restore language auto

Author: ealmloff

src/advanced.rs

@@ -22,6 +22,9 @@ use std::{borrow::Cow, fmt, ops::Range};
 #[derive(Debug, Clone, PartialEq, Eq)]
 #[non_exhaustive]
 pub enum HighlightError {
+    /// [`Language::Auto`] could not detect a supported language.
+    #[cfg(feature = "detection")]
+    LanguageDetectionFailed,
     /// The tree-sitter parser rejected the selected grammar.
     GrammarLoad {
         /// The language whose grammar failed to load.
@@ -92,6 +95,8 @@ impl HighlightError {
 impl fmt::Display for HighlightError {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
+            #[cfg(feature = "detection")]
+            Self::LanguageDetectionFailed => write!(f, "could not detect language"),
             Self::GrammarLoad { language, message } => {
                 write!(
                     f,
@@ -624,6 +629,7 @@ impl Buffer {
     /// ```
     pub fn new(language: Language, source: impl ToString) -> Result<Self, HighlightError> {
         let source = source.to_string();
+        let language = resolve_language(language, &source)?;
         let (mut parser, incremental) = Self::parser_for(language)?;
         let mut cursor = arborium_tree_sitter::QueryCursor::new();
         let (tree, spans) = Self::parse_source(
@@ -733,6 +739,7 @@ impl Buffer {
     /// assert_eq!(buffer.language(), Language::Rust);
     /// ```
     pub fn set_language(&mut self, language: Language) -> Result<(), HighlightError> {
+        let language = resolve_language(language, &self.source)?;
         if self.language == language {
             return Ok(());
         }
@@ -853,11 +860,23 @@ fn collect_spans(
     normalize_spans(raw)
 }
 
+#[cfg(feature = "runtime")]
+fn resolve_language(language: Language, _source: &str) -> Result<Language, HighlightError> {
+    #[cfg(feature = "detection")]
+    if language == Language::Auto {
+        return Language::detect(_source).ok_or(HighlightError::LanguageDetectionFailed);
+    }
+
+    Ok(language)
+}
+
 #[cfg(feature = "runtime")]
 fn grammar_for(language: Language) -> (arborium_tree_sitter::LanguageFn, &'static str) {
     // Rust is bundled with the `runtime` feature; everything else is opt-in via
     // its `lang-*` cargo feature (or the `all-languages` umbrella).
     match language {
+        #[cfg(feature = "detection")]
+        Language::Auto => unreachable!("auto language must be resolved before loading a grammar"),
         Language::Rust => (
             arborium::lang_rust::language(),
             arborium::lang_rust::HIGHLIGHTS_QUERY,

src/language.rs

@@ -26,6 +26,9 @@ macro_rules! define_languages {
         #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
         #[non_exhaustive]
         pub enum Language {
+            /// Automatically detect the language from the source text.
+            #[cfg(feature = "detection")]
+            Auto,
             $(
                 $(#[$attr])*
                 #[doc = concat!("Arborium slug `\"", $slug, "\"`.")]
@@ -44,6 +47,8 @@ macro_rules! define_languages {
             /// assert!(Language::ALL.contains(&Language::Rust));
             /// ```
             pub const ALL: &'static [Language] = &[
+                #[cfg(feature = "detection")]
+                Self::Auto,
                 $(
                     $(#[$attr])*
                     Self::$variant,
@@ -53,6 +58,8 @@ macro_rules! define_languages {
             /// Arborium slug for this language.
             pub const fn slug(self) -> &'static str {
                 match self {
+                    #[cfg(feature = "detection")]
+                    Self::Auto => "auto",
                     $(
                         $(#[$attr])*
                         Self::$variant => $slug,

src/lib.rs

@@ -229,6 +229,14 @@ pub struct SourceCode {
     language: Language,
 }
 
+/// Source-first builder for [`SourceCode`].
+#[cfg(feature = "runtime")]
+#[cfg_attr(docsrs, doc(cfg(feature = "runtime")))]
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct SourceCodeBuilder {
+    source: String,
+}
+
 #[cfg(feature = "runtime")]
 #[cfg_attr(docsrs, doc(cfg(feature = "runtime")))]
 impl SourceCode {
@@ -245,6 +253,13 @@ impl SourceCode {
         }
     }
 
+    /// Start a source-first builder.
+    pub fn builder(source: impl ToString) -> SourceCodeBuilder {
+        SourceCodeBuilder {
+            source: source.to_string(),
+        }
+    }
+
     /// Replace the language used to highlight this source.
     ///
     /// To set the language from a runtime slug, use [`Language::from_slug`]
@@ -277,6 +292,15 @@ impl SourceCode {
     }
 }
 
+#[cfg(feature = "runtime")]
+#[cfg_attr(docsrs, doc(cfg(feature = "runtime")))]
+impl SourceCodeBuilder {
+    /// Finish the builder with an explicit language.
+    pub fn with_language(self, language: Language) -> SourceCode {
+        SourceCode::new(language, self.source)
+    }
+}
+
 #[cfg(feature = "runtime")]
 pub(crate) struct RawHighlightSpan {
     pub(crate) start: u32,