cards2articles.py

#!/usr/bin/env python

# time python pwb.py scripts/pagefromfile.py -notitle -force -file:/home/fgombault/Sync/duelyst/duelyst-api/wiki_articles.txt

import json
import re

def getCardImageName(p):
    return p['name'].replace(" ", "_")


def getRace(p):
    if (p['category'] == 'unit'):
        return p['type'] # should contain "Minion", "Arcanyst", "Golem"...
    else:
        return u''


def getType(p):
    return unicode.upper(p['category'][0]) + p['category'][1:]


def boldKeywords(str):
    output = ""
    matches = re.findall(r'<b>[^<>]*</b>', str)
    for m in matches:
        candidate = re.sub(r'<b>', '[[', re.sub(r'</b>', ']] ', m))
        if (not candidate in output):
            output = output + candidate
    words = output.split()
    return " ".join(sorted(set(words), key=words.index))


def getAbilities(p):
    if (p['category'] != 'unit'):
        return ""
    try:
        shortdesc = p['description'][0:p['description'].rindex('<br')]
    except ValueError:
        shortdesc = p['description']
    return boldKeywords(shortdesc)


def getTags(p):
    output = u"[[%s]] " % p['faction']
    output += boldKeywords(p['description'])

    # Find some keywords in the text
    if (re.search(r'Dispel', p['description'], re.IGNORECASE)):
        output += u' [[Dispel]]'
    if (re.search(r'Draw', p['description'], re.IGNORECASE)):
        output += u' [[Draw]]'

    # Unify some word variants
    output = re.sub(r"Stunned", "Stun", output)
    return output


def getCardSet(p):
    if (p['setName'] == 'Core'):
        return 'Core'
    if (p['setName'] == 'Basic'):
        return 'Core'
    if (p['setName'] == 'Ancients'):
        return 'Ancient Bonds'
    if (p['setName'] == 'ShimZar'):
        return "Denizens of Shim'Zar"
    if (p['setName'] == 'Unearthed'):
        return 'Unearthed Prophecy'
    if (p['setName'] == 'Immortal'):
        return 'Immortal Vanguard'
    if (p['setName'] == 'Mythron'):
        return 'Trials of Mythron'
    # default value
    return p['setName']


def getDescription(p):
    #TODO: replace *bold* words with [[wiki links]]
    output = re.sub(r'<b>', "<b>[[", re.sub(r'</b>', ']]</b>', p['description']))
    return output


def getCategories(p):
    cats = getTags(p)
    cats = re.sub(r'\[\[', "[[Category:", cats)
    if (getType(p) != ''):
        cats += " [[Category:%s]]" % getType(p)
    if (getRace(p) != ''):
        cats += " [[Category:%s]]" % getRace(p)
    return cats

with open('cards.json') as json_file:
    data = json.load(json_file)
    for p in data:
        if p['id'] < 1000000:
            print(u"{{-start-}}")
            print(u"'''Data:Cards/" + p['name'] + "'''")
            print(u"This page is maintained by bots. MANUAL CHANGES '''WILL''' BE OVERWRITTEN.")
            print(u'<onlyinclude>')
            print(u'{{Card Metadata')
            print(u'| name = ' + p['name'])
            print(u'| image = ' + getCardImageName(p))
            print(u'| set = ' + getCardSet(p))
            print(u'| faction = ' + p['faction'])
            print(u'| rarity = ' + p['rarity'])
            print(u'| type = ' + getType(p))
            print(u'| race = ' + getRace(p))
            try:
                print(u'| cost = %d' % p['mana'])
            except: KeyError
            try:
                print(u'| attack = %d' % p['attack'])
                print(u'| hp = %d' % p['hp'])
            except: KeyError
            print(u'| abilities = ' + getAbilities(p))
            print(u'| tags = ' + getTags(p))
            print(u'| desc = ' + getDescription(p))
            print(u'| link = ')
            print(u'| id = %d' % p['id'])
            print(u'| rotation = standard')
            print(u'| available = %s' % p['available'])
            print(u'| hidden = %s' % p['hidden'])
            print(u"}}")  # End of Card Metadata
            print(u'</onlyinclude>')
            print(getCategories(p))
            # print(u'\n{{Display Card Metadata}}')
            print(u"{{-stop-}}")