Obtém handle de redes sociais de candidatos (Twitter, Facebook, Instagram e Youtube) a partir de informações do Repositório de Dados Eleitorais do TSE

redes_sociais_candidatos2022.R

################################################################################
################################################################################
##########
##########  OBTER HANDLE DE REDES SOCIAIS DE CANDIDATOS
##########  A PARTIR DE INFORMACOES DO REPOSITORIO DO TSE

### Lucas Gelape

# Pacotes
library(dplyr)
library(readr)
library(stringr)
library(stringi)

################################################################################

##### BAIXAR E ABRIR DADOS DE 2022

download.file(
  "https://cdn.tse.jus.br/estatistica/sead/odsele/consulta_cand/rede_social_candidato_2022.zip",
  "redesocial2022.zip")

unzip("redesocial2022.zip",
      files = "rede_social_candidato_2022.csv")

file.remove("redesocial2022.zip")

redes2022 <- read_csv2(
  "rede_social_candidato_2022.csv",
  locale = locale(encoding = "latin1"))

################################################################################

##### IDENTIFICAR PRINCIPAIS REDES SOCIAIS E EXCLUIR DEMAIS LINHAS

# Selecionamos somente Instagram, Facebook, Twitter e Youtube
# Procedimentos semelhantes podem ser usados para outras redes

redes2022 <- redes2022 %>%
  # Mantem somente as quatro redes acima
  filter(
    str_detect(
      DS_URL, 
      paste("instagram", "facebook", "twitter", "youtube", sep = "|"))) %>%
  mutate(
    # Cria uma variavel que identifica qual a rede
    rede = case_when(
      str_detect(DS_URL, "facebook")  ~ "facebook",
      str_detect(DS_URL, "instagram") ~ "instagram",
      str_detect(DS_URL, "twitter")   ~ "twitter",
      str_detect(DS_URL, "youtube")   ~ "youtube"),
    
    # Guardar a URL_ORIGINAL pois vamos limpar DS_URL
    URL_ORIGINAL = DS_URL,
    
    # Limpeza inicial do texto da URL
    DS_URL = str_squish(DS_URL),
    DS_URL = tolower(DS_URL),
    DS_URL = stri_trans_general(DS_URL, "Latin-ASCII"))

################################################################################

##### LIMPAR TEXTO DESSAS QUATRO REDES

### YOUTUBE

youtube <- redes2022 %>%
  filter(rede == "youtube") %>%
  mutate(
    
    # Geral
    DS_URL = str_replace_all(DS_URL, "www,", "www\\."),
    DS_URL = str_replace_all(DS_URL, "/m\\.", "/"),
    DS_URL = str_remove_all(DS_URL, "https://"),
    DS_URL = str_remove_all(DS_URL, "http://"),
    DS_URL = str_remove_all(DS_URL, "www\\."),
    
    # Youtube
    DS_URL = ifelse(
      str_detect(DS_URL, "watch"),
      NA, DS_URL),
    
    DS_URL = str_remove_all(DS_URL, "youtube\\.com/channel/"),
    DS_URL = str_remove_all(DS_URL, "youtube\\.com/c/"),
    DS_URL = str_remove_all(DS_URL, "youtube\\.com/"),
    DS_URL = str_remove_all(DS_URL, "youtube\\.com\\.br/"),
    DS_URL = str_remove_all(DS_URL, "youtube\\/"),
    DS_URL = str_remove_all(DS_URL, "user/"),
    DS_URL = str_remove_all(DS_URL, "/about"),
    DS_URL = str_remove_all(DS_URL, "/featured"),
    
    # Remove tudo depois de /?
    DS_URL = str_remove_all(DS_URL, "(?=\\/\\?).*"),
    # Remove tudo depois de /?
    DS_URL = str_remove_all(DS_URL, "(?=\\?).*"),
    # Remove tudo depois de /
    DS_URL = str_remove_all(DS_URL, "(?=\\/).*"),
    # Remove tudo ate "- "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\- )"),
    # Remove tudo ate ": "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\: )"),
    
  )

##### INSTAGRAM

instagram <- redes2022 %>%
  filter(rede == "instagram") %>%
  mutate(
    
    DS_URL = ifelse(DS_URL == "instagram", NA, DS_URL),
    rede = ifelse(DS_URL == "instagram", NA, rede),
    
    # Geral
    DS_URL = str_replace_all(DS_URL, "www,", "www\\."),
    DS_URL = str_replace_all(DS_URL, "/m\\.", "/"),
    DS_URL = str_replace_all(DS_URL, "htpps", "https"),
    DS_URL = str_remove_all(DS_URL, "https://"),
    DS_URL = str_remove_all(DS_URL, "http://"),
    DS_URL = str_remove_all(DS_URL, "www\\."),
    DS_URL = str_remove_all(DS_URL, "\\.br"),
    
    # Instagram
    DS_URL = ifelse(
      str_detect(DS_URL, "instagram\\.com/p/"),
      NA, DS_URL),
    DS_URL = ifelse(
      str_detect(DS_URL, "\\l\\.instagram\\.com"),
      NA, DS_URL),
    DS_URL = ifelse(
      str_detect(DS_URL, "invites"),
      NA, DS_URL),
    DS_URL = ifelse(
      str_detect(DS_URL, "reel"),
      NA, DS_URL),
    
    DS_URL = str_remove_all(DS_URL, "instagram\\.com/accounts/login/\\?next=/"),
    DS_URL = str_remove_all(DS_URL, "instagram\\.com/account/login/\\?next=/"),
    DS_URL = str_remove_all(DS_URL, "instagram\\.com/:"),
    DS_URL = str_remove_all(DS_URL, "instagram\\.com/"),
    DS_URL = str_remove_all(DS_URL, "instagram/"),
    
    # Remove tudo depois de /?
    DS_URL = str_remove_all(DS_URL, "(?=\\/\\?).*"),
    # Remove tudo depois de ?
    DS_URL = str_remove_all(DS_URL, "(?=\\?).*"),
    # Remove tudo depois de /
    DS_URL = str_remove_all(DS_URL, "(?=\\/).*"),
    # Remove tudo ate "- "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\- )"),
    # Remove tudo ate ": "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\: )"),
    
    # Remove @
    DS_URL = str_remove_all(DS_URL, "\\@"),
    
  )

################################################################################

##### FACEBOOK

facebook <- redes2022 %>%
  filter(rede == "facebook") %>%
  mutate(
    
    DS_URL = ifelse(DS_URL == "facebook", NA, DS_URL),
    rede = ifelse(DS_URL == "facebook", NA, rede),
    
    # Geral
    DS_URL = str_replace_all(DS_URL, "www,", "www\\."),
    DS_URL = str_replace_all(DS_URL, "\\/m\\.", "\\/"),
    
    DS_URL = str_replace_all(DS_URL, "htpps", "https"),
    DS_URL = str_replace_all(DS_URL, "httos", "https"),
    
    DS_URL = str_remove_all(DS_URL, "https:\\/\\/"),
    DS_URL = str_remove_all(DS_URL, "http:\\/\\/"),
    DS_URL = str_remove_all(DS_URL, "https:"),
    DS_URL = str_remove_all(DS_URL, "https\\."),
    DS_URL = str_remove_all(DS_URL, "https"),
    DS_URL = str_remove_all(DS_URL, "www\\."),
    DS_URL = str_remove_all(DS_URL, "web\\."),
    
    # Facebook
    DS_URL = ifelse(
      str_detect(DS_URL, "groups"),
      NA, DS_URL),
    DS_URL = ifelse(
      str_detect(DS_URL, "story\\.php"),
      NA, DS_URL),
    DS_URL = ifelse(
      str_detect(DS_URL, "photo"),
      NA, DS_URL),
    
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/people/"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/search/top\\?q="),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/search/top/\\?q="),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/profile.php\\?id="),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/pg\\/"),
    DS_URL = str_remove_all(DS_URL, "pt-br\\.facebook\\.com/"),
    DS_URL = str_remove_all(DS_URL, "web\\.facebook\\.com/"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com\\.br/"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com/"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com\\.br"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com"),
    DS_URL = str_remove_all(DS_URL, "facebook\\.com\\\\"),
    DS_URL = str_remove_all(DS_URL, "www\\.facebook/"),
    DS_URL = str_remove_all(DS_URL, "facebook\\/"),
    DS_URL = str_remove_all(DS_URL, "- facebook"),
    DS_URL = str_remove_all(DS_URL, "facebook:"),
    DS_URL = str_remove_all(DS_URL, "facebook\\\\"),
    DS_URL = str_remove_all(DS_URL, "facebook@"),
    DS_URL = str_remove_all(DS_URL, "facebook "),
    
    # Remove tudo depois de /?
    DS_URL = str_remove_all(DS_URL, "(?=\\/\\?).*"),
    # Remove tudo depois de ?
    DS_URL = str_remove_all(DS_URL, "(?=\\?).*"),
    # Remove tudo depois de /
    DS_URL = str_remove_all(DS_URL, "(?=\\/).*"),
    # Remove tudo ate "- "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\- )"),
    # Remove tudo ate ": "
    DS_URL = str_remove_all(DS_URL, "^.*?(?<=\\: )"),
    # Remove tudo depois de &ref
    DS_URL = str_remove_all(DS_URL, "(?=&).*"),
    # Remove inicios de padroes identificados
    DS_URL = str_remove_all(DS_URL, "^:@"),
    DS_URL = str_remove_all(DS_URL, "^@"),
    DS_URL = str_remove_all(DS_URL, "^:"),
    DS_URL = str_remove_all(DS_URL, "^#!")
    
  )

################################################################################

##### TWITTER

twitter <- redes2022 %>%
  filter(rede == "twitter") %>%
  mutate(
    
    DS_URL = ifelse(DS_URL == "twitter", NA, DS_URL),
    rede = ifelse(DS_URL == "twitter", NA, rede),
    
    # Geral
    DS_URL = str_replace_all(DS_URL, "www,", "www\\."),
    DS_URL = str_replace_all(DS_URL, "\\/m\\.", "\\/"),
    DS_URL = str_replace_all(DS_URL, "htpps", "https"),
    DS_URL = str_replace_all(DS_URL, "httos", "https"),
    DS_URL = str_remove_all(DS_URL, "https:\\/\\/"),
    DS_URL = str_remove_all(DS_URL, "http:\\/\\/"),
    DS_URL = str_remove_all(DS_URL, "https:"),
    DS_URL = str_remove_all(DS_URL, "https\\."),
    DS_URL = str_remove_all(DS_URL, "https"),
    DS_URL = str_remove_all(DS_URL, "www\\."),
    
    # Twitter
    DS_URL = str_remove_all(DS_URL, "mobile\\."),
    DS_URL = str_remove_all(DS_URL, "twitter\\.com\\.br/"),
    DS_URL = str_remove_all(DS_URL, "twitter\\.com/\\:"),
    DS_URL = str_remove_all(DS_URL, "twitter\\.com/"),
    DS_URL = str_remove_all(DS_URL, "twitter: "),
    
    # Remove tudo depois de ?
    DS_URL = str_remove_all(DS_URL, "(?=\\?).*"),
    # Remove tudo depois de /
    DS_URL = str_remove_all(DS_URL, "(?=\\/).*"),
    
    # Transforma em NA aqueles que estao como home
    DS_URL = ifelse(DS_URL == "home", NA, DS_URL),
    
    # Remove @ e /
    DS_URL = str_remove_all(DS_URL, "@"),
    DS_URL = str_remove_all(DS_URL, "/"),
    
  )

##### FINAL: QUATRO REDES

redes4_2022 <- bind_rows(instagram, facebook, twitter, youtube) %>%
  mutate(
    # Remove espacos
    DS_URL = str_remove_all(DS_URL, " "),
    # Coloca as redes que estao em branco como NA
    DS_URL = ifelse(DS_URL == "", NA, DS_URL))

Oi @lgelape , será que httr::parse_url não te ajudaria nesse tratamento? Pelo que entendi você quer obter o id ou user do candidato na rede X e essa funçãozinha faz isso e mais algumas coisas. Já usei ela pra extrair só o hostname, por exemplo, fica como sugestão. Abraço!

Testei rapidamente aqui e ajuda demais, @schoulten. Solução bem mais simples e elegante. Vou tentar ver com calma se passa algo ou se resolve todos os problemas. Obrigado pela sugestão, não conhecia a função!