chore: updated old copy of mkrunewidth

4 files changed, 60 insertions, 197 deletions

diff --git a/.gitignore b/.gitignore
index fd958fb..8fb80bc 100644
--- a/.gitignore
+++ b/.gitignore
@@ -17,10 +17,10 @@ include/libimage.h
 sys/cc
 sys/nixos
 sys/libdraw
-sys/libfont
 sys/libimage
 sys/libterm
-sys/libutf8
+
+sys/libutf/*-14.0.0.c
 
 sys/cmd/muc
 sys/cmd/wm
diff --git a/sys/libutf/runetype-14.0.0.c b/sys/libutf/runetype-14.0.0.c
deleted file mode 100644
index 6f4469d..0000000
--- a/sys/libutf/runetype-14.0.0.c
+++ /dev/null
@@ -1,111 +0,0 @@
-#include <u.h>
-#include <libutf.h>
-
-static
-rune*
-rangesearch(rune c, rune *t, int n, int ne)
-{
-   rune *p;
-   int m;
-   while(n > 1) {
-   m = n >> 1;
-   p = t + m*ne;
-   if(c >= p[0]){
-       t = p;
-       n = n-m;
-   }else
-       n = m;
-   }
-   if(n && c >= t[0])
-       return t;
-   return 0;
-}
-
-static rune isspace_range[] = {
-	0x0009, 0x000d,
-	0x0085, 0x0085,
-};
-
-int
-utf8·isspace(rune c)
-{
-    rune *p;
-
-   p = rangesearch(c, isspace_range, arrlen(isspace_range)/2, 2);
-   if(p && c >= p[0] && c <= p[1])
-       return 1;
-   return 0;
-}
-
-int
-utf8·isdigit(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-int
-utf8·isalpha(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-int
-utf8·isupper(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-int
-utf8·islower(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-int
-utf8·istitle(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-int
-utf8·ispunct(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-rune
-utf8·toupper(rune c)
-{
-	rune *p;
-
-   return c;
-}
-
-rune
-utf8·tolower(rune c)
-{
-	rune *p;
-
-   return c;
-}
-
-rune
-utf8·totitle(rune c)
-{
-	rune *p;
-
-   return c;
-}
-
diff --git a/sys/libutf/runewidth-14.0.0.c b/sys/libutf/runewidth-14.0.0.c
deleted file mode 100644
index 113c35e..0000000
--- a/sys/libutf/runewidth-14.0.0.c
+++ /dev/null
@@ -1,71 +0,0 @@
-#include <u.h>
-#include <libutf.h>
-
-static
-rune*
-rangesearch(rune c, rune *t, int n, int ne)
-{
-   rune *p;
-   int m;
-   while(n > 1) {
-   m = n >> 1;
-   p = t + m*ne;
-   if(c >= p[0]){
-       t = p;
-       n = n-m;
-   }else
-       n = m;
-   }
-   if(n && c >= t[0])
-       return t;
-   return 0;
-}
-
-static rune width0_range[] = {
-	0x2028, 0x2029,
-};
-
-static int
-iswidth0(rune c)
-{
-    rune *p;
-
-   p = rangesearch(c, width0_range, arrlen(width0_range)/2, 2);
-   if(p && c >= p[0] && c <= p[1])
-       return 1;
-   return 0;
-}
-
-static rune width1_single[] = {
-	0x00ad,
-};
-
-static int
-iswidth1(rune c)
-{
-    rune *p;
-
-   p = rangesearch(c, width1_single, arrlen(width1_single), 1);
-   if(p && c == p[0])
-       return 1;
-   return 0;
-}
-
-static int
-iswidth2(rune c)
-{
-    rune *p;
-
-   return 0;
-}
-
-
-int
-utf8·runewidth(rune c)
-{
-   if(iswidth1(c))
-       return 1;
-   if(iswidth2(c))
-       return 2;
-   return 0;
-}
diff --git a/sys/libutf/vendor/mkrunewidth.c b/sys/libutf/vendor/mkrunewidth.c
index 43feb5e..14e6973 100644
--- a/sys/libutf/vendor/mkrunewidth.c
+++ b/sys/libutf/vendor/mkrunewidth.c
@@ -91,12 +91,27 @@ parse_category(char *path)
 
 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)
 {
     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 */
@@ -138,6 +147,43 @@ parse_eawidths(char *path)
 
 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)
 {
     int r, p=0, s=0;
@@ -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);