demographic-manager-sanitized/main.py

165 lines
5.5 KiB
Python
Raw Normal View History

2024-06-24 04:57:26 -05:00
import json
import common
2024-06-24 04:57:26 -05:00
from openpyxl import Workbook
from openpyxl.utils import get_column_letter
from openpyxl.styles import Font
2024-06-24 04:57:43 -05:00
def convert_x_y_to_col_id(x, y):
return get_column_letter(x) + str(y)
def convert_x_y_range_to_cols(x1, y1, x2, y2):
return "{}:{}".format(convert_x_y_to_col_id(x1, y1), convert_x_y_to_col_id(x2, y2))
2024-06-24 04:57:26 -05:00
def mapper(string):
# map to human-readable values
try:
2024-06-24 04:57:43 -05:00
return common.mapper_data[str(string)]
2024-06-24 04:57:26 -05:00
except KeyError:
return "WARNING: {}".format(string)
def aggregate_demographic_totals(payload, adultonly, childonly):
2024-06-24 04:57:26 -05:00
results = {"age": {}, "gender": {}, "ethnicity": {}, "race": {}}
for i in payload:
if adultonly and i["age"] == "c":
continue
if childonly and i["age"] == "a":
continue
2024-06-24 04:57:26 -05:00
# for every entry we get
for key in results.keys():
# for every field we have
results[key].setdefault(i[key], 0)
results[key][i[key]] += 1
return results
def get_totals_for_specific_attribute_on_specific_site(site, adultonly=False, childonly=False):
2024-06-24 04:57:26 -05:00
manifest = open("data/manifest.json")
manifest_data = json.loads(manifest.read())
record_to_be_used = manifest_data["sites"]
site_object = None
for i in manifest_data["sites"]:
if i["name"] == site:
site_object = i
if not site_object:
raise TypeError("site argument is not valid")
used_filename = site_object["usedrecord"]
site_specific_data_file = open("data/" + used_filename)
demographic_data_json = json.loads(site_specific_data_file.read())
payload = demographic_data_json["payload"]
return aggregate_demographic_totals(payload, adultonly, childonly)
2024-06-24 04:57:26 -05:00
def generate_table_for_attr(attr, totals):
2024-06-24 04:57:43 -05:00
header = [mapper(attr), "Count"]
data = []
2024-06-24 04:57:26 -05:00
print(totals)
2024-06-24 04:57:43 -05:00
2024-06-24 04:57:26 -05:00
for key, value in totals[attr].items():
print(key, value)
2024-06-24 04:57:43 -05:00
data.append([mapper(key), value])
data.sort()
data.insert(0, header)
2024-06-24 04:57:26 -05:00
2024-06-24 04:57:43 -05:00
return data
2024-06-24 04:57:26 -05:00
def write_dataarray_to_specific_cell(x, y, dataarray, worksheet):
i_c = 0
for i in dataarray:
j_c = 0
for j in i:
worksheet.cell(row=i_c + 1 + x, column=j_c + 1 + y, value=dataarray[i_c][j_c])
j_c += 1
i_c += 1
def generate_provider_string(providers):
2024-06-24 04:57:43 -05:00
used_providers = sorted(providers)
if len(used_providers) > 1:
before_and = ', '.join(used_providers[:-1])
if len(used_providers) > 2:
and_string = ', and {}'.format(used_providers[-1])
else:
and_string = ' and {}'.format(used_providers[-1])
2024-06-24 04:57:26 -05:00
return before_and + and_string
2024-06-24 04:57:43 -05:00
return used_providers[0]
2024-06-24 04:57:26 -05:00
def handle_writing_of_attrs(worksheet, totals, offset):
2024-06-24 04:57:26 -05:00
count = 0
maxlen = 0
2024-06-24 04:57:26 -05:00
for i in totals.keys():
tmp = generate_table_for_attr(i, totals)
maxlen = max(maxlen, len(tmp))
2024-06-24 04:57:26 -05:00
write_dataarray_to_specific_cell(offset, count * 3, tmp, worksheet)
2024-06-24 04:57:26 -05:00
count += 1
return maxlen
2024-06-24 04:57:26 -05:00
def adjust_column_width(worksheet, times):
for i in range(times):
worksheet.column_dimensions[get_column_letter((i * 3) + 1)].width = 20
def write_information_to_spreadsheet(totals, worksheet, offset):
2024-06-24 04:57:43 -05:00
return handle_writing_of_attrs(worksheet, totals, offset)
def handle_spreadsheet_decoration(ws, providers, persons, commleads, onsiteleads):
ws["A1"].value = "Information for {} site attendance.".format(ws.title)
ws["A2"].value = "Fields not present should be assumed 0."
ws["A3"].value = "Data submitted by {} in person after conclusion of site operations.".format(generate_provider_string(providers))
ws["A4"].value = "Interns present: {}".format(generate_provider_string(persons))
ws["A5"].value = "Communication Lead(s): {}".format(generate_provider_string(commleads))
ws["A6"].value = "On Site Lead(s): {}".format(generate_provider_string(onsiteleads))
ws.merge_cells("A1:E1")
ws.merge_cells("A2:E2")
ws.merge_cells("A3:H3")
ws.merge_cells("A4:H4")
ws.merge_cells("A5:H5")
ws.merge_cells("A6:H6")
2024-06-24 04:57:26 -05:00
fd = open("data/manifest.json", "r")
json_data = json.loads(fd.read())
wb = Workbook()
for site in json_data["sites"]:
providers = []
for record in site["records"]:
providers.append(record['submitter'])
persons = site["present"]
commleads = site["commleads"]
onsiteleads = site["onsiteleads"]
length = 8
ws = wb.create_sheet(site["name"])
ws.cell(row=length, column=1, value="Combined Totals")
2024-06-24 04:57:43 -05:00
ws.merge_cells(convert_x_y_range_to_cols(1, length, 11, length))
ws[convert_x_y_to_col_id(1, length)].font = Font(italic=True)
length += write_information_to_spreadsheet(get_totals_for_specific_attribute_on_specific_site(site["name"]), ws, length) + 2
ws.cell(row=length, column=1, value="Adults")
2024-06-24 04:57:43 -05:00
ws.merge_cells(convert_x_y_range_to_cols(1, length, 11, length))
ws[convert_x_y_to_col_id(1, length)].font = Font(italic=True)
length += write_information_to_spreadsheet(get_totals_for_specific_attribute_on_specific_site(site["name"], adultonly=True), ws, length) + 3
ws.cell(row=length, column=1, value="Children")
2024-06-24 04:57:43 -05:00
ws.merge_cells(convert_x_y_range_to_cols(1, length, 11, length))
ws[convert_x_y_to_col_id(1, length)].font = Font(italic=True)
length += write_information_to_spreadsheet(get_totals_for_specific_attribute_on_specific_site(site["name"], childonly=True), ws, length) + 3
adjust_column_width(ws, 4)
handle_spreadsheet_decoration(ws, providers, persons, commleads, onsiteleads)
2024-06-24 04:57:26 -05:00
del wb["Sheet"]
wb.save("test.xlsx")