From 01f18c4c6ee68cbd58944e21d1fe36991315a889 Mon Sep 17 00:00:00 2001
From: Jarno Mäkipää <jmakip87@gmail.com>
Date: Fri, 18 Oct 2019 20:21:57 +0300
Subject: cut: re-enable crunch_str on cut -C

Reason: unicolumns() does not print combining characters correctly
Combining characters follow the character which they modify.
https://www.cl.cam.ac.uk/~mgk25/unicode.html#comb

xterm renders cut test1.txt -C -1 now correctly
---
 tests/cut.test   |  3 +++
 toys/posix/cut.c | 38 +++++++-------------------------------
 2 files changed, 10 insertions(+), 31 deletions(-)

diff --git a/tests/cut.test b/tests/cut.test
index e475288a..8d8c4ba1 100755
--- a/tests/cut.test
+++ b/tests/cut.test
@@ -34,6 +34,9 @@ testing "-c a,b-c,d" "cut -c 3,5-7,10 abc.txt" "etwoh\npa:ba\nequi \n" "" ""
 toyonly testing "-c japan.txt" 'cut -c 3-6,9-12 "$FILES/utf8/japan.txt"' \
   "ガラスをられます\n" "" ""
 
+toyonly testing "-C test1.txt" 'cut -C -1 "$FILES/utf8/test1.txt"' \
+  "l̴̗̞̠\n" "" ""
+
 # substitute for awk
 toyonly testcmd "-DF" "-DF 2,7,5" \
   "said and your\nare\nis demand. supply\nforecast :\nyou you better,\n\nEm: Took hate\n" "" \
diff --git a/toys/posix/cut.c b/toys/posix/cut.c
index 9f7f7458..61b2b409 100644
--- a/toys/posix/cut.c
+++ b/toys/posix/cut.c
@@ -46,28 +46,6 @@ GLOBALS(
   regex_t reg;
 )
 
-// Return number of bytes to start of first column fitting in columns
-// invalid sequences are skipped/ignored
-int unicolumns(char *start, unsigned columns)
-{
-  int i, j = 0;
-  wchar_t wc;
-  char *s = start, *ss = start;
-
-  // Skip start, rounding down if we hit a multicolumn char
-  while (j<columns && (i = utf8towc(&wc, s, 4))) {
-    if (i<0) s++;
-    else {
-      s += i;
-      if (0<(i = wcwidth(wc))) {
-        if ((j += i)>columns) break;
-        ss = s;
-      }
-    }
-  }
-
-  return ss-start;
-}
 
 // Apply selections to an input line, producing output
 static void cut_line(char **pline, long len)
@@ -99,15 +77,13 @@ static void cut_line(char **pline, long len)
       // crunch_str() currently assumes that combining characters get
       // escaped, to provide an unambiguous visual representation.
       // This assumes the input string is null terminated.
-      //if (start) crunch_str(&s, start, 0, 0, 0);
-      //if (!*s) continue;
-      //start = s-line;
-      //ss = s;
-      //crunch_str(&ss, count, 0, 0, 0);
-      //count = ss-s;
-
-      s += unicolumns(s, start);
-      count = unicolumns(s, end-start);
+      if (start) crunch_str(&s, start, 0, 0, 0);
+      if (!*s) continue;
+      start = s-line;
+      ss = s;
+      crunch_str(&ss, count, 0, 0, 0);
+      count = ss-s;
+
     } else if (toys.optflags&FLAG_c) {
       wchar_t wc;
       char *sss;
-- 
cgit v1.2.3