fix!: make comment data spans encoding-independent

1 files changed, 16 insertions, 11 deletions

diff --git a/src/machine.rs b/src/machine.rs
index f00af0a..26e1652 100644
--- a/src/machine.rs
+++ b/src/machine.rs
@@ -950,11 +950,11 @@ where
         State::BogusComment => match slf.read_char()? {
             Some('>') => {
                 slf.state = State::Data;
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.position_before_match);
                 Ok(ControlToken::Continue)
             }
             None => {
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.position_before_match);
                 Ok(ControlToken::Eof)
             }
             Some('\0') => {
@@ -994,7 +994,7 @@ where
             Some('>') => {
                 slf.emit_error(Error::AbruptClosingOfEmptyComment);
                 slf.state = State::Data;
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.position_before_match);
                 Ok(ControlToken::Continue)
             }
             c => {
@@ -1008,15 +1008,19 @@ where
                 slf.state = State::CommentEnd;
                 Ok(ControlToken::Continue)
             }
-            Some('>') => {
+            Some(c @ '>') => {
                 slf.emit_error(Error::AbruptClosingOfEmptyComment);
                 slf.state = State::Data;
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(
+                    slf.position_before_match - slf.reader.len_of_char_in_current_encoding(c),
+                );
                 Ok(ControlToken::Continue)
             }
             None => {
                 slf.emit_error(Error::EofInComment);
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(
+                    slf.position_before_match - slf.reader.len_of_char_in_current_encoding('-'),
+                );
                 Ok(ControlToken::Eof)
             }
             c @ Some(_) => {
@@ -1033,6 +1037,7 @@ where
                 Ok(ControlToken::Continue)
             }
             Some('-') => {
+                slf.some_offset = slf.position_before_match;
                 slf.state = State::CommentEndDash;
                 Ok(ControlToken::Continue)
             }
@@ -1109,7 +1114,7 @@ where
             }
             None => {
                 slf.emit_error(Error::EofInComment);
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.some_offset);
                 Ok(ControlToken::Eof)
             }
             c => {
@@ -1122,7 +1127,7 @@ where
         State::CommentEnd => match slf.read_char()? {
             Some('>') => {
                 slf.state = State::Data;
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.some_offset);
                 Ok(ControlToken::Continue)
             }
             Some('!') => {
@@ -1135,7 +1140,7 @@ where
             }
             None => {
                 slf.emit_error(Error::EofInComment);
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.some_offset);
                 Ok(ControlToken::Eof)
             }
             c @ Some(_) => {
@@ -1157,12 +1162,12 @@ where
             Some('>') => {
                 slf.emit_error(Error::IncorrectlyClosedComment);
                 slf.state = State::Data;
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.some_offset);
                 Ok(ControlToken::Continue)
             }
             None => {
                 slf.emit_error(Error::EofInComment);
-                slf.emitter.emit_current_comment(slf.reader.position());
+                slf.emitter.emit_current_comment(slf.some_offset);
                 Ok(ControlToken::Eof)
             }
             c @ Some(_) => {