#!/usr/bin/python3

import elementtree.ElementTree as ET
import os, os.path

SrcEnc = 'utf_8'
TargetEnc = 'latin1'

def convertBillFile(srcFile, dstFile):
    data = str(open(srcFile, 'r').read(), SrcEnc)
    data = data.encode('utf-8')
    et = ET.XML(data)
    num = int(getDataFromXPath(et, 'vote-metadata/rollcall-num'))
    billName = getDataFromXPath(et, 'vote-metadata/legis-num')
    title = getDataFromXPath(et, 'vote-metadata/vote-desc')
    voteType = getDataFromXPath(et, 'vote-metadata/vote-question')
    dateTime = getDataFromXPath(et, 'vote-metadata/action-date') + ' ' + getDataFromXPath(et, 'vote-metadata/action-time')
    dstF = open(dstFile, 'w')
    dstF.write(("%d\n%s\n%s\n%s\n%s" % (num, billName, title, voteType, dateTime)).encode(TargetEnc, 'replace'))
    dstF.close()

def convertVotesFile(srcFile, dstFile, legisFile):
    data = str(open(srcFile, 'r').read(), SrcEnc)
    data = data.encode('utf-8')
    et = ET.XML(data)
    votes = et.findall('vote-data/recorded-vote')
    dstF = open(dstFile, 'w')
    legisExists = True
    if (not os.path.exists(legisFile)):
        legisExists = False
        legisF = open(legisFile, 'w')
    #print votes
    for vote in votes:
        legisTODO = vote.find('legislator').get('unaccented-name')
        #legis = vote.find('legislator').text.encode(TargetEnc, 'replace')
        v = vote.find('vote').text
        # Wow, this is annoying.
        if (v == 'Aye'):
            v = 'Yea'
        elif (v == 'No'):
            v = 'Nay'
        dstF.write("%s:%s\n" % (legisTODO, v))
        if (not legisExists):
            id = vote.find('legislator').get('name-id')
            party = vote.find('legislator').get('party')
            state = vote.find('legislator').get('state')
            legisTODO = vote.find('legislator').get('unaccented-name')
            legisF.write("%s:%s:%s:%s\n" % (legisTODO, id, party, state))
    dstF.close()
    if (not legisExists):
        legisF.close()

def getDataFromXPath(et, xpath):
    elem = et.find(xpath)
    if (elem != None and elem.text != None):
        return elem.text
    else:
        return ""

def main():
    srcDir = 'votes'
    dstDir = 'parsedVotes'
    #data = open(srcDir + '/2006/roll002.xml', 'r').read()
    #convertVotesFile(srcDir + '/2006/roll002.xml', dstDir + '/2006/roll002.votes.txt')
    dirs = os.listdir(srcDir)
    #dirs = []
    for dir in dirs:
        for file in os.listdir(os.path.join(srcDir, dir)):
            if (file.endswith('.xml')):
                print(os.path.join(srcDir, dir, file))
                convertBillFile(os.path.join(srcDir, dir, file), os.path.join(dstDir, dir, file[:-4] + '.bill.txt'))
                convertVotesFile(os.path.join(srcDir, dir, file), os.path.join(dstDir, dir, file[:-4] + '.votes.txt'), os.path.join(dstDir, dir, "legisinfo.txt"))
    #convertFile(srcDir + '/2006/roll001.xml', dstDir + '/2006/roll001.bill.txt')

if (__name__ == '__main__'):
    main()