User:PotatoBot/Code/7

!/usr/bin/python
-*- coding: utf-8 -*-

import pywikibot as w

import scripts.noreferences, re, mysave

from datetime import date

PotatoBot Task 7: Adds Glottolog codes to language infoboxes

def main():

datapage = 'Wikipedia:WikiProject Languages/Glottolog 2.2 language names'

errNobox, errIsSection, errMultiISO, errWrongGlotto, errMultiISOGlotto, errWrongISO, errNoISO = [], [], [], [], [], [], []

errIO, refAdded = ,

refbot = scripts.noreferences.NoReferencesBot(None, True)

for glottoline in w.Page(w.getSite(), datapage).get().splitlines():

# glottoline == ' aari1239 [aiw] Aari'

i = glottoline.find('[[ISO 639:')

if i > 1:

w.output('')

if '|' in glottoline[i+24:]:

pipe = glottoline.find('|', i+24)

langpage = mysave.resolveredir(w.Page(w.getSite(), glottoline[i+24 : pipe]))

glottoname = glottoline[pipe+1 : glottoline.find(']', pipe)]

else:

glottoname = glottoline[i+24 : glottoline.find(']', i+24)]

langpage = mysave.resolveredir(w.Page(w.getSite(), glottoname))

glottocode = glottoline[1 : glottoline.find(' ', 1)]

isocode = glottoline[i+10 : i+13]

w.output(langpage.title() + ' (' + isocode + '/' + glottocode + '):')

langbox = False

paramISO3, paramLC, paramLD, paramGlotto, paramGlottoref = , [], [], [],

if langpage.exists() and '#' not in langpage.title():

for template in langpage.templatesWithParams():

if template[0].title() == 'Template:Infobox language' or template[0].title() == 'Template:Infobox Language':

langbox = True

for param in template[1]:

pp = param.partition('=')

pname = pp[0].strip()

pval = pp[2].strip()

if pname == 'iso3':

paramISO3 = pval

paramLC.append(pval)

elif pname[:2] == 'lc' and pname[2:].isdigit():

paramLC.append(pval)

elif pname[:2] == 'ld' and pname[2:].isdigit():

paramLD.append('"' + pval + '"')

elif pname[:6] == 'glotto' and (len(pname) == 6 or pname[6:].isdigit()):

paramGlotto.append(pval)

elif pname == 'glottorefname':

paramGlottoref = pval

if len(paramLD) > 0:

if len(paramLD) == 1: paramLC += paramLD

else: paramLC.append('several dialect names')

entry = '#' + langpage.title() + ' (' + isocode + '/' + glottocode + ')'

if not langbox:

if '#' in langpage.title():

w.output(' \03{yellow}redirect to section\03{default}')

errIsSection.append(entry)

else:

w.output(' \03{yellow}no language infobox\03{default}')

errNobox.append(entry)

elif len(paramLC) > 1:

if len(paramGlotto) == 0:

w.output(' \03{yellow}multiple ISO codes, but no glottocode\03{default}')

errMultiISO.append(entry + ': ' + ', '.join(paramLC))

else:

w.output(' \03{yellow}multiple ISO codes and glottocode(s)\03{default}')

errMultiISOGlotto.append(entry + ': ' + ', '.join(paramLC) + ' / ' + ', '.join(paramGlotto))

else:

if len(paramGlotto) == 0:

langtext = langpage.get()

isoparam = re.search(r'\|\s*iso3\s*=[^\|\}]*', langtext)

if isoparam and paramISO3 != '':

if paramISO3 != isocode:

w.output(' \03{yellow}mismatching ISO code\03{default}')

errWrongISO.append(entry + ': ' + paramISO3)

else:

langtext = langtext[:isoparam.end()] + '|glotto=' + glottocode + '\n' +\

('|glottorefname=' + glottoname + '\n') * (paramGlottoref == '') + langtext[isoparam.end():]

summary = 'Add Glottolog code ' + glottocode

w.output(' NoReferencesBot says: ', newline = False)

if refbot.lacksReferences(langtext + ''):

langtext = refbot.addReferences(langtext)

summary += ' and <references />'

refAdded += entry + '\n'

errIO += mysave.savepage(langpage, langtext, '7', summary)

else:

w.output(' \03{yellow}no ISO code\03{default}')

errNoISO.append(entry)

elif len(paramGlotto) > 1 or paramGlotto[0] != glottocode:

w.output(' \03{yellow}mismatching glottocode(s)\03{default}')

errWrongGlotto.append(entry + ': ' + ', '.join(paramGlotto))

errNobox.sort()

errIsSection.sort()

errWrongGlotto.sort()

errMultiISO.sort()

errMultiISOGlotto.sort()

errWrongISO.sort()

errNoISO.sort()

mysave.savepage(w.Page(w.getSite(), 'User:PotatoBot/Lists/Glottolog log'),

'Log for the addition of Glottolog codes to language infoboxes (Task 7). Date: ' +\

mysave.fmtdate(date.today()) + '.\n\n' + "Syntax:" + '\n' + '#Some language (ISO 639-3 code/glottocode, from [[' +\

datapage + '|database]]): ISO codes and dialect names / glottocodes, from {{tl|Infobox language}} (if applicable)\n\n{{TOC right}}\n' +\

'=== No language infobox ===\n' + '\n'.join(errNobox) + '\n\n' +\

'=== Redirect to section ===\n' + '\n'.join(errIsSection) + '\n\n' +\

'=== One ISO code and mismatching glottocode(s) ===\n' + '\n'.join(errWrongGlotto) + '\n\n' +\

'=== Multiple ISO codes, but no glottocode ===\n' + '\n'.join(errMultiISO) + '\n\n' +\

'=== Multiple ISO codes and mismatching, missing or surplus glottocode(s) ===\n' + '\n'.join(errMultiISOGlotto) + '\n\n' +\

'=== Mismatching ISO code ===\n' + '\n'.join(errWrongISO) + '\n\n' +\

'=== No ISO code ===\n' + '\n'.join(errNoISO) + '\n\n' +\

'=== <references /> added ===\n' + refAdded +\

'=== I/O error ===\n' + errIO, '7', 'Creating Glottolog log')

if __name__ == "__main__":

try:

main()

finally:

w.stopme()