fix(extract): persist --no-gitignore instead of clobbering it (follow-up to #1979)

The PR always wrote gitignore=not no_gitignore into .graphify_build.json, so a
flag-less `graphify extract` after `--no-gitignore` reset it to True and the
git-ignored code silently disappeared again — the exact #1971 complaint. Write
False only when the flag is set (None = leave as-is, mirroring #1886 excludes),
and honor the persisted value for the run when the flag is absent.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

graphify/cli.py

@@ -2501,11 +2501,22 @@ def dispatch_command(cmd: str) -> None:
         graphify_out.mkdir(parents=True, exist_ok=True)
         # Persist corpus-shaping options so later update/watch/hook rebuilds
         # use the same file set as the initial extraction (#1886).
-        from graphify.watch import _write_build_config as _write_build_cfg
+        from graphify.watch import (
+            _write_build_config as _write_build_cfg,
+            _read_build_gitignore as _read_build_gi,
+        )
+        # #1971 persistence: an explicit --no-gitignore persists False; a later
+        # flag-less `graphify extract` must NOT clobber it back to True, which
+        # would make the git-ignored code silently disappear again (the exact
+        # complaint #1971 is about). Honor the persisted value for THIS run when
+        # the flag is absent (read before the write below), and write False only
+        # when the flag is set — None leaves the setting as-is, mirroring how
+        # #1886 persists --exclude.
+        _effective_gitignore = False if no_gitignore else _read_build_gi(graphify_out)
         _write_build_cfg(
             graphify_out,
             excludes=cli_excludes or None,
-            gitignore=not no_gitignore,
+            gitignore=False if no_gitignore else None,
         )
 
         stages = _StageTimer(cli_timing)
@@ -2555,7 +2566,7 @@ def dispatch_command(cmd: str) -> None:
                 manifest_path=str(manifest_path),
                 google_workspace=google_workspace or None,
                 extra_excludes=cli_excludes or None,
-                gitignore=not no_gitignore,
+                gitignore=_effective_gitignore,
             )
             files_by_type = detection.get("files", {})
             new_by_type = detection.get("new_files", {})
@@ -2583,7 +2594,7 @@ def dispatch_command(cmd: str) -> None:
                 google_workspace=google_workspace or None,
                 extra_excludes=cli_excludes or None,
                 cache_root=out_root,
-                gitignore=not no_gitignore,
+                gitignore=_effective_gitignore,
             )
             files_by_type = detection.get("files", {})
             code_files = [Path(p) for p in files_by_type.get("code", [])]

tests/test_extract_code_only_cli.py

@@ -80,3 +80,30 @@ def test_no_gitignore_indexes_vcs_ignored_code_but_keeps_graphifyignore(tmp_path
     assert any(source.endswith("proj/deep/generated/Gen.cs") for source in sources)
     assert any(source.endswith("local/Local.cs") for source in sources)
     assert not any(source.endswith("proj/hidden/Hidden.cs") for source in sources)
+
+
+def test_no_gitignore_setting_persists_across_flagless_extract(tmp_path):
+    """#1971 persistence: once --no-gitignore is set, a later flag-less
+    `graphify extract` must NOT clobber it back to honoring .gitignore (which
+    would make the git-ignored code silently disappear again)."""
+    repo = tmp_path / "repo"
+    gen = repo / "generated"
+    gen.mkdir(parents=True)
+    (repo / ".gitignore").write_text("generated/\n")
+    (repo / "app.py").write_text("def hello():\n    return 1\n")
+    (gen / "Gen.py").write_text("def gen():\n    return 2\n")
+
+    def _sources():
+        g = json.loads((repo / "graphify-out" / "graph.json").read_text())
+        return {Path(str(n.get("source_file", ""))).as_posix() for n in g["nodes"]}
+
+    r1 = _run(repo, "--no-gitignore", "--code-only", "--no-cluster")
+    assert r1.returncode == 0, r1.stderr
+    assert any(s.endswith("generated/Gen.py") for s in _sources())
+
+    # A plain flag-less re-extract must keep the git-ignored file (setting persisted).
+    r2 = _run(repo, "--code-only", "--no-cluster")
+    assert r2.returncode == 0, r2.stderr
+    assert any(s.endswith("generated/Gen.py") for s in _sources()), (
+        "flag-less re-extract clobbered the persisted --no-gitignore setting (#1971)"
+    )