Script para convertir los papeles de Bárcenas publicados por EL PAÍS a json.

parse_papeles.py

# -*- coding: utf8 -*-
import re
import json
from lxml import html
try:
    from urllib2 import urlopen
except ImportError:
    from urllib import urlopen

PAPELES_URL = "http://elpais.com/especiales/2013/caso_barcenas/" \
              "todos_los_papeles.html"
money_re = re.compile(u"([0-9\.]+) (ptas|\(¿\?\)|€)")


def parse_number(s):
    x = s.replace(".", "").replace(",", ".")
    m = re.match(money_re, x)
    if m:
        debe_float = float(m.groups()[0])
        debe_unit = m.groups()[1]
    else:
        debe_float = None
        debe_unit = None

    return debe_float, debe_unit


s = urlopen(PAPELES_URL).read()
tree = html.fromstring(s)

trs = tree.xpath("//table[@id='tabla_datos']/tbody/tr")


pagos = []
for tr in trs:
    tds = tr.getchildren()
    month = tds[0].text
    try:
        year = int(tds[1].text)
    except TypeError, ValueError:
        year = None
    form = tds[2].text
    concept = tds[3].text_content()
    debe_float, debe_unit = parse_number(tds[4].text_content())
    haber_float, haber_unit = parse_number(tds[5].text_content())
    saldo_float, saldo_unit = parse_number(tds[6].text_content())

    pagos.append(dict(mes=month,
                      anyo=year,
                      forma=form,
                      concepto=concept,
                      debe=debe_float,
                      debe_unidad=debe_unit,
                      haber=haber_float,
                      haber_unidad=haber_unit,
                      saldo=saldo_float,
                      saldo_unidad=saldo_unit
                ))

with open("papeles_barcenas.json", "w") as f:
    json.dump(pagos, f, indent=2, encoding="utf8")