summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMartin Fischer <martin@push-f.com>2023-08-16 17:07:06 +0200
committerMartin Fischer <martin@push-f.com>2023-08-19 06:41:55 +0200
commit681404e5036841ec45356f56f77cc5377f3640d9 (patch)
tree568f6cdd34e0366372ba1ad3c2d13f1b8e2b530c
parentc021afde2c2bd4b8bdd9b3eec4b1660cb0a896e5 (diff)
refactor: decouple html5lib_tests from html5tokenizer
Previously we mapped the test tokens to our own token type. Now we do the reverse, which makes more sense as it enables us to easily add more detailed fields to our own token variants without having to worry about these fields not being present in the html5lib test data. (An alternative would be to normalize the values of these fields to some arbitrary value so that PartialEq still holds but seeing such normalized fields in the diff printed by pretty_assertions on a test failure would be quite confusing).
-rw-r--r--html5lib_tests/Cargo.toml1
-rw-r--r--html5lib_tests/src/lib.rs237
-rw-r--r--integration_tests/tests/test_html5lib.rs44
3 files changed, 137 insertions, 145 deletions
diff --git a/html5lib_tests/Cargo.toml b/html5lib_tests/Cargo.toml
index 06fc6b8..66e4624 100644
--- a/html5lib_tests/Cargo.toml
+++ b/html5lib_tests/Cargo.toml
@@ -11,4 +11,3 @@ publish = false # prevent accidental publishes until it's ready to be published
[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
index c007317..6cf46db 100644
--- a/html5lib_tests/src/lib.rs
+++ b/html5lib_tests/src/lib.rs
@@ -1,5 +1,4 @@
-use html5tokenizer::{Attribute, Doctype, EndTag, Error, StartTag, Token};
-use serde::{de::Error as _, Deserialize};
+use serde::{de, Deserialize};
use std::collections::BTreeMap;
pub fn parse_tests(
@@ -9,120 +8,6 @@ pub fn parse_tests(
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")]
@@ -148,37 +33,127 @@ fn initial_states_default() -> Vec<InitialState> {
pub struct Test {
pub description: String,
pub input: String,
- pub output: ExpectedOutputTokens,
#[serde(default = "initial_states_default")]
pub initial_states: Vec<InitialState>,
+ #[serde(flatten)]
+ pub output: Output,
#[serde(default)]
double_escaped: bool,
#[serde(default)]
pub last_start_tag: Option<String>,
+}
+
+#[derive(Deserialize, PartialEq, Eq, Debug)]
+pub struct Output {
#[serde(default)]
- pub errors: Vec<ParseError>,
+ pub errors: Vec<Error>,
+ #[serde(rename = "output")]
+ pub tokens: Vec<Token>,
}
-#[derive(Debug, Eq, PartialEq)]
-pub struct ParseErrorInner(pub Error);
+#[derive(Debug, PartialEq, Eq)]
+pub enum Token {
+ Doctype {
+ name: Option<String>,
+ public_id: Option<String>,
+ system_id: Option<String>,
+ force_quirks: bool,
+ },
+ StartTag {
+ name: String,
+ attributes: BTreeMap<String, String>,
+ self_closing: bool,
+ },
+ EndTag {
+ name: String,
+ },
+ Comment(String),
+ Character(String),
+}
-impl<'de> Deserialize<'de> for ParseErrorInner {
+impl<'de> Deserialize<'de> for Token {
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))
+ deserializer.deserialize_seq(TokenVisitor)
+ }
+}
+
+#[derive(Deserialize)]
+enum TokenType {
+ #[serde(rename = "DOCTYPE")]
+ Doctype,
+ StartTag,
+ EndTag,
+ Comment,
+ Character,
+}
+
+struct TokenVisitor;
+
+impl<'de> de::Visitor<'de> for TokenVisitor {
+ type Value = Token;
+
+ fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+ formatter.write_str("an array describing a token")
+ }
+
+ fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
+ where
+ A: serde::de::SeqAccess<'de>,
+ {
+ let typ: TokenType = seq.next_element()?.ok_or(
+ de::Error::custom(
+ r#"expected first array element to be one of "DOCTYPE", "StartTag", "EndTag", "Comment" or "Character""#,
+ )
+ )?;
+
+ Ok(match typ {
+ TokenType::Doctype => Token::Doctype {
+ name: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("name"))?,
+ public_id: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("public_id"))?,
+ system_id: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("system_id"))?,
+ force_quirks: !seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("correctness"))?,
+ },
+ TokenType::StartTag => Token::StartTag {
+ name: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("name"))?,
+ attributes: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("attributes"))?,
+ self_closing: seq.next_element()?.unwrap_or_default(),
+ },
+ TokenType::EndTag => Token::EndTag {
+ name: seq
+ .next_element()?
+ .ok_or(de::Error::missing_field("name"))?,
+ },
+ TokenType::Comment => Token::Comment(
+ seq.next_element()?
+ .ok_or(de::Error::missing_field("data"))?,
+ ),
+ TokenType::Character => Token::Character(
+ seq.next_element()?
+ .ok_or(de::Error::missing_field("data"))?,
+ ),
+ })
}
}
#[derive(Deserialize, Debug, Eq, PartialEq)]
#[serde(rename_all = "camelCase")]
-pub struct ParseError {
- pub code: ParseErrorInner,
+pub struct Error {
+ pub code: String,
// TODO: lineno and column?
}
@@ -191,12 +166,12 @@ fn undo_double_escaping(mut test: Test) -> Test {
if test.double_escaped {
test.input = unescape(&test.input);
- test.output.0 = test
+ test.output.tokens = test
.output
- .0
+ .tokens
.into_iter()
.map(|token| match token {
- Token::String(x) => Token::String(unescape(&x)),
+ Token::Character(x) => Token::Character(unescape(&x)),
Token::Comment(x) => Token::Comment(unescape(&x)),
token => token,
})
diff --git a/integration_tests/tests/test_html5lib.rs b/integration_tests/tests/test_html5lib.rs
index 3236f0f..23adec0 100644
--- a/integration_tests/tests/test_html5lib.rs
+++ b/integration_tests/tests/test_html5lib.rs
@@ -1,6 +1,8 @@
use std::{fs::File, io::BufReader, path::Path};
-use html5lib_tests::{parse_tests, InitialState, ParseError, ParseErrorInner, Test};
+use html5lib_tests::{
+ parse_tests, Error as TestError, InitialState, Output, Test, Token as TestToken,
+};
use html5tokenizer::{InternalState, Reader, Token, Tokenizer};
use pretty_assertions::assert_eq;
@@ -107,22 +109,38 @@ fn run_test_inner<R: Reader>(
tokenizer.set_last_start_tag(last_start_tag);
}
- let mut actual_tokens = Vec::new();
- let mut actual_errors = Vec::new();
+ let mut actual = Output {
+ errors: Vec::new(),
+ tokens: Vec::new(),
+ };
for token in tokenizer {
let token = token.unwrap();
- if let Token::Error { error, .. } = token {
- actual_errors.push(ParseError {
- code: ParseErrorInner(error),
- });
- } else {
- actual_tokens.push(token);
- }
+ match token {
+ Token::Error { error, .. } => actual.errors.push(TestError {
+ code: error.to_string(),
+ }),
+ Token::StartTag(tag) => actual.tokens.push(TestToken::StartTag {
+ name: tag.name,
+ attributes: tag
+ .attributes
+ .into_iter()
+ .map(|(name, map_val)| (name, map_val.value))
+ .collect(),
+ self_closing: tag.self_closing,
+ }),
+ Token::EndTag(tag) => actual.tokens.push(TestToken::EndTag { name: tag.name }),
+ Token::String(data) => actual.tokens.push(TestToken::Character(data)),
+ Token::Comment(data) => actual.tokens.push(TestToken::Comment(data)),
+ Token::Doctype(doctype) => actual.tokens.push(TestToken::Doctype {
+ name: Some(doctype.name).filter(|name| !name.is_empty()),
+ public_id: doctype.public_identifier,
+ system_id: doctype.system_identifier,
+ force_quirks: doctype.force_quirks,
+ }),
+ };
}
- assert_eq!(test.output.0, actual_tokens);
-
- assert_eq!(test.errors, actual_errors);
+ assert_eq!(test.output, actual);
}