Add optional Google Workspace shortcut export

tests/test_detect.py

@@ -258,6 +258,40 @@ def test_classify_video_extensions():
     assert classify_file(Path("audio.m4a")) == FileType.VIDEO
 
 
+def test_classify_google_workspace_shortcuts():
+    assert classify_file(Path("notes.gdoc")) == FileType.DOCUMENT
+    assert classify_file(Path("budget.gsheet")) == FileType.DOCUMENT
+    assert classify_file(Path("deck.gslides")) == FileType.DOCUMENT
+
+
+def test_detect_skips_google_workspace_shortcuts_by_default(tmp_path):
+    (tmp_path / "notes.gdoc").write_text('{"doc_id":"doc-1"}', encoding="utf-8")
+
+    result = detect(tmp_path)
+
+    assert not result["files"]["document"]
+    assert any("Google Workspace shortcut skipped" in item for item in result["skipped_sensitive"])
+
+
+def test_detect_converts_google_workspace_shortcuts_when_enabled(tmp_path, monkeypatch):
+    shortcut = tmp_path / "notes.gdoc"
+    shortcut.write_text('{"doc_id":"doc-1"}', encoding="utf-8")
+
+    def fake_convert(path, out_dir, *, xlsx_to_markdown=None):
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out = out_dir / "notes_converted.md"
+        out.write_text("# Notes\n\nA converted Google Doc.", encoding="utf-8")
+        return out
+
+    monkeypatch.setattr("graphify.detect.convert_google_workspace_file", fake_convert)
+
+    result = detect(tmp_path, google_workspace=True)
+
+    assert len(result["files"]["document"]) == 1
+    assert result["files"]["document"][0].endswith("notes_converted.md")
+    assert result["total_words"] > 0
+
+
 def test_detect_includes_video_key(tmp_path):
     """detect() result always includes a 'video' key even with no video files."""
     (tmp_path / "main.py").write_text("x = 1")

tests/test_google_workspace.py

@@ -0,0 +1,81 @@
+from pathlib import Path
+
+import graphify.google_workspace as gw
+
+
+def test_read_google_shortcut_doc_id(tmp_path):
+    shortcut = tmp_path / "Planning.gdoc"
+    shortcut.write_text(
+        '{"url":"https://docs.google.com/document/d/doc-123/edit","doc_id":"doc-123","email":"me@example.com"}',
+        encoding="utf-8",
+    )
+
+    metadata = gw.read_google_shortcut(shortcut)
+
+    assert metadata["file_id"] == "doc-123"
+    assert metadata["account"] == "me@example.com"
+
+
+def test_read_google_shortcut_extracts_id_from_url(tmp_path):
+    shortcut = tmp_path / "Budget.gsheet"
+    shortcut.write_text(
+        '{"url":"https://docs.google.com/spreadsheets/d/sheet-456/edit?resourcekey=key-1"}',
+        encoding="utf-8",
+    )
+
+    metadata = gw.read_google_shortcut(shortcut)
+
+    assert metadata["file_id"] == "sheet-456"
+    assert metadata["resource_key"] == "key-1"
+
+
+def test_convert_gdoc_to_markdown_sidecar(tmp_path, monkeypatch):
+    shortcut = tmp_path / "Planning.gdoc"
+    shortcut.write_text(
+        '{"url":"https://docs.google.com/document/d/doc-123/edit","doc_id":"doc-123"}',
+        encoding="utf-8",
+    )
+
+    def fake_export(file_id, mime_type, output, resource_key=None):
+        assert file_id == "doc-123"
+        assert mime_type == "text/markdown"
+        output.write_text("# Planning\n\nExported doc text.", encoding="utf-8")
+
+    monkeypatch.setattr(gw, "_run_gws_export", fake_export)
+
+    out = gw.convert_google_workspace_file(shortcut, tmp_path / "converted")
+
+    assert out is not None
+    assert out.suffix == ".md"
+    content = out.read_text(encoding="utf-8")
+    assert 'source_type: "google_workspace"' in content
+    assert "# Planning" in content
+
+
+def test_convert_gsheet_uses_xlsx_markdown_callback(tmp_path, monkeypatch):
+    shortcut = tmp_path / "Budget.gsheet"
+    shortcut.write_text('{"doc_id":"sheet-456"}', encoding="utf-8")
+
+    def fake_export(file_id, mime_type, output, resource_key=None):
+        assert file_id == "sheet-456"
+        assert mime_type == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+        output.write_bytes(b"xlsx")
+
+    monkeypatch.setattr(gw, "_run_gws_export", fake_export)
+
+    out = gw.convert_google_workspace_file(
+        shortcut,
+        tmp_path / "converted",
+        xlsx_to_markdown=lambda path: "## Sheet: Main\n\n| A |\n| --- |\n| 1 |",
+    )
+
+    assert out is not None
+    assert "## Sheet: Main" in out.read_text(encoding="utf-8")
+
+
+def test_google_workspace_enabled_env(monkeypatch):
+    monkeypatch.setenv("GRAPHIFY_GOOGLE_WORKSPACE", "yes")
+    assert gw.google_workspace_enabled()
+
+    monkeypatch.setenv("GRAPHIFY_GOOGLE_WORKSPACE", "0")
+    assert not gw.google_workspace_enabled()