class CreateByImportFileForm(forms.Form):
"""Form for step 1 (specifying file) of the import wizard"""
# Basic Data
name = common.HiveIdentifierField(label=_t("Table Name"), required=True)
comment = forms.CharField(label=_t("Description"), required=False)
# File info
path = PathField(label=_t("Input File or Directory"))
load_data = forms.ChoiceField(required=True,
choices=[
("IMPORT", _("Import data")),
("EXTERNAL", _("Create External Table")),
("EMPTY", ("Leave Empty")),
],
help_text=_t("Select 'import' to load data from the file into the Hive warehouse directory after creation. "
"Select 'external' if the table is an external table and the data files should not be moved. " +
"Select 'empty' if the file should only be used to define the table schema but not loaded (table will be empty).")
)
def __init__(self, *args, **kwargs):
self.db = kwargs.pop('db', None)
super(CreateByImportFileForm, self).__init__(*args, **kwargs)
def clean_name(self):
return _clean_tablename(self.db, self.cleaned_data['name'])
def clean_path(self):
path = self.cleaned_data['path']
if path.lower().startswith(S3_ROOT):
path = path.lower().replace(S3_ROOT, S3A_ROOT)
if not path.endswith('/'):
path = '%s/' % path
return path
class SaveResultsForm(DependencyAwareForm):
"""Used for saving the query result data"""
SAVE_TYPES = (SAVE_TYPE_TBL, SAVE_TYPE_DIR) = (_('to a new table'),
_('to HDFS directory'))
save_target = forms.ChoiceField(required=True,
choices=common.to_choices(SAVE_TYPES),
widget=forms.RadioSelect)
target_table = common.HiveIdentifierField(
label=_t("Table Name"),
required=False,
help_text=_t("Name of the new table"))
target_dir = PathField(
label=_t("Results Location"),
required=False,
help_text=_t("Empty directory in HDFS to put the results"))
dependencies = [
('save_target', SAVE_TYPE_TBL, 'target_table'),
('save_target', SAVE_TYPE_DIR, 'target_dir'),
]
def clean_target_table(self):
tbl = self.cleaned_data.get('target_table')
if tbl:
try:
db_utils.meta_client().get_table("default", tbl)
raise forms.ValidationError(_('Table already exists'))
except hive_metastore.ttypes.NoSuchObjectException:
pass
return tbl
class LoadDataForm(forms.Form):
"""Form used for loading data into an existing table."""
path = PathField(label=_t("Path"))
overwrite = forms.BooleanField(required=False,
initial=False,
label=_t("Overwrite?"))
is_embeddable = forms.BooleanField(required=False, initial=False)
def __init__(self, table_obj, *args, **kwargs):
"""
@param table_obj is a hive_metastore.thrift Table object,
used to add fields corresponding to partition keys.
"""
super(LoadDataForm, self).__init__(*args, **kwargs)
self.partition_columns = dict()
for i, column in enumerate(table_obj.partition_keys):
# We give these numeric names because column names
# may be unpleasantly arbitrary.
name = "partition_%d" % i
char_field = forms.CharField(
required=True,
label=_t(
"%(column_name)s (partition key with type %(column_type)s)"
) % {
'column_name': column.name,
'column_type': column.type
})
self.fields[name] = char_field
self.partition_columns[name] = column.name
class SaveResultsForm(DependencyAwareForm):
"""Used for saving the query result data"""
SAVE_TYPES = (SAVE_TYPE_TBL, SAVE_TYPE_DIR) = ('to a new table',
'to HDFS directory')
save_target = forms.ChoiceField(required=True,
choices=common.to_choices(SAVE_TYPES),
widget=forms.RadioSelect,
initial=SAVE_TYPE_TBL)
target_table = common.HiveIdentifierField(
label=_t("Table Name"),
required=False,
help_text=_t("Name of the new table"))
target_dir = PathField(
label=_t("Results Location"),
required=False,
help_text=_t("Empty directory in HDFS to store results."))
dependencies = [
('save_target', SAVE_TYPE_TBL, 'target_table'),
('save_target', SAVE_TYPE_DIR, 'target_dir'),
]
def __init__(self, *args, **kwargs):
self.db = kwargs.pop('db', None)
self.fs = kwargs.pop('fs', None)
super(SaveResultsForm, self).__init__(*args, **kwargs)
def clean(self):
cleaned_data = super(SaveResultsForm, self).clean()
if cleaned_data.get('save_target') == SaveResultsForm.SAVE_TYPE_TBL:
tbl = cleaned_data.get('target_table')
if tbl:
try:
if self.db is not None:
self.db.get_table('default',
tbl) # Assumes 'default' DB
self._errors['target_table'] = self.error_class(
[_('Table already exists')])
del cleaned_data['target_table']
except Exception:
pass
elif cleaned_data['save_target'] == SaveResultsForm.SAVE_TYPE_DIR:
target_dir = cleaned_data['target_dir']
if not target_dir.startswith('/'):
self._errors['target_dir'] = self.error_class(
[_('Directory should start with /')])
elif self.fs.exists(target_dir):
self._errors['target_dir'] = self.error_class([
_('Directory already exists.')
]) # Overwrite destination directory content
return cleaned_data
class CreateByImportFileForm(forms.Form):
"""Form for step 1 (specifying file) of the import wizard"""
# Basic Data
name = common.HiveIdentifierField(label=_t("Table Name"), required=True)
comment = forms.CharField(label=_t("Description"), required=False)
# File info
path = PathField(label=_t("Input File"))
do_import = forms.BooleanField(
required=False,
initial=True,
label=_t("Import data from file"),
help_text=_t(
"Automatically load this file into the table after creation."))
def clean_name(self):
return _clean_tablename(self.cleaned_data['name'])