Belediye meclisi vs. Büyükşehir oyları anomali kontrol

sts_meclis_vs_buyuksehir.py

import urllib
import html5lib
from html5lib import treebuilders
from xml.etree import cElementTree

import sys
import time

import argparse

STS_URL = "http://sts.chp.org.tr/SonucDetay.aspx?sid=%s"
TAB_PATH = ".//{http://www.w3.org/1999/xhtml}div[@id='TabContainer_TabPanel%s']"
ROW_PATH = ".//{http://www.w3.org/1999/xhtml}div[@class='chp-vote-row']"

# 205407, 217637

LOCATIONS = {
    'akyurt': (205407, 205482),
    'altindag': (205483, 206246),
    'ayas': (206292, 206345),
    'bala': (206346, 206445),
    'beypazari': (206446, 206612),
    'camlidere': (206613, 206664),
    'cankaya': (206747, 208739),
    'cubuk': (208871, 209111),
    'elmadag': (209112, 209224),
    'etimesgut': (209277, 210240),
    'evren': (210272, 210287),
    'golbasi': (210288, 210555),
    'gudul': (210556, 210596),
    'haymana': (210597, 210738),
    'kalecik': (210739, 210807),
    'kazan': (210808, 210936),
    'kecioren': (211747, 211005),
    'kizilcahamam': (212890, 213039),
    'mamak': (213040, 214337),
    'nallihan': (214338, 214471),
    'polatli': (214472, 214786),
    'pursaklar': (214787, 215055),
    'sincan': (215056, 216122),
    'sereflikochisar': (216123, 216239),
    'yenimahalle': (216275, 217637),
    'HEPSI': (205407, 217637)
}

TABS = {
    'belediye_meclis': 3,
    'ilce': 2,
    'buyuksehir': 1
}

parser = argparse.ArgumentParser(description='Search task (retrieving in the background, saving data required for plotting')
parser.add_argument('--yer', required=True, choices=LOCATIONS.keys(), help='Yer (ilce)')
parser.add_argument('--esik', default=50, type=int, help='Anomali esik degeri (alt limit)')

def breakdown(row):
    imgs = row.findall(".//{http://www.w3.org/1999/xhtml}img")
    if not imgs:
        return {}
    img = imgs[0]
    src = img.attrib['src']
    lastPart = src.split('/')[-1]
    inputt = row.findall(".//{http://www.w3.org/1999/xhtml}input")[0]
    val = 0
    if inputt.attrib.has_key('value'):
        val = int(inputt.attrib['value'])
    return {'parti': lastPart.rstrip('.gif'), 'oy': val}

def getResults(tabDoc):
    rows = tabDoc.findall(ROW_PATH)
    results = {}
    for row in rows:
        r = breakdown(row)
        if not r:   break
        results[r['parti']] = r['oy']
    return results

def parse(feedHTML):
    p = html5lib.HTMLParser(tree=treebuilders.getTreeBuilder('etree', cElementTree))
    doc = p.parse(feedHTML)
    # locate topic list
    tabResults = {}
    for tabName, tabNo in TABS.items():
        tabDocs = doc.findall(TAB_PATH % tabNo)
        if not tabDocs:  continue
        tabDoc = tabDocs[0]
        tabResults[tabName] = getResults(tabDoc)
    return tabResults

def check(placeId, limit):
    url = STS_URL % placeId
    text = urllib.urlopen(url).read()
    tabResults = parse(text)
    if tabResults:
        meclisResults = tabResults['belediye_meclis']
        buyuksehirResults = tabResults['buyuksehir']
        for bKey,bVal in buyuksehirResults.items():
            if not bKey in meclisResults:   continue
            diff = meclisResults[bKey]-bVal
            if diff > limit:
                print "STS-ID %s --- %s --- belediye meclisi: %s, buyuksehir: %s --- FARK: %s" % (i, bKey.upper(), meclisResults[bKey], bVal, diff)
    time.sleep(1)

if __name__ == "__main__":
    args = parser.parse_args()
    l = LOCATIONS[args.yer]
    for i in range(l[0],l[1]):
        check(placeId=i, limit=args.esik)