Lecture et écriture des fichiers Excel + quelques bugs
- Mise en place de la lecture de la liste des membres à partir d'un fichier Excel - Mise en place de l'écriture de la liste des membres décédés dans un fichier Excel - Correction du bug: le traitement ne rend pas la main à l'IHM losque terminé - Vider le texte de l'IHM lorsqu'on clique sur OK
This commit is contained in:
parent
9febc4cc55
commit
308ed11077
27
install.py
27
install.py
|
|
@ -26,6 +26,27 @@ except ImportError:
|
|||
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
print(p.communicate())
|
||||
|
||||
url = "https://git.roflcopter.fr/sdjgeek/purge-registres-deces-insee/-/archive/v5.0/purge-registres-deces-insee-v5.0.zip"
|
||||
myfile = requests.get(url)
|
||||
open('purge-registres-deces-insee-v5.0.zip', 'wb').write(myfile.content)
|
||||
try:
|
||||
import pandas
|
||||
except ImportError:
|
||||
print("Installing pandas")
|
||||
p = subprocess.Popen([sys.executable, "-m", "pip", "install", "-U", "pandas"],
|
||||
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
print(p.communicate())
|
||||
|
||||
try:
|
||||
import xlrd
|
||||
except ImportError:
|
||||
print("Installing xlrd")
|
||||
p = subprocess.Popen([sys.executable, "-m", "pip", "install", "-U", "xlrd"],
|
||||
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
print(p.communicate())
|
||||
|
||||
try:
|
||||
import numpy
|
||||
except ImportError:
|
||||
print("Installing numpy")
|
||||
p = subprocess.Popen([sys.executable, "-m", "pip", "install", "-U", "numpy"],
|
||||
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
print(p.communicate())
|
||||
|
||||
|
|
|
|||
|
|
@ -8,6 +8,7 @@ Classe d'accès aux données du site de l'Église
|
|||
|
||||
import os
|
||||
import pandas as pd
|
||||
import numpy as np
|
||||
|
||||
from membre_base import MembreBase, MembreProvider
|
||||
|
||||
|
|
@ -25,12 +26,23 @@ class ExcelIn(MembreProvider):
|
|||
|
||||
def __init__(self, provider, row):
|
||||
super().__init__(provider)
|
||||
self.r_id = str(row["id"])
|
||||
self.r_first_name = str(row["first_name"]).strip().split(' ')[0].upper()
|
||||
self.r_last_name = str(row["last_name"]).strip().split(' ')[0].upper()
|
||||
self.r_maiden_name = str(row["maiden_name"]).strip().split(' ')[0].upper() if type(row["maiden_name"]) == np.str else None
|
||||
self.r_annee = str(row["annee"])
|
||||
self.r_mois = str(row["mois"])
|
||||
self.r_jour = str(row["jour"])
|
||||
self.r_ville = "<VIDE>"
|
||||
self.r_sexe = "F" if str(row["status"]) == "Female" else "M"
|
||||
|
||||
def get_name(self):
|
||||
return os.path.basename(self.excel_path)
|
||||
|
||||
def load(self):
|
||||
self.dataframe = pd.read_excel(self.excel_path)
|
||||
self.dataframe = pd.read_excel(self.excel_path, skiprows=[0, 1], usecols="B:E,G:I,K", header=None,
|
||||
names=["last_name", "first_name", "maiden_name", "id", "jour", "mois", "annee",
|
||||
"status"])
|
||||
return len(self.dataframe)
|
||||
|
||||
def get_member_list(self):
|
||||
|
|
|
|||
|
|
@ -0,0 +1,30 @@
|
|||
"""
|
||||
Copyright (c) 2020 Sdj Geek
|
||||
Voir le fichier LICENSE
|
||||
|
||||
Classe d'accès aux données INSEE dans la base SQLite
|
||||
|
||||
"""
|
||||
|
||||
import pandas as pd
|
||||
|
||||
|
||||
class ExcelOut:
|
||||
|
||||
def __init__(self, out_path):
|
||||
self.out_path = out_path
|
||||
self.data = list()
|
||||
|
||||
def add_member(self, member):
|
||||
self.data.append({
|
||||
'nom_registres': member.get_nom_registres(),
|
||||
'nom_insee': member.get_nom_insee(),
|
||||
'mrn': member.r_id,
|
||||
'death_year': member.i_annee_deces,
|
||||
'death_month': member.i_mois_deces,
|
||||
'death_day': member.i_jour_deces
|
||||
})
|
||||
|
||||
def generate_output(self):
|
||||
df = pd.DataFrame(self.data)
|
||||
df.to_excel(self.out_path)
|
||||
|
|
@ -37,38 +37,50 @@ class MainApplication(tk.Frame):
|
|||
self.entry_bdd_insee.grid(row=0, column=1, sticky='ew')
|
||||
self.button_bdd_insee = tk.Button(self.frame, text="...", command=self.command_button_bdd_insee)
|
||||
self.button_bdd_insee.grid(row=0, column=2, sticky='w')
|
||||
|
||||
# Sélection fichier Excel
|
||||
self.label_fichier_excel = tk.Label(self.frame, text="Fichier Excel")
|
||||
self.label_fichier_excel.grid(row=1, column=0, sticky='e')
|
||||
self.value_fichier_excel = tk.StringVar()
|
||||
self.entry_fichier_excel = tk.Entry(self.frame, state='disabled', textvariable=self.value_fichier_excel)
|
||||
self.entry_fichier_excel.grid(row=1, column=1, sticky='ew')
|
||||
self.button_fichier_excel = tk.Button(self.frame, text="...", command=self.command_button_fichier_excel)
|
||||
self.button_fichier_excel.grid(row=1, column=2, sticky='w')
|
||||
|
||||
# Sélection répertoire sortie
|
||||
self.label_dir_out = tk.Label(self.frame, text="Répertoire de sortie")
|
||||
self.label_dir_out.grid(row=1, column=0, sticky='e')
|
||||
self.label_dir_out.grid(row=2, column=0, sticky='e')
|
||||
self.value_dir_out = tk.StringVar()
|
||||
self.entry_dir_out = tk.Entry(self.frame, state='disabled', textvariable=self.value_dir_out)
|
||||
self.entry_dir_out.grid(row=1, column=1, sticky='ew')
|
||||
self.entry_dir_out.grid(row=2, column=1, sticky='ew')
|
||||
self.button_dir_out = tk.Button(self.frame, text="...", command=self.command_button_dir_out)
|
||||
self.button_dir_out.grid(row=1, column=2, sticky='w')
|
||||
self.button_dir_out.grid(row=2, column=2, sticky='w')
|
||||
|
||||
# Sélection des unités à traiter
|
||||
self.label_units = tk.Label(self.frame, text="Unités à purger")
|
||||
self.label_units.grid(row=2, column=0, sticky='e')
|
||||
self.label_units.grid(row=3, column=0, sticky='e')
|
||||
self.value_units = tk.StringVar()
|
||||
self.entry_units = tk.Entry(self.frame, textvariable=self.value_units)
|
||||
self.entry_units.grid(row=2, column=1, sticky='ew', columnspan="2")
|
||||
self.entry_units.grid(row=3, column=1, sticky='ew', columnspan="2")
|
||||
|
||||
# Bouton validation
|
||||
self.button_valid = tk.Button(self.frame, text="Ok", command=self.command_button_valid)
|
||||
self.button_valid.grid(row=3, column=0, columnspan=3)
|
||||
self.button_valid.grid(row=4, column=0, columnspan=3)
|
||||
|
||||
# Barre de progression
|
||||
self.progressbar = ttk.Progressbar(self.frame, orient=tk.HORIZONTAL, mode='determinate')
|
||||
self.progressbar.grid(row=4, column=0, columnspan=3, sticky='ew')
|
||||
self.progressbar.grid(row=5, column=0, columnspan=3, sticky='ew')
|
||||
|
||||
# Affichage des logs
|
||||
self.text_log = tk.Text(self.frame, state='disabled')
|
||||
self.text_log.grid(row=0, column=3, rowspan=5, sticky='nesw')
|
||||
self.text_log.grid(row=0, column=3, rowspan=6, sticky='nesw')
|
||||
|
||||
def command_button_bdd_insee(self):
|
||||
self.value_bdd_insee.set(tkfiledialog.askopenfilename(title="Fichier de l'INSEE"))
|
||||
|
||||
def command_button_fichier_excel(self):
|
||||
self.value_fichier_excel.set(tkfiledialog.askopenfilename(title="Fichier Excel"))
|
||||
|
||||
def command_button_dir_out(self):
|
||||
self.value_dir_out.set(tkfiledialog.askdirectory(title="Répertoire de sortie"))
|
||||
|
||||
|
|
@ -77,6 +89,11 @@ class MainApplication(tk.Frame):
|
|||
self.text_log.insert(tk.END, text)
|
||||
self.text_log.configure(state='disabled')
|
||||
|
||||
def clear_log(self):
|
||||
self.text_log.configure(state='normal')
|
||||
self.text_log.delete('1.0', tk.END)
|
||||
self.text_log.configure(state='disabled')
|
||||
|
||||
def watch(self):
|
||||
if self.run:
|
||||
if self.pipe.poll():
|
||||
|
|
@ -84,48 +101,52 @@ class MainApplication(tk.Frame):
|
|||
if message.get('step', False):
|
||||
self.progressbar.step(message['step'])
|
||||
elif message.get('text', False):
|
||||
print(message['text'])
|
||||
self.add_log(message['text'])
|
||||
elif message.get('running', False):
|
||||
elif 'running' in message:
|
||||
self.run = message['running']
|
||||
elif message.get('set_max', False):
|
||||
self.progressbar["value"] = 0
|
||||
self.progressbar['maximum'] = message['set_max']
|
||||
self.parent.after(100, self.watch)
|
||||
else:
|
||||
print("Recherche terminée")
|
||||
self.add_log("\nRecherche terminée\n")
|
||||
self.button_valid.configure(state='normal')
|
||||
|
||||
def command_button_valid(self):
|
||||
self.button_valid.configure(state='disabled')
|
||||
self.clear_log()
|
||||
self.run = True
|
||||
(conn1, conn2) = Pipe()
|
||||
self.pipe = conn1
|
||||
Worker(conn2, self.value_bdd_insee.get(), self.value_dir_out.get(), self.value_units.get().split(',')).start()
|
||||
unite = self.value_units.get().split(',')
|
||||
if unite == ['']:
|
||||
unite = None
|
||||
Worker(conn2, self.value_bdd_insee.get(), self.value_fichier_excel.get(), self.value_dir_out.get(), unite).start()
|
||||
self.watch()
|
||||
|
||||
|
||||
class Worker(Process):
|
||||
def __init__(self, pipe, bdd_insee, dir_out, units):
|
||||
def __init__(self, pipe, bdd_insee, fichier_excel, dir_out, units):
|
||||
Process.__init__(self)
|
||||
self.pipe = pipe
|
||||
self.bdd_insee = bdd_insee
|
||||
self.fichier_excel = fichier_excel
|
||||
self.dir_out = dir_out
|
||||
self.units = units
|
||||
|
||||
def tracker(self, step=None, text=None, set_max=None, running=None):
|
||||
if step:
|
||||
if step is not None:
|
||||
self.pipe.send({'step': step})
|
||||
elif text:
|
||||
elif text is not None:
|
||||
self.pipe.send({'text': text})
|
||||
elif set_max:
|
||||
elif set_max is not None:
|
||||
self.pipe.send({'set_max': set_max})
|
||||
elif running:
|
||||
elif running is not None:
|
||||
self.pipe.send({'running': running})
|
||||
|
||||
def run(self):
|
||||
trouver_decedes(chemin_base_donnees=self.bdd_insee,
|
||||
excel_path=self.fichier_excel,
|
||||
chemin_repertoire_sortie=self.dir_out,
|
||||
numeros_unites=self.units,
|
||||
tracker=self.tracker)
|
||||
|
|
|
|||
|
|
@ -34,6 +34,9 @@ class MembreBase(ABC):
|
|||
nom_registres = f"{nom_registres} née {self.r_maiden_name.upper()}"
|
||||
return nom_registres
|
||||
|
||||
def get_nom_insee(self):
|
||||
return f"{self.i_last_name}, {self.i_first_name}"
|
||||
|
||||
def set_insee(self, insee):
|
||||
self.i_first_name = insee.first_name
|
||||
self.i_last_name = insee.last_name
|
||||
|
|
@ -57,7 +60,7 @@ class MembreBase(ABC):
|
|||
Le membre {self.get_nom_registres()} ({self.r_id}),
|
||||
né{feminin} le {self.r_jour:0>2}/{self.r_mois:0>2}/{self.r_annee:0>4} à {self.r_ville.upper()}
|
||||
semble être décédé{feminin}.
|
||||
Dans le fichier de l'INSEE on peut trouver {self.i_last_name}, {self.i_first_name}
|
||||
Dans le fichier de l'INSEE on peut trouver {self.get_nom_insee()}
|
||||
né{feminin} le {self.i_jour_naissance:0>2}/{self.i_mois_naissance:0>2}/{self.i_annee_naissance:0>4} à {self.i_ville_naissance}
|
||||
décédé{feminin} le {self.i_jour_deces:0>2}/{self.i_mois_deces:0>2}/{self.i_annee_deces:0>4} à {self.i_ville_deces}
|
||||
"""
|
||||
|
|
|
|||
|
|
@ -45,7 +45,7 @@ class SiteEglise(MembreProvider):
|
|||
def complete(self):
|
||||
if not self.completed:
|
||||
member_profile = self.provider.get_member_profile(self.r_id)
|
||||
self.r_maiden_name = member_profile['individual']['maidenNameGroup']['name1']['family']
|
||||
self.r_maiden_name = member_profile['individual']['maidenNameGroup']['name1']['family'].strip().split(' ')[0].upper()
|
||||
self.r_ville = member_profile['individual']['birthPlace']
|
||||
if not self.r_ville:
|
||||
self.r_ville = "<VIDE>"
|
||||
|
|
|
|||
|
|
@ -15,6 +15,8 @@ import os
|
|||
|
||||
from bdd_insee import BddInsee
|
||||
from site_eglise import SiteEglise
|
||||
from excel_in import ExcelIn
|
||||
from excel_out import ExcelOut
|
||||
|
||||
|
||||
def default_tracker(step=None, text=None, set_max=None, running=None):
|
||||
|
|
@ -42,29 +44,35 @@ def trouver_decedes(chemin_base_donnees, numeros_unites, chemin_repertoire_sorti
|
|||
# Initialiser les fournisseurs de liste de membres
|
||||
fournisseurs_membres = list()
|
||||
if excel_path:
|
||||
raise NotImplementedError
|
||||
for unite in numeros_unites:
|
||||
fournisseurs_membres.append(SiteEglise(unite, cookie_path))
|
||||
fournisseurs_membres.append(ExcelIn(excel_path))
|
||||
if numeros_unites:
|
||||
for unite in numeros_unites:
|
||||
fournisseurs_membres.append(SiteEglise(unite, cookie_path))
|
||||
# Boucler sur la liste fournisseurs
|
||||
for member_provider in fournisseurs_membres:
|
||||
tracker(set_max=member_provider.load())
|
||||
tracker(text=member_provider.get_name())
|
||||
tracker(text=f"Recherche dans {member_provider.get_name()}\n")
|
||||
# Récupérer la liste des membres
|
||||
members = member_provider.get_member_list()
|
||||
# Préparer le fichier de sortie
|
||||
output_file = os.path.join(chemin_repertoire_sortie, f"liste_membres_decedes_{member_provider.get_name()}.txt")
|
||||
with open(output_file, 'w') as out_file:
|
||||
out_file.write("Les lieux dans le fichier de l'INSEE sont donnés en Code Officiel Géographique en vigueur au moment de la prise en compte du décès\n")
|
||||
# Boucler sur la liste des membres
|
||||
for member in members:
|
||||
query = base_insee.find_person(member.r_first_name, member.r_last_name, member.r_maiden_name,
|
||||
member.r_annee, member.r_mois, member.r_jour)
|
||||
for person in query:
|
||||
member.set_insee(person)
|
||||
text = member.get_texte_decede()
|
||||
tracker(text=text)
|
||||
out_file.write(text)
|
||||
tracker(step=1)
|
||||
# Préparer les fichiers de sortie
|
||||
output_base_name = os.path.join(chemin_repertoire_sortie, f"liste_membres_decedes_{member_provider.get_name()}")
|
||||
output_txt = open(output_base_name + ".txt", 'w')
|
||||
output_xls = ExcelOut(output_base_name + ".xlsx")
|
||||
output_txt.write("Les lieux dans le fichier de l'INSEE sont donnés en Code Officiel Géographique en vigueur au moment de la prise en compte du décès\n")
|
||||
# Boucler sur la liste des membres
|
||||
for member in members:
|
||||
query = base_insee.find_person(member.r_first_name, member.r_last_name, member.r_maiden_name,
|
||||
member.r_annee, member.r_mois, member.r_jour)
|
||||
for person in query:
|
||||
member.set_insee(person)
|
||||
text = member.get_texte_decede()
|
||||
tracker(text=text)
|
||||
output_txt.write(text)
|
||||
output_xls.add_member(member)
|
||||
tracker(step=1)
|
||||
# Clore les fichiers de sortie
|
||||
output_txt.close()
|
||||
output_xls.generate_output()
|
||||
tracker(running=False)
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -1,3 +1,6 @@
|
|||
peewee
|
||||
requests
|
||||
browser_cookie3
|
||||
browser_cookie3
|
||||
pandas
|
||||
xlrd
|
||||
numpy
|
||||
|
|
|
|||
Loading…
Reference in New Issue