Fix issue where empty files were literally "[Empty file]"

Author: holtskinner

src/gitingest/extract.py

@@ -6,7 +6,7 @@
 import re
 from pathlib import Path
 
-from gitingest.schemas.filesystem import SEPARATOR
+from gitingest.schemas.filesystem import EMPTY_FILE, SEPARATOR
 from gitingest.utils.logging_config import get_logger
 
 logger = get_logger(__name__)
@@ -103,6 +103,8 @@ def extract(digest_path: str | Path, output_dir: str | Path = ".") -> None:
             target_file_path.parent.mkdir(parents=True, exist_ok=True)
 
             try:
+                if file_content == EMPTY_FILE:
+                    file_content = ""
                 with target_file_path.open("w", encoding="utf-8") as f:
                     f.write(file_content)
                 logger.debug(f"Extracted: {target_file_path}")

src/gitingest/schemas/filesystem.py

@@ -15,7 +15,7 @@
     from pathlib import Path
 
 SEPARATOR = "=" * 48  # Tiktoken, the tokenizer openai uses, counts 2 tokens if we have more than 48
-
+EMPTY_FILE = "[Empty file]"
 
 class FileSystemNodeType(Enum):
     """Enum representing the type of a file system node (directory or file)."""
@@ -140,7 +140,7 @@ def content(self) -> str:  # pylint: disable=too-many-return-statements
             return "Error reading file"
 
         if chunk == b"":
-            return "[Empty file]"
+            return EMPTY_FILE
 
         if not _decodes(chunk, "utf-8"):
             return "[Binary file]"

tests/test_extract.py

@@ -70,7 +70,7 @@ def test_extract_empty_file(tmp_path: Path) -> None:
 
     extracted_file = output_dir / "empty.txt"
     assert extracted_file.exists()
-    assert extracted_file.read_text(encoding="utf-8") == "[Empty file]"
+    assert extracted_file.read_text(encoding="utf-8") == ""
 
 
 def test_extract_binary_file_placeholder(tmp_path: Path) -> None: