feat(editor): recognize AsciiDoc files with basic highlighting (#731)

Register .adoc/.asciidoc as AsciiDoc in the language catalog (backend registry and in-memory backend) with // line comments, and add a lightweight CodeMirror stream mode for headings, attributes, delimited blocks, emphasis, inline code, URLs and comments. Refs #708.

Co-authored-by: Bl0ck154 <36800583+Bl0ck154@users.noreply.github.com>

crates/ide-language/src/registry.rs

@@ -8,6 +8,7 @@ use std::sync::Arc;
 
 /// Built-in definitions, embedded at compile time. Add new languages here.
 const BUILTIN: &[(&str, &str)] = &[
+    ("asciidoc", include_str!("../../../languages/asciidoc.toml")),
     ("bat", include_str!("../../../languages/bat.toml")),
     ("c", include_str!("../../../languages/c.toml")),
     ("clojure", include_str!("../../../languages/clojure.toml")),
@@ -373,6 +374,7 @@ mod tests {
     /// `name`s exported by `@codemirror/language-data` or built into `src/editor/core/cm/languages.ts`
     /// (the UI resolves `ui-mode` against these).
     const UI_MODES: &[&str] = &[
+        "AsciiDoc",
         "Batch",
         "C",
         "C++",
@@ -532,6 +534,12 @@ mod tests {
         let reg = reg();
         assert_eq!(detect(&reg, "src/main.rs"), "rust");
         assert_eq!(detect(&reg, "a/b.TS"), "typescript");
+        for path in ["guide.adoc", "guide.asciidoc", "GUIDE.ADOC"] {
+            assert_eq!(detect(&reg, path), "asciidoc");
+        }
+        let asciidoc = reg.get(&LanguageId::new("asciidoc")).unwrap();
+        assert_eq!(asciidoc.ui_mode.as_deref(), Some("AsciiDoc"));
+        assert_eq!(asciidoc.line_comment.as_deref(), Some("//"));
         assert_eq!(reg.detect(Path::new("script"), Some("#!/usr/bin/env python3")).as_str(), "python");
         assert_eq!(detect(&reg, "README"), "plaintext");
         assert_eq!(detect(&reg, "unknown.zzz"), "plaintext");

languages/asciidoc.toml

@@ -0,0 +1,6 @@
+id = "asciidoc"
+name = "AsciiDoc"
+extensions = ["adoc", "asciidoc"]
+line-comment = "//"
+indent = { tab-size = 2, insert-spaces = true }
+ui-mode = "AsciiDoc"

src/editor/core/backend.ts

@@ -119,6 +119,8 @@ const EXT_LANG: Record<string, string> = {
   go: "go",
   json: "json",
   md: "markdown",
+  adoc: "asciidoc",
+  asciidoc: "asciidoc",
   css: "css",
   html: "html",
   yml: "yaml",
@@ -218,6 +220,15 @@ const MEM_LANGUAGES: LanguageInfo[] = [
     blockComment: null,
     indent: { tabSize: 2, insertSpaces: true },
   },
+  {
+    id: "asciidoc",
+    name: "AsciiDoc",
+    extensions: ["adoc", "asciidoc"],
+    uiMode: "AsciiDoc",
+    lineComment: "//",
+    blockComment: null,
+    indent: { tabSize: 2, insertSpaces: true },
+  },
   {
     id: "markdown",
     name: "Markdown",

src/editor/core/cm/asciidoc.ts

@@ -0,0 +1,50 @@
+import { StreamLanguage, type StreamParser } from "@codemirror/language";
+
+type State = { block: string | null };
+
+const parser: StreamParser<State> = {
+  startState: () => ({ block: null }),
+  copyState: ({ block }) => ({ block }),
+  token(stream, state) {
+    if (stream.sol()) {
+      const line = stream.string;
+      const delimiter = line.trim();
+      if (state.block) {
+        if (delimiter === state.block) {
+          state.block = null;
+          stream.skipToEnd();
+          return "meta";
+        }
+        stream.skipToEnd();
+        return "monospace";
+      }
+      if (/^={1,6}\s+\S/.test(line)) {
+        stream.skipToEnd();
+        return "heading";
+      }
+      if (/^:[\w-]+:/.test(line)) {
+        stream.skipToEnd();
+        return "attributeName";
+      }
+      if (["----", "....", "____", "++++"].includes(delimiter)) {
+        state.block = delimiter;
+        stream.skipToEnd();
+        return "meta";
+      }
+      if (/^\s*\/\//.test(line)) {
+        stream.skipToEnd();
+        return "comment";
+      }
+    }
+    if (stream.eatSpace()) return null;
+    if (stream.match(/^https?:\/\/[^\s[\]<>]+/)) return "link";
+    if (stream.match(/^\*[^*\n]+\*/)) return "strong";
+    if (stream.match(/^_[^_\n]+_/)) return "emphasis";
+    if (stream.match(/^\x60[^\x60\n]+\x60/)) return "monospace";
+    stream.next();
+    return null;
+  },
+  languageData: { commentTokens: { line: "//" } },
+};
+
+export const asciidocLanguage = StreamLanguage.define(parser);

src/editor/core/cm/languages.test.ts

@@ -3,6 +3,54 @@ import { highlightTree, tagHighlighter, tags as t } from "@lezer/highlight";
 import { findLanguageDescription, loadSupport } from "./languages";
 
 describe("language modes", () => {
+  it("recognizes AsciiDoc files and loads their lightweight syntax mode", async () => {
+    for (const path of ["/docs/guide.adoc", "/docs/guide.asciidoc"]) {
+      const description = findLanguageDescription("AsciiDoc", path);
+      expect(description?.name).toBe("AsciiDoc");
+      const support = await loadSupport(description!);
+      expect(support?.language).toBeTruthy();
+    }
+  });
+
+  it("assigns highlight tags to AsciiDoc syntax", async () => {
+    const support = await loadSupport(findLanguageDescription("AsciiDoc", "/docs/guide.adoc")!);
+    const doc = [
+      "= Guide",
+      ":toc: left",
+      "----",
+      "const example = true;",
+      "----",
+      "*bold* _italic_ `code` https://example.com",
+      "// Comment",
+    ].join("\n");
+    const highlighter = tagHighlighter([
+      { tag: t.heading, class: "heading" },
+      { tag: t.attributeName, class: "attribute" },
+      { tag: t.meta, class: "meta" },
+      { tag: t.monospace, class: "monospace" },
+      { tag: t.strong, class: "strong" },
+      { tag: t.emphasis, class: "emphasis" },
+      { tag: t.link, class: "link" },
+      { tag: t.comment, class: "comment" },
+    ]);
+    const highlighted: [string, string][] = [];
+    highlightTree(support!.language.parser.parse(doc), highlighter, (from, to, style) => {
+      highlighted.push([doc.slice(from, to), style]);
+    });
+    expect(highlighted).toEqual([
+      ["= Guide", "heading"],
+      [":toc: left", "attribute"],
+      ["----", "meta"],
+      ["const example = true;", "monospace"],
+      ["----", "meta"],
+      ["*bold*", "strong"],
+      ["_italic_", "emphasis"],
+      ["`code`", "monospace"],
+      ["https://example.com", "link"],
+      ["// Comment", "comment"],
+    ]);
+  });
+
   it("styles `?.` like `.` in JavaScript and TypeScript", async () => {
     const highlighter = tagHighlighter([{ tag: t.derefOperator, class: "deref" }]);
     for (const [mode, path] of [

src/editor/core/cm/languages.ts

@@ -68,6 +68,11 @@ const OWN_MODES = [
     filename: /^(?:CMakeLists\.txt|.*\.cmake(?:\.in)?)$/,
     load: async () => new LanguageSupport((await import("./cmake")).cmakeLanguage),
   }),
+  LanguageDescription.of({
+    name: "AsciiDoc",
+    extensions: ["adoc", "asciidoc"],
+    load: () => import("./asciidoc").then((m) => new LanguageSupport(m.asciidocLanguage)),
+  }),
   LanguageDescription.of({
     name: "Shell",
     load: async () => {
@@ -245,7 +250,7 @@ const DEDENT_STATEMENT = /^\s*(?:pass|break|continue|return)\b/;
 const KEYWORD_OPEN_LANGS = new Set(["elixir", "ruby", "lua"]);
 const KEYWORD_OPEN =
   /\b(?:do(?:\s*\|[^|]*\|)?|then|->)$|^\s*(?:def|defp|defmodule|class|module|function|local\s+function|if|unless|elsif|elseif|else|while|until|for|case|begin|cond|with|receive|try)\b/;
-const NO_INDENT_LANGS = new Set(["plaintext", "markdown", "log", "diff", "ignore", "gitattributes"]);
+const NO_INDENT_LANGS = new Set(["plaintext", "markdown", "asciidoc", "log", "diff", "ignore", "gitattributes"]);
 
 function opensBlock(code: string, language: string | undefined): boolean {
   if (!code) return false;