spans: add span tests

6 files changed, 88 insertions, 7 deletions

diff --git a/Cargo.toml b/Cargo.toml
index a9cb994..13f8e5d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -11,6 +11,7 @@ version = "0.2.1"
 include = ["src/**/*", "LICENSE", "README.md"]
 
 [dev-dependencies]
+codespan-reporting = "0.11.1"
 pretty_assertions = "1.0.0"
 serde = { version = "1.0.130", features = ["derive"] }
 serde_json = "1.0.71"
diff --git a/src/lib.rs b/src/lib.rs
index 40f9588..4b74d61 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -9,7 +9,7 @@ mod error;
 mod machine;
 mod never;
 mod reader;
-mod spans;
+pub mod spans;
 mod tokenizer;
 mod utils;
 
diff --git a/src/spans.rs b/src/spans.rs
index 9d02940..72b30c0 100644
--- a/src/spans.rs
+++ b/src/spans.rs
@@ -1,3 +1,4 @@
+//! Source code spans.
 use std::{
     collections::{BTreeSet, VecDeque},
     marker::PhantomData,
@@ -8,13 +9,18 @@ use crate::{Doctype, Emitter, EndTag, Error, Reader, StartTag, Token};
 
 type Span = std::ops::Range<usize>;
 
+/// A trait to be implemented by readers that track their own position.
 pub trait GetPos {
+    /// Returns the byte index of the current position.
     fn get_pos(&self) -> usize;
 }
 
-struct PosTracker<R> {
-    reader: R,
-    position: usize,
+/// Wraps a [`Reader`] so that it implements [`GetPos`].
+pub struct PosTracker<R> {
+    /// The wrapped reader.
+    pub reader: R,
+    /// The current position.
+    pub position: usize,
 }
 
 impl<R> GetPos for PosTracker<R> {
@@ -145,12 +151,15 @@ impl<R: GetPos> Emitter<R> for SpanEmitter<R> {
 
     fn init_start_tag(&mut self, reader: &R) {
         self.current_token = Some(Token::StartTag(StartTag {
-            name_span: reader.get_pos()..reader.get_pos(),
+            name_span: reader.get_pos() - 1..reader.get_pos() - 1,
             ..Default::default()
         }));
     }
-    fn init_end_tag(&mut self, _reader: &R) {
-        self.current_token = Some(Token::EndTag(Default::default()));
+    fn init_end_tag(&mut self, reader: &R) {
+        self.current_token = Some(Token::EndTag(EndTag {
+            name_span: reader.get_pos() - 1..reader.get_pos() - 1,
+            ..Default::default()
+        }));
         self.seen_attributes.clear();
     }
 
diff --git a/tests/span-tests/demo.html b/tests/span-tests/demo.html
new file mode 100644
index 0000000..900f74c
--- /dev/null
+++ b/tests/span-tests/demo.html
@@ -0,0 +1 @@
+this is a tag: <h1>test</h1>
diff --git a/tests/span-tests/demo.out b/tests/span-tests/demo.out
new file mode 100644
index 0000000..98d6766
--- /dev/null
+++ b/tests/span-tests/demo.out
@@ -0,0 +1,7 @@
+note:
+  ┌─ test.html:1:17
+  │
+1 │ this is a tag: <h1>test</h1>
+  │                 ^^       ^^ end tag
+  │                 │
+  │                 start tag
diff --git a/tests/test_spans.rs b/tests/test_spans.rs
new file mode 100644
index 0000000..a3d1c96
--- /dev/null
+++ b/tests/test_spans.rs
@@ -0,0 +1,63 @@
+use std::include_str;
+
+use codespan_reporting::{
+    self,
+    diagnostic::{Diagnostic, Label},
+    files::SimpleFiles,
+    term::{self, termcolor::Buffer},
+};
+use html5gum::{
+    spans::{PosTracker, SpanEmitter},
+    Readable, Token, Tokenizer,
+};
+
+#[test]
+fn test() {
+    let html = include_str!("span-tests/demo.html");
+
+    let mut files = SimpleFiles::new();
+    let file_id = files.add("test.html", html);
+    let mut labels = Vec::new();
+
+    for token in Tokenizer::new_with_emitter(
+        PosTracker {
+            reader: html.to_reader(),
+            position: 0,
+        },
+        SpanEmitter::default(),
+    )
+    .infallible()
+    {
+        if let Token::StartTag(tag) = token {
+            labels.push(Label::primary(file_id, tag.name_span).with_message("start tag"));
+        } else if let Token::EndTag(tag) = token {
+            labels.push(Label::primary(file_id, tag.name_span).with_message("end tag"));
+        }
+    }
+
+    let diagnostic = Diagnostic::note().with_labels(labels);
+
+    let mut writer = Buffer::no_color();
+    let config = codespan_reporting::term::Config::default();
+    term::emit(&mut writer, &config, &files, &diagnostic).unwrap();
+
+    let actual = remove_trailing_spaces(std::str::from_utf8(writer.as_slice()).unwrap());
+    let expected = include_str!("span-tests/demo.out");
+
+    if actual != expected {
+        println!(
+            "EXPECTED:\n{banner}\n{expected}{banner}\n\nACTUAL OUTPUT:\n{banner}\n{actual}{banner}",
+            banner = "-".repeat(30),
+            expected = expected,
+            actual = actual
+        );
+        panic!("failed");
+    }
+}
+
+fn remove_trailing_spaces(text: &str) -> String {
+    text.lines()
+        .map(|l| l.trim_end())
+        .collect::<Vec<_>>()
+        .join("\n")
+}