=4,
parent=PARA_SERIES,
required=False),
pckg.parameter_declaration(identifier=PARA_SERIES_LABEL,
name='Label',
data_type=pckg.DT_STRING,
index=5,
parent=PARA_SERIES,
required=False),
pckg.parameter_declaration(identifier=PARA_XAXIS,
name='X-Axis',
data_type=pckg.DT_RECORD,
index=6,
required=False),
pckg.parameter_declaration(identifier=PARA_XAXIS_COLUMN,
name='Column',
data_type=pckg.DT_COLUMN_ID,
index=7,
parent=PARA_XAXIS,
required=False),
pckg.parameter_declaration(identifier=PARA_XAXIS_RANGE,
name='Range',
data_type=pckg.DT_STRING,
index=8,
parent=PARA_XAXIS,
required=False),
pckg.parameter_declaration(identifier=PARA_CHART,
name='Chart',
data_type=pckg.DT_RECORD,
index=9),
pckg.parameter_declaration(
identifier=PARA_CHART_TYPE,
name='Type',
data_type=pckg.DT_STRING,
index=10,
values=[
pckg.enum_value(value='Area Chart'),
pckg.enum_value(value='Bar Chart', is_default=True),
pckg.enum_value(value='Line Chart'),
pckg.enum_value(value='Scatter Plot')
],
parent=PARA_CHART),
pckg.parameter_declaration(identifier=PARA_CHART_GROUPED,
name='Grouped',
data_type=pckg.DT_BOOL,
index=11,
parent=PARA_CHART)
],
format=[
pckg.constant_format('CREATE'),
pckg.constant_format('PLOT'),
pckg.variable_format(pckg.PARA_NAME),
pckg.constant_format('FOR'),
pckg.variable_format(pckg.PARA_DATASET)
])
# Command identifier (unique within the package)
SCALA_CODE = 'code'
# Scala source code parameter
PARA_SCALA_SOURCE = 'source'
"""Define the scala cell command structure."""
SCALA_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_SCALA,
commands=[
pckg.command_declaration(
identifier=SCALA_CODE,
name='Scala Script',
parameters=[
pckg.parameter_declaration(identifier=PARA_SCALA_SOURCE,
name='Scala Code',
data_type=pckg.DT_CODE,
language='scala',
index=0)
],
format=[pckg.variable_format(PARA_SCALA_SOURCE)])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------
def export_package(filename, format='YAML'):
"""Write package specification to the given file.
pckg.command_declaration(
identifier=MIMIR_GEOCODE,
name='Geocode',
parameters=[
pckg.para_dataset(0),
pckg.parameter_declaration(identifier=PARA_HOUSE_NUMBER,
name='House Nr.',
data_type=pckg.DT_COLUMN_ID,
index=1,
required=False),
pckg.parameter_declaration(identifier=PARA_STREET,
name='Street',
data_type=pckg.DT_COLUMN_ID,
index=2,
required=False),
pckg.parameter_declaration(identifier=PARA_CITY,
name='City',
data_type=pckg.DT_COLUMN_ID,
index=3,
required=False),
pckg.parameter_declaration(identifier=PARA_STATE,
name='State',
data_type=pckg.DT_COLUMN_ID,
index=4,
required=False),
pckg.parameter_declaration(identifier=PARA_GEOCODER,
name='Geocoder',
data_type=pckg.DT_STRING,
index=5,
values=[
pckg.enum_value(
value='GOOGLE',
is_default=True),
pckg.enum_value(value='OSM')
]),
para_materialize_input(6)
],
format=[
pckg.constant_format('GEOCODE'),
pckg.variable_format(pckg.PARA_DATASET),
pckg.constant_format('COLUMNS'),
pckg.optional_format(PARA_HOUSE_NUMBER,
prefix='HOUSE_NUMBER='),
pckg.optional_format(PARA_STREET, prefix='STREET='),
pckg.optional_format(PARA_CITY, prefix='CITY='),
pckg.optional_format(PARA_STATE, prefix='STATE='),
pckg.constant_format('USING'),
pckg.variable_format(PARA_GEOCODER)
]),
# Command identifier (unique within the package)
PYTHON_CODE = 'code'
# Python source code parameter
PYTHON_SOURCE = 'source'
"""Define the python cell command structure."""
PYTHON_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_PYTHON,
commands=[
pckg.command_declaration(identifier=PYTHON_CODE,
name='Python Script',
parameters=[
pckg.parameter_declaration(
identifier=PYTHON_SOURCE,
name='Python Code',
data_type=pckg.DT_CODE,
language='python',
index=0)
],
format=[pckg.variable_format(PYTHON_SOURCE)])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------
def export_package(filename, format='YAML'):
"""Write package specification to the given file.
# Command identifier (unique within the package)
MARKDOWN_CODE = 'code'
# Markdown source code parameter
PARA_MARKDOWN_SOURCE = 'source'
"""Define the markdown cell command structure."""
MARKDOWN_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_MARKDOWN,
commands=[
pckg.command_declaration(
identifier=MARKDOWN_CODE,
name='Markdown Script',
parameters=[
pckg.parameter_declaration(identifier=PARA_MARKDOWN_SOURCE,
name='Markdown Code',
data_type=pckg.DT_CODE,
language='markdown',
index=0)
],
format=[pckg.variable_format(PARA_MARKDOWN_SOURCE)])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------
def export_package(filename, format='YAML'):
"""Write package specification to the given file.
# Format
OUTPUT_FORMAT = 'format'
"""Define the python cell command structure."""
PYTHON_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_PYTHON,
commands=[
pckg.command_declaration(
identifier=PYTHON_CODE,
name='Python Script',
parameters=[
pckg.parameter_declaration(identifier=PYTHON_SOURCE,
name='Python Code',
data_type=pckg.DT_CODE,
language='python',
index=0),
pckg.parameter_declaration(identifier=OUTPUT_FORMAT,
name='Output Format',
data_type=pckg.DT_STRING,
index=1,
default_value=OUTPUT_TEXT,
hidden=True,
required=False)
],
format=[pckg.variable_format(PYTHON_SOURCE)])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------
pckg.command_declaration(
identifier=VIZUAL_LOAD,
name='Load Dataset',
suggest=False,
parameters=[
pckg.parameter_declaration(identifier=pckg.PARA_NAME,
name='Dataset Name',
data_type=pckg.DT_STRING,
index=0),
pckg.parameter_declaration(identifier=PARA_FILE,
name='Source File',
data_type=pckg.DT_FILE_ID,
index=1),
pckg.parameter_declaration(
PARA_LOAD_FORMAT,
name='Load Format',
data_type=pckg.DT_STRING,
values=[
pckg.enum_value(value='csv',
text='CSV',
is_default=True),
pckg.enum_value(value='json', text='JSON'),
pckg.enum_value(
value='mimir.exec.spark.datasource.pdf',
text='PDF'),
pckg.enum_value(
value=
'mimir.exec.spark.datasource.google.spreadsheet',
text='Google Sheet'),
pckg.enum_value(value='com.databricks.spark.xml',
text='XML'),
pckg.enum_value(value='com.crealytics.spark.excel',
text='Excel'),
pckg.enum_value(value='jdbc', text='JDBC Source'),
pckg.enum_value(value='text', text='Text'),
pckg.enum_value(value='parquet', text='Parquet'),
pckg.enum_value(value='orc', text='ORC'),
pckg.enum_value(value='org.vizierdb.publish.source',
text='Published Dataset')
],
index=2,
required=True),
pckg.parameter_declaration(
PARA_SCHEMA,
name='Schema (leave blank to guess)',
data_type=pckg.DT_LIST,
index=3,
required=False),
pckg.parameter_declaration(PARA_SCHEMA_COLUMN,
name='Column Name',
data_type=pckg.DT_STRING,
index=4,
parent=PARA_SCHEMA,
required=False),
pckg.parameter_declaration(
PARA_SCHEMA_TYPE,
name='Data Type',
data_type=pckg.DT_STRING,
index=5,
parent=PARA_SCHEMA,
required=False,
values=[
pckg.enum_value(value=t,
text=t,
is_default=(t == "string"))
for t in [
"string", "real", "float", "boolean", "short",
"date", "timestamp", "int", "long", "byte"
]
]),
pckg.parameter_declaration(PARA_INFER_TYPES,
name='Infer Types',
data_type=pckg.DT_BOOL,
index=6,
default_value=True,
hidden=True,
required=False),
pckg.parameter_declaration(PARA_DETECT_HEADERS,
name='File Has Headers',
data_type=pckg.DT_BOOL,
index=7,
default_value=True,
required=False),
pckg.parameter_declaration(
PARA_LOAD_DSE,
name='Data Source Error Annotations',
data_type=pckg.DT_BOOL,
hidden=True,
index=8,
required=False),
pckg.parameter_declaration(PARA_LOAD_OPTIONS,
name='Load Options',
data_type=pckg.DT_LIST,
index=9,
required=False),
pckg.parameter_declaration(PARA_LOAD_OPTION_KEY,
name='Option Key',
data_type=pckg.DT_STRING,
index=10,
parent=PARA_LOAD_OPTIONS,
required=False),
pckg.parameter_declaration(PARA_LOAD_OPTION_VALUE,
name='Option Value',
data_type=pckg.DT_STRING,
index=11,
parent=PARA_LOAD_OPTIONS,
required=False)
],
format=[
pckg.constant_format('LOAD'),
pckg.constant_format('DATASET'),
pckg.variable_format(pckg.PARA_NAME),
pckg.constant_format('FROM'),
pckg.variable_format(PARA_FILE)
],
),
# Command identifier (unique within the package)
R_CODE = 'code'
# R source code parameter
PARA_R_SOURCE = 'source'
"""Define the r cell command structure."""
R_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_R,
commands=[
pckg.command_declaration(identifier=R_CODE,
name='R Script',
parameters=[
pckg.parameter_declaration(
identifier=PARA_R_SOURCE,
name='R Code',
data_type=pckg.DT_CODE,
language='r',
index=0)
],
format=[pckg.variable_format(PARA_R_SOURCE)])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------
def export_package(filename, format='YAML'):
"""Write package specification to the given file.
PARA_OUTPUT_DATASET = 'output_dataset'
PARA_SQL_SOURCE = 'source'
"""Define SQL command structure."""
SQL_COMMANDS = pckg.package_declaration(
identifier=PACKAGE_SQL,
commands=[
pckg.command_declaration(
identifier=SQL_QUERY,
name='SQL Query',
parameters=[
pckg.parameter_declaration(identifier=PARA_SQL_SOURCE,
name='SQL Code',
data_type=pckg.DT_CODE,
language='sql',
index=0),
pckg.parameter_declaration(identifier=PARA_OUTPUT_DATASET,
name='Output Dataset',
data_type=pckg.DT_STRING,
index=1,
required=False)
],
format=[
pckg.variable_format(PARA_SQL_SOURCE),
pckg.optional_format(PARA_OUTPUT_DATASET, prefix='AS ')
])
])
# ------------------------------------------------------------------------------
# Helper Methods
# ------------------------------------------------------------------------------