2 files changed, 83 insertions, 17 deletions

diff --git a/src/helper.c b/src/helper.c
new file mode 100644
index 0000000..2749daa
--- /dev/null
+++ b/src/helper.c
@@ -0,0 +1,78 @@
+#include "../defs/all.h"
+#include "../defs/widechar.h"
+
+/* this reads one utf81 character from the start of bytes
+   it puts the character's number in *codepoint and returns how many bytes it used (its always 1 or more)
+   invalid input it stores U+FFFD (�) and returns 1, so the caller always moves forward. */
+static int utf8_decode(const unsigned char *bytes, unsigned long *codepoint){
+    unsigned long value = bytes[0];
+    unsigned long minallowed = 0;   //rejects overlong encodings 
+    int sequence_length;
+    int byte_index;
+
+    if(value < 0x80){
+        *codepoint = value;
+        return 1;
+    } else if(value >= 0xC2 && value <= 0xDF){
+        sequence_length = 2;
+        value &= 0x1F;
+        minallowed = 0x80;
+    } else if(value >= 0xE0 && value <= 0xEF){
+        sequence_length = 3;
+        value &= 0x0F;
+        minallowed = 0x800;
+    } else if(value >= 0xF0 && value <= 0xF4){
+        sequence_length = 4;
+        value &= 0x07;
+        minallowed = 0x10000;
+    } else {
+        *codepoint = 0xFFFD;
+        return 1;
+    }
+
+    for(byte_index = 1; byte_index < sequence_length; byte_index++){
+        if((bytes[byte_index] & 0xC0) != 0x80){
+            *codepoint = 0xFFFD;
+            return 1;
+        }
+        value = (value << 6) | (bytes[byte_index] & 0x3F);
+    }
+
+    if(value < minallowed || value > 0x10FFFF || (value >= 0xD800 && value <= 0xDFFF)){
+        *codepoint = 0xFFFD;
+        return 1;
+    }
+
+    *codepoint = value;
+    return sequence_length;
+}
+
+/* this turns the library's return value into a column count so 0 1 or 2 */
+static int codepoint_width(unsigned long codepoint){
+    int width = widechar_wcwidth(codepoint);
+    if(width > 0)
+        return width;
+
+    switch(width){
+        case widechar_ambiguous:     
+        case widechar_private_use:
+            return 1;
+        case widechar_widened_in_9:
+            return 2;
+        default:                     
+            return 0;
+    }
+}
+
+/* how many terminal columns a string takes up */
+int displaywidth(const char *text){
+    const unsigned char *current_byte = (const unsigned char *)text;
+    int total_width = 0;
+
+    while(*current_byte){
+        unsigned long codepoint;
+        current_byte += utf8_decode(current_byte, &codepoint);
+        total_width += codepoint_width(codepoint);
+    }
+    return total_width;
+}
diff --git a/src/main.c b/src/main.c
index 1445292..8c0fda3 100644
--- a/src/main.c
+++ b/src/main.c
@@ -63,7 +63,7 @@ char **wraptext(const char *text, int max, int *count){
 
     char *tok = strtok(copy, " \t\n");   
     while(tok != NULL){ 
-        if(strlen(current_line)+strlen(tok) < (size_t)max){
+        if((size_t)displaywidth(current_line)+(size_t)displaywidth(tok) < (size_t)max){
             if(current_line[0] != '\0'){
                 strcat(current_line, " ");
             }
@@ -101,7 +101,7 @@ char **getbubblelines(char *text){
 
     size_t content_width = 0;
     for (int i = 0; templines[i] != NULL; i++) {
-        size_t current_len = strlen(templines[i]);
+        size_t current_len = displaywidth(templines[i]);
         if (current_len > content_width)
             content_width = current_len;
     }
@@ -128,13 +128,13 @@ char **getbubblelines(char *text){
     strcat(bubblelines[0], top_right);
     
     for (int i = 0; lines[i] != NULL; i++) {
-        int line_width = strlen(lines[i]);
+        int line_width = displaywidth(lines[i]);
         int padding_needed = (bubble_width - 2) - line_width;
         if (padding_needed < 0) padding_needed = 0;
         int left_padding = padding_needed / 2;
         int right_padding = padding_needed - left_padding;
-
-        bubblelines[i + 1] = malloc(strlen(vertical) * 2 + 2 + line_width + padding_needed + 1);
+        
+        bubblelines[i + 1] = malloc(strlen(vertical) * 2 + 2 + strlen(lines[i]) + padding_needed + 1);
         sprintf(bubblelines[i + 1], "%s %*s%s%*s %s",vertical, left_padding, "", lines[i], right_padding, "", vertical);
     }
 
@@ -177,18 +177,6 @@ static void list_all_art_styles(void){
     printf("\nUse --style <number> to select a specific style, or omit for random selection.\n");
 }
 
-
-//display width in columns: counts utf 8 codepoints (skips continuation bytes)
-//assumes every character is 1 column wide
-//emojis and other bullshit won't work bu tthat's okay since who the fuck will try to put '🥺' into this
-static int displaywidth(const char *s){
-    int w = 0;
-    for(; *s; s++)
-        if(((unsigned char)*s & 0xC0) != 0x80)
-            w++;
-    return w;
-}
-
 void draw_miku_says(char *text, int style){
     char **speech_bubble_lines = getbubblelines(text);
     if(speech_bubble_lines == NULL)