def _load_data(val_row,
societies=None,
sources=None,
variables=None,
descriptions=None):
ext_id = val_row.get('soc_id')
if ext_id not in societies:
logging.warn(
"Attempting to load values for %s but no Society object exists, skipping"
% ext_id)
return
society = societies[ext_id]
variable_id = val_row['VarID']
variable = variables.get(
var_number_to_label(val_row['Dataset'], variable_id))
if variable is None:
logging.warn("Could not find variable %s for society %s" %
(variable_id, society.name))
return
v = dict(variable=variable,
society=society,
source=get_source(val_row['Dataset']),
coded_value=val_row['Code'],
code=descriptions.get((variable.id, val_row['Code'].strip())),
focal_year=val_row['Year'],
comment=val_row['Comment'],
subcase=val_row['SubCase'])
if variable.data_type == 'Continuous' and val_row[
'Code'] and val_row['Code'] != 'NA':
v['coded_value_float'] = float(val_row['Code'])
refs = set()
for r in val_row['EthnoReferences'].split(";"):
r = r.strip()
author, year = None, None
m = BINFORD_REF_PATTERN.match(r)
if m:
author, year = m.group('author').strip(), m.group('year')
if author.endswith(','):
author = author[:-1].strip()
else:
ref_short = r.split(",")
if len(ref_short) == 2:
author = ref_short[0].strip()
year = ref_short[1].strip().split(':')[0]
if author and year:
ref = sources.get((author, year))
if ref:
refs.add(ref.id)
else: # pragma: no cover
logging.warn(
"Could not find reference %s, %s in database, skipping reference"
% (author, year))
return v, refs
def load_codes(items):
count = 0
for row in items:
label = var_number_to_label(row['Dataset'], row['VarID'])
variable = CulturalVariable.objects.filter(label=label).first()
if variable:
code_description, created = CulturalCodeDescription.objects.get_or_create(
variable=variable, code=row['Code'])
code_description.description = row['CodeDescription']
code_description.short_description = row['ShortName']
code_description.save()
logging.info(("Created CulturalCodeDescription: %s" %
code_description).decode('utf8'))
count += 1
else:
logging.warn("Missing variable in database: %s" % label)
return count
def load_codes(items):
count = 0
for row in items:
label = var_number_to_label(row['Dataset'], row['VarID'])
variable = CulturalVariable.objects.filter(label=label).first()
if variable:
code_description, created = CulturalCodeDescription.objects.get_or_create(
variable=variable, code=row['Code'])
code_description.description = row['CodeDescription']
code_description.short_description = row['ShortName']
code_description.save()
logging.info(
("Created CulturalCodeDescription: %s" % code_description).decode('utf8'))
count += 1
else:
logging.warn("Missing variable in database: %s" % label)
return count
def load_var(var_dict, categories):
label = var_number_to_label(var_dict['Dataset'], var_dict['VarID'])
variable, created = CulturalVariable.objects.get_or_create(
label=label, source=get_source(var_dict['Dataset']))
variable.name = var_dict['VarTitle']
variable.codebook_info = var_dict['VarDefinition']
variable.data_type = var_dict['VarType']
assert variable.data_type in ['Continuous', 'Categorical', 'Ordinal']
variable.units = "" if 'Units' not in var_dict else var_dict['Units']
for c in map(clean_category, var_dict['IndexCategory'].split(',')):
index_category = categories.get(c)
if not index_category:
index_category = categories[c] = CulturalCategory.objects.create(name=c)
logging.info("Created CulturalCategory: %s" % c)
if index_category not in variable.index_categories.all():
variable.index_categories.add(index_category)
variable.save()
logging.info("Created CulturalVariable: %s" % label)
logging.info("Saved variable %s - %s" % (label, variable.name))
return True
def load_var(var_dict, categories):
label = var_number_to_label(var_dict['Dataset'], var_dict['VarID'])
variable, created = CulturalVariable.objects.get_or_create(
label=label, source=get_source(var_dict['Dataset']))
variable.name = var_dict['VarTitle']
variable.codebook_info = var_dict['VarDefinition']
variable.data_type = var_dict['VarType']
assert variable.data_type in ['Continuous', 'Categorical', 'Ordinal']
variable.units = "" if 'Units' not in var_dict else var_dict['Units']
for c in map(clean_category, var_dict['IndexCategory'].split(',')):
index_category = categories.get(c)
if not index_category:
index_category = categories[c] = CulturalCategory.objects.create(
name=c)
logging.info("Created CulturalCategory: %s" % c)
if index_category not in variable.index_categories.all():
variable.index_categories.add(index_category)
variable.save()
logging.info("Created CulturalVariable: %s" % label)
logging.info("Saved variable %s - %s" % (label, variable.name))
return True
