R/cat_page.R

In leojoubert/WikiSocio: A MediaWiki API Wrapper

#' List all members of a category
#'
#' @param x Category's name or id
#' @param domain The domain where the wiki is located
#' @param namespace The namespace of contributions
#'
#' @return A data-frame
#' @export
#' 
#' @importFrom magrittr %>% %<>%
#'
cat_page <- function(x, domain = "fr", namespace = "0") {
  
  result <- matrix(ncol = 5)
  
  query = list(
    action = "query", 
    list = "categorymembers", 
    cmlimit = "max",
    namespace = namespace,
    cmprop = "ids|title|type|timestamp",
    redirects = "")
  
  if(is.numeric(x)) {
    query["cmpageid"] <- x
  } else {
    query["cmtitle"] <- x
  }
  
  repeat {
    
    exec <- exec_query(query, domain)
    
    tmp <- unlist(exec$query$categorymembers) %>%
      matrix(ncol = 5, byrow = T)
    
    result <- rbind(result, tmp)
    
    if("continue" %in% names(exec)) {
      query["cmcontinue"] <- exec$continue$cmcontinue
      print("test")
    } else {
      break
    }
    
  }
  
  result <- as.data.frame(result[-1, ], stringsAsFactors = FALSE)
  
  # Mise en forme
  result[, 1] %<>% as.numeric()
  result[, 2] %<>% as.numeric()
  result[, 4] %<>% as.factor()
  result[, 5] %<>% strptime("%Y-%m-%dT%H:%M:%SZ") %>%
    as.POSIXct()
  
  names(result) <- c("pageid", "ns", "title", "type", "inclusion_date")
  
  result
}