def tree2table(data, transpose=False, head=None):
"""build table from data.
The table will be multi-level (main-rows and sub-rows), if:
1. there is more than one column
2. each cell within a row is a list or tuple
If any of the paths contain tuples/lists, these are
expanded to extra columns as well.
If head is given, only first head rows are output.
returns matrix, row_headers, col_headers
"""
logger = Component.get_logger()
labels = getPaths(data)
if len(labels) < 2:
raise ValueError("expected at least two levels for building table, got %i: %s" %
(len(labels), str(labels)))
effective_labels = count_levels(labels)
# subtract last level (will be expanded) and 1 for row header
effective_cols = sum(effective_labels[:-1]) - 1
col_headers = [""] * effective_cols + labels[-1]
ncols = len(col_headers)
paths = list(itertools.product(*labels[1:-1]))
header_offset = effective_cols
matrix = []
logger.debug(
"Datatree.buildTable: creating table with %i columns" %
(len(col_headers)))
# the following can be made more efficient
# by better use of indices
row_offset = 0
row_headers = []
# iterate over main rows
for x, row in enumerate(labels[0]):
first = True
for xx, path in enumerate(paths):
# get data - skip if there is None
work = getLeaf(data, (row,) + path)
if isinstance(work, pandas.DataFrame):
if work.empty:
continue
else:
if not work:
continue
row_data = [""] * ncols
# add row header only for first row (if there are sub-rows)
if first:
if type(row) in Utils.ContainerTypes:
row_headers.append(row[0])
for z, p in enumerate(row[1:]):
row_data[z] = p
else:
row_headers.append(row)
first = False
else:
row_headers.append("")
# enter data for the first row
for z, p in enumerate(path):
row_data[z] = p
# check for multi-level rows
is_container = True
max_rows = None
for y, column in enumerate(labels[-1]):
if column not in work:
continue
if type(work[column]) not in Utils.ContainerTypes:
is_container = False
break
if max_rows == None:
max_rows = len(work[column])
elif max_rows != len(work[column]):
raise ValueError("multi-level rows - unequal lengths: %i != %i" %
(max_rows, len(work[column])))
# add sub-rows
if is_container:
# multi-level rows
for z in range(max_rows):
for y, column in enumerate(labels[-1]):
try:
row_data[
y + header_offset] = Utils.quote_rst(work[column][z])
except KeyError:
pass
if z < max_rows - 1:
matrix.append(row_data)
row_headers.append("")
row_data = [""] * ncols
else:
# single level row
for y, column in enumerate(labels[-1]):
try:
row_data[
y + header_offset] = Utils.quote_rst(work[column])
except KeyError:
pass
matrix.append(row_data)
if head and len(matrix) >= head:
break
if transpose:
row_headers, col_headers = col_headers, row_headers
matrix = list(zip(*matrix))
# convert headers to string (might be None)
row_headers = [str(x) for x in row_headers]
col_headers = [str(x) for x in col_headers]
return matrix, row_headers, col_headers
