Wikipedysta:BotOks/AntyColorbox

# -*- coding: utf-8  -*-

import pdb

import string
import sys
import io
import time
from re import search, split, match, sub
import re

sys.path.insert(0, 'pywikipedia')
sys.path.insert(0, '.')

del sys

import wikipedia, config

log = None
operations = None
page_name = None

unrecognized = []

class Early(Exception):
    None

def translate(color, is_last):
    global unrecognized
    #              hex short/long,    textual,    similar/misspellings
    if   color in ['#00f', '#0000ff', 'blue',     '#00008b', '#0000cc', '#0000cd', '#0000dd', '#1c80c6', '#00bfff', '#0099ff', '#007fff', '#00cfff', '#87cefa', '#84cfff', '#2898ff', '#3333ff', '#023d8e', '#054aa7', '#0033ff', '#003fff']: # #00008b uznaję za niebieski, jako że niedokładnie odpowiada granatowi i uznaję, że ktoś kombinował z dobraniem barwy
        return u'niebieskie' if is_last else u'niebiesko-'
    elif color in ['#fff', '#ffffff', 'white',    'whtie', '#fffffff', 'ffffff']:
        return u'białe' if is_last else u'biało-'
    elif color in ['#000', '#000000', 'black',    '#111', '#111111']:
        return u'czarne' if is_last else u'czarno-'
    elif color in ['#ff0', '#ffff00', 'yellow',   '#fff000', '#fbfc71', '#fef602', '#fff300', '#fff301', '#fdee00', '#ef9b0f']:
        return u'żółte' if is_last else u'żółto-'
    elif color in ['#f00', '#ff0000', 'red',      '#ff0010', '#ff2000', '#ff2400', '#ff3030', '#cc0000', '#dd0000']:
        return u'czerwone' if is_last else u'czerwono-'
    elif color in [        '#008000', 'green',    '#0f0', '#00ff00', '#008800', '#008744', '#008600', '#009900', '#009600', '#006400', '#006600', '#006633', '#228b22', '#3a9d23', '#00c040', '#0f8c0a', '#017b2e']:
        return u'zielone' if is_last else u'zielono-'
    elif color in [        '#000080', 'navy', 'darkblue']:
        return u'granatowe' if is_last else u'granatowo-'
    elif color in ['#ffa500',         'orange',   '#f60', '#ff6600', '#ff8c00', '#ffa000', '#ff8800', '#ff7300', '#ff6c00', '#b803ff', '#fe7f00', '#ec5302']:
        return u'pomarańczowe' if is_last else u'pomarańczowo-'
    elif color in [                   'violet',   '#b803ff']:
        return u'fioletowe' if is_last else u'fioletowo-'
    elif color in [        '#800000']:
        return u'bordowe' if is_last else u'bordowo-'
    elif color in [        '#800080', 'purple']:
        return u'purpurowe' if is_last else u'purpurowo-'
    elif color in [                   'gold']:
        return u'złote' if is_last else u'złoto-'
    elif color in [                   'silver']:
        return u'srebrne' if is_last else u'srebrno-'
    else:
        unrecognized.append(color)
        raise Exception("Can't translate " + color + ".")

RE_INNER_TPL  = ur"\{\{[^}]*\}\}"
RE_LINK       = ur"\[\[[^]]*\]\]"
RE_PARAM      = ur"\|(?:%s|%s|[^{}|])*" % (RE_INNER_TPL, RE_LINK)

RE_KS_INFOBOX = ur"\{\{Klub sportowy infobox\s*(%s)*\}\}" % RE_PARAM
RE_COLORBOX   = lambda capture_color: ur"\{\{\s*colorbox\s*\|(%s#?[a-z0-9]+)\s*\}\}" % (capture_color and ur"?P<color>" or ur"?:")

barwy_found = False

def fix_param(p):
    global barwy_found, page_name, operations
    if not search(u"=", p.group(0)):
        return p.group(0) # Czasami brakuje znaku =, tego błędu nie poprawiamy.
    name_part, val_part = p.group(0).split(u'=', 1)
    if name_part[1:].strip() != u"barwy":
        return p.group(0)
    barwy_found = True
    out = u""
    colorbox_found = False
    other_text_found = False

    tokenized = split(ur"(%s)" % RE_COLORBOX(False), val_part, flags=re.I)
    tokenized.reverse()
    for i in range(len(tokenized)):
        token = tokenized[i]
        m = match(ur"\A%s\Z" % RE_COLORBOX(True), token, flags=re.I)
        if m:
            out = translate(m.group('color').lower(), not colorbox_found) + out
            colorbox_found = True
        elif match(ur"\A\s*\Z", token):
            if i in [0, len(tokenized) - 1]:
                out = token + out
        else:
            other_text_found = True

    if colorbox_found and other_text_found:
        raise Exception(u"Extra text present: '%s'" % val_part.strip())
    elif colorbox_found:
        operations.write(u"%s %s -> %s\n" % (page_name.ljust(25, '.'), val_part.strip(), out.strip()))
        return u"=".join((name_part, u"".join(out)))
    else:
        raise Early(u"No colorbox found, just text: '%s'" % val_part.strip())

def fix_template(t):
    return sub(RE_PARAM, fix_param, t.group(0), flags=re.I)

def fix(src):
    global barwy_found
    barwy_found = False
    new = sub(RE_KS_INFOBOX, fix_template, src, flags=re.I)
    if barwy_found:
        return new
    else:
        raise Early(u"No parameter 'barwy' found")

def run(page):
    global page_name
    if page.title().find(':') < 0:
        time.sleep(1)
        old = page.get()
        page_name = page.title()

        try:
            new = fix(old)
            page.put(new,  maxTries=1, comment=u"Robot naprawia szablon: Wyrzucenie colorboksów z infoboksów, zlecenie [[WP:ZDBOT]] (''Usunięcie colorboksu z szablonów klubów sportowych (barwy klubu), być może innych podobnych''), szczegóły [[Wikipedysta:BotOks/AntyColorbox]].")
            msg = u"DONE:  %s" % page.title()
            time.sleep(5)
        except Early as ex:
            msg = u"GOOD:  %s %s" % (page.title().ljust(25, '.'), unicode(ex))
        except Exception as ex:
            msg = u"FAIL:  %s %s" % (page.title().ljust(25, '.'), unicode(ex))
    else:
        msg = u"OMIT:  %s" % page.title()
    print msg
    log.write(msg + '\n')


def main():
    global operations, log, unrecognized
    try:
        log = io.open('logs.txt', 'w')
        operations = io.open('operations.txt', 'a')
        klub_tpl = u"Szablon:Klub sportowy infobox"
        colb_tpl = u"Szablon:Colorbox"

        wikipedia.handleArgs()
        site = wikipedia.getSite()

        klub_page = wikipedia.Page(site, klub_tpl)
        colb_page = wikipedia.Page(site, colb_tpl)

        colb_refs = [page for page in colb_page.getReferences()]

        intersection = [page for page in klub_page.getReferences() if page in colb_refs]

        for page in intersection:
           run(page)

        unrecognized.sort()
        unrecognized_rows = map(lambda s: u"|{{colorbox|%s}}\n|%s\n|-\n" % (s, s), unrecognized)
        unrecognized_text = u"".join([u"{|\n! Wzorzec !! Zapis\n|-\n"] + unrecognized_rows + [u"|}"])
        unrec_page = wikipedia.Page(site, u"Wikipedysta:BotOks/AntyColorbox/doInterpretacji")
        unrec_page.put(unrecognized_text, maxTries=1, comment=u"Robot nie poradził sobie z następującymi kolorami.")

    finally:
        log.close()
        operations.close()

if __name__ == "__main__":
    try:
        main()
    finally:
        wikipedia.stopme()