Parser.py

from lxml import html
import requests, os


wowhead = "http://www.wowhead.com/item="
wantedStats = ['Agility or Strength or Intellect', 'Strength', 'Agility', 'Intellect', 'Attack Power', 'Spell Power', 'Haste', 'Critical Strike', 'Mastery', 'Versatility', 'Multistrike'] 

def getItemNumber(itemNumber=114979):
    if itemNumber == 114979:
        itemNumber = raw_input("Please enter the item number you wish to test:\n\t")

    try:
        return int(itemNumber)
    except:
        print "You did not enter a valid item number"
        itemNumber = getItemNumber()
        
def getFileName(fileName='blah'):
    if fileName == 'blah':
        fileName = raw_input("Please enter the Character-Spec that you want to test:\n\t")

    if fileName in os.listdir(os.getcwd()):
        return fileName
    else:
        print "You did not enter a valid file name"
        fileName = getFileName('blah')
        
def getItemScore(formattedStats):
    itemValue = 0
    for x in formattedStats:
        if x != formattedStats[len(formattedStats)-1]:
            statValue = x.split(' ')[0]
            statName = x.replace(statValue + ' ', '').replace('[', '').replace(']', '')
            if statName in wantedStats[0]:
                itemValue += statRatios[wantedStats[0]] * float(statValue)
            else:
                itemValue += statRatios[statName] * float(statValue)
    return itemValue

def getStatRatios():
    statWeights = {}
    character = open(getFileName(), 'r+')
    for line in character:
        for stat in wantedStats:
            if stat in line:
                statWeights[stat] = float(line.split(' ')[0])
                if stat in wantedStats[0]:
                    statWeights[wantedStats[0]] = float(line.split(' ')[0])
    character.close()
    return statWeights
            
def queryItem(itemQueryNumber):
    url=wowhead+"%s/"%itemQueryNumber
    page = requests.get(url)
    tree = html.fromstring(page.content)
    mainStat = tree.xpath('//span/text()')
    name = str(tree.xpath('//title/text()')[0]).split('-')[0]
    
    formattedStats = []
    for x in mainStat:        
        potentialStat = x
        try:
            potentialStat = str(x)
            if 'Strength' in potentialStat or 'Agility' in potentialStat or 'Intellect' in potentialStat:
                potentialStat.replace('+','').replace('[','').replace(']','')
                formattedStats.append(potentialStat)
            if ') Set' in potentialStat and ' set ' not in name:
                name += '!!! has a set bonus!!!'
            if 'Use' in potentialStat and ' use 'not in name:
                name += '!!! has an on use effect!!!'
            if 'Equip:' in potentialStat and ' on proc 'not in name:
                name += '!!! has an on proc effect!!!'
        except:
            potentialStat = x.encode('ascii', 'ignore')
            formattedStats.append(potentialStat)
        
    formattedStats.append(name)
    return formattedStats

statRatios = getStatRatios()
    
itemQueryOne = getItemNumber()
statListOne = queryItem(itemQueryOne)
itemScoreOne = getItemScore(statListOne)

itemQueryTwo = getItemNumber()
statListTwo = queryItem(itemQueryTwo)
itemScoreTwo = getItemScore(statListTwo)

print "\n\t" + statListOne[len(statListOne)-1] + "\n\t vs. \n\t" + statListTwo[len(statListTwo)-1]

if itemScoreOne > itemScoreTwo:
    print "\n%s is better"%statListOne[len(statListOne)-1]
else:
    print "\n%s is better"%statListTwo[len(statListTwo)-1]