From 1455834a50b8b6a15567e971db664fe7a6cdfaf6 Mon Sep 17 00:00:00 2001
From: Nicholas Noll <nbnoll@eml.cc>
Date: Fri, 29 Oct 2021 17:46:41 -0700
Subject: fix(unicode): emoji widths and readline now moves in a more unicode
 aware manner

---
 sys/libunicode/vendor/mkrunewidth.c | 71 ++++++++++++++++++++++++++++++-------
 1 file changed, 58 insertions(+), 13 deletions(-)

(limited to 'sys/libunicode/vendor/mkrunewidth.c')

diff --git a/sys/libunicode/vendor/mkrunewidth.c b/sys/libunicode/vendor/mkrunewidth.c
index 43feb5e..14e6973 100644
--- a/sys/libunicode/vendor/mkrunewidth.c
+++ b/sys/libunicode/vendor/mkrunewidth.c
@@ -89,6 +89,21 @@ parse_category(char *path)
     io·close(utf8);
 }
 
+static
+void
+coderange(char *field, int *l, int *r)
+{
+    char *s;
+
+    if(!(s = strstr(field, "..")))
+        *l=*r=codepoint(field);
+    else{
+        *s++ = 0, *s++ = 0;
+        *l=codepoint(field);
+        *r=codepoint(s);
+    }
+}
+
 static
 void
 parse_eawidths(char *path)
@@ -96,7 +111,7 @@ parse_eawidths(char *path)
     int at, w;
     int l, c, r;
     io·Stream *utf8;
-    char *s, *field[2], line[1024];
+    char *field[2], line[1024];
 
     utf8 = io·open(path, "r");
     while((at=parse(utf8, arrlen(field), field, arrlen(line), line)) != ParseEOF){
@@ -118,13 +133,7 @@ parse_eawidths(char *path)
             panicf("malformed east asian width class: %s\n", field[1]);
         }
 
-        if(!(s = strstr(field[0], "..")))
-            l=codepoint(field[0]), r=l;
-        else{
-            *s++ = 0, *s++ = 0;
-            l=codepoint(field[0]);
-            r=codepoint(s);
-        }
+        coderange(field[0], &l, &r);
 
         for(c=l; c <= r; c++){
             /* ensure it only exists in one table */
@@ -136,6 +145,43 @@ parse_eawidths(char *path)
     io·close(utf8);
 }
 
+static
+void
+parse_emoji(char *path)
+{
+    int at, w;
+    int l, c, r;
+    io·Stream *utf8;
+    char *s, *field[2], line[1024];
+
+    utf8 = io·open(path, "r");
+    while((at=parse(utf8, arrlen(field), field, arrlen(line), line)) != ParseEOF){
+        if(at == ParseSkip)
+            continue;
+
+        /* only override emoji presentation */
+        if(!strstr(field[1], "Emoji_Presentation"))
+            continue;
+
+        /* trim trailing space */
+        for(s=field[0]; *s; s++){
+            if(*s == ' ')
+                *s = 0;
+        }
+
+        coderange(field[0], &l, &r);
+
+        for(c=l; c <= r; c++){
+            table.width[0][c] = 0;
+            table.width[1][c] = 0;
+            table.width[2][c] = 1;
+        }
+    }
+
+    io·close(utf8);
+}
+
+/* output functions */
 static
 void
 maketable(char *label, char *table, int pairs, int onlyranges)
@@ -204,8 +250,6 @@ maketable(char *label, char *table, int pairs, int onlyranges)
         "\n");
 }
 
-
-
 // -----------------------------------------------------------------------
 // main point of entry
 
@@ -213,7 +257,7 @@ static
 void
 usage(void)
 {
-    fprintf(stderr, "usage: mkrunewidth <UnicodeData.txt> <EastAsianWidth.txt>\n");
+    fprintf(stderr, "usage: mkrunewidth <UnicodeData.txt> <EastAsianWidth.txt> <EmojiData.txt>\n");
     exit(1);
 }
 
@@ -241,11 +285,12 @@ main(int argc, char *argv[])
     ARGBEGIN{
     }ARGEND;
 
-    if(argc != 2)
+    if(argc != 3)
         usage();
 
     parse_category(*argv++);
-    parse_eawidths(*argv);
+    parse_eawidths(*argv++);
+    parse_emoji(*argv);
 
     /* overrides */
     SETW0(0x2028);
-- 
cgit v1.2.1