fcn-toolbox/ccoop-resplit

#! /usr/bin/env python3
# -*- coding: utf-8 -*-

# Moulinette permettant de produire des relevés de comptes mensuels
# au format CSV à partir d'exports CSV venant de l'interface Web du
# crédit coopératif et ayant des chevauchements.

import os, sys, json
import csv
from collections import OrderedDict
from datetime import datetime
import hashlib
import locale
locale.setlocale(locale.LC_ALL, 'fr_FR.UTF-8')

class CsvStatementParser(object):
    
    def __init__(self):
        self.lines = OrderedDict()
        self.fieldnames = None
        self.date_fieldname = "Date"
        self.overlap_detector = {}
        self.first_ops = {}
        self.last_ops = {}
        self.daterange = [datetime.now(), datetime.fromordinal(1)]
        self.dups = dict() # holds counters for duplicate lines
    
    
    def parse(self, filename):
        with open(filename, encoding='iso-8859-1') as csvfile:
            reader = csv.DictReader(csvfile, delimiter=';')
            if self.fieldnames is None:
                # Le premier fichier parcourru détermine les noms de
                # colonnes attendus dans les prochains fichiers.
                self.fieldnames = [k for k in reader.fieldnames if k != '']
                
                # On identifie également la permière colonne qui
                # ressemble à une date, elle servira ensuite de clef
                # d'indexation.
                for fname in self.fieldnames:
                    if "date" in fname.lower():
                        self.date_fieldname = fname
                        break
            
            if self.fieldnames != [k for k in reader.fieldnames if k != '']:
                print("""Fichier ignoré : %s. Cause: does not have the expected column names.
  Found:    %s
  Expected: %s
""" % (filename, ",".join(reader.fieldnames), ",".join(self.fieldnames)))
            else:
                self._parse_file(filename, reader)
    
    
    def _parse_file(self, filename, reader):
        self.dups = dict() # Duplicate counters must be reset for each file
        print("Lecture du fichier %s" % os.path.basename(filename))
        for row in reader:
            opdate = datetime.strptime(row[self.date_fieldname], '%d/%m/%Y')
            ophash = datetime.strftime(opdate, '%Y-%m-%d') + hashlib.md5(json.dumps(row).encode()).hexdigest()
            # Special use case: one file contains multiple identical lines.
            # Then we append a counter to the duplicate ophash.
            if ophash in self.dups:
#                print("    *** doublon trouvé dans '{}': {}".format(filename, ';'.join(row.values())))
                self.dups[ophash] = self.dups[ophash] + 1
                ophash = ophash + "-" + str(self.dups[ophash])
                # print("   We have now :\n  {}\n  {}".format("\n  ".join([h + "   // " + "".join(v.values()) for h,v in self.lines.items() if h.startswith(ophash[0:10])]), ophash + " // " + "".join(row.values()))) # XXX DEBUG
            else:
                self.dups[ophash] = 0
            self.lines[ophash] = {k:v for k,v in row.items() if k != ''}
            # Adjust dateranges
            if opdate < self.daterange[0]:
                self.daterange[0] = opdate
            if opdate > self.daterange[1]:
                self.daterange[1] = opdate
            # Prepare overlap detection
            if ophash not in self.overlap_detector:
                self.overlap_detector[ophash] = set()
            self.overlap_detector[ophash].add(filename)
            # Remember first line of each CSV file
            if filename not in self.first_ops:
                self.first_ops[filename] = ophash
        # Remember last line of each CSV file
        if filename not in self.last_ops:
            self.last_ops[filename] = ophash
        # CSV files are sometimes sorted by date ASC and sometimes
        # sorted by date DESC. So we may need to swap first_op and last_op.
        if (int(self.first_ops[filename][0:10].replace('-', '')) > int(self.last_ops[filename][0:10].replace('-', ''))):
            tmp = self.first_ops[filename]
            self.first_ops[filename] = self.last_ops[filename]
            self.last_ops[filename] = tmp
    
    
    def dump_full(self, output_filename):
        with open(output_filename, 'w') as outfile:
            writer = csv.DictWriter(outfile, self.fieldnames, delimiter=';')
            writer.writeheader()
            for line in reversed(sorted(self.lines.items())):
                writer.writerow(line[1])
        print("Relevé intégral généré dans le fichier %s" % os.path.abspath(output_filename))
    
    
    def dump_monthly_reports(self, outputdir):
        firstmonth = int('{:%Y%m}'.format(self.daterange[0])) + 1
        lastmonth  = int('{:%Y%m}'.format(self.daterange[1])) - 1
        if firstmonth >= lastmonth:
            print("Impossible de générer des relevés mensuels car la plage de dates traitée est trop petite.")
            return
        curmonth = firstmonth
        def __openfile__(curmonth):
            fname = "releve_{0}.csv".format(curmonth)
            outfile = open(os.path.join(outputdir, fname), 'w')
            writer = csv.DictWriter(outfile, self.fieldnames, delimiter=';')
            writer.writeheader()
            return outfile, writer
        outfile, writer = __openfile__(curmonth)
        writer = csv.DictWriter(outfile, self.fieldnames, delimiter=';')
        for line in sorted(self.lines.items()):
            month = int(line[0][0:4] + line[0][5:7])
            if month < curmonth:
                continue
            if month > lastmonth:
                break
            if month > curmonth:
                outfile.close()
                curmonth = month
                if month in self.badmonths:
                    outfile, writer = __openfile__(str(curmonth) + "_potentiellement_incomplet")
                else:
                    outfile, writer = __openfile__(curmonth)
                
            writer.writerow(line[1])
        outfile.close()
        print("Relevés mensuels générés dans le dossier %s" % os.path.abspath(outputdir))
    
    
    def check_overlaps(self):
        """
        Helps finding possible missing operations if exported CSV files
        are not "contiguous".
        """
        self.badmonths = set()
        print("\nRecherche de chevauchements, car les chevauchements de fichiers CSV c'est bien, ça confirme qu'il n'y a pas d'écritures manquantes...")
        for filename, first_op in self.first_ops.items():
            if first_op in self.overlap_detector:
                otherfiles = [v for v in self.overlap_detector.get(first_op)]
                otherfiles.remove(filename)
                if len(otherfiles) > 0:
                    # Eliminate files having the same first_op
                    otherfiles[:] = [candidate for candidate in otherfiles if self.first_ops[candidate] != first_op]
                if len(otherfiles) == 0 and first_op[0:10] != "{0:%Y-%m-%d}".format(self.daterange[0]):
                    self.badmonths.add(int(first_op[0:7].replace('-', '')))
                    print("Attention. Il y a peut-être des écritures manquantes avant le %s (fichier %s)." % (first_op[0:10], os.path.basename(filename)))
        
        for filename, last_op in self.last_ops.items():
            if last_op in self.overlap_detector:
                otherfiles = [v for v in self.overlap_detector.get(last_op)]
                otherfiles.remove(filename)
                if len(otherfiles) > 0:
                    # Eliminate files having the same last_op
                    otherfiles[:] = [candidate for candidate in otherfiles if self.last_ops[candidate] != last_op]
                if len(otherfiles) == 0 and last_op[0:10] != "{0:%Y-%m-%d}".format(self.daterange[1]):
                    self.badmonths.add(int(last_op[0:7].replace('-', '')))
                    print("Attention. Il y a peut-être des écritures manquantes après le %s (fichier %s)." % (last_op[0:10], os.path.basename(filename)))
        print("")


def start_cli(dirpath, outputdir):
    # Lecture des fichiers CSV présents dans le dossier
    p = CsvStatementParser()
    for f in sorted(os.listdir(dirpath)):
        if f.endswith('.csv') or f.endswith('.CSV'):
            p.parse(os.path.join(dirpath, f))
    print("Les écritures lues s'étalent entre le {0:%d %B %Y} et le {1:%d %B %Y}.".format(p.daterange[0], p.daterange[1]))
    
    # Recherche de chevauchements
    p.check_overlaps()
    
    # Générer un relevé intégral et des relevés mensuels
    suffix = "_{0:%Y-%m-%d}__{1:%Y-%m-%d}".format(p.daterange[0], p.daterange[1])
    if len(p.badmonths): suffix += "_avec_des_trous"
    p.dump_full(os.path.join(outputdir, "integral%s.csv" % suffix))
    p.dump_monthly_reports(outputdir)


if __name__ == '__main__':
    if len(sys.argv) < 2:
        print("Erreur. Merci de préciser le chemin du dossier où se trouvent les fichiers CSV à analyser.")
        print("Usage:")
        print("  %s exports_csv/ csv_mensuels/" % sys.argv[0])
        sys.exit(1)
    inputdir = sys.argv[1]
    if len(sys.argv) > 2:
        outputdir = sys.argv[2]
    else:
        outputdir = os.path.join(inputdir, "outputdir")
    # Création d'un dossier output si besoin
    if not os.path.isdir(outputdir):
        os.makedirs(outputdir)
    start_cli(inputdir, outputdir)