evolution immobilier par dept (source castor)

evol.py

#!/usr/bin/env python3
# -*- coding: utf8 -*-
from pyquery import PyQuery as P
from archery import mdict
from archery.trait import Copier
from pyaml import dumps
import re

by3 = lambda s: "".join(
    [ " " * int((i+1)%3==0) + x for i,x in enumerate(s[::-1])][::-1]).strip()

class ToxicSet(Copier,set):
    def __iadd__(x,y):
        x|=y
        return x
    def __add__(x,y):
        return x|y

s = re.compile(r'''% (?P<dpt>\d{2})(?P<ville>\d+).*-.*(?:(?P<type>Vends|Maison|Studio|Appartement|Propriété|Ferme|Terrain|Immeuble|Hôtel| ,))(?:.*\s(?P<piece>\d+)? pièces)?(?:[,\-] (?P<surface>\d+)m2?)? \((?P<de>\d+)€ -> (?P<a>\d+)€\)''', re.DOTALL).search

q=P("https://www.castorus.com/activite.php")

all_ = list(map(P,q("li.nodeborde")))

l = list(map(mdict,
    filter(
        bool,
        map(
            lambda t:s(t) and s(t).groupdict(),
            map(
                lambda e:e.text_content(), q("li.nodeborde"))))))
all_dpt = sum(map(
    lambda e: dict({
        e["dpt"] : mdict({
            'dpt' : ToxicSet([e["dpt"],]),
            'type' : mdict({e['type'] :1}),
            'total' :1,
            'piece' : mdict({e["piece"] or "0" : 1}),
            'prix':mdict(
                de=int(e["de"]), a=int(e["a"]))
            })}), l),
            mdict({}))
pp = lambda e:print(dumps(e, indent=4).decode("utf8"))
def pp(e):
    print("-"* 30)
    mmax = max(e.values())
    for k in sorted(map(lambda i:int(i) if i.isdecimal() else i, e.keys())):
        print("%13s :" % k, end='')
        print(" %6.2f%% %s" % (e[str(k)]*100, (int(e[str(k)]*50.0/mmax) * "*")))
    print("-"* 30)
import locale
def format(elt):
    print("repartition par type")
    totr = sum(elt["type"].values())
    pp(1.0 * elt["type"] / totr)
    print("repartition par piece")
    totr = sum(elt["piece"].values())
    pp(1.0 * elt["piece"] / totr)
    print("evolution", end=" : ")
    print("%+5.2f%%" % (100.0 * (elt["prix"]["a"] - elt["prix"]["de"]) / elt["prix"]["de"]), end="")
    print("(%s€/%s€)" % (
        by3(str(elt["prix"]["a"] - elt["prix"]["de"])),
        by3(str(elt["prix"]["a"]))))


total=mdict()
for _,dpt in all_dpt.items():
    print("\ndepartement  %s" % _)
    total += dpt
    format(dpt)

print("Pour la France")
format(total)