Remove BOM coding from svg files

src/matchers/text.rs

@@ -44,6 +44,7 @@ pub fn is_html(buf: &[u8]) -> bool {
 pub fn is_xml(buf: &[u8]) -> bool {
     let val: &[u8] = b"<?xml";
     let buf = trim_start_whitespaces(buf);
+    let buf = trim_start_byte_order_marks(buf);
     starts_with_ignore_ascii_case(buf, val)
 }
 
@@ -61,6 +62,25 @@ fn trim_start_whitespaces(mut buf: &[u8]) -> &[u8] {
     buf
 }
 
+/// Strip whitespaces at the beginning of the buffer.
+fn trim_start_byte_order_marks(mut buf: &[u8]) -> &[u8] {
+    while !buf.is_empty() {
+        if buf.len() >= 3 {
+            match (buf[0], buf[1], buf[2]) {
+                (0xEF, 0xBB, 0xBF) => buf = &buf[3..],// UTF-8
+                _ => break,
+            }
+        } else if buf.len() >= 2 {
+            match (buf[0], buf[1]) {
+                (0xFE, 0xFF) => buf = &buf[2..],// UTF-16 BE
+                (0xFF, 0xFE) => buf = &buf[2..],// UTF-16 BE
+                _ => break,
+            }
+        }
+    }
+    buf
+}
+
 fn starts_with_ignore_ascii_case(buf: &[u8], needle: &[u8]) -> bool {
     buf.len() >= needle.len() && buf[..needle.len()].eq_ignore_ascii_case(needle)
 }

testdata/sample2.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="utf-8"?>
+<doc>
+  <assembly>
+    <name>System.Runtime</name>
+  </assembly>
+</doc>

tests/text.rs

@@ -4,4 +4,6 @@ test_format!(Text, "text/html", "html", html, "sample.html");
 
 test_format!(Text, "text/xml", "xml", xml, "sample.xml");
 
+test_format!(Text, "text/xml", "xml", xml2, "sample2.xml");
+
 test_format!(Text, "text/x-shellscript", "sh", sh, "sample.sh");