disclaw/tests/unit/split-for-discord.test.ts

import { describe, it, expect } from "vitest";
import { splitForDiscord } from "../../src/discord/split";

const DISCORD_HARD_LIMIT = 2000;

describe("splitForDiscord", () => {
  // (a) Short message below limit → returned as single unchanged chunk
  it("returns a single chunk for short messages", () => {
    const text = "Hello, world!";
    const result = splitForDiscord(text);
    expect(result).toHaveLength(1);
    expect(result[0]).toBe(text);
  });

  // (b) Text exactly at the limit → single chunk
  it("returns a single chunk when text is exactly at the limit", () => {
    const text = "a".repeat(1900);
    const result = splitForDiscord(text, 1900);
    expect(result).toHaveLength(1);
    expect(result[0]).toBe(text);
  });

  // (c) Long plain text → split at line boundaries
  it("splits long plain text at line boundaries", () => {
    const line = "x".repeat(100);
    // 25 lines = 25*100 + 24 newlines = 2524 chars → must split with limit 1900
    const text = Array.from({ length: 25 }, () => line).join("\n");
    const result = splitForDiscord(text, 1900);
    expect(result.length).toBeGreaterThan(1);
    // Reassembled text must equal original
    expect(result.join("\n")).toBe(text);
  });

  // (d) Code block spanning a split → fence is closed and reopened with lang tag
  it("closes and reopens fenced code blocks across chunk boundaries", () => {
    const preamble = "Here is some code:\n";
    const fenceOpen = "```typescript\n";
    // Fill enough code lines to force a split inside the fence
    const codeLine = "const x = 1; // " + "y".repeat(80) + "\n";
    const codeLines = Array.from({ length: 20 }, () => codeLine).join("");
    const fenceClose = "```\n";
    const epilogue = "End of message.";

    const text = preamble + fenceOpen + codeLines + fenceClose + epilogue;
    const result = splitForDiscord(text, 1900);

    expect(result.length).toBeGreaterThan(1);

    // Every chunk that starts mid-fence must open with ```typescript
    // Every chunk that ends mid-fence must close with ```
    for (let i = 0; i < result.length - 1; i++) {
      const chunk = result[i];
      // Count fence toggles to determine if we end inside a fence
      const fenceToggles = (chunk.match(/^```/gm) ?? []).length;
      if (fenceToggles % 2 === 1) {
        // Odd number means we opened but didn't close — the splitter must close it
        expect(chunk.trimEnd()).toMatch(/```$/);
        // And the next chunk must reopen with the same language
        expect(result[i + 1]).toMatch(/^```typescript/);
      }
    }
  });

  // (e) No chunk exceeds 2000 characters
  it("never produces a chunk exceeding 2000 characters", () => {
    const line = "z".repeat(150);
    const text = Array.from({ length: 50 }, () => line).join("\n");
    const result = splitForDiscord(text, 1900);
    for (const chunk of result) {
      expect(chunk.length).toBeLessThanOrEqual(DISCORD_HARD_LIMIT);
    }
  });

  // (f) Empty input → no crash, returns consistent value
  it("handles empty input without crashing", () => {
    expect(() => splitForDiscord("")).not.toThrow();
    const result = splitForDiscord("");
    expect(Array.isArray(result)).toBe(true);
  });

  // Additional: fence language tag is preserved correctly
  it("preserves the fence language tag when reopening after a split", () => {
    const limit = 200;
    const header = "Intro\n";
    const fenceOpen = "```python\n";
    // Enough lines to exceed 200 chars inside the fence
    const codeLine = "print('hello')\n";
    const codeLines = Array.from({ length: 20 }, () => codeLine).join("");
    const fenceClose = "```";

    const text = header + fenceOpen + codeLines + fenceClose;
    const result = splitForDiscord(text, limit);

    // Find the first chunk that ends with ``` (closed fence)
    const closingChunk = result.find(
      (c, i) => i < result.length - 1 && c.trimEnd().endsWith("```")
    );
    expect(closingChunk).toBeDefined();

    // The chunk following a mid-fence close should start with ```python
    const idx = result.indexOf(closingChunk!);
    expect(result[idx + 1]).toMatch(/^```python/);
  });

  // Regression: fence closure "\n```" must not push a chunk beyond the 2000 hard limit
  it("does not exceed 2000 chars even when fence closure is appended at limit boundary", () => {
    const fenceOpen = "```typescript\n";
    // Line long enough that current is near 2000 when fence closure is added
    const longLine = "x".repeat(1984);
    const nextLine = "y".repeat(10);
    const fenceClose = "\n```";

    const text = fenceOpen + longLine + "\n" + nextLine + fenceClose;
    const result = splitForDiscord(text, 2000);

    for (const chunk of result) {
      expect(chunk.length).toBeLessThanOrEqual(DISCORD_HARD_LIMIT);
    }
  });
});