From 01ec0de76fd7ce1186a374c6bc4b0a51ebc7e93a Mon Sep 17 00:00:00 2001
From: flavis <christian@flavigny.de>
Date: Tue, 10 Aug 2021 20:18:23 +0200
Subject: [PATCH] replace non ascii characters by unicode points

---
 R/parse.R  | 4 ++--
 R/repair.R | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/R/parse.R b/R/parse.R
index 520c211..d7fb267 100644
--- a/R/parse.R
+++ b/R/parse.R
@@ -153,10 +153,10 @@ parse_speech <- function(speech_xml, date) {
             }
         } else if (xml_name(node) == "kommentar") {
             # comments are of the form
-            # <kommentar>(blabla [Fraktion] – blabla liasdf – bla)</kommentar>
+            # <kommentar>(blabla [Fraktion] \u2013 blabla liasdf \u2013 bla)</kommentar>
             xml_text(node) %>%
                 str_sub(2, -2) %>%
-                str_split("–") %>%
+                str_split("\u2013") %>%
                 `[[`(1) %>%
                 lapply(parse_comment, speech_id = speech_id, on_speaker = cur_speaker) ->
                 cs
diff --git a/R/repair.R b/R/repair.R
index c52db56..01be4a0 100644
--- a/R/repair.R
+++ b/R/repair.R
@@ -64,7 +64,7 @@ repair_talks <- function(talks) {
 #' 
 #' returns a lookup table
 lookup_speaker <- function(tb, speaker, name_variable) {
-    tobereplaced <- "[-–—‑­­-­­­ ]"
+    tobereplaced <- "[\u002D\u2013\u2014\u2011\u00AD ]"
     speaker %>%
         unite(name, prename, lastname, sep=".*") %>%
         mutate(name = str_replace_all(name, tobereplaced, ".*")) ->