From 43d1e58652adc875b6dced525c3d6c544456a52d Mon Sep 17 00:00:00 2001
From: Orestis Floros <orestisflo@gmail.com>
Date: Thu, 30 Apr 2020 21:22:52 +0200
Subject: [PATCH] file_contents: Handle utf-8 characters

Fixes #410
---
 include/i3status.h        |  1 +
 src/print_file_contents.c | 35 ++++++++++++++++++++++++++++++-----
 2 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/include/i3status.h b/include/i3status.h
index e0dff2b6..c3bdb5eb 100644
--- a/include/i3status.h
+++ b/include/i3status.h
@@ -31,6 +31,7 @@ extern char *pct_mark;
 
 #define BEGINS_WITH(haystack, needle) (strncmp(haystack, needle, strlen(needle)) == 0)
 #define max(a, b) ((a) > (b) ? (a) : (b))
+#define min(x, y) ((x) < (y) ? (x) : (y))
 
 #define DEFAULT_SINK_INDEX UINT32_MAX
 #define COMPOSE_VOLUME_MUTE(vol, mute) ((vol) | ((mute) ? (1 << 30) : 0))
diff --git a/src/print_file_contents.c b/src/print_file_contents.c
index 65813f5e..b8cc4d00 100644
--- a/src/print_file_contents.c
+++ b/src/print_file_contents.c
@@ -24,19 +24,44 @@ static void *scalloc(size_t size) {
 
 void print_file_contents(yajl_gen json_gen, char *buffer, const char *title, const char *path, const char *format, const char *format_bad, const int max_chars) {
     const char *walk = format;
+    /* Each utf-8 character is a maximum of 4 bytes */
+    const int max_bytes = 4 * max_chars;
     char *outwalk = buffer;
-    char *buf = scalloc(max_chars * sizeof(char) + 1);
+    char *buf = scalloc(max_bytes * sizeof(char) + 1);
 
-    int n = -1;
     int fd = open(path, O_RDONLY);
 
     INSTANCE(path);
 
     if (fd > -1) {
-        n = read(fd, buf, max_chars);
-        if (n != -1) {
-            buf[n] = '\0';
+        read(fd, buf, max_bytes);
+
+        int i = 0;
+        int actual_chars = 0;
+        while (i < max_bytes) {
+            const unsigned char c = buf[i];
+
+            if (c == 0) {
+                break;
+            }
+            /* Count utf-8 characters, by only inspecting the first byte,
+             * skipping the rest. For reference see:
+             * https://tools.ietf.org/html/rfc3629#page-4 */
+            if (c < 128 /* 0xxxxxxx */) {
+                i++;
+            } else if (c < 0xe0 /* 110xxxxx */) {
+                i += 2;
+            } else if (c < 0xf0 /* 1110xxxx */) {
+                i += 3;
+            } else if (c < 0xf5 /* 11110xxx */) {
+                i += 4;
+            }
+            if (++actual_chars >= max_chars) {
+                buf[min(i, max_bytes)] = '\0';
+                break;
+            }
         }
+
         (void)close(fd);
         START_COLOR("color_good");
     } else if (errno != 0) {