From cb4e9cf7cfc1612c67f16bfabadbdf442cd380fe Mon Sep 17 00:00:00 2001
From: Martin Fischer <martin@push-f.com>
Date: Thu, 17 Aug 2023 16:12:51 +0200
Subject: fix: fix lots of position off-by-ones

Previously the PosTrackingReader always mysteriously subtracted 1
from the current position ... this wasn't sound at all ... the machine
just happens to often call `Tokenizer::unread_char` ... but not always.
E.g. for proper comments it didn't which resulted in their offset and
spans being off-by-one, which is fixed by this commit (see test_spans.rs).
---
 src/emitter.rs   | 7 +++----
 src/machine.rs   | 8 ++++----
 src/offset.rs    | 2 +-
 src/tokenizer.rs | 8 ++++----
 4 files changed, 12 insertions(+), 13 deletions(-)

(limited to 'src')

diff --git a/src/emitter.rs b/src/emitter.rs
index 1f60f70..90aa4db 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -310,7 +310,7 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
                 *self_closing = true;
             }
             Token::EndTag(_) => {
-                self.emit_error(Error::EndTagWithTrailingSolidus, offset);
+                self.emit_error(Error::EndTagWithTrailingSolidus, offset - 1);
             }
             _ => {
                 debug_assert!(false);
@@ -378,9 +378,8 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
             },
         ));
     }
-    fn init_attribute_value(&mut self, offset: O, quoted: bool) {
-        self.current_attribute.as_mut().unwrap().1.value_span =
-            offset + quoted as usize..offset + quoted as usize;
+    fn init_attribute_value(&mut self, offset: O, _quoted: bool) {
+        self.current_attribute.as_mut().unwrap().1.value_span = offset..offset;
     }
 
     fn push_attribute_name(&mut self, s: &str) {
diff --git a/src/machine.rs b/src/machine.rs
index deb3983..a58a754 100644
--- a/src/machine.rs
+++ b/src/machine.rs
@@ -694,7 +694,7 @@ where
                 Ok(ControlToken::Continue)
             }
             Some(x) => {
-                slf.emitter.init_attribute_name(slf.reader.position());
+                slf.emitter.init_attribute_name(slf.reader.position() - 1);
                 slf.state = State::AttributeName;
                 slf.unread_char(Some(x));
                 Ok(ControlToken::Continue)
@@ -747,7 +747,7 @@ where
                 Ok(ControlToken::Eof)
             }
             Some(x) => {
-                slf.emitter.init_attribute_name(slf.reader.position());
+                slf.emitter.init_attribute_name(slf.reader.position() - 1);
                 slf.state = State::AttributeName;
                 slf.unread_char(Some(x));
                 Ok(ControlToken::Continue)
@@ -775,7 +775,7 @@ where
             }
             c => {
                 slf.emitter
-                    .init_attribute_value(slf.reader.position(), false);
+                    .init_attribute_value(slf.reader.position() - 1, false);
                 slf.state = State::AttributeValueUnquoted;
                 slf.unread_char(c);
                 Ok(ControlToken::Continue)
@@ -952,7 +952,7 @@ where
             }
             c => {
                 slf.emit_error(Error::IncorrectlyOpenedComment);
-                slf.emitter.init_comment(slf.reader.position());
+                slf.emitter.init_comment(slf.reader.position() - 1);
                 slf.state = State::BogusComment;
                 slf.unread_char(c);
                 Ok(ControlToken::Continue)
diff --git a/src/offset.rs b/src/offset.rs
index 8809366..3152c78 100644
--- a/src/offset.rs
+++ b/src/offset.rs
@@ -68,7 +68,7 @@ impl<R> PosTrackingReader<R> {
 
 impl<R> Position<usize> for PosTrackingReader<R> {
     fn position(&self) -> usize {
-        self.position - 1
+        self.position
     }
 }
 
diff --git a/src/tokenizer.rs b/src/tokenizer.rs
index 02a4d62..e8a8908 100644
--- a/src/tokenizer.rs
+++ b/src/tokenizer.rs
@@ -106,7 +106,7 @@ impl From<State> for InternalState {
     }
 }
 
-impl<R: Reader + Position<O>, O, E: Emitter<O>> Tokenizer<R, O, E> {
+impl<R: Reader + Position<O>, O: Offset, E: Emitter<O>> Tokenizer<R, O, E> {
     /// Test-internal function to override internal state.
     ///
     /// Only available with the `integration-tests` feature which is not public API.
@@ -123,7 +123,7 @@ impl<R: Reader + Position<O>, O, E: Emitter<O>> Tokenizer<R, O, E> {
     /// Just a helper method for the machine.
     #[inline]
     pub(crate) fn emit_error(&mut self, error: Error) {
-        self.emitter.emit_error(error, self.reader.position());
+        self.emitter.emit_error(error, self.reader.position() - 1);
     }
 
     /// Assuming the _current token_ is an end tag, return true if all of these hold. Return false otherwise.
@@ -140,14 +140,14 @@ impl<R: Reader + Position<O>, O, E: Emitter<O>> Tokenizer<R, O, E> {
 
     #[inline]
     pub(crate) fn init_start_tag(&mut self) {
-        self.emitter.init_start_tag(self.reader.position());
+        self.emitter.init_start_tag(self.reader.position() - 1);
         self.current_tag_name.clear();
         self.is_start_tag = true;
     }
 
     #[inline]
     pub(crate) fn init_end_tag(&mut self) {
-        self.emitter.init_end_tag(self.reader.position());
+        self.emitter.init_end_tag(self.reader.position() - 1);
         self.current_tag_name.clear();
         self.is_start_tag = false;
     }
-- 
cgit v1.2.3