Version 2.1.1 (#142)

Author: cdgriffith

CHANGELOG.md

@@ -1,6 +1,11 @@
 Changelog
 =========
 
+Version 2.1.1
+-------------
+
+- Fixing #141 deep scan no longer overrides valid binary format matches (e.g. JPEG) for files without extensions (thanks to marph91)
+
 Version 2.1.0
 -------------
 

puremagic/main.py

@@ -35,7 +35,7 @@
     )
 
 __author__ = "Chris Griffith"
-__version__ = "2.1.0"
+__version__ = "2.1.1"
 __all__ = [
     "magic_file",
     "magic_string",
@@ -212,13 +212,12 @@ def perform_magic(header: bytes, footer: bytes, mime: bool, ext=None, filename=N
         raise PureValueError("Input was empty")
     infos = identify_all(header, footer, ext)
     if filename and os.path.isfile(filename) and os.getenv("PUREMAGIC_DEEPSCAN") != "0":
-        results = run_deep_scan(infos, filename, header, footer, raise_on_none=True)
-        if results:
-            if results[0].extension == "":
-                raise PureError("Could not identify file")
+        results = run_deep_scan(infos, filename, header, footer, raise_on_none=not infos)
+        if results and results[0].extension != "":
             if mime:
                 return results[0].mime_type
             return results[0].extension or ""
+        # Deep scan returned empty extension or no results — fall through to original matches
     if not infos:
         raise PureError("Could not identify file")
     info = infos[0]
@@ -556,13 +555,18 @@ def run_deep_scan(
 
     # No specific scanner matched — try the catch-all text scanner
     # Only override when existing matches are very low confidence (e.g. 2-byte BOM signatures)
-    if matches[0].confidence < 0.5:
+    # Only let the catch-all text scanner override when existing matches are
+    # generic text types (e.g. BOM-only signatures).  If the magic database
+    # already identified a specific non-text file type, trust it over a generic text guess.
+    best_mime = matches[0].mime_type or ""
+    is_generic = best_mime.startswith("text/") or best_mime == "application/octet-stream" or not best_mime
+    if matches[0].confidence < 0.5 and is_generic:
         try:
             result = catch_all_deep_scan(filename, head, foot)
         except Exception:
             pass
         else:
-            if result and result.confidence > matches[0].confidence:
+            if result and result.extension and result.confidence > matches[0].confidence:
                 return [
                     PureMagicWithConfidence(
                         confidence=result.confidence,

test/test_common_extensions.py

@@ -269,7 +269,6 @@ def test_riff_wav_mime():
     assert mime == "audio/wav"
 
 
-
 def test_cfbf_doc():
     """CFBF scanner correctly identifies Word .doc"""
     ext = puremagic.from_file(os.path.join(OFFICE_DIR, "test.doc"))

test/test_scanners.py

@@ -1,5 +1,5 @@
 import puremagic
-from test.common import OFFICE_DIR, SYSTEM_DIR, AUDIO_DIR
+from test.common import IMAGE_DIR, OFFICE_DIR, SYSTEM_DIR, AUDIO_DIR
 from puremagic.scanners import python_scanner, json_scanner, sndhdr_scanner
 
 sample_text = b"""Lorem ipsum dolor sit amet, consectetur adipiscing elit,{ending}
@@ -104,6 +104,25 @@ def test_eml_scanner():
     assert results[0].confidence == 1.0
 
 
+def test_jpg_without_extension():
+    # GH #141: JPEG file without extension should still be identified as image/jpeg
+    import struct
+
+    data = b"\xff\xd8\xff\xe0"
+    data += struct.pack(">H", 16)
+    data += b"JFIF\x00\x01\x01\x00"
+    data += struct.pack(">HH", 1, 1)
+    data += b"\x00\x00\xff\xd9"
+
+    no_ext_file = IMAGE_DIR / "test_jpeg_no_ext"
+    no_ext_file.write_bytes(data)
+    try:
+        result = puremagic.from_file(no_ext_file, mime=True)
+        assert result == "image/jpeg", f"Expected image/jpeg, got {result}"
+    finally:
+        no_ext_file.unlink()
+
+
 def test_sndhdr_scanner():
     # Test the sndhdr scanner with sndr file
     sndr_file = AUDIO_DIR / "test.sndr"