From 5c6a95ddfe4168094c4d7eeaac9206242048781f Mon Sep 17 00:00:00 2001
From: Luke Craig <lacraig3@gmail.com>
Date: Mon, 11 Aug 2025 14:19:24 -0400
Subject: [PATCH] Add --skip-binary option to CLI and update save_repo_to_text
 to handle binary files

---
 repo_to_text/cli/cli.py   |  8 +++++++-
 repo_to_text/core/core.py | 17 ++++++++++++-----
 tests/test_cli.py         |  3 ++-
 3 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/repo_to_text/cli/cli.py b/repo_to_text/cli/cli.py
index 911dd1a..6276f43 100644
--- a/repo_to_text/cli/cli.py
+++ b/repo_to_text/cli/cli.py
@@ -74,6 +74,11 @@ def parse_args() -> argparse.Namespace:
         help="List of files or directories to ignore in both tree and content sections. "
         "Supports wildcards (e.g., '*')."
     )
+    parser.add_argument(
+        '--skip-binary',
+        action='store_true',
+        help='Skip binary files in the output.'
+    )
     return parser.parse_args()
 
 def main() -> NoReturn:
@@ -95,7 +100,8 @@ def main() -> NoReturn:
                 path=args.input_dir,
                 output_dir=args.output_dir,
                 to_stdout=args.stdout,
-                cli_ignore_patterns=args.ignore_patterns
+                cli_ignore_patterns=args.ignore_patterns,
+                skip_binary=args.skip_binary
             )
 
         logging.debug('repo-to-text script finished')
diff --git a/repo_to_text/core/core.py b/repo_to_text/core/core.py
index ccc9460..b5255ea 100644
--- a/repo_to_text/core/core.py
+++ b/repo_to_text/core/core.py
@@ -261,7 +261,8 @@ def save_repo_to_text(
         path: str = '.',
         output_dir: Optional[str] = None,
         to_stdout: bool = False,
-        cli_ignore_patterns: Optional[List[str]] = None
+        cli_ignore_patterns: Optional[List[str]] = None,
+        skip_binary: bool = False
     ) -> str:
     """Save repository structure and contents to a text file or multiple files."""
     # pylint: disable=too-many-locals
@@ -285,7 +286,8 @@ def save_repo_to_text(
         gitignore_spec,
         content_ignore_spec,
         tree_and_content_ignore_spec,
-        maximum_word_count_per_file
+        maximum_word_count_per_file,
+        skip_binary
     )
 
     if to_stdout:
@@ -352,11 +354,12 @@ def save_repo_to_text(
         return output_filepaths[0]
     return ""
 
-def _read_file_content(file_path: str) -> str:
+def _read_file_content(file_path: str, skip_binary: bool = False) -> str:
     """Read file content, handling binary files and broken symlinks.
     
     Args:
         file_path: Path to the file to read
+        skip_binary: Whether to skip binary files
         
     Returns:
         str: File content or appropriate message for special cases
@@ -365,6 +368,9 @@ def _read_file_content(file_path: str) -> str:
         with open(file_path, 'r', encoding='utf-8') as f:
             return f.read()
     except UnicodeDecodeError:
+        if skip_binary:
+            logging.debug('Skipping binary file: %s', file_path)
+            return "binary content skipped"
         logging.debug('Handling binary file contents: %s', file_path)
         with open(file_path, 'rb') as f_bin:
             binary_content: bytes = f_bin.read()
@@ -386,7 +392,8 @@ def generate_output_content(
         gitignore_spec: Optional[PathSpec],
         content_ignore_spec: Optional[PathSpec],
         tree_and_content_ignore_spec: Optional[PathSpec],
-        maximum_word_count_per_file: Optional[int] = None
+        maximum_word_count_per_file: Optional[int] = None,
+        skip_binary: bool = False
     ) -> List[str]:
     """Generate the output content for the repository, potentially split into segments."""
     # pylint: disable=too-many-arguments
@@ -453,7 +460,7 @@ def generate_output_content(
             cleaned_relative_path = relative_path.replace('./', '', 1)
             
             _add_chunk_to_output(f'\n<content full_path="{cleaned_relative_path}">\n')
-            file_content = _read_file_content(file_path)
+            file_content = _read_file_content(file_path, skip_binary)
             _add_chunk_to_output(file_content)
             _add_chunk_to_output('\n</content>\n')
 
diff --git a/tests/test_cli.py b/tests/test_cli.py
index 7382ce7..924ceb4 100644
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -84,7 +84,8 @@ def test_main_normal_execution(mock_save_repo: MagicMock) -> None:
             path='.',
             output_dir=None,
             to_stdout=True,
-            cli_ignore_patterns=None
+            cli_ignore_patterns=None,
+            skip_binary=False
         )
 
 @patch('repo_to_text.cli.cli.create_default_settings_file')