117_Distribuidores.pdf REE file parsing. Creates a csv with extracted information. Can generate a sql for GISCE-ERP ref2 update. It may be useful for 117_Comercilizadores.pdf file

parse_REE_agent.py

# -*- coding: utf-8 -*-

# pdftotext 117_Distribuidores.pdf creates 117_Distribuidores.txt
# pdftotext 117_Comercializadores.pdf creates 117_Comercializadores.txt


import csv
import sys
import re


agent = 'comer'
#agent = 'distri'

numbertest = True
assql = False

if agent == 'distri':
    filename = '117_Distribuidores.txt'
else:
    filename = '117_Comercializadores.txt'
    assql = False

lines = []

# matches i.e '1 de 11'
num_pag = re.compile(r'[0-9]+ de [0-9]+')

with open(filename, 'r') as fitxer:
    csvfile = csv.reader(fitxer, delimiter='|')
    for row in csvfile:
        if row:
            if num_pag.search(row[0]):
                for i in range(0, 1):
                    lines.pop()
                continue
            if len(row[0]) > 70:
                # too long row (on comer)
                lines.append(row[0][:70])
                lines.append(row[0][69:])
            elif 'Baja a' in row[0]:
                parts = row[0].split('B', 2)
                lines.append(parts[0])
                lines.append('B' + parts[1])
            else:
                lines.append(row[0])

agents = [lines[x:x + 5] for x in range(0, len(lines), 5)]
sql = "UPDATE res_partner SET ref2='{0}' WHERE ref='{1}';"

for agent in agents:
    if numbertest and '0' not in agent[0] and '1' not in agent[0]:
        continue
    if not assql:
        sys.stdout.write(';'.join(agent))
    else:
        sys.stdout.write(sql.format(agent[2], agent[0]))
    sys.stdout.write("\n")
    sys.stdout.flush()