Quick and Dirty Parser for D&D Weapons

weaponParse.py

import sqlite3
from bs4 import BeautifulSoup
from collections import defaultdict


conn = sqlite3.connect("C:\Users\Phill\Dropbox\code\python\compendium\items.db")
c = conn.cursor()

count = 0
for row in c.execute("SELECT * FROM weapondata LIMIT 5"):
    soup = BeautifulSoup(row[2])

    wpn = defaultdict(lambda: None)

    wpn["id"] = row[0]
    wpn["name"] = row[1]

    isDamage = False
    
    for detail in soup.div.contents:
        sdetail = detail.encode("utf-8")

        if sdetail.startswith("<h1"):
            continue
        if sdetail == "<br/>":
            continue
        
        if isDamage:
            isDamage = False
            wpn["damage"] = detail[2:]
            continue
        
        
        weaponSkill = ("Simple", "Military","Superior", "Improvised")
        isSkills = reduce(lambda res, sk: res or sk, map(lambda x: str(detail.encode("utf-8")).startswith(x) and str(detail.encode("utf-8")).endswith("weapon"), weaponSkill))
        
        if isSkills == True:
            words = detail.split(" ")
            wpn["skill"] = words[0]
            wpn["numHands"] = words[1]
            wpn["range"] = words[2]
            continue;

        if sdetail.endswith("gp"):
            wpn["cost"] = float(sdetail.replace("Cost:", "").replace("\xe2\x80\x94", "0").replace("gp",""))
            continue

        if sdetail.startswith("<b>Damage</b>"):
            isDamage = True
            continue

        if sdetail.startswith("Proficient:"):
            wpn["proficient"] = int(sdetail.replace("Proficient:", ""))
            continue

        if sdetail.startswith("Weight:"):
            wpn["weight"] = float(sdetail.replace("Weight:", "")
                                .replace("lb.","")
                                .replace("1\xe2\x80\x93","")
                                .replace("\xe2\x80\x94","0")
                                .replace("6\xe2\x80\x93",""))
            continue
        
        if sdetail.startswith("Range:"):
            if sdetail.count("-") > 0:
                continue
            wpn["srange"] = int(sdetail.replace("Range:", "").split("/")[0])
            wpn["lrange"] = int(sdetail.replace("Range:", "").split("/")[1])
            continue

        if sdetail.endswith("("):
            print "PROPERTYGROUPFOUND!"
        print detail  
    print wpn

conn.commit()
c.close()