1 files changed, 78 insertions, 0 deletions
diff --git a/src/tokenizer/error.rs b/src/tokenizer/error.rs
new file mode 100644
index 0000000..89eed2a
--- /dev/null
+++ b/src/tokenizer/error.rs
@@ -0,0 +1,78 @@
+//! Types to represent the parser errors that can occur.
+use std::fmt::Display;
+
+#[derive(PartialEq, Eq, Debug)]
+#[non_exhaustive]
+pub enum Error {
+    AttributesOnEndTag,
+    SelfClosingEndTag,
+    DuplicateAttribute,
+    BadCharacter(char),
+    UnexpectedCharacter(char, InternalState),
+    UnexpectedEOF(InternalState),
+    CharRef(CharRefError),
+}
+
+/// Allows Error variants to include the internal tokenizer state without making it public.
+#[derive(PartialEq, Eq, Debug)]
+pub struct InternalState(pub(crate) crate::tokenizer::states::State);
+
+impl Display for Error {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Error::AttributesOnEndTag => write!(f, "attributes on an end tag"),
+            Error::SelfClosingEndTag => write!(f, "self-closing end tag"),
+            Error::DuplicateAttribute => write!(f, "duplicate attribute"),
+            Error::BadCharacter(char) => write!(f, "bad character {:?}", char),
+            Error::UnexpectedCharacter(char, state) => {
+                write!(
+                    f,
+                    "unexpected character: saw {:?} in state {:?}",
+                    char, state.0
+                )
+            }
+            Error::UnexpectedEOF(state) => write!(f, "unexpected EOF in state {:?}", state.0),
+            Error::CharRef(error) => error.fmt(f),
+        }
+    }
+}
+
+#[derive(PartialEq, Eq, Debug)]
+#[non_exhaustive]
+pub enum CharRefError {
+    MissingSemicolon,
+    NumericCharRefWithoutDigits,
+    NumericCharRefInvalid(u32),
+    EofInNumericCharRef,
+    EofAfterNumberSign,
+    EqualsSignAfterCharRefInAttr,
+    InvalidNamedCharRef,
+}
+
+impl Display for CharRefError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            CharRefError::NumericCharRefWithoutDigits => {
+                write!(f, "numeric character reference without digits")
+            }
+            CharRefError::MissingSemicolon => {
+                write!(f, "semicolon missing after character reference")
+            }
+            CharRefError::NumericCharRefInvalid(num) => {
+                write!(f, "invalid numeric character reference value 0x{:06X}", num)
+            }
+            CharRefError::EofInNumericCharRef => {
+                write!(f, "EOF in numeric character reference")
+            }
+            CharRefError::EofAfterNumberSign => {
+                write!(f, "EOF after '#' in character reference")
+            }
+            CharRefError::EqualsSignAfterCharRefInAttr => {
+                write!(f, "equals sign after character reference in attribute")
+            }
+            CharRefError::InvalidNamedCharRef => {
+                write!(f, "invalid named character reference")
+            }
+        }
+    }
+}