De Wiki do Arch Linux Brasil

#!/usr/bin/env python
# coding: utf-8

#### Baixa um arquivo texto com cordenadas de um mapa e imprime na tela
#### o código kml (xml usado no googlemaps)
 
##	Autor:	Audren Cezar
##	E-mail: audrencezar arroba gmail ponto com
## 	Última alteração:	2008-07-05
## 	Depende de:
##		python

from string import Template
import re
from urllib2 import urlopen

class Text2Kml:
    #<latitude> <longitude> <nome> # <cidade>, <estado>
    #<latitude> <longitude> [[User:<nome_de_usuario_no_wiki>|<nome>]] # <cidade>, <estado>
    regex = re.compile(r'\* (?P<lat>-?\d+\.\d+) (?P<lon>-?\d+\.\d+) (\")?(\[\[User\:(?P<wikiusername>[^\|]+)\|)?(?P<name>.+?)(\]\])?(\"?) # +(?P<description>.*)')
    
    header_template = Template("""<?xml version="1.0" encoding="UTF-8"?>
<kml xmlns="http://earth.google.com/kml/2.2">
<Document>
  <name>${mapname}</name>
  <description><![CDATA[${mapdescription}]]]></description>
  <Style id="style1">
${style}
  </Style>
  """)

    placemark_template = Template("""  <Placemark>
    <name>${name}</name>
    <description><![CDATA[<div dir="ltr">${description}</div>]]></description>
    <styleUrl>#style1</styleUrl>
    <Point>
      <coordinates>${lon},${lat},0.000000</coordinates>
    </Point>
  </Placemark>""")

    footer = """</Document>
</kml>
"""
    wikilink_template = Template('; <a href="${wikiurl}${wikiusername}">${name}</a>')

    def __init__(self, url, mapname, mapdescription, wikiurl=None, style=""):
        self.url = url
        self.wikiurl = wikiurl
        self.mapname = mapname
        self.mapdescription = mapdescription
        self.style = style
       
    def parse(self):
        print(self.header_template.substitute({
            'mapname': self.mapname,
            'mapdescription': self.mapdescription,
            'style': self.style
        }))
        
        file = urlopen(self.url)
        for line in file:
            res = self.regex.match(line)
            if res:
                d = res.groupdict()
                if self.wikiurl and d['wikiusername']:
                    d['wikiurl'] = self.wikiurl
                    d['description'] += self.wikilink_template.substitute(d)
                print(self.placemark_template.substitute(d))
        file.close()

    def parsey(self):
        yield self.header_template.substitute({
            'mapname': self.mapname,
            'mapdescription': self.mapdescription,
            'style': self.style
        })
        
        file = urlopen(self.url)
        for line in file:
            res = self.regex.match(line)
            if res:
                d = res.groupdict()
                if self.wikiurl and d['wikiusername']:
                    d['wikiurl'] = self.wikiurl
                    d['description'] += self.wikilink_template.substitute(d)
                yield self.placemark_template.substitute(d)
        file.close()

        
        yield self.footer
            
if __name__ == "__main__":
    style = """<IconStyle>
      <Icon>
        <href>http://img379.imageshack.us/img379/8628/arch2jb7.png</href>
      </Icon>
    </IconStyle>"""
    
    parser = Text2Kml(
        "http://wiki.archlinux-br.org/Especial:Exportar/ArchersNoBrasil",
        "ArchersNoBrasil",
        "Usuários Arch Linux no Brasil",
        "http://wiki.archlinux-br.org/Usuário:",
        style)
    
    #parser.parse()
    # uso .xml porque alguns servidores não reconhecem a extensão .kml
    wfile = open("ArchersNoBrasil.kml.xml", 'w')
    for p in parser.parsey():
        wfile.write(p)
    wfile.close()