use SpanContents instead of SourceCode for language detection

src/handlers/graphical.rs

@@ -482,7 +482,7 @@ impl GraphicalReportHandler {
             .map(|(label, st)| FancySpan::new(label.label().map(String::from), *label.inner(), st))
             .collect::<Vec<_>>();
 
-        let mut highlighter_state = self.highlighter.start_highlighter_state(source);
+        let mut highlighter_state = self.highlighter.start_highlighter_state(&*contents);
 
         // The max number of gutter-lines that will be active at any given
         // point. We need this to figure out indentation, so we do one loop

src/highlighters/blank.rs

@@ -1,5 +1,7 @@
 use owo_colors::Style;
 
+use crate::SpanContents;
+
 use super::{Highlighter, HighlighterState};
 
 /// The default syntax highlighter. It applies `Style::default()` to input text.
@@ -10,7 +12,7 @@ pub struct BlankHighlighter;
 impl Highlighter for BlankHighlighter {
     fn start_highlighter_state<'h>(
         &'h self,
-        _source: &dyn crate::SourceCode,
+        _source: &dyn SpanContents<'_>,
     ) -> Box<dyn super::HighlighterState + 'h> {
         Box::new(BlankHighlighterState)
     }

src/highlighters/mod.rs

@@ -13,7 +13,7 @@
 
 use std::{ops::Deref, sync::Arc};
 
-use crate::SourceCode;
+use crate::SpanContents;
 use owo_colors::Styled;
 
 #[cfg(feature = "syntect-highlighter")]
@@ -24,22 +24,22 @@ mod blank;
 #[cfg(feature = "syntect-highlighter")]
 mod syntect;
 
-/// A syntax highlighter for highlighting miette [SourceCode] snippets.
+/// A syntax highlighter for highlighting miette [`SourceCode`](crate::SourceCode) snippets.
 pub trait Highlighter {
-    /// Creates a new [HighlighterState] to begin parsing and highlighting
-    /// a [SourceCode] snippet.
+    ///  Creates a new [HighlighterState] to begin parsing and highlighting
+    /// a [SpanContents].
     ///
     /// The [GraphicalReportHandler](crate::GraphicalReportHandler) will call
-    /// this method at the start of rendering a [Diagnostic](crate::Diagnostic).
+    /// this method at the start of rendering a [SpanContents].
     ///
-    /// The source is provided as input only so that the Highlighter can detect
-    /// language syntax and make other initialization decisions prior
+    /// The [SpanContents] is provided as input only so that the [Highlighter]
+    /// can detect language syntax and make other initialization decisions prior
     /// to highlighting, but it is not intended that the Highlighter begin
     /// highlighting at this point. The returned [HighlighterState] is
     /// responsible for the actual rendering.
     fn start_highlighter_state<'h>(
         &'h self,
-        source: &dyn SourceCode,
+        source: &dyn SpanContents<'_>,
     ) -> Box<dyn HighlighterState + 'h>;
 }
 

src/highlighters/syntect.rs

@@ -14,7 +14,7 @@ use owo_colors::{Rgb, Style, Styled};
 
 use crate::{
     highlighters::{Highlighter, HighlighterState},
-    SourceCode,
+    SpanContents,
 };
 
 use super::BlankHighlighterState;
@@ -41,9 +41,9 @@ impl Default for SyntectHighlighter {
 impl Highlighter for SyntectHighlighter {
     fn start_highlighter_state<'h>(
         &'h self,
-        source: &dyn SourceCode,
+        source: &dyn SpanContents<'_>,
     ) -> Box<dyn HighlighterState + 'h> {
-        if let Some(syntax) = self.get_syntax_from_source(source) {
+        if let Some(syntax) = self.detect_syntax(source) {
             let highlighter = syntect::Highlighter::new(&self.theme);
             let parse_state = syntect::ParseState::new(syntax);
             let highlight_state =
@@ -81,19 +81,26 @@ impl SyntectHighlighter {
     }
 
     /// Determine syntect SyntaxReference to use for given SourceCode
-    fn get_syntax_from_source(&self, source: &dyn SourceCode) -> Option<&syntect::SyntaxReference> {
-        if let Some(language) = source.language() {
-            self.syntax_set.find_syntax_by_name(language)
-        } else if let Some(name) = source.name() {
-            if let Some(ext) = Path::new(name).extension() {
-                self.syntax_set
-                    .find_syntax_by_extension(ext.to_string_lossy().as_ref())
-            } else {
-                None
-            }
-        } else {
-            None
+    fn detect_syntax(&self, contents: &dyn SpanContents<'_>) -> Option<&syntect::SyntaxReference> {
+        // use language if given
+        if let Some(language) = contents.language() {
+            return self.syntax_set.find_syntax_by_name(language);
         }
+        // otherwise try to use any file extension provided in the name
+        if let Some(name) = contents.name() {
+            if let Some(ext) = Path::new(name).extension() {
+                return self
+                    .syntax_set
+                    .find_syntax_by_extension(ext.to_string_lossy().as_ref());
+            }
+        }
+        // finally, attempt to guess syntax based on first line
+        return self.syntax_set.find_syntax_by_first_line(
+            &std::str::from_utf8(contents.data())
+                .ok()?
+                .split('\n')
+                .next()?,
+        );
     }
 }
 

src/lib.rs

@@ -655,13 +655,14 @@
 //! automatically use the [`syntect`] crate to highlight the `#[source_code]`
 //! field of your [`Diagnostic`].
 //!
-//! Syntax detection with [`syntect`] is handled by checking 2 methods on the [`SourceCode`] trait, in order:
-//! * [language()](SourceCode::language) - Provides the name of the language
+//! Syntax detection with [`syntect`] is handled by checking 2 methods on the [`SpanContents`] trait, in order:
+//! * [language()](SpanContents::language) - Provides the name of the language
 //!   as a string. For example `"Rust"` will indicate Rust syntax highlighting.
-//!   You can set the language of a [`NamedSource`] via the
-//!   [`with_language`](NamedSource::with_language) method.
-//! * [name()](SourceCode::name) - In the absence of an explicitly set
-//!   language, the source code's name is assumed to contain a file name or file path.
+//!   You can set the language of the [`SpanContents`] produced by a
+//!   [`NamedSource`] via the [`with_language`](NamedSource::with_language)
+//!   method.
+//! * [name()](SpanContents::name) - In the absence of an explicitly set
+//!   language, the name is assumed to contain a file name or file path.
 //!   The highlighter will check for a file extension at the end of the name and
 //!   try to guess the syntax from that.
 //!

src/named_source.rs

@@ -36,7 +36,7 @@ impl NamedSource {
         &*self.source
     }
 
-    /// Sets the [`language`](SourceCode::language) for this source code.
+    /// Sets the [`language`](SpanContents::language) for this source code.
     pub fn with_language(mut self, language: impl Into<String>) -> Self {
         self.language = Some(language.into());
         self
@@ -50,27 +50,20 @@ impl SourceCode for NamedSource {
         context_lines_before: usize,
         context_lines_after: usize,
     ) -> Result<Box<dyn SpanContents<'a> + 'a>, MietteError> {
-        let contents = self
-            .inner()
-            .read_span(span, context_lines_before, context_lines_after)?;
-        Ok(Box::new(MietteSpanContents::new_named(
+        let inner_contents =
+            self.inner()
+                .read_span(span, context_lines_before, context_lines_after)?;
+        let mut contents = MietteSpanContents::new_named(
             self.name.clone(),
-            contents.data(),
-            *contents.span(),
-            contents.line(),
-            contents.column(),
-            contents.line_count(),
-        )))
-    }
-
-    fn name(&self) -> Option<&str> {
-        Some(&self.name)
-    }
-
-    fn language(&self) -> Option<&str> {
-        match &self.language {
-            Some(language) => Some(language),
-            None => self.source.language(),
+            inner_contents.data(),
+            *inner_contents.span(),
+            inner_contents.line(),
+            inner_contents.column(),
+            inner_contents.line_count(),
+        );
+        if let Some(language) = &self.language {
+            contents = contents.with_language(language);
         }
+        Ok(Box::new(contents))
     }
 }

src/protocol.rs

@@ -240,20 +240,6 @@ pub trait SourceCode: Send + Sync {
         context_lines_before: usize,
         context_lines_after: usize,
     ) -> Result<Box<dyn SpanContents<'a> + 'a>, MietteError>;
-
-    /// Optional method. The name of this source, if any
-    fn name(&self) -> Option<&str> {
-        None
-    }
-
-    /// Optional method. The language name for this source code, if any.
-    /// This is used to drive syntax highlighting.
-    ///
-    /// Examples: Rust, TOML, C
-    ///
-    fn language(&self) -> Option<&str> {
-        None
-    }
 }
 
 /// A labeled [`SourceSpan`].
@@ -447,6 +433,15 @@ pub trait SpanContents<'a> {
     fn column(&self) -> usize;
     /// Total number of lines covered by this `SpanContents`.
     fn line_count(&self) -> usize;
+
+    /// Optional method. The language name for this source code, if any.
+    /// This is used to drive syntax highlighting.
+    ///
+    /// Examples: Rust, TOML, C
+    ///
+    fn language(&self) -> Option<&str> {
+        None
+    }
 }
 
 /**
@@ -466,6 +461,8 @@ pub struct MietteSpanContents<'a> {
     line_count: usize,
     // Optional filename
     name: Option<String>,
+    // Optional language
+    language: Option<String>,
 }
 
 impl<'a> MietteSpanContents<'a> {
@@ -484,6 +481,7 @@ impl<'a> MietteSpanContents<'a> {
             column,
             line_count,
             name: None,
+            language: None,
         }
     }
 
@@ -503,8 +501,15 @@ impl<'a> MietteSpanContents<'a> {
             column,
             line_count,
             name: Some(name),
+            language: None,
         }
     }
+
+    /// Sets the [`language`](SourceCode::language) for syntax highlighting.
+    pub fn with_language(mut self, language: impl Into<String>) -> Self {
+        self.language = Some(language.into());
+        self
+    }
 }
 
 impl<'a> SpanContents<'a> for MietteSpanContents<'a> {
@@ -526,6 +531,9 @@ impl<'a> SpanContents<'a> for MietteSpanContents<'a> {
     fn name(&self) -> Option<&str> {
         self.name.as_deref()
     }
+    fn language(&self) -> Option<&str> {
+        self.language.as_deref()
+    }
 }
 
 /// Span within a [`SourceCode`]