improve applause party parsing

R/helpers.R

@@ -1,4 +1,5 @@
 `%$%` <- function(f, x) f(x)
 `%.%` <- function(f, g) function(...) f(g(...))
+flip <- function(f) function(x, y) f(y, x)
 
 clear_na <- function(xs) xs[!is.na(xs)]

R/parse.R

@@ -137,12 +137,14 @@ parse_rede <- function(rede_xml) {
          comments = comments)
 }
 
-fraktionspattern <- "BÜNDNIS 90/DIE GRÜNEN|CDU/CSU|AfD|SPD|DIE LINKE|FDP"
+fraktionspattern <- "BÜNDNIS(SES)?\\W*90/DIE\\W*GRÜNEN|CDU/CSU|AfD|SPD|DIE LINKE|FDP|LINKEN"
+fraktionsnames <- c("BÜNDNIS 90/DIE GRÜNEN", "CDU/CSU", "AfD", "SPD", "DIE LINKE", "FDP")
 
 parse_comment <- function(comment, rede_id, on_redner) {
     base <- c(rede_id = rede_id, on_redner = on_redner)
     str_extract_all(comment, fraktionspattern) %>%
         `[[`(1) %>%
+        sapply(partial(flip(head), 1) %.% agrep, x=fraktionsnames, max=0.2, value=T) %>%
         str_c(collapse=",") ->
         by
     # classify comment