def getAdditionalInfo(jardin): jardinInfo = {} soup = request.get_content_parsed(jardin["additionalInfo"]) detalles = parse(soup) jardinInfo["url"] = jardin["additionalInfo"] jardinInfo["info"] = detalles return jardinInfo
def getAdditionalInfo(diputado): url = "https://www.hcdiputados-ba.gov.ar/includes/undiputado.php?c_codigo={}&urlImagen=&incluyeIncludes=includes/".format(diputado["codigo"]) soup = request.get_content_parsed(url) diputadoInfo = parse(soup) diputadoInfo["img"] = diputado["img"] diputadoInfo["codigo"] = diputado["codigo"] return diputadoInfo
def main(): data = {} soup = request.get_content_parsed("https://guia-capital-federal.escuelasyjardines.com.ar/guia-jardines-de-infantes-en-capital-federal-belgrano.htm") paginas = parsePaginator(soup) data["jardines"] = parseJardines(soup) data["paginas"] = paginas for pagina in paginas: data["jardines"] = data["jardines"] + parseJardinesPerPage(pagina) files.save_as_json_2('jardines.json', data)
def getAdditionalInfo(senador): url = "https://www.senadodecatamarca.gob.ar" + senador["additionalInfo"] soup = request.get_content_parsed(url) senadorInfo = parse(soup) senadorInfo["img"] = senador["img"] senadorInfo["additionalInfo"] = senador["additionalInfo"] senadorInfo["bloque"] = senador["bloque"] senadorInfo["distrito"] = senador["distrito"] senadorInfo["nombre"] = senador["nombre"] return senadorInfo
def parseCandidato(nombreCandidato): data = {} soup = request.get_content_parsed("http://www.senadoctes.gov.ar/" + nombreCandidato) tds = soup.find_all("td", class_="arryabaazul") data["img"] = "http://www.senadoctes.gov.ar/" + tds[0].find("img")["src"] strongTags = tds[1].find_all("font") data["nombre"] = strongTags[0].text.replace( "\n", "").strip().split(":")[1].strip() data["mandato"] = strongTags[2].text.replace("\n", "") return data
def main(): soup = request.get_content_parsed( "http://www.legiscba.gob.ar/legisladores/") candidates = parse(soup) files.save_as_json('sources/legisladoresCordoba.json', candidates)
def getDiputadosByBloque(url, nombreBloque): soup = request.get_content_parsed(url) return __parse(soup, nombreBloque)
def main(): soup = request.get_content_parsed( "http://www.legislaturachaco.gov.ar/sitio/legisladores.php") candidates = parse(soup) files.save_as_json('sources/legisladoresChaco.json', candidates)
def main(): soup = request.get_content_parsed("https://www.hcdiputados-ba.gov.ar/index.php?id=diputados&id_menu=mandatovigente") candidates = parse(soup) files.save_as_json('sources/buenosAires/diputadosBuenosAires.aux.json', candidates)
def main(): soup = request.get_content_parsed( "http://www.senado-ba.gov.ar/Senadores.aspx") senadores = parse(soup) files.save_as_json('sources/provinciales/senadoresBuenosAires.json', senadores)
def parseJardinesPerPage(pagina): soup = request.get_content_parsed(pagina) return parseJardines(soup)
def main(): soup = request.get_content_parsed( "http://www.legischubut2.gov.ar/index.php?option=com_content&view=article&id=1254&Itemid=171" ) candidates = parse(soup) files.save_as_json('sources/chubut/diputados.json', candidates)
def main(): soup = request.get_content_parsed( "http://www.legislaturadetucuman.gob.ar/infoseccion.php?seccion=0") candidates = parse(soup) files.save_as_json('sources/legisladoresTucuman.json', candidates)
def main(): soup = request.get_content_parsed("http://www.senadodecatamarca.gob.ar/informacion-institucional/senadores") candidates = parse(soup) files.save_as_json('sources/catamarca/senadores.aux.json', candidates)