fixed issue with file listing

Author: Andrei Neagu

packages/service-library/src/servicelib/archiving_utils/_errors.py

@@ -16,6 +16,6 @@ class CouldNotRunCommandError(ArchiveError):
 class TableHeaderNotFoundError(ArchiveError):
     msg_template = (
         "Excepted to detect a table header since files were detected "
-        "lines_with_file_name='{lines_with_file_name}'. "
+        "file_lines='{file_lines}'. "
         "Command output:\n{command_output}"
     )

packages/service-library/src/servicelib/archiving_utils/_interface_7zip.py

@@ -38,7 +38,6 @@
 _ALL_DONE_RE: Final[re.Pattern] = re.compile(r"Everything is Ok", re.IGNORECASE)
 
 _TOKEN_TABLE_HEADER_START: Final[str] = "------------------- "
-_TOKEN_FILE_PERMISSIONS: Final[str] = " ..... "
 
 _7ZIP_EXECUTABLE: Final[Path] = Path("/usr/bin/7z")
 
@@ -250,24 +249,35 @@ async def _compressed_bytes(byte_progress: NonNegativeInt) -> None:
             await shutil_move(f"{destination}.zip", destination)
 
 
+def _is_folder(line: str) -> bool:
+    folder_attribute = line[20]
+    return folder_attribute == "D"
+
+
 def _extract_file_names_from_archive(command_output: str) -> set[str]:
     file_name_start: NonNegativeInt | None = None
 
+    entries_lines: list[str] = []
+    can_add_to_entries: bool = False
+
+    # extract all lines containing files or folders
     for line in command_output.splitlines():
         if line.startswith(_TOKEN_TABLE_HEADER_START):
             file_name_start = line.rfind(" ") + 1
-            break
+            can_add_to_entries = not can_add_to_entries
+            continue
+
+        if can_add_to_entries:
+            entries_lines.append(line)
 
-    lines_with_file_name: list[str] = [
-        line for line in command_output.splitlines() if _TOKEN_FILE_PERMISSIONS in line
-    ]
+    file_lines: list[str] = [line for line in entries_lines if not _is_folder(line)]
 
-    if lines_with_file_name and file_name_start is None:
+    if file_lines and file_name_start is None:
         raise TableHeaderNotFoundError(
-            lines_with_file_name=lines_with_file_name, command_output=command_output
+            file_lines=file_lines, command_output=command_output
         )
 
-    return {line[file_name_start:] for line in lines_with_file_name}
+    return {line[file_name_start:] for line in file_lines}
 
 
 async def unarchive_dir(

packages/service-library/tests/archiving_utils/test_archiving__interface_7zip.py

@@ -66,7 +66,21 @@ async def test_archive_unarchive(
     mixed_file_types: Path, archive_path: Path, unpacked_archive: Path, compress: bool
 ):
     await archive_dir(mixed_file_types, archive_path, compress=compress)
+    await unarchive_dir(archive_path, unpacked_archive)
+
 
+@pytest.fixture
+def empty_folder(tmp_path: Path) -> Path:
+    path = tmp_path / "empty_folder"
+    path.mkdir()
+    return path
+
+
+@pytest.mark.parametrize("compress", [True, False])
+async def test_archive_unarchive_empty_folder(
+    empty_folder: Path, archive_path: Path, unpacked_archive: Path, compress: bool
+):
+    await archive_dir(empty_folder, archive_path, compress=compress)
     await unarchive_dir(archive_path, unpacked_archive)
 
 
@@ -75,6 +89,8 @@ async def test_archive_unarchive(
     [
         ("list_edge_case.txt", 3),
         ("list_stdout.txt", 674),
+        ("list_broken_format.txt", 22),
+        ("list_empty_archive.txt", 0),
     ],
 )
 def test__extract_file_names_from_archive(

packages/service-library/tests/data/archive_utils/list_broken_format.txt

@@ -0,0 +1,35 @@
+7-Zip (z) 24.09 (x64) : Copyright (c) 1999-2024 Igor Pavlov : 2024-11-28
+ 64-bit locale=C.UTF-8 Threads:12 OPEN_MAX:1048576, ASM
+Scanning the drive for archives:
+1 file, 2089391770 bytes (1993 MiB)
+Listing archive: /tmp/simcorefiles/3f8abbcc-923b-429b-953f-827df68b715f/input_1/PromlematicZipFile.zip
+--
+Path = /tmp/simcorefiles/3f8abbcc-923b-429b-953f-827df68b715f/input_1/PromlematicZipFile.zip
+Type = zip
+Physical Size = 2089391770
+   Date      Time    Attr         Size   Compressed  Name
+------------------- ----- ------------ ------------  ------------------------
+2024-12-11 15:31:40 ....A     45388221     45392151  b0_pair.nii.gz
+2024-12-11 15:31:32 ....A          326           14  bvals
+2024-12-11 15:31:32 ....A         2610         1228  bvecs
+2024-12-11 15:31:52 ....A          192          111  diff2struct_fsl.mat
+2024-12-11 15:31:52 ....A          191          110  diff2struct_fsl_inverse.mat
+2024-12-11 15:31:52 ....A          848          367  diff2struct_mrtrix.txt
+2024-12-11 15:31:46 ....A   1124081680   1009101417  dwi_preproc.mif
+2024-12-11 15:31:52 ....A   1008525390   1008612664  dwi_preproc.nii.gz
+2024-12-11 15:31:52 ....A          154          118  electrode_coords_coreg.txt
+2024-12-11 15:31:52 ....A        57871        36566  mask_brain_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16811          703  mask_lh_hippo_body_coreg.nii.gz
+2024-12-11 15:31:52 ....A        17304         1324  mask_lh_hippo_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16903          821  mask_lh_hippo_head_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16679          540  mask_lh_hippo_tail_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16779          670  mask_rh_hippo_body_coreg.nii.gz
+2024-12-11 15:31:52 ....A        17232         1246  mask_rh_hippo_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16840          752  mask_rh_hippo_head_coreg.nii.gz
+2024-12-11 15:31:52 ....A        16699          582  mask_rh_hippo_tail_coreg.nii.gz
+2024-12-11 15:31:38 ....A     15068179     15069543  mean_b0.nii.gz
+2024-12-11 11:49:22 ....A           54           54  simulation_parameters.txt
+2024-11-22 12:21:58 ....A           94           80  stimulation_parameters.txt
+2024-12-11 15:31:40 ....A     11166326     11168031  T1_coreg.nii.gz
+------------------- ----- ------------ ------------  ------------------------
+2024-12-11 15:31:52         2204427383   2089389092  22 files

packages/service-library/tests/data/archive_utils/list_empty_archive.txt

@@ -0,0 +1,17 @@
+7-Zip (z) 24.09 (x64) : Copyright (c) 1999-2024 Igor Pavlov : 2024-11-28
+ 64-bit locale=en_US.UTF-8 Threads:12 OPEN_MAX:1024, ASM
+
+Scanning the drive for archives:
+1 file, 22 bytes (1 KiB)
+
+Listing archive: /tmp/pytest-of-silenthk/pytest-692/test_archive_unarchive_empty_f1/mixed_types_dir.zip
+
+--
+Path = /tmp/pytest-of-silenthk/pytest-692/test_archive_unarchive_empty_f1/mixed_types_dir.zip
+Type = zip
+Physical Size = 22
+
+   Date      Time    Attr         Size   Compressed  Name
+------------------- ----- ------------ ------------  ------------------------
+------------------- ----- ------------ ------------  ------------------------
+                                     0            0  0 files