def __init__(self, *arg):
# read raw rows and definition
super().__init__(*arg)
# label rows
self.label()
# allow call like rs.data.annual_df()
self.data = DictsAsDataframes(self.dicts())
#check for duplicates
try_dfa = self.data.annual_df()
class CoreRowSystem(InputDefinition):
"""Data structure and functions to manupulate raw data and pasring specification.
Main mthods are:
dicts()
label()
save()
"""
def __init__(self, *arg):
# read raw rows and definition
super().__init__(*arg)
# label rows
self.label()
# allow call like rs.data.annual_df()
self.data = DictsAsDataframes(self.dicts())
#check for duplicates
try_dfa = self.data.annual_df()
def dicts(self):
return dicts_as_stream(self)
def named_dicts(self, name):
for d in self.dicts():
if d['varname'] == name:
yield d
def get_header_and_desc_dicts(self):
for spec in self.segments:
for k,v in spec.header_dict.items():
yield {"_head":v[0], '_desc':k}
def save(self):
self.save_as_default()
def save_as_default(self):
self.dump_dicts_to_db(db = DefaultDatabase())
self.toc()
return self
def save_as_test(self):
self.dump_dicts_to_db(db = TrialDatabase())
return self
def dump_dicts_to_db(self, db):
# WARNING: call order matters, cannot call db.save_data_dicts(), causes error
db.save_headlabel_description_dicts(gen = self.get_header_and_desc_dicts())
db.save_data_dicts(gen = self.dicts())
def label(self):
self._assign_parsing_specification_by_row()
self._run_label_adjuster()
def _run_label_adjuster(self):
"""Label rows using markup information from self.specs[i].header_dict
and .unit_dict. Stores labels in self.labels[i].
"""
cur_label = UnknownLabel()
for i, head in self.row_heads:
# change cur_label at text rows, not starting with year number
if not is_year(head):
cur_label = adjust_labels(textline=head, incoming_label=cur_label,
dict_headline=self.specs[i].header_dict,
dict_unit=self.specs[i].unit_dict)
self.labels[i] = Label(cur_label.head, cur_label.unit)
def _assign_parsing_specification_by_row(self):
"""Write appropriate parsing specification from selfsegments[0] or self.segments
to self.specs[i] based on segments[i].start_line and .end_line
"""
switch = _SegmentState(self.segments)
# no segment information - all rows have default parsing specification
if len(self.segments) == 1:
for i, head in self.row_heads:
self.specs[i] = self.segments[0]
# segment information is supplied, will check row_heads and compare it with
else:
for i, head in self.row_heads:
# are we in the default spec?
if switch.in_segment:
# we are in custom spec. do we have to switch to the default spec?
switch.update_if_leaving_custom_segment(head)
# maybe it is also a start of a new custom spec?
switch.update_if_entered_custom_segment(head)
# should we switch to custom spec?
else:
switch.update_if_entered_custom_segment(head)
#finished adjusting specification for i-th row
self.specs[i] = switch.current_spec
