Merge pull request #178 from DigitalSlideArchive/170-recursive-redaction

Allow redacting on subdirs

imagedephi/main.py

@@ -112,20 +112,22 @@ def imagedephi(
     type=click.Path(exists=True, file_okay=False, readable=True, writable=True, path_type=Path),
 )
 @click.option("--rename/--skip-rename", default=True)
+@click.option("-r", "--recursive", is_flag=True, help="Redact images in subdirectories.")
 @click.pass_obj
 def run(
     obj: ImagedephiContext,
     input_path: Path,
     output_dir: Path,
     rename: bool,
+    recursive: bool,
     verbose,
     quiet,
     log_file,
 ):
     """Perform the redaction of images."""
     if verbose or quiet or log_file:
         set_logging_config(verbose, quiet, log_file)
-    redact_images(input_path, output_dir, obj.override_rule_set, rename)
+    redact_images(input_path, output_dir, obj.override_rule_set, rename, recursive=recursive)
 
 
 @imagedephi.command

imagedephi/redact/redact.py

@@ -37,7 +37,7 @@ def get_base_rules():
         return base_rule_set
 
 
-def iter_image_files(directory: Path) -> Generator[Path, None, None]:
+def iter_image_files(directory: Path, recursive: bool = False) -> Generator[Path, None, None]:
     """Given a directory return an iterable of available images."""
     for child in sorted(directory.iterdir()):
         # Use first four bits to check if its a tiff file
@@ -49,6 +49,8 @@ def iter_image_files(directory: Path) -> Generator[Path, None, None]:
             else:
                 if data in (b"II\x2a\x00", b"MM\x00\x2a", b"II\x2b\x00", b"MM\x00\x2b"):
                     yield child
+        elif child.is_dir() and recursive:
+            yield from iter_image_files(child, recursive)
 
 
 def create_redact_dir(base_output_dir: Path) -> Path:
@@ -71,13 +73,16 @@ def redact_images(
     override_rules: Ruleset | None = None,
     rename: bool = True,
     overwrite: bool = False,
+    recursive: bool = False,
 ) -> None:
     base_rules = get_base_rules()
     output_file_name_base = (
         override_rules.output_file_name if override_rules else base_rules.output_file_name
     )
     # Convert to a list in order to get the length
-    images_to_redact = list(iter_image_files(input_path) if input_path.is_dir() else [input_path])
+    images_to_redact = list(
+        iter_image_files(input_path, recursive) if input_path.is_dir() else [input_path]
+    )
     output_file_counter = 1
     output_file_max = len(images_to_redact)
     redact_dir = create_redact_dir(output_dir)
@@ -97,16 +102,22 @@ def redact_images(
                     redaction_plan.report_missing_rules()
                 else:
                     redaction_plan.execute_plan()
+                    output_parent_dir = redact_dir
+                    if recursive:
+                        output_parent_dir = Path(
+                            str(image_file).replace(str(input_path), str(redact_dir), 1)
+                        ).parent
+                        output_parent_dir.mkdir(parents=True, exist_ok=True)
                     output_path = (
                         _get_output_path(
                             image_file,
-                            redact_dir,
+                            output_parent_dir,
                             output_file_name_base,
                             output_file_counter,
                             output_file_max,
                         )
                         if rename
-                        else redact_dir / image_file.name
+                        else output_parent_dir / image_file.name
                     )
                     redaction_plan.save(output_path, overwrite)
                     if output_file_counter == output_file_max:

tests/test_e2e.py

@@ -151,3 +151,24 @@ def test_e2e_rename_flag(cli_runner, data_dir: Path, tmp_path: Path, rename: boo
         else tmp_path / "Redacted_2023-05-12_12-12-53" / "test_image.tif"
     )
     assert output_file_name.exists()
+
+
+@freeze_time("2024-01-04 10:48:00")
+@pytest.mark.timeout(5)
+@pytest.mark.parametrize(
+    "recursive,rename", [(True, True), (True, False), (False, False), (False, True)]
+)
+def test_e2e_recursive(cli_runner, data_dir: Path, tmp_path: Path, recursive: bool, rename: bool):
+    args = ["run", str(data_dir / "input"), "--output-dir", str(tmp_path)]
+    if recursive:
+        args.append("--recursive")
+    if rename:
+        args.append("--skip-rename")
+    result = cli_runner.invoke(main.imagedephi, args)
+
+    assert result.exit_code == 0
+    output_subdir = tmp_path / "Redacted_2024-01-04_10-48-00" / "svs"
+    assert output_subdir.exists() == recursive
+
+    if recursive:
+        assert len(list(output_subdir.iterdir()))