From 291eea1bfaa364fcfb920fecec550ce22541bea9 Mon Sep 17 00:00:00 2001
From: Babz <babz@tfnux.org>
Date: Sat, 18 Sep 2021 21:44:06 +0200
Subject: [PATCH] fix string padding with variable-length chars (utf-8)

---
 CMakeLists.txt |  1 +
 src/term.c     | 14 +++-----------
 src/ui.c       | 22 ++++++++++++----------
 src/utf8.c     | 28 ++++++++++++++++++++++++++++
 src/utf8.h     |  7 +++++++
 5 files changed, 51 insertions(+), 21 deletions(-)
 create mode 100644 src/utf8.c
 create mode 100644 src/utf8.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 5d6a406..c8f54d4 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -24,6 +24,7 @@ set(SOURCES
   src/syscalls.S
   src/job.c
   src/wren_utils.c
+  src/utf8.c
 )
 set(SOURCES_WREN
   wren/src/wren_compiler.c
diff --git a/src/term.c b/src/term.c
index 53b2c54..c5a5c57 100644
--- a/src/term.c
+++ b/src/term.c
@@ -8,6 +8,8 @@
 #include <stdarg.h>
 #include <string.h>
 
+#include "utf8.h"
+
 extern font_t uf5x7;
 
 struct tcell {
@@ -36,17 +38,7 @@ static int tgrid_sets(int *row, int *col, int fg, int bg, const char *s) {
         if (*col >= UNS_TERM_COLS)
             return i;
 
-        // detect utf-8 char length
-        unsigned char lb = s[i];
-        int charlen = 0;
-        if ((lb & 0x80) == 0) // lead bit is zero, must be a single ascii
-            charlen = 1;
-        else if ((lb & 0xE0) == 0xC0) // 110x xxxx
-            charlen = 2;
-        else if ((lb & 0xF0) == 0xE0) // 1110 xxxx
-            charlen = 3;
-        else if ((lb & 0xF8) == 0xF0) // 1111 0xxx
-            charlen = 4;
+        int charlen = charlen_utf8(s[i]);
 
         char unichar[4];
         int j;
diff --git a/src/ui.c b/src/ui.c
index 7357fbf..6c5cef2 100644
--- a/src/ui.c
+++ b/src/ui.c
@@ -1,13 +1,15 @@
 #include "ui.h"
 
-#include "term.h"
+#include <printf.h>
+#include <string.h>
+
 #include <gint/display.h>
 #include <gint/kmalloc.h>
 #include <gint/mmu.h>
 #include <gint/rtc.h>
 
-#include <printf.h>
-#include <string.h>
+#include "term.h"
+#include "utf8.h"
 
 static void date_str(char *buf) {
     rtc_time_t t;
@@ -17,24 +19,24 @@ static void date_str(char *buf) {
 }
 
 static void str_pad(char *buf, const char *prefix, const char *suffix, int wanted_len) {
-    const int l1 = strlen(prefix);
-    const int l2 = strlen(suffix);
+    const size_t l1 = strlen(prefix);
+    const size_t l2 = strlen(suffix);
 
-    const int toadd = wanted_len - (l1 + l2);
+    const size_t toadd = wanted_len - (strlen_utf8(prefix) + strlen_utf8(suffix));
 
-    for (int i = 0; i < l1; i++) {
+    for (size_t i = 0; i < l1; i++) {
         buf[i] = prefix[i];
     }
 
-    for (int i = 0; i < toadd; i++) {
+    for (size_t i = 0; i < toadd; i++) {
         buf[l1 + i] = ' ';
     }
 
-    for (int i = 0; i < l2; i++) {
+    for (size_t i = 0; i < l2; i++) {
         buf[l1 + toadd + i] = suffix[i];
     }
 
-    buf[wanted_len] = '\0';
+    buf[l1 + l2 + toadd] = '\0';
 }
 
 void set_statusbar(int tick_ctr, int shift_state, int alpha_state, int battery) {
diff --git a/src/utf8.c b/src/utf8.c
new file mode 100644
index 0000000..cf07e00
--- /dev/null
+++ b/src/utf8.c
@@ -0,0 +1,28 @@
+#include "utf8.h"
+
+int charlen_utf8(char c) {
+    int charlen = 0;
+
+    if ((c & 0x80) == 0) // lead bit is zero, must be a single ascii
+        charlen = 1;
+    else if ((c & 0xE0) == 0xC0) // 110x xxxx
+        charlen = 2;
+    else if ((c & 0xF0) == 0xE0) // 1110 xxxx
+        charlen = 3;
+    else if ((c & 0xF8) == 0xF0) // 1111 0xxx
+        charlen = 4;
+
+    return charlen;
+}
+
+int strlen_utf8(const char *str) {
+    int utf8_len = 0;
+    int i = 0;
+
+    while (str[i] != '\0') {
+        i += charlen_utf8(str[i]);
+        utf8_len++;
+    }
+
+    return utf8_len;
+}
\ No newline at end of file
diff --git a/src/utf8.h b/src/utf8.h
new file mode 100644
index 0000000..46acea4
--- /dev/null
+++ b/src/utf8.h
@@ -0,0 +1,7 @@
+#ifndef UNS_UTF8_H
+#define UNS_UTF8_H
+
+int charlen_utf8(char c);
+int strlen_utf8(const char *str);
+
+#endif // #ifndef UNS_UTF8_H
\ No newline at end of file