From 987a92002ff1d56634e43b5243d7f731e2ef38f6 Mon Sep 17 00:00:00 2001
From: Minijackson <minijackson@riseup.net>
Date: Sun, 23 Feb 2020 19:59:23 +0100
Subject: clang-parser: fix comment parsing with UTF-8

---
 src/parser/clang/parsing.rs | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

(limited to 'src')

diff --git a/src/parser/clang/parsing.rs b/src/parser/clang/parsing.rs
index 82f0a94..f772d4f 100644
--- a/src/parser/clang/parsing.rs
+++ b/src/parser/clang/parsing.rs
@@ -265,6 +265,7 @@ pub fn parse_comment(raw: String) -> String {
     let mut result = String::new();
 
     'parse_loop: loop {
+        // After the beginning of line comment delimiter / embellishment
         let maybe_space = chars.next();
         let mut empty_line = false;
         match maybe_space {
@@ -278,9 +279,12 @@ pub fn parse_comment(raw: String) -> String {
             Some(ch) => result.push(ch),
         }
 
+        // After the whitespace following the comment delimiter / embellishment
         if !empty_line {
             let rest = chars.as_str();
-            match rest.find('\n') {
+            // Not using .find() here because it returns the byte offset instead of the character
+            // offset
+            match rest.chars().position(|c| c == '\n') {
                 None => {
                     result.push_str(rest);
                     break;
@@ -302,7 +306,7 @@ pub fn parse_comment(raw: String) -> String {
             }
         };
 
-        // `chars` is right after the first non whitespace character
+        // `chars` is right after the first non whitespace character of the line
         match style {
             CommentStyle::Starred if first_non_ws_ch == '*' => {
                 if &chars.as_str()[..1] == "/" {
-- 
cgit v1.2.3