introduce Error enum

1 files changed, 18 insertions, 22 deletions

diff --git a/src/tokenizer/mod.rs b/src/tokenizer/mod.rs
index 4511cf8..78101f6 100644
--- a/src/tokenizer/mod.rs
+++ b/src/tokenizer/mod.rs
@@ -9,6 +9,7 @@
 
 //! The HTML5 tokenizer.
 
+use self::error::InternalState;
 pub use self::interface::{Attribute, Doctype, Tag, TagKind, Token};
 use self::interface::{CharacterTokens, EOFToken, NullCharacterToken, ParseError};
 use self::interface::{CommentToken, DoctypeToken, EndTag, StartTag, TagToken};
@@ -21,9 +22,9 @@ use self::states::{Rawtext, Rcdata, ScriptData, ScriptDataEscaped};
 
 use self::char_ref::{CharRef, CharRefTokenizer};
 
+use crate::error::Error;
 use crate::util::{smallcharset::SmallCharSet, str::lower_ascii_letter};
 
-use std::borrow::Cow::{self, Borrowed};
 use std::collections::BTreeMap;
 use std::default::Default;
 use std::mem::replace;
@@ -34,6 +35,7 @@ use crate::util::buffer_queue::{FromSet, NotFromSet, SetResult};
 pub use states::RawKind;
 
 mod char_ref;
+pub mod error;
 mod interface;
 mod states;
 
@@ -292,8 +294,7 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
                 _ => false,
             }
         {
-            let msg = format!("Bad character {}", c);
-            self.emit_error(Cow::Owned(msg));
+            self.emit_error(Error::BadCharacter(c));
         }
 
         #[cfg(feature = "spans")]
@@ -400,24 +401,14 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
     }
 
     fn bad_char_error(&mut self) {
-        let msg = format_if!(
-            self.opts.exact_errors,
-            "Bad character",
-            "Saw {} in state {:?}",
+        self.emit_error(Error::UnexpectedCharacter(
             self.current_char,
-            self.state
-        );
-        self.emit_error(msg);
+            InternalState(self.state),
+        ));
     }
 
     fn bad_eof_error(&mut self) {
-        let msg = format_if!(
-            self.opts.exact_errors,
-            "Unexpected EOF",
-            "Saw EOF in state {:?}",
-            self.state
-        );
-        self.emit_error(msg);
+        self.emit_error(Error::UnexpectedEOF(InternalState(self.state)));
     }
 
     fn emit_char(&mut self, c: char) {
@@ -444,10 +435,10 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
             }
             EndTag => {
                 if !self.current_tag_attrs.is_empty() {
-                    self.emit_error(Borrowed("Attributes on an end tag"));
+                    self.emit_error(Error::AttributesOnEndTag);
                 }
                 if self.current_tag_self_closing {
-                    self.emit_error(Borrowed("Self-closing end tag"));
+                    self.emit_error(Error::SelfClosingEndTag);
                 }
             }
         }
@@ -547,7 +538,7 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
         };
 
         if dup {
-            self.emit_error(Borrowed("Duplicate attribute"));
+            self.emit_error(Error::DuplicateAttribute);
             self.current_attr_name.clear();
             self.current_attr_value.clear();
         } else {
@@ -606,7 +597,7 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
         self.get_char(input);
     }
 
-    fn emit_error(&mut self, error: Cow<'static, str>) {
+    fn emit_error(&mut self, error: Error) {
         self.process_token_and_continue(ParseError(error));
     }
 }
@@ -2451,11 +2442,16 @@ mod test {
     #[test]
     #[cfg(feature = "named-entities")]
     fn named_entities() {
+        use crate::error::{CharRefError, Error};
+
         let opts = opts();
         let vector = vec![String::from("&amp;\r\n"), String::from("&aamp;\r\n")];
         let expected = vec![
             (3, CharacterTokens("&\n".into())),
-            (3, ParseError("Invalid character reference".into())),
+            (
+                3,
+                ParseError(Error::CharRef(CharRefError::InvalidNamedCharRef)),
+            ),
             (4, CharacterTokens("&aamp;\n".into())),
         ];
         let results = tokenize(vector, opts);