char-slop/berry

git clone https://git.t4t.associates/char-slop/berry

Charlotte Somcode: support exact text replacements alongside hashline edits0ad1bcc2026-10-09

import { assert, assertEquals, assertMatch, assertRejects } from "@std/assert";
import { decodeBase64 } from "@std/encoding/base64";
import { PNG, tempDir } from "@berry/util";
import { readFile, stat, symlink, writeFile } from "node:fs/promises";
import { join } from "node:path";
import { localEnvironment } from "@berry/api/environment.ts";
import { testExtension } from "@berry/api/testing.ts";
import code from "./mod.ts";

type Fixture = {
  cwd: string;
  path: string;
  call(name: string, args?: Record<string, unknown>, signal?: AbortSignal): Promise<string>;
};

function test(name: string, run: (f: Fixture) => Promise<void>, content: string = "") {
  Deno.test(name, async () => {
    await using dir = await tempDir();
    const path = join(dir.path, "file.txt");
    await writeFile(path, content);
    const extension = testExtension(code, { env: localEnvironment(dir.path) });
    await run({
      cwd: dir.path,
      path,
      call: async (name, args = {}, signal) =>
        await extension.call(name, { path: "file.txt", ...args } as never, { signal }) as string,
    });
  });
}

const anchors = (output: string) =>
  output.split("\n").filter((row) => row.includes("│")).map((row) => row.split("│")[0]);
const unanchored = (output: string) =>
  output.replace(/^([+-])[A-Za-z]{4}│/gm, "$1").split("\n").slice(2);

test("diffs carry the new lines' hashes and keep replacement text literal", async (f) => {
  const [old] = anchors(await f.call("read"));
  assertMatch(old, /^[A-Za-z]{4}$/);
  const output = await f.call("edit", { edits: [{ old: [old], new: ["ABCD│literal"] }] });
  assert(output.includes(`-${old}│original`));
  assertMatch(output, /^\+[A-Za-z]{4}│ABCD│literal$/m);
}, "original\n");

test(
  "batched diffs merge nearby context without duplicating it and separate distant hunks",
  async (f) => {
    const old = anchors(await f.call("read"));
    const output = await f.call("edit", {
      edits: [
        { old: [old[16]], new: ["last"] },
        { old: [old[2]], new: ["first", "extra"] },
        { old: [old[5]], new: [] },
      ],
    });
    assertEquals(unanchored(output), [
      "@@ original line 1 @@",
      " line 0",
      " line 1",
      "-line 2",
      "+first",
      "+extra",
      " line 3",
      " line 4",
      "-line 5",
      " line 6",
      " line 7",
      " line 8",
      "@@ original line 14 @@",
      " line 13",
      " line 14",
      " line 15",
      "-line 16",
      "+last",
      " line 17",
      " line 18",
      " line 19",
    ]);
  },
  Array.from({ length: 20 }, (_, i) => `line ${i}`).join("\n"),
);

test("references survive external line movement", async (f) => {
  const [first, target, last] = anchors(await f.call("read"));
  await writeFile(f.path, "inserted\nlast\nfirst\ntarget\n");
  assertEquals(anchors(await f.call("read")).slice(1), [last, first, target]);
  await f.call("edit", { edits: [{ old: [target], new: ["changed"] }] });
  assertEquals(await readFile(f.path, "utf8"), "inserted\nlast\nfirst\nchanged\n");
}, "first\ntarget\nlast\n");

test("duplicate lines need unique adjacent context, which is preserved", async (f) => {
  const original = "left\nsame\nright\nleft\nsame\nother\n";
  const [left, same, right, leftAgain, sameAgain] = anchors(await f.call("read"));
  assertEquals(left, leftAgain);
  assertEquals(same, sameAgain);
  await assertRejects(
    () => f.call("edit", { edits: [{ old: [same], new: ["changed"] }] }),
    Error,
    "ambiguous",
  );
  await assertRejects(
    () => f.call("edit", { edits: [{ before: [left], old: [same], new: ["changed"] }] }),
    Error,
    "ambiguous",
  );
  assertEquals(await readFile(f.path, "utf8"), original);
  await f.call("edit", {
    edits: [{ before: [left], old: [same], after: [right], new: ["changed"] }],
  });
  assertEquals(await readFile(f.path, "utf8"), "left\nchanged\nright\nleft\nsame\nother\n");
}, "left\nsame\nright\nleft\nsame\nother\n");

test(
  "batches resolve against the original, regardless of order or line count changes",
  async (f) => {
    const [one, two, three, four] = anchors(await f.call("read"));
    await f.call("edit", {
      edits: [{ old: [four], new: ["last"] }, { old: [one], new: ["first", "extra"] }],
    });
    assertEquals(await readFile(f.path, "utf8"), "first\nextra\ntwo\nthree\nlast\n");
    assertEquals(anchors(await f.call("read")).slice(2, 4), [two, three]);
  },
  "one\ntwo\nthree\nfour\n",
);

test("a stale interior line refuses the entire batch without writing", async (f) => {
  const [one, two, three, four] = anchors(await f.call("read"));
  const external = "one\nexternal\nthree\nfour\n";
  await writeFile(f.path, external);
  await assertRejects(
    () =>
      f.call("edit", {
        edits: [{ old: [four], new: ["would change"] }, {
          old: [one, two, three],
          new: ["replacement"],
        }],
      }),
    Error,
    "stale",
  );
  assertEquals(await readFile(f.path, "utf8"), external);
}, "one\ntwo\nthree\nfour\n");

test("hashes include trailing whitespace and the entire line", async (f) => {
  const long = "x".repeat(1000);
  const original = `text\ntext \n${long}a\n${long}b\n`;
  await writeFile(f.path, original);
  const hashes = anchors(await f.call("read"));
  assertEquals(new Set(hashes).size, 4);
  await writeFile(f.path, original.replace("text \n", "text  \n"));
  await assertRejects(
    () => f.call("edit", { edits: [{ old: [hashes[1]], new: ["changed"] }] }),
    Error,
    "stale",
  );
});

test("overlapping ranges and same-position insertions are refused", async (f) => {
  const original = "one\ntwo\nthree\n";
  const [one, two, three] = anchors(await f.call("read"));
  for (
    const edits of [
      [{ old: [one, two], new: [] }, { old: [two, three], new: [] }],
      [{ before: [one], old: [], new: ["x"] }, { after: [two], old: [], new: ["y"] }],
    ]
  ) {
    await assertRejects(() => f.call("edit", { edits }), Error, "Overlapping");
    assertEquals(await readFile(f.path, "utf8"), original);
  }
}, "one\ntwo\nthree\n");

test("context may overlap other edits, but is always checked against the original", async (f) => {
  const [one, two, three] = anchors(await f.call("read"));
  await f.call("edit", {
    edits: [
      { old: [one], after: [two], new: ["first"] },
      { before: [one], old: [two], after: [three], new: ["second"] },
    ],
  });
  assertEquals(await readFile(f.path, "utf8"), "first\nsecond\nthree\n");
}, "one\ntwo\nthree\n");

test("insert before/after, delete, and reuse hashes returned by edit", async (f) => {
  const [middle] = anchors(await f.call("read"));
  const diff = await f.call("edit", {
    edits: [{ old: [], after: [middle], new: ["first"] }, {
      before: [middle],
      old: [],
      new: ["last"],
    }],
  });
  assertEquals(await readFile(f.path, "utf8"), "first\nmiddle\nlast\n");
  const [first, last] = anchors(diff).map((anchor) => anchor.slice(1));
  await f.call("edit", { edits: [{ old: [first, middle, last], new: [] }] });
  assertEquals(await readFile(f.path, "utf8"), "");
}, "middle\n");

test("empty files can be seeded, but nonempty files require a selector", async (f) => {
  assertMatch(await f.call("read"), /Empty file/);
  await f.call("edit", { edits: [{ old: [], new: ["seed"] }] });
  assertEquals(await readFile(f.path, "utf8"), "seed\n");
  await assertRejects(
    () => f.call("edit", { edits: [{ old: [], new: ["unsafe"] }] }),
    Error,
    "context",
  );
});

test("BOM, mixed line endings, and missing final newline survive edits", async (f) => {
  const [one, two, three] = anchors(await f.call("read"));
  await f.call("edit", { edits: [{ old: [one], new: ["first"] }] });
  assertEquals(await readFile(f.path, "utf8"), "\uFEFFfirst\r\ntwo\nthree");
  await f.call("edit", { edits: [{ before: [three], old: [], new: ["last"] }] });
  assertEquals(await readFile(f.path, "utf8"), "\uFEFFfirst\r\ntwo\nthree\r\nlast");
  assertEquals(anchors(await f.call("read"))[1], two);
}, "\uFEFFone\r\ntwo\nthree");

test("blank replacement lines are not deletion, including at unterminated EOF", async (f) => {
  const [, two] = anchors(await f.call("read"));
  await f.call("edit", { edits: [{ old: [two], new: [""] }] });
  assertEquals(await readFile(f.path, "utf8"), "one\n\n");
}, "one\ntwo");

test("no-op edits do not write, and replacement text is literal", async (f) => {
  const [one] = anchors(await f.call("read"));
  const before = await stat(f.path);
  assertMatch(await f.call("edit", { edits: [{ old: [one], new: ["one"] }] }), /No changes/);
  assertEquals((await stat(f.path)).mtimeMs, before.mtimeMs);
  await f.call("edit", { edits: [{ old: [one], new: [`${one}│one`, "one", "one"] }] });
  assertEquals(await readFile(f.path, "utf8"), `${one}│one\none\none\n`);
}, "one\n");

test("invalid text and embedded newlines are refused without changing bytes", async (f) => {
  for (
    const content of [
      new Uint8Array([0xff, 0xfe, 65, 0]),
      new Uint8Array([97, 0, 98]),
      new Uint8Array([0xff]),
    ]
  ) {
    await writeFile(f.path, content);
    await assertRejects(() => f.call("read"), Error, "UTF-8");
    await assertRejects(
      () => f.call("edit", { edits: [{ old: [], new: ["unsafe"] }] }),
      Error,
      "UTF-8",
    );
    assertEquals(new Uint8Array(await readFile(f.path)), content);
  }
  await writeFile(f.path, "one\n");
  const [one] = anchors(await f.call("read"));
  for (const line of ["two\nthree", "two\rthree", "two\0three"]) {
    await assertRejects(() => f.call("edit", { edits: [{ old: [one], new: [line] }] }));
    assertEquals(await readFile(f.path, "utf8"), "one\n");
  }
});

test("paged reads provide the same anchors and respect line and byte limits", async (f) => {
  const page = await f.call("read");
  assertEquals(anchors(page).length, 2000);
  assertMatch(page, /offset=2001/);
  const small = await f.call("read", { offset: 2, limit: 2 });
  assertEquals(anchors(small), anchors(page).slice(1, 3));
  assertMatch(small, /offset=4/);
  assertEquals(anchors(await f.call("read", { offset: 2001 })).length, 5);
  await assertRejects(() => f.call("read", { offset: 2006 }), Error, "beyond");

  await writeFile(f.path, ("x".repeat(30_000) + "\n").repeat(3));
  const bytes = await f.call("read");
  assertEquals(anchors(bytes).length, 1);
  assertMatch(bytes, /offset=2/);
  await writeFile(f.path, "x".repeat(60_000));
  await assertRejects(() => f.call("read"), Error, "exceeds");
}, Array.from({ length: 2005 }, (_, i) => i.toString(36)).join("\n"));

test("parallel edits through symlink aliases do not overwrite each other", async (f) => {
  const alias = f.path + ".link";
  await symlink(f.path, alias);
  const [one, two] = anchors(await f.call("read"));
  await Promise.all([
    f.call("edit", { edits: [{ old: [one], new: ["first"] }] }),
    f.call("edit", { path: alias, edits: [{ old: [two], new: ["second"] }] }),
  ]);
  assertEquals(await readFile(f.path, "utf8"), "first\nsecond\n");
}, "one\ntwo\n");

test("images resolve to an image block", async (f) => {
  await writeFile(join(f.cwd, "pixel.png"), decodeBase64(PNG));
  assertEquals(await f.call("read", { path: "pixel.png" }), {
    type: "image",
    data: PNG,
    mimeType: "image/png",
  } as unknown as string);
});

test("aborted calls do not write", async (f) => {
  const [one] = anchors(await f.call("read"));
  await assertRejects(() =>
    f.call("edit", { edits: [{ old: [one], new: ["changed"] }] }, AbortSignal.abort())
  );
  assertEquals(await readFile(f.path, "utf8"), "one\n");
}, "one\n");

test("text edits replace exact substrings and return reusable line hashes", async (f) => {
  const output = await f.call("edit", {
    edits: [{ oldText: "const x = 1;\nconst y", newText: 'const x = "$&";\nconst z' }],
  });
  assertEquals(await readFile(f.path, "utf8"), 'const x = "$&";\nconst z = 2;\n');
  const added = anchors(output).filter((anchor) => anchor.startsWith("+")).map((a) => a.slice(1));
  await f.call("edit", { edits: [{ old: [added[1]], new: ["done"] }] });
  assertEquals(await readFile(f.path, "utf8"), 'const x = "$&";\ndone\n');
}, "const x = 1;\nconst y = 2;\n");

test("text and hash edits share original coordinates even within one line", async (f) => {
  const [heading] = anchors(await f.call("read"));
  await f.call("edit", {
    edits: [
      { oldText: "right", newText: "R" },
      { old: [heading], new: ["new heading", "extra"] },
      { oldText: "left", newText: "longer left" },
    ],
  });
  assertEquals(await readFile(f.path, "utf8"), "new heading\nextra\nlonger left + R\n");
}, "heading\nleft + right\n");

test("ambiguous text includes overlapping occurrences and can be disambiguated", async (f) => {
  await assertRejects(
    () => f.call("edit", { edits: [{ oldText: "aa", newText: "x" }] }),
    Error,
    "ambiguous",
  );
  assertEquals(await readFile(f.path, "utf8"), "aaa\n");
  await f.call("edit", { edits: [{ oldText: "aa\n", newText: "x\n" }] });
  assertEquals(await readFile(f.path, "utf8"), "ax\n");
}, "aaa\n");

test(
  "missing text refuses the entire batch rather than matching earlier replacements",
  async (f) => {
    await assertRejects(
      () =>
        f.call("edit", {
          edits: [{ oldText: "one", newText: "two" }, { oldText: "two", newText: "three" }],
        }),
      Error,
      "not found",
    );
    assertEquals(await readFile(f.path, "utf8"), "one\n");
  },
  "one\n",
);

test("overlap between text and hash ranges refuses the entire batch", async (f) => {
  const [line] = anchors(await f.call("read"));
  await assertRejects(
    () =>
      f.call("edit", {
        edits: [{ old: [line], new: ["whole"] }, { oldText: "part", newText: "piece" }],
      }),
    Error,
    "Overlapping",
  );
  assertEquals(await readFile(f.path, "utf8"), "a part of a line\n");
}, "a part of a line\n");

test("empty oldText seeds only empty files and each entry must choose one form", async (f) => {
  await f.call("edit", { edits: [{ oldText: "", newText: "seed" }] });
  assertEquals(await readFile(f.path, "utf8"), "seed");
  await assertRejects(() => f.call("edit", { edits: [{ oldText: "", newText: "unsafe" }] }));
  await assertRejects(() =>
    f.call("edit", { edits: [{ oldText: "seed", newText: "text", old: [], new: ["hash"] }] })
  );
  assertEquals(await readFile(f.path, "utf8"), "seed");
});

test(
  "text edits preserve BOM and literal line endings, including deleting through EOF",
  async (f) => {
    const before = await stat(f.path);
    assertMatch(
      await f.call("edit", { edits: [{ oldText: "one", newText: "one" }] }),
      /No changes/,
    );
    assertEquals((await stat(f.path)).mtimeMs, before.mtimeMs);
    await f.call("edit", {
      edits: [{ oldText: "one\r\ntwo\n", newText: "first\nsecond\r\n" }],
    });
    assertEquals(await readFile(f.path, "utf8"), "\uFEFFfirst\nsecond\r\nthree");
    await f.call("edit", { edits: [{ oldText: "\r\nthree", newText: "" }] });
    assertEquals(await readFile(f.path, "utf8"), "\uFEFFfirst\nsecond");
  },
  "\uFEFFone\r\ntwo\nthree",
);

test("hash replacement can keep an unterminated line and add a following line", async (f) => {
  const [line] = anchors(await f.call("read"));
  await f.call("edit", { edits: [{ old: [line], new: ["one", "two"] }] });
  assertEquals(await readFile(f.path, "utf8"), "one\ntwo");
}, "one");