hateimparlament/R/repair.R

fraktionen <- c("AFD" = "AfD",
                "BÜNDNIS90/" = "BÜNDNIS 90 / DIE GRÜNEN",
                "BÜNDNIS90/DIEGRÜNEN" = "BÜNDNIS 90 / DIE GRÜNEN",
                "FRAKTIONSLOS" = "Fraktionslos",
                "DIELINKE" = "DIE LINKE",
                "SPD" = "SPD",
                "CDU/CSU" = "CDU/CSU",
                "FDP" = "FDP")

repair_fraktion <- function(fraktion) {
    cleaned <- str_to_upper %$% str_replace_all(fraktion, "\\s", "")
    fraktionen[cleaned]
}

# takes vector of titel and keeps longest
longest_titel <- function(titel) {
    if (all(is.na(titel))) NA_character_
    else titel[which.max %$% str_length(titel)]
}

# takes character vector, removes duplicates and collapses
collect_unique <- function(xs) xs %>% clear_na() %>% unique() %>% str_c(collapse="&") %>% na_if("")

# expects a tibble of redner and repairs
repair_redner <- function(redner) {
    if (nrow(redner) == 0) return(redner)
    redner %>% mutate(fraktion = Vectorize(repair_fraktion)(fraktion)) %>% # fix fraktion
        group_by(id, vorname, nachname) %>%
        summarize(fraktion = collect_unique(fraktion),
                  titel = longest_titel(titel),
                  rolle_kurz = collect_unique(str_squish(rolle_kurz)),
                  rolle_lang = collect_unique(str_squish(rolle_lang)))
}

repair_reden <- function(reden) {
    if (nrow(reden) == 0) return(reden)
    # TODO: fill with content
    reden
}

repair_talks <- function(talks) {
    if (nrow(talks) == 0) return(talks)
    # TODO: fill with content
    talks
}

#' Repair parsed tables
#'
#' @export
repair <- function(parse_output) {
    list(redner = repair_redner(parse_output$redner),
         reden = repair_reden(parse_output$reden),
         talks = repair_talks(parse_output$talks))
}