From 362083b4fb85789e18a912ea0322b7f55365826e Mon Sep 17 00:00:00 2001
From: Zhe Yu <zcabzyu@ucl.ac.uk>
Date: Wed, 14 May 2025 09:30:41 +0800
Subject: [PATCH] fix(cli): make sure arg parsers use correct defaults.

---
 src/vectorcode/cli_utils.py | 25 +++++++++++++++++++------
 tests/test_cli_utils.py     |  6 ++++++
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/src/vectorcode/cli_utils.py b/src/vectorcode/cli_utils.py
index b1198890..651016a2 100644
--- a/src/vectorcode/cli_utils.py
+++ b/src/vectorcode/cli_utils.py
@@ -172,22 +172,27 @@ async def merge_from(self, other: "Config") -> "Config":
 
 
 def get_cli_parser():
+    __default_config = Config()
     shared_parser = argparse.ArgumentParser(add_help=False)
     chunking_parser = argparse.ArgumentParser(add_help=False)
     chunking_parser.add_argument(
-        "--overlap", "-o", type=float, help="Ratio of overlaps between chunks."
+        "--overlap",
+        "-o",
+        type=float,
+        default=__default_config.overlap_ratio,
+        help="Ratio of overlaps between chunks.",
     )
     chunking_parser.add_argument(
         "-c",
         "--chunk_size",
         type=int,
-        default=-1,
+        default=__default_config.chunk_size,
         help="Size of chunks (-1 for no chunking).",
     )
     chunking_parser.add_argument(
         "--encoding",
         type=str,
-        default="utf8",
+        default=__default_config.encoding,
         help="Encoding used by the files. See https://docs.python.org/3/library/codecs.html#standard-encodings for supported encodings. Use `_auto` for automatic encoding detection.",
     )
     shared_parser.add_argument(
@@ -262,10 +267,18 @@ def get_cli_parser():
     )
     query_parser.add_argument("query", nargs="+", help="Query keywords.")
     query_parser.add_argument(
-        "--multiplier", "-m", type=int, default=-1, help="Query multiplier."
+        "--multiplier",
+        "-m",
+        type=int,
+        default=__default_config.query_multiplier,
+        help="Query multiplier.",
     )
     query_parser.add_argument(
-        "-n", "--number", type=int, default=1, help="Number of results to retrieve."
+        "-n",
+        "--number",
+        type=int,
+        default=__default_config.n_result,
+        help="Number of results to retrieve.",
     )
     query_parser.add_argument(
         "--exclude", nargs="*", help="Files to exclude from query results."
@@ -281,7 +294,7 @@ def get_cli_parser():
         choices=list(i.value for i in QueryInclude),
         nargs="+",
         help="What to include in the final output.",
-        default=["path", "document"],
+        default=__default_config.include,
     )
 
     subparsers.add_parser("drop", parents=[shared_parser], help="Remove a collection.")
diff --git a/tests/test_cli_utils.py b/tests/test_cli_utils.py
index 940f87a9..d00057ed 100644
--- a/tests/test_cli_utils.py
+++ b/tests/test_cli_utils.py
@@ -489,6 +489,12 @@ async def test_parse_cli_args_chunks():
         assert config.overlap_ratio == 0.5
         assert config.chunk_size == 100
 
+    with patch("sys.argv", ["vectorcode", "chunks", "file.py"]):
+        config = await parse_cli_args()
+        assert config.action == CliAction.chunks
+        assert config.overlap_ratio == Config().overlap_ratio
+        assert config.chunk_size == Config().chunk_size
+
 
 @pytest.mark.asyncio
 async def test_config_import_from_hnsw():