Show removed SQL comments in Git diffs (#6569)

Hunk content resembling patch headers was discarded before line parsing. Metadata is already ignored before the first hunk, so preserve every added and removed line once a hunk starts.

Closes #6564

Co-authored-by: Yu Han <73627805+Mar-garet@users.noreply.github.com>

packages/server/src/server/daemon-e2e/checkout-diff-content.e2e.test.ts

@@ -0,0 +1,87 @@
+import { execFileSync } from "node:child_process";
+import { mkdtempSync, rmSync, writeFileSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { expect, test } from "vitest";
+import { createDaemonTestContext } from "../test-utils/index.js";
+
+function git(cwd: string, ...args: string[]): string {
+  return execFileSync("git", args, { cwd, encoding: "utf8" });
+}
+
+test.each(["uncommitted", "base", "commit"] as const)(
+  "%s diff preserves removed SQL comments and their deletion counts",
+  async (mode) => {
+    const ctx = await createDaemonTestContext();
+    const cwd = mkdtempSync(join(tmpdir(), "checkout-diff-content-"));
+
+    try {
+      git(cwd, "init", "-b", "main");
+      git(cwd, "config", "user.name", "Test");
+      git(cwd, "config", "user.email", "test@example.com");
+      writeFileSync(join(cwd, "query.sql"), "-- old comment\nSELECT 1;\n");
+      writeFileSync(join(cwd, "control.sql"), "SELECT 1;\n");
+      git(cwd, "add", ".");
+      git(cwd, "-c", "commit.gpgsign=false", "commit", "-m", "initial");
+      writeFileSync(join(cwd, "query.sql"), "-- new comment\nSELECT 1;\n");
+      writeFileSync(join(cwd, "control.sql"), "SELECT 2;\n");
+
+      expect(git(cwd, "diff", "--numstat", "HEAD", "--", "query.sql").trim()).toBe(
+        "1\t1\tquery.sql",
+      );
+
+      if (mode !== "uncommitted") {
+        git(cwd, "checkout", "-b", "feature");
+        git(cwd, "add", ".");
+        git(cwd, "-c", "commit.gpgsign=false", "commit", "-m", "edit");
+      }
+
+      const files =
+        mode === "commit"
+          ? [
+              (
+                await ctx.client.getCommitFileDiff(
+                  cwd,
+                  git(cwd, "rev-parse", "HEAD").trim(),
+                  "query.sql",
+                )
+              ).file,
+              (
+                await ctx.client.getCommitFileDiff(
+                  cwd,
+                  git(cwd, "rev-parse", "HEAD").trim(),
+                  "control.sql",
+                )
+              ).file,
+            ]
+          : (await ctx.client.getCheckoutDiff(cwd, { mode, baseRef: "main" })).files;
+
+      const control = files.find((file) => file?.path === "control.sql");
+      expect(control).toMatchObject({ additions: 1, deletions: 1 });
+      expect(control?.hunks.flatMap((hunk) => hunk.lines)).toEqual(
+        expect.arrayContaining([
+          expect.objectContaining({ type: "remove", content: "SELECT 1;" }),
+          expect.objectContaining({ type: "add", content: "SELECT 2;" }),
+        ]),
+      );
+
+      const query = files.find((file) => file?.path === "query.sql");
+      expect
+        .soft(query)
+        .toMatchObject({ additions: 1, deletions: 1, isNew: false, isDeleted: false });
+      expect
+        .soft(query?.hunks.flatMap((hunk) => hunk.lines))
+        .toEqual(
+          expect.arrayContaining([
+            expect.objectContaining({ type: "remove", content: "-- old comment" }),
+            expect.objectContaining({ type: "add", content: "-- new comment" }),
+            expect.objectContaining({ type: "context", content: "SELECT 1;" }),
+          ]),
+        );
+    } finally {
+      await ctx.cleanup();
+      rmSync(cwd, { recursive: true, force: true });
+    }
+  },
+  60_000,
+);

packages/server/src/server/utils/diff-highlighter.test.ts

@@ -190,6 +190,34 @@ index 1234567..abcdefg 100644
 `;
 
 describe("parseDiff", () => {
+  it.each(["-- comment", "++ marker"])(
+    "preserves hunk content resembling patch headers: %s",
+    (content) => {
+      const files = parseDiff(`diff --git a/example.txt b/example.txt
+index 1234567..abcdefg 100644
+--- a/example.txt
++++ b/example.txt
+@@ -1 +1 @@
+-${content} before
++${content} after
+`);
+
+      expect(files).toHaveLength(1);
+      expect(files[0]).toMatchObject({
+        path: "example.txt",
+        isNew: false,
+        isDeleted: false,
+        additions: 1,
+        deletions: 1,
+      });
+      expect(files[0].hunks[0].lines).toEqual([
+        { type: "header", content: "@@ -1 +1 @@" },
+        { type: "remove", content: `${content} before` },
+        { type: "add", content: `${content} after` },
+      ]);
+    },
+  );
+
   it("parses a simple diff with one hunk", () => {
     const files = parseDiff(SIMPLE_DIFF);
 

packages/server/src/server/utils/diff-highlighter.ts

@@ -79,16 +79,6 @@ function extractPathFromDiffHeader(lines: string[]): string {
   return "unknown";
 }
 
-function isMetadataLine(line: string): boolean {
-  return (
-    line.startsWith("index ") ||
-    line.startsWith("--- ") ||
-    line.startsWith("+++ ") ||
-    line.startsWith("new file mode") ||
-    line.startsWith("deleted file mode")
-  );
-}
-
 function parseHunkHeader(line: string): DiffHunk | null {
   const hunkMatch = line.match(/^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@/);
   if (!hunkMatch) return null;
@@ -116,8 +106,6 @@ function parseSectionBody(lines: string[]): ParsedSectionBody {
   for (let i = 1; i < lines.length; i++) {
     const line = lines[i];
 
-    if (isMetadataLine(line)) continue;
-
     const newHunk = parseHunkHeader(line);
     if (newHunk) {
       if (currentHunk) hunks.push(currentHunk);