Extract every patch segment and decode byte lines

src/extractcode/patch.py

@@ -31,6 +31,15 @@
 LOG = logging.getLogger(__name__)
 
 
+def _as_text(line):
+    if not isinstance(line, bytes):
+        return line
+    try:
+        return line.decode("utf-8")
+    except UnicodeDecodeError:
+        return line.decode("latin-1")
+
+
 def extract(location, target_dir):
     """
     Extract each patch of a patch file at `location` as files in a target_dir
@@ -74,7 +83,7 @@ def extract(location, target_dir):
         with open(subfile_path, "w") as subfile:
             subfile.write("\n".join(text))
 
-        return []
+    return []
 
 
 def is_patch(location, include_extracted=False):
@@ -140,5 +149,5 @@ def patch_info(location):
     for ptch in patchset.items:
         src = fileutils.as_posixpath(ptch.source.strip())
         tgt = fileutils.as_posixpath(ptch.target.strip())
-        text = [l.strip() for l in patch_text(ptch) if l]
-        yield src, tgt, text
+        lines = [_as_text(line).strip() for line in patch_text(ptch) if line]
+        yield src, tgt, lines

tests/test_patch.py

@@ -1615,3 +1615,23 @@ def test_patch_info_patch_patches_xvidcap_xorg_patch(self):
         test_file = self.get_test_loc("patch/patches/xvidcap-xorg.patch")
         expected_file = self.get_test_loc("patch/patches/xvidcap-xorg.patch.expected")
         check_patch(test_file, expected_file)
+
+
+def test_extract_writes_a_subfile_for_each_patch_segment(tmp_path):
+    test_file = os.path.join(
+        os.path.dirname(__file__), "data", "patch", "patches", "dnsmasq_2.63-1.diff"
+    )
+    warnings = patch.extract(test_file, str(tmp_path))
+    extracted = [p for p in tmp_path.rglob("*") if p.is_file()]
+    assert warnings == []
+    assert len(extracted) == 19
+
+
+def test_extract_handles_bytes_patch_lines(tmp_path):
+    test_file = os.path.join(
+        os.path.dirname(__file__), "data", "patch", "patches", "libmediainfo-0.7.43.diff"
+    )
+    warnings = patch.extract(test_file, str(tmp_path))
+    extracted = [p for p in tmp_path.rglob("*") if p.is_file()]
+    assert warnings == []
+    assert len(extracted) == 1