class ColumnTypeForm(DependencyAwareForm):
"""
Form used to specify a column during table creation
"""
dependencies = [
("column_type", "array", "array_type"),
("column_type", "map", "map_key_type"),
("column_type", "map", "map_value_type"),
]
column_name = common.HiveIdentifierField(required=True)
column_type = forms.ChoiceField(required=True,
choices=common.to_choices(sorted(HIVE_TYPES)),
initial=HIVE_TYPES[0])
array_type = forms.ChoiceField(required=False,
choices=common.to_choices(sorted(HIVE_PRIMITIVE_TYPES)),
initial=HIVE_PRIMITIVE_TYPES[0],
label="Array Value Type")
map_key_type = forms.ChoiceField(required=False,
choices=common.to_choices(sorted(HIVE_PRIMITIVE_TYPES)),
initial=HIVE_PRIMITIVE_TYPES[0],
help_text="Specify if column_type is map.")
map_value_type = forms.ChoiceField(required=False,
choices=common.to_choices(sorted(HIVE_PRIMITIVE_TYPES)),
initial=HIVE_PRIMITIVE_TYPES[0],
help_text="Specify if column_type is map.")
class CreateTableFromFileForm(forms.Form):
"""
Form used in the create table from file page
"""
# Basic Data
name = common.HiveIdentifierField(label="Table Name", required=False)
comment = forms.CharField(label="Description", required=False)
path = filebrowser.forms.PathField(label="Input File")
# csv/tsv files
encoding = UnicodeEncodingField()
delimiter = ChoiceOrOtherField(required=False, initial=TERMINATOR_CHOICES[0][0], choices=TERMINATOR_CHOICES)
replace_delimiter_with = ChoiceOrOtherField(required=False, initial=TERMINATOR_CHOICES[0][0], choices=TERMINATOR_CHOICES,
label="Replace delimiter with")
read_column_headers = forms.BooleanField(required=False, initial=True,
label="Read column headers",
help_text="")
import_data = forms.BooleanField(required=False, initial=True,
label="Import data",
help_text="")
autodetect_delimiter = forms.BooleanField(required=False, initial=True,
label="Autodetect delimiter",
help_text="")
ignore_whitespaces = forms.BooleanField(required=False, initial=False,
label="Ignore whitespaces",
help_text="")
ignore_tabs = forms.BooleanField(required=False, initial=False,
label="Ignore tabs",
help_text="")
single_line_comment = forms.CharField(label="Single line comment", required=False)
java_style_comments = forms.BooleanField(required=False, initial=False,
label="Java-style comments",
help_text="")
apply_excel_dialect = forms.BooleanField(required=False, initial=True,
label="Read column headers",
help_text="")
# xls/xlsx files
xls_sheet = ChoiceFieldExtended(label="Sheet", required=False)
xls_cell_range = forms.CharField(label="Cell range", required=False)
xls_read_column_headers = forms.BooleanField(required=False, initial=True,
label="Read column headers",
help_text="")
class CreateDatabaseForm(DependencyAwareForm):
"""
Form used in the create database page
"""
dependencies = []
# Basic Data
db_name = common.HiveIdentifierField(label="Database Name", required=True)
comment = forms.CharField(label="Description", required=False)
# External if not true
use_default_location = forms.BooleanField(required=False, initial=True, label="Use default location")
external_location = forms.CharField(required=False, help_text="Path to HDFS directory or file of database data.")
dependencies += [("use_default_location", False, "external_location")]
def __init__(self, *args, **kwargs):
DependencyAwareForm.__init__(self, *args, **kwargs)
self.database_list = []
def clean_db_name(self):
return _clean_databasename(self.database_list, self.cleaned_data['db_name'])
class PartitionTypeForm(forms.Form):
column_name = common.HiveIdentifierField(required=True)
column_type = forms.ChoiceField(required=True, choices=common.to_choices(sorted(HIVE_PRIMITIVE_TYPES)),
initial=HIVE_PRIMITIVE_TYPES[0])
class CreateTableForm(DependencyAwareForm):
"""
Form used in the create table page
"""
dependencies = []
# Basic Data
name = common.HiveIdentifierField(label="Table Name", required=True)
comment = forms.CharField(label="Description", required=False)
# Row Formatting
row_format = forms.ChoiceField(required=True,
choices=common.to_choices(["Delimited", "SerDe"]),
initial="Delimited")
# Delimited Row
# These initials are per LazySimpleSerDe.DefaultSeparators
field_terminator = ChoiceOrOtherField(required=False, initial=TERMINATOR_CHOICES[0][0],
choices=TERMINATOR_CHOICES)
collection_terminator = ChoiceOrOtherField(required=False, initial=TERMINATOR_CHOICES[1][0],
choices=TERMINATOR_CHOICES)
map_key_terminator = ChoiceOrOtherField(required=False, initial=TERMINATOR_CHOICES[2][0],
choices=TERMINATOR_CHOICES)
dependencies += [
("row_format", "Delimited", "field_terminator"),
("row_format", "Delimited", "collection_terminator"),
("row_format", "Delimited", "map_key_terminator"),
]
# Serde Row
serde_name = forms.CharField(required=False, label="SerDe Name")
serde_properties = forms.CharField(
required=False,
help_text="Comma-separated list of key-value pairs, eg., 'p1=v1, p2=v2'")
dependencies += [
("row_format", "SerDe", "serde_name"),
("row_format", "SerDe", "serde_properties"),
]
# File Format
file_format = forms.ChoiceField(required=False, initial="TextFile",
choices=common.to_choices(["TextFile", "SequenceFile", "InputFormat"]),
widget=forms.RadioSelect)
input_format_class = forms.CharField(required=False, label="InputFormat Class")
output_format_class = forms.CharField(required=False, label="OutputFormat Class")
dependencies += [
("file_format", "InputFormat", "input_format_class"),
("file_format", "InputFormat", "output_format_class"),
]
# External?
use_default_location = forms.BooleanField(required=False, initial=True,
label="Use default location")
external_location = forms.CharField(required=False, help_text="Path to HDFS directory or file of table data.")
dependencies += [
("use_default_location", False, "external_location")
]
def __init__(self, *args, **kwargs):
DependencyAwareForm.__init__(self, *args, **kwargs)
self.table_list = []
def clean_field_terminator(self):
return _clean_terminator(self.cleaned_data.get('field_terminator'))
def clean_collection_terminator(self):
return _clean_terminator(self.cleaned_data.get('collection_terminator'))
def clean_map_key_terminator(self):
return _clean_terminator(self.cleaned_data.get('map_key_terminator'))
def clean_name(self):
return _clean_tablename(self.table_list, self.cleaned_data['name'])