refactor: split off reusable html5lib_tests crate

2 files changed, 266 insertions, 0 deletions

diff --git a/html5lib_tests/Cargo.toml b/html5lib_tests/Cargo.toml
new file mode 100644
index 0000000..06fc6b8
--- /dev/null
+++ b/html5lib_tests/Cargo.toml
@@ -0,0 +1,14 @@
+[package]
+name = "html5lib_tests"
+authors = ["Martin Fischer <martin@push-f.com>", "Markus Unterwaditzer <markus-honeypot@unterwaditzer.net>"]
+description = "Deserializable types for the .test files from html5lib-tests."
+version = "0.0.0"
+edition = "2021"
+publish = false # prevent accidental publishes until it's ready to be published
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+serde = { version = "1.0.130", features = ["derive"] }
+serde_json = "1.0.71"
+html5tokenizer = { path = ".." } # TODO: get rid of this dependency
diff --git a/html5lib_tests/src/lib.rs b/html5lib_tests/src/lib.rs
new file mode 100644
index 0000000..5678b0d
--- /dev/null
+++ b/html5lib_tests/src/lib.rs
@@ -0,0 +1,252 @@
+use html5tokenizer::{Attribute, Doctype, EndTag, Error, StartTag, Token};
+use serde::{de::Error as _, Deserialize};
+use std::collections::BTreeMap;
+
+pub fn parse_tests(
+    reader: impl std::io::Read,
+) -> Result<impl Iterator<Item = Test>, serde_json::Error> {
+    let Tests { tests } = serde_json::from_reader(reader)?;
+    Ok(tests.into_iter().map(undo_double_escaping))
+}
+
+pub struct ExpectedOutputTokens(pub Vec<Token<()>>);
+
+impl<'de> Deserialize<'de> for ExpectedOutputTokens {
+    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
+    where
+        D: serde::Deserializer<'de>,
+    {
+        // this macro is a horrible way to define a type that deserializes only from a particular
+        // string. Together with serde(untagged) this gives us really flexible enum tagging with really
+        // terrible error messages.
+        macro_rules! def_const {
+            ($str:expr, $ty:ident) => {
+                #[derive(Deserialize)]
+                enum $ty {
+                    #[serde(rename = $str)]
+                    $ty,
+                }
+            };
+        }
+
+        def_const!("DOCTYPE", DoctypeConst);
+        def_const!("StartTag", StartTagConst);
+        def_const!("EndTag", EndTagConst);
+        def_const!("Comment", CommentConst);
+        def_const!("Character", CharacterConst);
+
+        type Attributes = BTreeMap<String, String>;
+
+        #[derive(Deserialize)]
+        #[serde(untagged)]
+        enum OutputToken {
+            // "DOCTYPE", name, public_id, system_id, correctness
+            Doctype(
+                DoctypeConst,
+                Option<String>,
+                Option<String>,
+                Option<String>,
+                bool,
+            ),
+            // "StartTag", name, attributes, self_closing
+            StartTag(StartTagConst, String, Attributes),
+            StartTag2(StartTagConst, String, Attributes, bool),
+            // "EndTag", name
+            EndTag(EndTagConst, String),
+            // "Comment", data
+            Comment(CommentConst, String),
+            // "Character", data
+            Character(CharacterConst, String),
+        }
+
+        Ok(ExpectedOutputTokens(
+            Vec::deserialize(deserializer)?
+                .into_iter()
+                .map(|output_token| match output_token {
+                    OutputToken::Doctype(
+                        _,
+                        name,
+                        public_identifier,
+                        system_identifier,
+                        correctness,
+                    ) => Token::Doctype(Doctype {
+                        name: name.unwrap_or_default(),
+                        public_identifier,
+                        system_identifier,
+                        force_quirks: !correctness,
+                    }),
+                    OutputToken::StartTag(_, name, attributes) => Token::StartTag(StartTag {
+                        self_closing: false,
+                        name,
+                        attributes: attributes
+                            .into_iter()
+                            .map(|(k, v)| {
+                                (
+                                    k,
+                                    Attribute {
+                                        value: v,
+                                        ..Default::default()
+                                    },
+                                )
+                            })
+                            .collect(),
+                        name_span: (),
+                    }),
+                    OutputToken::StartTag2(_, name, attributes, self_closing) => {
+                        Token::StartTag(StartTag {
+                            self_closing,
+                            name,
+                            attributes: attributes
+                                .into_iter()
+                                .map(|(k, v)| {
+                                    (
+                                        k,
+                                        Attribute {
+                                            value: v,
+                                            ..Default::default()
+                                        },
+                                    )
+                                })
+                                .collect(),
+                            name_span: (),
+                        })
+                    }
+                    OutputToken::EndTag(_, name) => Token::EndTag(EndTag {
+                        name,
+                        name_span: (),
+                    }),
+                    OutputToken::Comment(_, data) => Token::Comment(data),
+                    OutputToken::Character(_, data) => Token::String(data),
+                })
+                .collect::<Vec<Token<()>>>(),
+        ))
+    }
+}
+
+#[derive(Debug, Deserialize)]
+pub enum InitialState {
+    #[serde(rename = "Data state")]
+    Data,
+    #[serde(rename = "PLAINTEXT state")]
+    PlainText,
+    #[serde(rename = "RCDATA state")]
+    RcData,
+    #[serde(rename = "RAWTEXT state")]
+    RawText,
+    #[serde(rename = "Script data state")]
+    ScriptData,
+    #[serde(rename = "CDATA section state")]
+    CdataSection,
+}
+
+fn initial_states_default() -> Vec<InitialState> {
+    vec![InitialState::Data]
+}
+
+#[derive(Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct Test {
+    pub description: String,
+    pub input: String,
+    pub output: ExpectedOutputTokens,
+    #[serde(default = "initial_states_default")]
+    pub initial_states: Vec<InitialState>,
+    #[serde(default)]
+    double_escaped: bool,
+    #[serde(default)]
+    pub last_start_tag: Option<String>,
+    #[serde(default)]
+    pub errors: Vec<ParseError>,
+}
+
+#[derive(Debug, Eq, PartialEq)]
+pub struct ParseErrorInner(pub Error);
+
+impl<'de> Deserialize<'de> for ParseErrorInner {
+    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
+    where
+        D: serde::Deserializer<'de>,
+    {
+        let str_err = String::deserialize(deserializer)?;
+        let err: Error = str_err
+            .parse()
+            .map_err(|_| D::Error::custom(&format!("failed to deserialize error: {}", str_err)))?;
+        Ok(ParseErrorInner(err))
+    }
+}
+
+#[derive(Deserialize, Debug, Eq, PartialEq)]
+#[serde(rename_all = "camelCase")]
+pub struct ParseError {
+    pub code: ParseErrorInner,
+    // TODO: lineno and column?
+}
+
+#[derive(Deserialize)]
+struct Tests {
+    tests: Vec<Test>,
+}
+
+fn undo_double_escaping(mut test: Test) -> Test {
+    test.input = if test.double_escaped {
+        unescape(&test.input)
+    } else {
+        test.input
+    };
+
+    test.output = if test.double_escaped {
+        ExpectedOutputTokens(
+            test.output
+                .0
+                .into_iter()
+                .map(|token| match token {
+                    Token::String(x) => Token::String(unescape(&x)),
+                    Token::Comment(x) => Token::Comment(unescape(&x)),
+                    token => token,
+                })
+                .collect(),
+        )
+    } else {
+        ExpectedOutputTokens(test.output.0)
+    };
+    test
+}
+
+/// Implements the escape sequences described in the tokenizer tests of html5lib-tests (and nothing
+/// more)
+fn unescape(data: &str) -> String {
+    let mut stream = data.chars();
+    let mut rv = String::new();
+
+    loop {
+        match stream.next() {
+            Some('\\') => (),
+            Some(x) => {
+                rv.push(x);
+                continue;
+            }
+            None => break,
+        }
+
+        match stream.next() {
+            Some('u') => (),
+            x => panic!("unexpected escape: {:?}", x),
+        }
+
+        let orig_len = rv.len();
+
+        for _ in 0..4 {
+            rv.push(match stream.next() {
+                Some(x) => x,
+                None => panic!("unexpected eof after \\u"),
+            });
+        }
+
+        let c = u32::from_str_radix(&rv[orig_len..], 16).expect("failed to parse as hex");
+        let c = char::from_u32(c).expect("bad character");
+        rv.truncate(orig_len);
+        rv.push(c);
+    }
+
+    rv
+}