feat: support formatting of languages embedded in tagged template literals (Rust API Only) (#701)

Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>

Author: kt3k

Cargo.lock

@@ -31,6 +31,7 @@ harness = false
 
 [dependencies]
 anyhow = "1.0.64"
+capacity_builder = "0.5.0"
 deno_ast = { version = "0.46.3", features = ["view"] }
 dprint-core = { version = "0.67.4", features = ["formatting"] }
 dprint-core-macros = "0.1.0"
@@ -41,5 +42,8 @@ serde_json = { version = "1.0", optional = true }
 
 [dev-dependencies]
 dprint-development = "0.10.1"
+dprint-plugin-sql = "0.2.0"
+malva = "0.11.1"
+markup_fmt = "0.19.0"
 pretty_assertions = "1.3.0"
 serde_json = { version = "1.0" }

Cargo.toml

@@ -10,8 +10,17 @@ use crate::swc::ensure_no_specific_syntax_errors;
 
 use super::configuration::Configuration;
 use super::generation::generate;
+pub use super::generation::ExternalFormatter;
 use super::swc::parse_swc_ast;
 
+pub struct FormatTextOptions<'a> {
+  pub path: &'a Path,
+  pub extension: Option<&'a str>,
+  pub text: String,
+  pub config: &'a Configuration,
+  pub external_formatter: Option<&'a ExternalFormatter>,
+}
+
 /// Formats a file.
 ///
 /// Returns the file text or an error when it failed to parse.
@@ -35,19 +44,32 @@ use super::swc::parse_swc_ast;
 /// // now format many files (it is recommended to parallelize this)
 /// let files_to_format = vec![(PathBuf::from("path/to/file.ts"), "const  t  =  5 ;")];
 /// for (file_path, file_text) in files_to_format {
-///     let result = format_text(&file_path, None, file_text.into(), &config);
+///     let result = format_text(FormatTextOptions {
+///         path: &file_path,
+///         extension: None,
+///         text: file_text.into(),
+///         config: &config,
+///         external_formatter: None,
+///     });
 ///     // save result here...
 /// }
 /// ```
-pub fn format_text(file_path: &Path, file_extension: Option<&str>, file_text: String, config: &Configuration) -> Result<Option<String>> {
+pub fn format_text(options: FormatTextOptions) -> Result<Option<String>> {
+  let FormatTextOptions {
+    path: file_path,
+    extension: file_extension,
+    text: file_text,
+    config,
+    external_formatter,
+  } = options;
   if super::utils::file_text_has_ignore_comment(&file_text, &config.ignore_file_comment_text) {
     Ok(None)
   } else {
     let had_bom = file_text.starts_with("\u{FEFF}");
     let file_text = if had_bom { file_text[3..].to_string() } else { file_text };
     let file_text: Arc<str> = file_text.into();
     let parsed_source = parse_swc_ast(file_path, file_extension, file_text)?;
-    match inner_format(&parsed_source, config)? {
+    match inner_format(&parsed_source, config, external_formatter)? {
       Some(new_text) => Ok(Some(new_text)),
       None => {
         if had_bom {
@@ -61,20 +83,20 @@ pub fn format_text(file_path: &Path, file_extension: Option<&str>, file_text: St
 }
 
 /// Formats an already parsed source. This is useful as a performance optimization.
-pub fn format_parsed_source(source: &ParsedSource, config: &Configuration) -> Result<Option<String>> {
+pub fn format_parsed_source(source: &ParsedSource, config: &Configuration, external_formatter: Option<&ExternalFormatter>) -> Result<Option<String>> {
   if super::utils::file_text_has_ignore_comment(source.text(), &config.ignore_file_comment_text) {
     Ok(None)
   } else {
     ensure_no_specific_syntax_errors(source)?;
-    inner_format(source, config)
+    inner_format(source, config, external_formatter)
   }
 }
 
-fn inner_format(parsed_source: &ParsedSource, config: &Configuration) -> Result<Option<String>> {
+fn inner_format(parsed_source: &ParsedSource, config: &Configuration, external_formatter: Option<&ExternalFormatter>) -> Result<Option<String>> {
   let result = dprint_core::formatting::format(
     || {
       #[allow(clippy::let_and_return)]
-      let print_items = generate(parsed_source, config);
+      let print_items = generate(parsed_source, config, external_formatter);
       // println!("{}", print_items.get_as_text());
       print_items
     },
@@ -91,7 +113,7 @@ fn inner_format(parsed_source: &ParsedSource, config: &Configuration) -> Result<
 pub fn trace_file(file_path: &Path, file_text: &str, config: &Configuration) -> dprint_core::formatting::TracingResult {
   let parsed_source = parse_swc_ast(file_path, None, file_text.into()).unwrap();
   ensure_no_specific_syntax_errors(&parsed_source).unwrap();
-  dprint_core::formatting::trace_printing(|| generate(&parsed_source, config), config_to_print_options(file_text, config))
+  dprint_core::formatting::trace_printing(|| generate(&parsed_source, config, None), config_to_print_options(file_text, config))
 }
 
 fn config_to_print_options(file_text: &str, config: &Configuration) -> PrintOptions {
@@ -111,9 +133,15 @@ mod test {
   fn strips_bom() {
     for input_text in ["\u{FEFF}const t = 5;\n", "\u{FEFF}const t =   5;"] {
       let config = crate::configuration::ConfigurationBuilder::new().build();
-      let result = format_text(&std::path::PathBuf::from("test.ts"), None, input_text.into(), &config)
-        .unwrap()
-        .unwrap();
+      let result = format_text(FormatTextOptions {
+        path: &std::path::PathBuf::from("test.ts"),
+        extension: None,
+        text: input_text.into(),
+        config: &config,
+        external_formatter: None,
+      })
+      .unwrap()
+      .unwrap();
       assert_eq!(result, "const t = 5;\n");
     }
   }

src/format_text.rs

@@ -18,11 +18,41 @@ use super::*;
 use crate::configuration::*;
 use crate::utils::Stack;
 
+/// A callback that will be called when encountering certain tagged templates.
+///
+/// Currently supports `css`, `html` and `sql` tagged templated.
+///
+/// Examples:
+/// ```ignore
+/// const styles = css`color: red;`;
+///
+/// const markup = html`<html>
+///   <body>
+///     <h1>Hello!<h1>
+///   </body>
+/// </html>`;
+///
+/// const query = sql`
+/// SELECT
+///   *
+/// FROM
+///   users
+/// WHERE
+///   active IS TRUE;
+/// ```
+///
+/// External formatter should return `None` if it doesn't understand given `MediaType`, in such
+/// cases the templates will be left as they are.
+///
+/// Only templates with no interpolation are supported.
+pub type ExternalFormatter = dyn Fn(MediaType, String, &Configuration) -> Option<String>;
+
 pub struct Context<'a> {
   pub media_type: MediaType,
   pub program: Program<'a>,
   pub config: &'a Configuration,
   pub comments: CommentTracker<'a>,
+  pub external_formatter: Option<&'a ExternalFormatter>,
   pub token_finder: TokenFinder<'a>,
   pub current_node: Node<'a>,
   pub parent_stack: Stack<Node<'a>>,
@@ -43,12 +73,20 @@ pub struct Context<'a> {
 }
 
 impl<'a> Context<'a> {
-  pub fn new(media_type: MediaType, tokens: &'a [TokenAndSpan], current_node: Node<'a>, program: Program<'a>, config: &'a Configuration) -> Context<'a> {
+  pub fn new(
+    media_type: MediaType,
+    tokens: &'a [TokenAndSpan],
+    current_node: Node<'a>,
+    program: Program<'a>,
+    config: &'a Configuration,
+    external_formatter: Option<&'a ExternalFormatter>,
+  ) -> Context<'a> {
     Context {
       media_type,
       program,
       config,
       comments: CommentTracker::new(program, tokens),
+      external_formatter,
       token_finder: TokenFinder::new(program),
       current_node,
       parent_stack: Default::default(),

src/generation/context.rs

@@ -25,13 +25,20 @@ use super::*;
 use crate::configuration::*;
 use crate::utils;
 
-pub fn generate(parsed_source: &ParsedSource, config: &Configuration) -> PrintItems {
+pub fn generate(parsed_source: &ParsedSource, config: &Configuration, external_formatter: Option<&ExternalFormatter>) -> PrintItems {
   // eprintln!("Leading: {:?}", parsed_source.comments().leading_map());
   // eprintln!("Trailing: {:?}", parsed_source.comments().trailing_map());
 
   parsed_source.with_view(|program| {
     let program_node = program.into();
-    let mut context = Context::new(parsed_source.media_type(), parsed_source.tokens(), program_node, program, config);
+    let mut context = Context::new(
+      parsed_source.media_type(),
+      parsed_source.tokens(),
+      program_node,
+      program,
+      config,
+      external_formatter,
+    );
     let mut items = gen_node(program_node, &mut context);
     items.push_condition(if_true(
       "endOfFileNewLine",
@@ -3001,6 +3008,97 @@ fn gen_spread_element<'a>(node: &SpreadElement<'a>, context: &mut Context<'a>) -
   items
 }
 
+/// Formats the tagged template literal using an external formatter.
+/// Detects the type of embedded language automatically.
+fn maybe_gen_tagged_tpl_with_external_formatter<'a>(node: &TaggedTpl<'a>, context: &mut Context<'a>) -> Option<PrintItems> {
+  let external_formatter = context.external_formatter.as_ref()?;
+  let media_type = detect_embedded_language_type(node)?;
+
+  // First creates text with placeholders for the expressions.
+  let placeholder_text = "dpr1nt_";
+  let text = capacity_builder::StringBuilder::<String>::build(|builder| {
+    let expr_len = node.tpl.exprs.len();
+    for (i, quasi) in node.tpl.quasis.iter().enumerate() {
+      builder.append(quasi.raw().as_str());
+      if i < expr_len {
+        builder.append(placeholder_text);
+        if i < 10 {
+          // increase chance all placeholders have the same length
+          builder.append("0");
+        }
+        builder.append(i); // give each placeholder a unique name so the formatter doesn't remove duplicates
+        builder.append("_d");
+      }
+    }
+  })
+  .unwrap();
+
+  // Then formats the text with the external formatter.
+  let formatted_tpl = external_formatter(media_type, text, context.config)?;
+
+  let mut items = PrintItems::new();
+  items.push_sc(sc!("`"));
+  items.push_signal(Signal::NewLine);
+  items.push_signal(Signal::StartIndent);
+  let mut index = 0;
+  for line in formatted_tpl.lines() {
+    let mut pos = 0;
+    let mut parts = line.split(placeholder_text).enumerate().peekable();
+    while let Some((i, part)) = parts.next() {
+      let end = pos + part.len();
+      if i > 0 {
+        pos += part.find("_d").unwrap() + 2;
+      }
+      let text = &line[pos..end];
+      if !text.is_empty() {
+        items.push_string(text.to_string());
+      }
+      if parts.peek().is_some() {
+        items.push_sc(sc!("${"));
+        items.extend(gen_node(node.tpl.exprs[index].into(), context));
+        items.push_sc(sc!("}"));
+        pos = end + placeholder_text.len();
+        index += 1;
+      }
+    }
+    items.push_signal(Signal::NewLine);
+  }
+  items.push_signal(Signal::FinishIndent);
+  items.push_sc(sc!("`"));
+  Some(items)
+}
+
+/// Detects the type of embedded language in a tagged template literal.
+fn detect_embedded_language_type<'a>(node: &TaggedTpl<'a>) -> Option<MediaType> {
+  match node.tag {
+    Expr::Ident(ident) => {
+      match ident.sym().as_str() {
+        "css" => Some(MediaType::Css),   // css`...`
+        "html" => Some(MediaType::Html), // html`...`
+        "sql" => Some(MediaType::Sql),   // sql`...`
+        _ => None,
+      }
+    }
+    Expr::Member(member_expr) => {
+      if let Expr::Ident(ident) = member_expr.obj {
+        if ident.sym().as_str() == "styled" {
+          return Some(MediaType::Css); // styled.foo`...`
+        }
+      }
+      None
+    }
+    Expr::Call(call_expr) => {
+      if let Callee::Expr(Expr::Ident(ident)) = call_expr.callee {
+        if ident.sym().as_str() == "styled" {
+          return Some(MediaType::Css); // styled(Button)`...`
+        }
+      }
+      None
+    }
+    _ => None,
+  }
+}
+
 fn gen_tagged_tpl<'a>(node: &TaggedTpl<'a>, context: &mut Context<'a>) -> PrintItems {
   let use_space = context.config.tagged_template_space_before_literal;
   let mut items = gen_node(node.tag.into(), context);
@@ -3017,6 +3115,11 @@ fn gen_tagged_tpl<'a>(node: &TaggedTpl<'a>, context: &mut Context<'a>) -> PrintI
     items.extend(generated_between_comments);
   }
 
+  if let Some(formatted_tpl) = maybe_gen_tagged_tpl_with_external_formatter(node, context) {
+    items.push_condition(conditions::indent_if_start_of_line(formatted_tpl));
+    return items;
+  }
+
   items.push_condition(conditions::indent_if_start_of_line(gen_node(node.tpl.into(), context)));
   items
 }

src/generation/generate.rs

@@ -12,4 +12,5 @@ use context::*;
 use generate_types::*;
 use tokens::*;
 
+pub use context::ExternalFormatter;
 pub use generate::generate;

src/generation/mod.rs

@@ -18,6 +18,8 @@ mod utils;
 
 pub use format_text::format_parsed_source;
 pub use format_text::format_text;
+pub use format_text::ExternalFormatter;
+pub use format_text::FormatTextOptions;
 
 #[cfg(feature = "tracing")]
 pub use format_text::trace_file;

src/lib.rs

@@ -359,6 +359,6 @@ Merge conflict marker encountered. at file:///test.ts:6:1
     // a source file that had a non-fatal diagnostic
     let parsed_source = parse_inner_no_diagnostic_check(&file_path, None, text.into()).unwrap();
     let config = ConfigurationBuilder::new().build();
-    assert_eq!(crate::format_parsed_source(&parsed_source, &config).err().unwrap().to_string(), expected);
+    assert_eq!(crate::format_parsed_source(&parsed_source, &config, None).err().unwrap().to_string(), expected);
   }
 }

src/swc.rs

@@ -60,7 +60,15 @@ impl SyncPluginHandler<Configuration> for TypeScriptPluginHandler {
 
   fn format(&mut self, request: SyncFormatRequest<Configuration>, _format_with_host: impl FnMut(SyncHostFormatRequest) -> FormatResult) -> FormatResult {
     let file_text = String::from_utf8(request.file_bytes)?;
-    super::format_text(request.file_path, None, file_text, request.config).map(|maybe_text| maybe_text.map(|t| t.into_bytes()))
+    super::format_text(super::FormatTextOptions {
+      path: request.file_path,
+      extension: None,
+      text: file_text,
+      config: request.config,
+      // todo: support this in Wasm
+      external_formatter: None,
+    })
+    .map(|maybe_text| maybe_text.map(|t| t.into_bytes()))
   }
 }