feat: label code fences so Vicinae can colour them (#10)

src/__tests__/code.test.ts

@@ -0,0 +1,129 @@
+import { describe, expect, it } from "vitest";
+import { inferLanguage, labelCodeFences } from "../lib/code";
+
+describe("inferLanguage", () => {
+  it("returns nothing for empty or whitespace-only code", () => {
+    expect(inferLanguage("")).toBeUndefined();
+    expect(inferLanguage("   \n  ")).toBeUndefined();
+  });
+
+  describe("shell", () => {
+    it("recognises a command as the first word", () => {
+      expect(inferLanguage("pwd")).toBe("bash");
+      expect(inferLanguage("git status")).toBe("bash");
+      expect(inferLanguage("brew install ripgrep")).toBe("bash");
+    });
+
+    it("recognises an explicit prompt marker", () => {
+      expect(inferLanguage("$ git status")).toBe("bash");
+      expect(inferLanguage("> ls -la")).toBe("bash");
+    });
+
+    it("recognises a home-relative path", () => {
+      expect(inferLanguage("~/Documents/Code")).toBe("bash");
+    });
+
+    it("recognises a pipeline between known commands", () => {
+      expect(inferLanguage("git log --oneline | head -20")).toBe("bash");
+    });
+
+    it("ignores comments when looking at the first word", () => {
+      expect(inferLanguage("# check the branch\ngit branch --show-current")).toBe("bash");
+    });
+  });
+
+  describe("json", () => {
+    it("recognises a parseable object", () => {
+      expect(inferLanguage('{\n  "model": "gpt-oss"\n}')).toBe("json");
+    });
+
+    it("recognises a parseable array", () => {
+      expect(inferLanguage('["a", "b"]')).toBe("json");
+    });
+
+    it("does not claim malformed braces", () => {
+      expect(inferLanguage("{ not actually json")).not.toBe("json");
+    });
+  });
+
+  describe("typescript", () => {
+    it("recognises type annotations", () => {
+      expect(inferLanguage("function add(a: number, b: number) {\n  return a + b;\n}")).toBe("ts");
+    });
+
+    it("recognises interfaces and type aliases", () => {
+      expect(inferLanguage("interface User {\n  id: string;\n}")).toBe("ts");
+      expect(inferLanguage("type Id = string | number;")).toBe("ts");
+    });
+
+    it("does not label plain JavaScript, which is ambiguous", () => {
+      // Guessing wrong costs the highlighting, so this stays unlabelled.
+      expect(inferLanguage("console.log('hi')")).toBeUndefined();
+      expect(inferLanguage("const x = [1, 2, 3].map((n) => n * 2);")).toBeUndefined();
+    });
+  });
+
+  describe("python", () => {
+    it("recognises a function definition", () => {
+      expect(inferLanguage("def greet(name):\n    return f'hi {name}'")).toBe("python");
+    });
+
+    it("recognises imports and self", () => {
+      expect(inferLanguage("from pathlib import Path")).toBe("python");
+      expect(inferLanguage("class A:\n    def go(self):\n        pass")).toBe("python");
+    });
+  });
+
+  it("leaves genuinely ambiguous content alone", () => {
+    expect(inferLanguage("some plain text\nacross two lines")).toBeUndefined();
+  });
+});
+
+describe("labelCodeFences", () => {
+  it("labels a bare fence", () => {
+    const labelled = labelCodeFences("```\ngit status\n```");
+    expect(labelled).toBe("```bash\ngit status\n```");
+  });
+
+  it("leaves an already labelled fence exactly as the model wrote it", () => {
+    expect(labelCodeFences("```rust\nfn main() {}\n```")).toBe("```rust\nfn main() {}\n```");
+  });
+
+  it("labels each block independently", () => {
+    const labelled = labelCodeFences(["```", "pwd", "```", "", "```", '{"a": 1}', "```"].join("\n"));
+
+    expect(labelled).toContain("```bash");
+    expect(labelled).toContain("```json");
+  });
+
+  it("leaves an unlabellable block bare rather than guessing", () => {
+    const input = "```\njust some words\n```";
+    expect(labelCodeFences(input)).toBe(input);
+  });
+
+  it("preserves surrounding prose", () => {
+    const labelled = labelCodeFences("Before:\n\n```\nls\n```\n\nAfter.");
+
+    expect(labelled).toContain("Before:");
+    expect(labelled).toContain("After.");
+    expect(labelled).toContain("```bash");
+  });
+
+  it("handles a fence that is never closed", () => {
+    expect(labelCodeFences("```\ngit log")).toContain("```bash");
+  });
+
+  it("is a no-op on markdown with no fences", () => {
+    const input = "Just a paragraph with `inline code` in it.";
+    expect(labelCodeFences(input)).toBe(input);
+  });
+
+  it("is idempotent", () => {
+    const once = labelCodeFences("```\nls\n```");
+    expect(labelCodeFences(once)).toBe(once);
+  });
+
+  it("preserves indentation of a nested fence", () => {
+    expect(labelCodeFences("  ```\n  ls\n  ```")).toContain("  ```bash");
+  });
+});