From 0f93d0eb8911ec079683ee092f6cc47522765725 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Mon, 30 Oct 2023 16:32:32 +0300
Subject: [PATCH 01/10] bump 0.0.7_dev

---
 src/nbmetaclean/version.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/nbmetaclean/version.py b/src/nbmetaclean/version.py
index 62c73e6..6f26183 100644
--- a/src/nbmetaclean/version.py
+++ b/src/nbmetaclean/version.py
@@ -1 +1 @@
-__version__ = "0.0.6"  # pragma: no cover
+__version__ = "0.0.7_dev"  # pragma: no cover

From 90609a71f2789ed61cd45f3ca81d209842df43d5 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Thu, 2 Nov 2023 11:29:15 +0300
Subject: [PATCH 02/10] typing

---
 src/nbmetaclean/clean.py | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index fd25a68..8758ef6 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -5,7 +5,8 @@
 from pathlib import Path
 from typing import Optional, Union
 
-from .core import read_nb, write_nb, PathOrStr
+from nbmetaclean.core import read_nb, write_nb
+
 from .typing import NbNode, Metadata
 
 NB_METADATA_PRESERVE_MASKS = [
@@ -14,11 +15,11 @@
 
 
 def filter_meta_mask(
-    nb_meta: Union[Metadata, str],
+    nb_meta: Union[str, int, Metadata],
     mask: [tuple[str, ...]] = None,
-) -> Union[Metadata, str]:
+) -> Union[str, int, Metadata]:
     """Filter metadata by mask. If no mask return empty dict."""
-    if isinstance(nb_meta, str) or mask == ():
+    if isinstance(nb_meta, (str, int)) or mask == ():
         return nb_meta
     if mask is None:
         return {}
@@ -37,9 +38,9 @@ def filter_metadata(
     """Clean notebooknode metadata."""
     if masks is None:
         return {}
-    filtered_meta = {}
+    filtered_meta: Metadata = {}
     for mask in masks:
-        filtered_meta.update(filter_meta_mask(nb_meta, mask))
+        filtered_meta.update(filter_meta_mask(nb_meta, mask))  # type: ignore
     return filtered_meta
 
 
@@ -84,7 +85,7 @@ def clean_nb(
     clear_execution_count: bool = True,
     clear_outputs: bool = False,
     preserve_nb_metadata_masks: Optional[list[tuple[str, ...]],] = None,
-    preserve_cell_metadata_mask: Optional[str] = None,
+    preserve_cell_metadata_mask: Optional[tuple[str, ...]] = None,
 ) -> tuple[NbNode, bool]:
     """Clean notebook - metadata, execution_count, outputs.
 
@@ -118,7 +119,7 @@ def clean_nb(
 
 
 def clean_nb_file(
-    path: Union[PathOrStr, list[PathOrStr]],
+    path: Union[Path, list[Path]],
     clear_nb_metadata: bool = True,
     clear_cell_metadata: bool = True,
     clear_execution_count: bool = True,

From 414467d608ddfc1dbee5a849c70bcd6c1bd9b4d2 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 3 Nov 2023 12:06:09 +0300
Subject: [PATCH 03/10] try exept on clean

---
 src/nbmetaclean/app.py   |  6 ++++-
 src/nbmetaclean/clean.py | 49 ++++++++++++++++++++++++----------------
 tests/test_clean.py      | 13 ++++++-----
 3 files changed, 41 insertions(+), 27 deletions(-)

diff --git a/src/nbmetaclean/app.py b/src/nbmetaclean/app.py
index cb257b0..f6dba39 100644
--- a/src/nbmetaclean/app.py
+++ b/src/nbmetaclean/app.py
@@ -41,13 +41,17 @@ def app() -> None:
             print(f"{path} not exists!")
     if not cfg.silent:
         print(f"notebooks to check: {len(nb_files)} ")
-    cleaned = clean_nb_file(
+    cleaned, errors = clean_nb_file(
         nb_files,
         silent=cfg.silent,
         preserve_timestamp=not cfg.not_pt,
     )
     if not cfg.silent:
         print(f"cleaned nbs: {len(cleaned)}")
+        if errors:
+            print(f"with errors: {len(errors)}")
+            for nb, exc in errors:
+                print(f"{nb}: {exc}")
 
 
 if __name__ == "__main__":
diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index 8758ef6..478c61b 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -126,7 +126,7 @@ def clean_nb_file(
     clear_outputs: bool = False,
     preserve_timestamp: bool = True,
     silent: bool = False,
-) -> list[Path]:
+) -> tuple[list[Path], list[tuple[Path, Exception]]]:
     """Clean metadata and execution count from notebook.
 
     Args:
@@ -139,28 +139,37 @@ def clean_nb_file(
         silent (bool): Silent mode. Defaults to False.
 
     Returns:
-        List[Path]: List of cleaned notebooks
+        tuple[List[Path], List[TuplePath]]: List of cleaned notebooks, list of notebooks with errors.
     """
     if not isinstance(path, list):
         path = [path]
     cleaned: list[Path] = []
+    errors: list[tuple[Path, Exception]] = []
     to_clean = len(path)
     for num, filename in enumerate(path):
-        nb = read_nb(filename)
-        nb, result = clean_nb(
-            nb,
-            clear_execution_count=clear_execution_count,
-            clear_outputs=clear_outputs,
-            clear_nb_metadata=clear_nb_metadata,
-            clear_cell_metadata=clear_cell_metadata,
-        )
-        if result:
-            cleaned.append(filename)
-            if preserve_timestamp:
-                stat = filename.stat()
-            write_nb(nb, filename)
-            if preserve_timestamp:
-                os.utime(filename, (stat.st_atime, stat.st_mtime))
-            if not silent:
-                print(f"done {num + 1} of {to_clean}: {filename}")
-    return cleaned
+        try:
+            nb = read_nb(filename)
+        except Exception as ex:
+            errors.append((filename, ex))
+            continue
+        try:
+            nb, result = clean_nb(
+                nb,
+                clear_execution_count=clear_execution_count,
+                clear_outputs=clear_outputs,
+                clear_nb_metadata=clear_nb_metadata,
+                clear_cell_metadata=clear_cell_metadata,
+            )
+            if result:
+                cleaned.append(filename)
+                if preserve_timestamp:
+                    stat = filename.stat()
+                write_nb(nb, filename)
+                if preserve_timestamp:
+                    os.utime(filename, (stat.st_atime, stat.st_mtime))
+                if not silent:
+                    print(f"done {num + 1} of {to_clean}: {filename}")
+        except Exception as ex:
+            errors.append((filename, ex))
+            continue
+    return cleaned, errors
diff --git a/tests/test_clean.py b/tests/test_clean.py
index e2277b0..35352ad 100644
--- a/tests/test_clean.py
+++ b/tests/test_clean.py
@@ -164,8 +164,9 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
 
     # clean meta, leave execution_count
-    cleaned = clean_nb_file(test_nb_path, clear_execution_count=False)
+    cleaned, errors = clean_nb_file(test_nb_path, clear_execution_count=False)
     assert len(cleaned) == 1
+    assert len(errors) == 0
     nb = read_nb(cleaned[0])
     assert nb["metadata"] == nb_clean["metadata"]
     assert nb["cells"][1]["execution_count"] == 1
@@ -173,7 +174,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
 
     # clean meta, execution_count
     # path as list
-    cleaned = clean_nb_file([test_nb_path])
+    cleaned, errors = clean_nb_file([test_nb_path])
     captured = capsys.readouterr()
     out = captured.out
     assert out.startswith("done")
@@ -183,7 +184,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
     assert nb == nb_clean
 
     # try clean cleaned
-    cleaned = clean_nb_file(test_nb_path)
+    cleaned, errors = clean_nb_file(test_nb_path)
     assert len(cleaned) == 0
     captured = capsys.readouterr()
     out = captured.out
@@ -191,7 +192,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
 
     # silent
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
-    cleaned = clean_nb_file(test_nb_path, silent=True)
+    cleaned, errors = clean_nb_file(test_nb_path, silent=True)
     assert len(cleaned) == 1
     captured = capsys.readouterr()
     assert not captured.out.strip()
@@ -210,7 +211,7 @@ def test_clean_nb_file_timestamp(tmp_path: Path):
     assert test_nb_stat.st_atime == nb_stat.st_atime
     assert test_nb_stat.st_mtime == nb_stat.st_mtime
 
-    cleaned = clean_nb_file(test_nb_path)
+    cleaned, errors = clean_nb_file(test_nb_path)
     assert len(cleaned) == 1
     cleaned_stat = cleaned[0].stat()
     assert True
@@ -219,7 +220,7 @@ def test_clean_nb_file_timestamp(tmp_path: Path):
     # dont preserve timestamp
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
     os.utime(test_nb_path, (nb_stat.st_atime, nb_stat.st_mtime))
-    cleaned = clean_nb_file(test_nb_path, preserve_timestamp=False)
+    cleaned, errors = clean_nb_file(test_nb_path, preserve_timestamp=False)
     assert len(cleaned) == 1
     cleaned_stat = cleaned[0].stat()
     assert True

From de066781f7376c9a8926d2d56a835569f12fc284 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 3 Nov 2023 13:53:23 +0300
Subject: [PATCH 04/10] add catch errors

---
 src/nbmetaclean/clean.py | 36 ++++++++++++++++--------------------
 tests/test_clean.py      | 25 +++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 20 deletions(-)

diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index 478c61b..b2f5bb3 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -152,24 +152,20 @@ def clean_nb_file(
         except Exception as ex:
             errors.append((filename, ex))
             continue
-        try:
-            nb, result = clean_nb(
-                nb,
-                clear_execution_count=clear_execution_count,
-                clear_outputs=clear_outputs,
-                clear_nb_metadata=clear_nb_metadata,
-                clear_cell_metadata=clear_cell_metadata,
-            )
-            if result:
-                cleaned.append(filename)
-                if preserve_timestamp:
-                    stat = filename.stat()
-                write_nb(nb, filename)
-                if preserve_timestamp:
-                    os.utime(filename, (stat.st_atime, stat.st_mtime))
-                if not silent:
-                    print(f"done {num + 1} of {to_clean}: {filename}")
-        except Exception as ex:
-            errors.append((filename, ex))
-            continue
+        nb, result = clean_nb(
+            nb,
+            clear_execution_count=clear_execution_count,
+            clear_outputs=clear_outputs,
+            clear_nb_metadata=clear_nb_metadata,
+            clear_cell_metadata=clear_cell_metadata,
+        )
+        if result:
+            cleaned.append(filename)
+            if preserve_timestamp:
+                stat = filename.stat()
+            write_nb(nb, filename)
+            if preserve_timestamp:
+                os.utime(filename, (stat.st_atime, stat.st_mtime))
+            if not silent:
+                print(f"done {num + 1} of {to_clean}: {filename}")
     return cleaned, errors
diff --git a/tests/test_clean.py b/tests/test_clean.py
index 35352ad..ffaaa1e 100644
--- a/tests/test_clean.py
+++ b/tests/test_clean.py
@@ -194,10 +194,33 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
     cleaned, errors = clean_nb_file(test_nb_path, silent=True)
     assert len(cleaned) == 1
+    assert len(errors) == 0
     captured = capsys.readouterr()
     assert not captured.out.strip()
 
 
+def test_clean_nb_file_errors(capsys: CaptureFixture[str], tmp_path: Path):
+    """test clean_nb_file, errors"""
+    path = tmp_path / "wrong_name"
+    cleaned, errors = clean_nb_file(path)
+    assert len(cleaned) == 0
+    assert len(errors) == 1
+    assert errors[0][0] == path
+    assert "No such file or directory" in str(errors[0][1])
+    captured = capsys.readouterr()
+    assert not captured.out
+    assert not captured.err
+    with path.open("w", encoding="utf-8") as fh:
+        fh.write("wrong nb")
+    cleaned, errors = clean_nb_file(path)
+    assert "wrong_name" in str(errors[0])
+    assert len(cleaned) == 0
+    assert len(errors) == 1
+    captured = capsys.readouterr()
+    assert not captured.out
+    assert not captured.err
+
+
 def test_clean_nb_file_timestamp(tmp_path: Path):
     """test clean_nb_file, timestamp"""
     path = Path("tests/test_nbs")
@@ -213,6 +236,7 @@ def test_clean_nb_file_timestamp(tmp_path: Path):
 
     cleaned, errors = clean_nb_file(test_nb_path)
     assert len(cleaned) == 1
+    assert len(errors) == 0
     cleaned_stat = cleaned[0].stat()
     assert True
     assert cleaned_stat.st_mtime == test_nb_stat.st_mtime
@@ -222,6 +246,7 @@ def test_clean_nb_file_timestamp(tmp_path: Path):
     os.utime(test_nb_path, (nb_stat.st_atime, nb_stat.st_mtime))
     cleaned, errors = clean_nb_file(test_nb_path, preserve_timestamp=False)
     assert len(cleaned) == 1
+    assert len(errors) == 0
     cleaned_stat = cleaned[0].stat()
     assert True
     assert cleaned_stat.st_mtime != nb_stat.st_mtime

From d0efd63676390ff455ceb35024c61d9186655899 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 13:46:09 +0300
Subject: [PATCH 05/10] change hidden argument

---
 src/nbmetaclean/core.py  | 12 +++++++-----
 tests/test_read_write.py | 11 +++++++----
 2 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/src/nbmetaclean/core.py b/src/nbmetaclean/core.py
index 70b3074..8b24a7d 100644
--- a/src/nbmetaclean/core.py
+++ b/src/nbmetaclean/core.py
@@ -51,14 +51,14 @@ def write_nb(
 def get_nb_names(
     path: Optional[PathOrStr] = None,
     recursive: bool = True,
-    filter_hidden: bool = True,
+    hidden: bool = False,
 ) -> list[Path]:
     """Return list of notebooks from `path`. If no `path` return notebooks from current folder.
 
     Args:
         path (Union[Path, str, None]): Path for nb or folder with notebooks.
         recursive bool: Recursive search.
-        filter_hidden bool: Filter hidden paths.
+        hidden bool: Skip or not hidden paths, defaults to False.
 
     Raises:
         sys.exit: If filename or dir not exists or not nb file.
@@ -75,14 +75,16 @@ def get_nb_names(
         result = []
         for item in nb_path.iterdir():
             if item.is_file() and item.suffix == ".ipynb":
-                if filter_hidden and item.name.startswith("."):
+                if not hidden and item.name.startswith("."):
                     continue
                 result.append(item)
             if item.is_dir():
                 if recursive:
-                    if filter_hidden and item.name.startswith("."):
+                    if not hidden and item.name.startswith("."):
                         continue
-                    result.extend(get_nb_names(item, recursive, filter_hidden))
+                    if "checkpoint" in item.name:
+                        continue
+                    result.extend(get_nb_names(item, recursive, hidden))
 
         return result
 
diff --git a/tests/test_read_write.py b/tests/test_read_write.py
index 5b07b55..55ee39c 100644
--- a/tests/test_read_write.py
+++ b/tests/test_read_write.py
@@ -45,16 +45,19 @@ def test_write_nb(tmp_path: Path):
 def test_get_nb_names():
     """test get_nb_names"""
     path = Path("tests/test_nbs")
+    # filename as argument
     file = path / "test_nb_1.ipynb"
     names = get_nb_names(file)
     assert len(names) == 1
     names.sort(key=lambda x: x.name)
     assert names[0] == file
+    # path as argument
     names = get_nb_names(path)
     assert len(names) == 2
     names.sort(key=lambda x: x.name)
     assert names[0] == file
-    names = get_nb_names(path, filter_hidden=False)
+    # path as argument. add hidden files
+    names = get_nb_names(path, hidden=True)
     assert len(names) == 3
     try:
         get_nb_names("wrong_name")
@@ -78,7 +81,7 @@ def test_get_nb_names_recursive_hidden(tmp_path: Path):
         pass
     files = get_nb_names(tmp_path)
     assert len(files) == 1
-    files = get_nb_names(tmp_path, filter_hidden=False)
+    files = get_nb_names(tmp_path, hidden=True)
     assert len(files) == 2
     # add simple file
     with open((tmp_path / "simple"), "w", encoding="utf-8") as _:
@@ -95,7 +98,7 @@ def test_get_nb_names_recursive_hidden(tmp_path: Path):
         pass
     files = get_nb_names(tmp_path)
     assert len(files) == 2
-    files = get_nb_names(tmp_path, filter_hidden=False)
+    files = get_nb_names(tmp_path, hidden=True)
     assert len(files) == 4
 
     files = get_nb_names(tmp_path, recursive=False)
@@ -108,7 +111,7 @@ def test_get_nb_names_recursive_hidden(tmp_path: Path):
         pass
     with open((hid_dir / ".tst").with_suffix(suffix), "w", encoding="utf-8") as _:
         pass
-    files = get_nb_names(tmp_path, filter_hidden=False)
+    files = get_nb_names(tmp_path, hidden=True)
     assert len(files) == 6
     files = get_nb_names(tmp_path)
     assert len(files) == 2

From 1bdce1c4a1235f186585e862a1cf331f11d6c0b8 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 17:30:07 +0300
Subject: [PATCH 06/10] cfg for clean

---
 src/nbmetaclean/app.py   |   8 ++-
 src/nbmetaclean/clean.py | 146 +++++++++++++++++++++++----------------
 tests/test_clean.py      |  99 +++++++++++++++++++++-----
 3 files changed, 173 insertions(+), 80 deletions(-)

diff --git a/src/nbmetaclean/app.py b/src/nbmetaclean/app.py
index f6dba39..bc2e4b8 100644
--- a/src/nbmetaclean/app.py
+++ b/src/nbmetaclean/app.py
@@ -1,7 +1,7 @@
 import argparse
 from pathlib import Path
 
-from .clean import clean_nb_file
+from .clean import CleanConfig, clean_nb_file
 from .core import get_nb_names
 
 parser = argparse.ArgumentParser(
@@ -43,8 +43,10 @@ def app() -> None:
         print(f"notebooks to check: {len(nb_files)} ")
     cleaned, errors = clean_nb_file(
         nb_files,
-        silent=cfg.silent,
-        preserve_timestamp=not cfg.not_pt,
+        CleanConfig(
+            silent=cfg.silent,
+            preserve_timestamp=not cfg.not_pt,
+        ),
     )
     if not cfg.silent:
         print(f"cleaned nbs: {len(cleaned)}")
diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index b2f5bb3..604373b 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 import copy
+from dataclasses import dataclass
 import os
 
 from pathlib import Path
@@ -9,9 +10,34 @@
 
 from .typing import NbNode, Metadata
 
-NB_METADATA_PRESERVE_MASKS = [
-    ("language_info", "name"),
-]
+NB_METADATA_PRESERVE_MASKS = (("language_info", "name"),)
+
+
+@dataclass
+class CleanConfig:
+    """Clean config.
+
+    Args:
+        clear_nb_metadata (bool, optional): Clear notebook metadata. Defaults to True.
+        clear_cell_metadata (bool, optional): Clear cell metadata. Defaults to False.
+        clear_execution_count (bool, optional): Clear cell execution count. Defaults to True.
+        clear_outputs (bool, optional): Clear cell outputs. Defaults to False.
+        preserve_timestamp (bool, optional): Preserve timestamp. Defaults to True.
+        silent (bool, optional): Silent mode. Defaults to False.
+        nb_metadata_preserve_mask (Optional[tuple[str, ...]], optional):
+            Preserve mask for notebook metadata. Defaults to None.
+        cell_metadata_preserve_mask (Optional[tuple[str, ...]], optional):
+            Preserve mask for cell metadata. Defaults to None.
+    """
+
+    clear_nb_metadata: bool = True
+    clear_cell_metadata: bool = False
+    clear_execution_count: bool = True
+    clear_outputs: bool = False
+    preserve_timestamp: bool = True
+    silent: bool = False
+    nb_metadata_preserve_mask: Optional[tuple[str, ...]] = None
+    cell_metadata_preserve_mask: Optional[tuple[str, ...]] = None
 
 
 def filter_meta_mask(
@@ -44,49 +70,59 @@ def filter_metadata(
     return filtered_meta
 
 
-def clean_cell_metadata(
+def clean_cell(
     cell: NbNode,
-    clear_execution_count: bool = True,
-    clear_outputs: bool = False,
-    preserve_cell_metadata_mask: Optional[list[tuple[str, ...]]] = None,
+    cfg: CleanConfig,
 ) -> bool:
-    """Clean cell metadata."""
+    """Clean cell: optionally metadata, execution_count and outputs."""
     changed = False
-    if metadata := cell.get("metadata", None):
-        old_metadata = copy.deepcopy(metadata)
-        cell["metadata"] = filter_metadata(metadata, preserve_cell_metadata_mask)
-        if cell["metadata"] != old_metadata:
-            changed = True
-    if clear_outputs and cell.get("outputs"):
-        cell["outputs"] = []
-        changed = True
-    if clear_execution_count and cell.get("execution_count"):
+
+    if cfg.clear_cell_metadata:
+        if metadata := cell.get("metadata", None):
+            old_metadata = copy.deepcopy(metadata)
+            cell["metadata"] = filter_metadata(
+                metadata, cfg.cell_metadata_preserve_mask
+            )
+            if cell["metadata"] != old_metadata:
+                changed = True
+
+    if cfg.clear_execution_count and cell.get("execution_count"):
         cell["execution_count"] = None
         changed = True
-    if outputs := cell.get("outputs"):
-        for output in outputs:
-            if clear_execution_count and output.get("execution_count", None):
-                output["execution_count"] = None
+
+    if cell.get("outputs"):
+        if cfg.clear_outputs:
+            cell["outputs"] = []
+            changed = True
+        elif cfg.clear_cell_metadata or cfg.clear_execution_count:
+            result = clean_outputs(cell["outputs"], cfg)
+            if result:
+                changed = True
+
+    return changed
+
+
+def clean_outputs(outputs: list[NbNode], cfg: CleanConfig) -> bool:
+    """Clean outputs."""
+    changed = False
+    for output in outputs:
+        if cfg.clear_execution_count and output.get("execution_count", None):
+            output["execution_count"] = None
+            changed = True
+        if cfg.clear_cell_metadata and (metadata := output.get("metadata", None)):
+            old_metadata = copy.deepcopy(metadata)
+            output["metadata"] = filter_metadata(
+                metadata, cfg.cell_metadata_preserve_mask
+            )
+            if output["metadata"] != old_metadata:
                 changed = True
-            if metadata := output.get("metadata", None):
-                old_metadata = copy.deepcopy(metadata)
-                output["metadata"] = filter_metadata(
-                    metadata, preserve_cell_metadata_mask
-                )
-                if output["metadata"] != old_metadata:
-                    changed = True
     return changed
 
 
 def clean_nb(
     nb: NbNode,
-    clear_nb_metadata: bool = True,
-    clear_cell_metadata: bool = True,
-    clear_execution_count: bool = True,
-    clear_outputs: bool = False,
-    preserve_nb_metadata_masks: Optional[list[tuple[str, ...]],] = None,
-    preserve_cell_metadata_mask: Optional[tuple[str, ...]] = None,
-) -> tuple[NbNode, bool]:
+    cfg: CleanConfig,
+) -> bool:
     """Clean notebook - metadata, execution_count, outputs.
 
     Args:
@@ -98,41 +134,36 @@ def clean_nb(
         bool: True if changed.
     """
     changed = False
-    if clear_nb_metadata and (metadata := nb.get("metadata")):
+    if cfg.clear_nb_metadata and (metadata := nb.get("metadata")):
         old_metadata = copy.deepcopy(metadata)
-        masks = preserve_nb_metadata_masks or NB_METADATA_PRESERVE_MASKS
+        masks = (
+            cfg.nb_metadata_preserve_mask or NB_METADATA_PRESERVE_MASKS
+        )  # todo: merge or replace?
         nb["metadata"] = filter_metadata(metadata, masks=masks)
         if nb["metadata"] != old_metadata:
             changed = True
-    if clear_cell_metadata:
+    if cfg.clear_cell_metadata or cfg.clear_execution_count or cfg.clear_outputs:
         for cell in nb["cells"]:
-            result = clean_cell_metadata(
+            result = clean_cell(
                 cell,
-                clear_execution_count=clear_execution_count,
-                clear_outputs=clear_outputs,
-                preserve_cell_metadata_mask=preserve_cell_metadata_mask,
+                cfg,
             )
             if result:
                 changed = True
 
-    return nb, changed
+    return changed
 
 
 def clean_nb_file(
     path: Union[Path, list[Path]],
-    clear_nb_metadata: bool = True,
-    clear_cell_metadata: bool = True,
-    clear_execution_count: bool = True,
-    clear_outputs: bool = False,
-    preserve_timestamp: bool = True,
-    silent: bool = False,
+    cfg: Optional[CleanConfig] = None,
 ) -> tuple[list[Path], list[tuple[Path, Exception]]]:
     """Clean metadata and execution count from notebook.
 
     Args:
         path (Union[str, PosixPath]): Notebook filename or list of names.
         clear_nb_metadata (bool): Clear notebook metadata. Defaults to True.
-        clear_cell_metadata (bool): Clear cell metadata. Defaults to True.
+        clear_cell_metadata (bool): Clear cell metadata. Defaults to False.
         clear_outputs (bool): Clear outputs. Defaults to False.
         preserve_timestamp (bool): Preserve timestamp. Defaults to True.
         clear_execution_count (bool, optional): Clean execution count. Defaults to True.
@@ -141,6 +172,8 @@ def clean_nb_file(
     Returns:
         tuple[List[Path], List[TuplePath]]: List of cleaned notebooks, list of notebooks with errors.
     """
+    if cfg is None:
+        cfg = CleanConfig()
     if not isinstance(path, list):
         path = [path]
     cleaned: list[Path] = []
@@ -152,20 +185,17 @@ def clean_nb_file(
         except Exception as ex:
             errors.append((filename, ex))
             continue
-        nb, result = clean_nb(
+        result = clean_nb(
             nb,
-            clear_execution_count=clear_execution_count,
-            clear_outputs=clear_outputs,
-            clear_nb_metadata=clear_nb_metadata,
-            clear_cell_metadata=clear_cell_metadata,
+            cfg,
         )
         if result:
             cleaned.append(filename)
-            if preserve_timestamp:
+            if cfg.preserve_timestamp:
                 stat = filename.stat()
             write_nb(nb, filename)
-            if preserve_timestamp:
+            if cfg.preserve_timestamp:
                 os.utime(filename, (stat.st_atime, stat.st_mtime))
-            if not silent:
+            if not cfg.silent:
                 print(f"done {num + 1} of {to_clean}: {filename}")
     return cleaned, errors
diff --git a/tests/test_clean.py b/tests/test_clean.py
index ffaaa1e..e8a7fd3 100644
--- a/tests/test_clean.py
+++ b/tests/test_clean.py
@@ -6,7 +6,8 @@
 
 from nbmetaclean.clean import (
     NB_METADATA_PRESERVE_MASKS,
-    clean_cell_metadata,
+    CleanConfig,
+    clean_cell,
     clean_nb,
     clean_nb_file,
     filter_meta_mask,
@@ -60,7 +61,13 @@ def test_clean_cell_metadata():
     assert not cell.get("metadata")
     assert cell.get("execution_count") == 1
     cell["metadata"] = {"some key": "some value"}
-    changed = clean_cell_metadata(cell, clear_outputs=True)
+    changed = clean_cell(
+        cell,
+        cfg=CleanConfig(
+            clear_outputs=True,
+            clear_cell_metadata=True,
+        ),
+    )
     assert changed
     assert not cell.get("outputs")
     assert not cell.get("metadata")
@@ -73,24 +80,35 @@ def test_clean_cell_metadata():
         "some key": "some value",
         "some other key": "some value",
     }
-    changed = clean_cell_metadata(
+    changed = clean_cell(
         cell,
-        clear_execution_count=False,
-        preserve_cell_metadata_mask=[("some key",)],
+        CleanConfig(
+            clear_execution_count=False,
+            clear_cell_metadata=True,
+            cell_metadata_preserve_mask=(("some key",),),
+        ),
     )
     assert changed
     assert cell["outputs"][0]["metadata"] == {"some key": "some value"}
     assert cell["metadata"] == {"some key": "some value"}
     assert cell["execution_count"] == 1
 
-    # clear outputs, same mask -> no changes meta, clear ex
-    changed = clean_cell_metadata(cell, preserve_cell_metadata_mask=[("some key",)])
+    # clear outputs, same mask -> no changes meta, clear execution_count
+    changed = clean_cell(
+        cell,
+        cfg=CleanConfig(),
+    )
     assert changed
     assert cell["execution_count"] is None
     assert cell["metadata"] == {"some key": "some value"}
 
     # clear execution_count, metadata
-    changed = clean_cell_metadata(cell)
+    changed = clean_cell(
+        cell,
+        cfg=CleanConfig(
+            clear_cell_metadata=True,
+        ),
+    )
     assert changed
     assert not cell["outputs"][0]["metadata"]
     assert not cell["execution_count"]
@@ -98,12 +116,49 @@ def test_clean_cell_metadata():
     assert not cell["outputs"][0]["metadata"]
 
 
+def test_clean_cell():
+    """test clean_cel"""
+    test_nb = read_nb("tests/test_nbs/.test_nb_2_meta.ipynb")
+
+    # nothing to clean.
+    cell = copy.deepcopy(test_nb.get("cells")[1])
+    assert cell.get("outputs")
+    assert not cell.get("metadata")
+    assert cell.get("execution_count") == 1
+    result = clean_cell(cell, CleanConfig(clear_execution_count=False))
+    assert not result
+
+    # clean cell metadata, cell without metadata
+    cell["metadata"] = {}
+    result = clean_cell(cell, CleanConfig(clear_cell_metadata=True))
+    assert result
+    assert not cell.get("metadata")
+    assert cell.get("outputs")
+
+    # clear output metadata
+    cell["outputs"][0]["metadata"] = {"some key": "some value"}
+    result = clean_cell(
+        cell,
+        CleanConfig(
+            clear_cell_metadata=True,
+            cell_metadata_preserve_mask=(("some key",),),
+        ),
+    )
+    assert not result
+    assert cell["outputs"][0].get("metadata") == {"some key": "some value"}
+
+
 def test_clean_cell_metadata_markdown():
     """test clean_cell_metadata with markdown cell"""
     test_nb = read_nb("tests/test_nbs/.test_nb_2_meta.ipynb")
     cell = copy.deepcopy(test_nb["cells"][0])
     cell["metadata"] = {"some key": "some value"}
-    changed = clean_cell_metadata(cell)
+    changed = clean_cell(
+        cell,
+        cfg=CleanConfig(
+            clear_cell_metadata=True,
+        ),
+    )
     assert changed
     assert not cell["metadata"]
 
@@ -117,7 +172,7 @@ def test_clean_nb():
     assert nb["cells"][1]["execution_count"] == 1
     assert nb["cells"][1]["outputs"][0]["execution_count"] == 1
     assert nb["metadata"]
-    nb, result = clean_nb(nb)
+    result = clean_nb(nb, cfg=CleanConfig())
     assert result is True
     assert nb["cells"][1]["execution_count"] is None
     assert nb["cells"][1]["outputs"][0]["execution_count"] is None
@@ -125,12 +180,15 @@ def test_clean_nb():
     assert nb == nb_clean
 
     # # try clean cleaned
-    nb, result = clean_nb(nb_clean)
+    result = clean_nb(nb_clean, cfg=CleanConfig())
     assert not result
 
     # # clean metadata, leave execution_count
     nb = read_nb(nb_path)
-    nb, result = clean_nb(nb, clear_execution_count=False)
+    result = clean_nb(
+        nb,
+        cfg=CleanConfig(clear_execution_count=False),
+    )
     assert result
     assert nb["cells"][1]["execution_count"] == 1
     assert nb["cells"][1]["outputs"][0]["execution_count"] == 1
@@ -139,7 +197,7 @@ def test_clean_nb():
     # clean nb metadata, leave cells metadata
     nb = read_nb(nb_path)
     nb["cells"][1]["metadata"] = {"some key": "some value"}
-    nb, result = clean_nb(nb, clear_cell_metadata=False)
+    result = clean_nb(nb, CleanConfig(clear_execution_count=False))
     assert result
     assert nb["metadata"] == nb_clean["metadata"]
     assert nb["cells"][1]["metadata"] == {"some key": "some value"}
@@ -148,7 +206,7 @@ def test_clean_nb():
     # clean cells metadata, leave nb metadata
     nb = read_nb(nb_path)
     nb_meta = copy.deepcopy(nb["metadata"])
-    nb, result = clean_nb(nb, clear_nb_metadata=False)
+    result = clean_nb(nb, CleanConfig(clear_nb_metadata=False))
     assert result
     assert nb["metadata"] == nb_meta
     assert nb["cells"][1]["execution_count"] is None
@@ -164,7 +222,10 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
 
     # clean meta, leave execution_count
-    cleaned, errors = clean_nb_file(test_nb_path, clear_execution_count=False)
+    cleaned, errors = clean_nb_file(
+        test_nb_path,
+        cfg=CleanConfig(clear_execution_count=False),
+    )
     assert len(cleaned) == 1
     assert len(errors) == 0
     nb = read_nb(cleaned[0])
@@ -174,7 +235,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
 
     # clean meta, execution_count
     # path as list
-    cleaned, errors = clean_nb_file([test_nb_path])
+    cleaned, errors = clean_nb_file([test_nb_path], CleanConfig())
     captured = capsys.readouterr()
     out = captured.out
     assert out.startswith("done")
@@ -184,7 +245,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
     assert nb == nb_clean
 
     # try clean cleaned
-    cleaned, errors = clean_nb_file(test_nb_path)
+    cleaned, errors = clean_nb_file(test_nb_path, CleanConfig())
     assert len(cleaned) == 0
     captured = capsys.readouterr()
     out = captured.out
@@ -192,7 +253,7 @@ def test_clean_nb_file(tmp_path: Path, capsys: CaptureFixture[str]):
 
     # silent
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
-    cleaned, errors = clean_nb_file(test_nb_path, silent=True)
+    cleaned, errors = clean_nb_file(test_nb_path, CleanConfig(silent=True))
     assert len(cleaned) == 1
     assert len(errors) == 0
     captured = capsys.readouterr()
@@ -244,7 +305,7 @@ def test_clean_nb_file_timestamp(tmp_path: Path):
     # dont preserve timestamp
     test_nb_path = write_nb(read_nb(path / nb_name), tmp_path / nb_name)
     os.utime(test_nb_path, (nb_stat.st_atime, nb_stat.st_mtime))
-    cleaned, errors = clean_nb_file(test_nb_path, preserve_timestamp=False)
+    cleaned, errors = clean_nb_file(test_nb_path, CleanConfig(preserve_timestamp=False))
     assert len(cleaned) == 1
     assert len(errors) == 0
     cleaned_stat = cleaned[0].stat()

From cdf1fced84c9a940516221707a753a4622b24c86 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 18:03:25 +0300
Subject: [PATCH 07/10] add authors to mask

---
 src/nbmetaclean/clean.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index 604373b..f5e8e8d 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -10,7 +10,10 @@
 
 from .typing import NbNode, Metadata
 
-NB_METADATA_PRESERVE_MASKS = (("language_info", "name"),)
+NB_METADATA_PRESERVE_MASKS = (
+    ("language_info", "name"),
+    ("authors",),
+)
 
 
 @dataclass

From d97974c43472e9f08d7dd35d368a30baa8888d77 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 18:06:08 +0300
Subject: [PATCH 08/10] sign nbs

---
 tests/test_nbs/.test_nb_2_meta.ipynb | 6 ++++++
 tests/test_nbs/test_nb_1.ipynb       | 6 ++++++
 tests/test_nbs/test_nb_2_clean.ipynb | 6 ++++++
 3 files changed, 18 insertions(+)

diff --git a/tests/test_nbs/.test_nb_2_meta.ipynb b/tests/test_nbs/.test_nb_2_meta.ipynb
index b75ce89..1a024c9 100644
--- a/tests/test_nbs/.test_nb_2_meta.ipynb
+++ b/tests/test_nbs/.test_nb_2_meta.ipynb
@@ -29,6 +29,12 @@
   }
  ],
  "metadata": {
+  "authors": [
+   {
+    "github": "https://github.com/ayasyrev",
+    "name": "Andrei Yasyrev"
+   }
+  ],
   "kernelspec": {
    "display_name": "nbmetaclean",
    "language": "python",
diff --git a/tests/test_nbs/test_nb_1.ipynb b/tests/test_nbs/test_nb_1.ipynb
index 5c6deca..4c229c7 100644
--- a/tests/test_nbs/test_nb_1.ipynb
+++ b/tests/test_nbs/test_nb_1.ipynb
@@ -14,6 +14,12 @@
   }
  ],
  "metadata": {
+  "authors": [
+   {
+    "github": "https://github.com/ayasyrev",
+    "name": "Andrei Yasyrev"
+   }
+  ],
   "language_info": {
    "name": "python"
   }
diff --git a/tests/test_nbs/test_nb_2_clean.ipynb b/tests/test_nbs/test_nb_2_clean.ipynb
index f316a23..82d52cc 100644
--- a/tests/test_nbs/test_nb_2_clean.ipynb
+++ b/tests/test_nbs/test_nb_2_clean.ipynb
@@ -29,6 +29,12 @@
   }
  ],
  "metadata": {
+  "authors": [
+   {
+    "github": "https://github.com/ayasyrev",
+    "name": "Andrei Yasyrev"
+   }
+  ],
   "language_info": {
    "name": "python"
   }

From c9c24fa95dc0839e901161d4cc2a4deabd1533cb Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 19:13:52 +0300
Subject: [PATCH 09/10] mask merge

---
 src/nbmetaclean/clean.py | 21 ++++++++++++++-------
 tests/test_clean.py      | 26 ++++++++++++++++++++++++++
 tests/test_read_write.py |  3 ++-
 3 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/src/nbmetaclean/clean.py b/src/nbmetaclean/clean.py
index f5e8e8d..747a302 100644
--- a/src/nbmetaclean/clean.py
+++ b/src/nbmetaclean/clean.py
@@ -4,7 +4,7 @@
 import os
 
 from pathlib import Path
-from typing import Optional, Union
+from typing import Iterable, Optional, Union
 
 from nbmetaclean.core import read_nb, write_nb
 
@@ -31,6 +31,8 @@ class CleanConfig:
             Preserve mask for notebook metadata. Defaults to None.
         cell_metadata_preserve_mask (Optional[tuple[str, ...]], optional):
             Preserve mask for cell metadata. Defaults to None.
+        mask_merge (bool, optional): Merge masks. Add new mask to default.
+            If False - use new mask. Defaults to True.
     """
 
     clear_nb_metadata: bool = True
@@ -39,13 +41,14 @@ class CleanConfig:
     clear_outputs: bool = False
     preserve_timestamp: bool = True
     silent: bool = False
-    nb_metadata_preserve_mask: Optional[tuple[str, ...]] = None
-    cell_metadata_preserve_mask: Optional[tuple[str, ...]] = None
+    nb_metadata_preserve_mask: Optional[Iterable[tuple[str, ...]]] = None
+    cell_metadata_preserve_mask: Optional[Iterable[tuple[str, ...]]] = None
+    mask_merge: bool = True
 
 
 def filter_meta_mask(
     nb_meta: Union[str, int, Metadata],
-    mask: [tuple[str, ...]] = None,
+    mask: Optional[Iterable[tuple[str, ...]]] = None,
 ) -> Union[str, int, Metadata]:
     """Filter metadata by mask. If no mask return empty dict."""
     if isinstance(nb_meta, (str, int)) or mask == ():
@@ -139,9 +142,13 @@ def clean_nb(
     changed = False
     if cfg.clear_nb_metadata and (metadata := nb.get("metadata")):
         old_metadata = copy.deepcopy(metadata)
-        masks = (
-            cfg.nb_metadata_preserve_mask or NB_METADATA_PRESERVE_MASKS
-        )  # todo: merge or replace?
+        masks = NB_METADATA_PRESERVE_MASKS
+        if cfg.nb_metadata_preserve_mask:
+            if not cfg.mask_merge:
+                masks = cfg.nb_metadata_preserve_mask
+            else:
+                masks = cfg.nb_metadata_preserve_mask + masks
+
         nb["metadata"] = filter_metadata(metadata, masks=masks)
         if nb["metadata"] != old_metadata:
             changed = True
diff --git a/tests/test_clean.py b/tests/test_clean.py
index e8a7fd3..f0e8f6b 100644
--- a/tests/test_clean.py
+++ b/tests/test_clean.py
@@ -51,6 +51,32 @@ def test_new_metadata():
     assert new_meta == {"language_info": {"name": "python"}}
 
 
+def test_clean_nb_metadata():
+    """test clean_nb_metadata"""
+    test_nb = read_nb("tests/test_nbs/test_nb_2_clean.ipynb")
+    cfg = CleanConfig()
+    result = clean_nb(test_nb, cfg)
+    assert not result
+
+    # add metadata, new filter, mask not merged
+    test_nb["metadata"]["some key"] = "some value"
+    cfg.nb_metadata_preserve_mask = (("some key",),)
+    cfg.mask_merge = False
+    result = clean_nb(test_nb, cfg)
+    assert result
+    assert test_nb["metadata"] == {"some key": "some value"}
+
+    # add metadata, new filter, mask merged
+    test_nb = read_nb("tests/test_nbs/test_nb_2_clean.ipynb")
+    test_nb["metadata"]["some_key"] = {"key_1": 1, "key_2": 2}
+    cfg.nb_metadata_preserve_mask = (("some_key", "key_1"),)
+    cfg.mask_merge = True
+    result = clean_nb(test_nb, cfg)
+    assert result
+    assert test_nb["metadata"]["authors"][0]["name"] == "Andrei Yasyrev"
+    assert test_nb["metadata"]["some_key"] == {"key_1": 1}
+
+
 def test_clean_cell_metadata():
     """test clean_cell_metadata"""
     test_nb = read_nb("tests/test_nbs/.test_nb_2_meta.ipynb")
diff --git a/tests/test_read_write.py b/tests/test_read_write.py
index 55ee39c..964959d 100644
--- a/tests/test_read_write.py
+++ b/tests/test_read_write.py
@@ -8,7 +8,8 @@ def test_read_nb():
     file = Path("tests/test_nbs/test_nb_1.ipynb")
     nb = read_nb(file)
     assert isinstance(nb, dict)
-    assert nb["metadata"] == {"language_info": {"name": "python"}}
+    assert nb["metadata"]["language_info"] == {"name": "python"}
+    assert nb["metadata"]["authors"][0]["name"] == "Andrei Yasyrev"
     assert nb["nbformat"] == 4
     assert nb["nbformat_minor"] == 2
     cells = nb["cells"]

From c294eb69ce1497a14e4fd06e0397aadf4fc3ece8 Mon Sep 17 00:00:00 2001
From: ayasyrev <a.yasyrev@gmail.com>
Date: Fri, 17 Nov 2023 19:19:32 +0300
Subject: [PATCH 10/10] bump 0.0.7

---
 src/nbmetaclean/version.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/nbmetaclean/version.py b/src/nbmetaclean/version.py
index 6f26183..5382d2c 100644
--- a/src/nbmetaclean/version.py
+++ b/src/nbmetaclean/version.py
@@ -1 +1 @@
-__version__ = "0.0.7_dev"  # pragma: no cover
+__version__ = "0.0.7"  # pragma: no cover