From 182467c1d393dee2081ff80f1c049cb145f23123 Mon Sep 17 00:00:00 2001
From: Titus Wormer <tituswormer@gmail.com>
Date: Mon, 20 Jun 2022 18:53:38 +0200
Subject: Add support for BOM

---
 src/subtokenize.rs |  4 ++++
 src/tokenizer.rs   | 10 ++++++++++
 2 files changed, 14 insertions(+)

(limited to 'src')

diff --git a/src/subtokenize.rs b/src/subtokenize.rs
index 4a29a01..0623a37 100644
--- a/src/subtokenize.rs
+++ b/src/subtokenize.rs
@@ -15,6 +15,10 @@ pub fn subtokenize(events: Vec<Event>, codes: &[Code]) -> (Vec<Event>, bool) {
     let mut link_to_info: HashMap<usize, (usize, usize, usize)> = HashMap::new();
     let mut done = true;
 
+    if events.is_empty() {
+        return (events, true);
+    }
+
     while index < events.len() {
         let event = &events[index];
 
diff --git a/src/tokenizer.rs b/src/tokenizer.rs
index d31c8c5..c0a7105 100644
--- a/src/tokenizer.rs
+++ b/src/tokenizer.rs
@@ -680,10 +680,20 @@ fn attempt_impl(
 // To do: handle BOM at start?
 pub fn as_codes(value: &str) -> Vec<Code> {
     let mut codes: Vec<Code> = vec![];
+    let mut at_start = true;
     let mut at_carriage_return = false;
     let mut column = 1;
 
     for char in value.chars() {
+        if at_start {
+            if char == '\u{feff}' {
+                // Ignore.
+                continue;
+            }
+
+            at_start = false;
+        }
+
         // Send a CRLF.
         if at_carriage_return && '\n' == char {
             at_carriage_return = false;
-- 
cgit