def get_cols(self, lines):
"""Initialize the header Column objects from the table ``lines``.
Based on the previously set Header attributes find or create the column names.
Sets ``self.cols`` with the list of Columns. This list only includes the actual
requested columns after filtering by the include_names and exclude_names
attributes. See ``self.names`` for the full list.
:param lines: list of table lines
:returns: list of table Columns
"""
if self.names is None:
# No column names supplied so first try to get them from NumPy structured array
try:
self.names = lines.dtype.names
except AttributeError:
# Still no col names available so auto-generate them
try:
first_data_vals = next(iter(lines))
except StopIteration:
raise core.InconsistentTableError('No data lines found so cannot autogenerate column names')
n_data_cols = len(first_data_vals)
self.names = [self.auto_format % i for i in range(1, n_data_cols+1)]
names = set(self.names)
if self.include_names is not None:
names.intersection_update(self.include_names)
if self.exclude_names is not None:
names.difference_update(self.exclude_names)
self.cols = [core.Column(name=x, index=i) for i, x in enumerate(self.names)
if x in names]
# ``lines`` could be one of: NumPy recarray, DictLikeNumpy obj, Python
# list of lists. If NumPy recarray then set col.type accordingly. In
# the other two cases convert the data values to strings so the usual
# data converter processing will get the correct type.
if core.has_numpy and isinstance(lines, numpy.ndarray):
for col in self.cols:
type_name = lines[col.name].dtype.name
if 'int' in type_name:
col.type = core.IntType
elif 'float' in type_name:
col.type = core.FloatType
elif 'string' in type_name:
col.type = core.StrType
else:
basic_reader = core._get_reader(Reader=basic.BasicReader,
names=[col.name for col in self.cols], quotechar="'")
data_in = tuple(' '.join(repr(vals[col.index]) for col in self.cols)
for vals in lines)
data_out = basic_reader.read(data_in)
for col, data_col in zip(self.cols, basic_reader.cols):
col.type = data_col.type
def get_cols(self, lines):
"""Initialize the header Column objects from the table ``lines``.
Based on the previously set Header attributes find or create the column names.
Sets ``self.cols`` with the list of Columns. This list only includes the actual
requested columns after filtering by the include_names and exclude_names
attributes. See ``self.names`` for the full list.
:param lines: list of table lines
:returns: list of table Columns
"""
if self.names is None:
# No column names supplied so first try to get them from NumPy structured array
try:
self.names = lines.dtype.names
except AttributeError:
# Still no col names available so auto-generate them
try:
first_data_vals = next(iter(lines))
except StopIteration:
raise core.InconsistentTableError(
'No data lines found so cannot autogenerate column names'
)
n_data_cols = len(first_data_vals)
self.names = [
self.auto_format % i for i in range(1, n_data_cols + 1)
]
self._set_cols_from_names()
# ``lines`` could be one of: NumPy recarray, DictLikeNumpy obj, Python
# list of lists. If NumPy recarray then set col.type accordingly. In
# the other two cases convert the data values to strings so the usual
# data converter processing will get the correct type.
if core.has_numpy and isinstance(lines, numpy.ndarray):
for col in self.cols:
type_name = lines[col.name].dtype.name
if 'int' in type_name:
col.type = core.IntType
elif 'float' in type_name:
col.type = core.FloatType
elif 'str' in type_name:
col.type = core.StrType
else:
# lines is a list of lists or DictLikeNumpy.
col_types = {}
col_indexes = [col.index for col in self.cols]
for vals in lines:
for col_index in col_indexes:
val = vals[col_index]
col_type_set = col_types.setdefault(col_index, set())
col_type_set.add(get_val_type(val))
for col in self.cols:
col.type = get_lowest_type(col_types[col.index])
def get_lines(self, table, names):
"""Get the lines from the ``table`` input.
:param table: table input
:param names: list of column names (only used for dict of lists to set column order)
:returns: list of lines
"""
try:
# If table is dict-like this will return the first key.
# If table is list-like this will return the first row.
first_row_or_key = next(iter(table))
except TypeError:
# Not iterable, is it an asciitable Reader instance?
if isinstance(table, core.BaseReader):
lines = table.table
else:
# None of the possible choices so raise exception
raise TypeError(
'Input table must be iterable or else be a Reader object')
else:
# table is iterable, now see if it is dict-like or list-like
try:
# If first_row_or_key is a row (in the case that table is
# list-like) then this will raise exception
table[first_row_or_key]
except (TypeError, IndexError, ValueError):
# Table is list-like (python list-of-lists or numpy recarray)
lines = table
else:
# Table is dict-like. Turn this into a DictLikeNumpy that has
# an API similar to a numpy recarray.
lines = core.DictLikeNumpy(table)
if names is None:
lines.dtype.names = sorted(lines.keys())
else:
lines.dtype.names = names
# ``lines`` could now be one of the following iterable objects:
# - NumPy recarray
# - DictLikeNumpy object
# - Python list of lists
return lines
def get_lines(self, table, names):
"""Get the lines from the ``table`` input.
:param table: table input
:param names: list of column names (only used for dict of lists to set column order)
:returns: list of lines
"""
try:
# If table is dict-like this will return the first key.
# If table is list-like this will return the first row.
first_row_or_key = next(iter(table))
except TypeError:
# Not iterable, is it an asciitable Reader instance?
if isinstance(table, core.BaseReader):
lines = table.table
else:
# None of the possible choices so raise exception
raise TypeError('Input table must be iterable or else be a Reader object')
else:
# table is iterable, now see if it is dict-like or list-like
try:
# If first_row_or_key is a row (in the case that table is
# list-like) then this will raise exception
table[first_row_or_key]
except (TypeError, IndexError):
# Table is list-like (python list-of-lists or numpy recarray)
lines = table
else:
# Table is dict-like. Turn this into a DictLikeNumpy that has
# an API similar to a numpy recarray.
lines = core.DictLikeNumpy(table)
if names is None:
lines.dtype.names = sorted(lines.keys())
else:
lines.dtype.names = names
# ``lines`` could now be one of the following iterable objects:
# - NumPy recarray
# - DictLikeNumpy object
# - Python list of lists
return lines