def clean_cells(data: Dict[str, Any], nb_source: str,
remove_outputs: bool) -> None:
"""Remove empty cells and clean code cells.
Args:
data: Object representing a parsed JSON notebook.
nb_source: JSON string of entire notebook contents.
remove_outputs: Boolean True to remove code cell outputs, False to keep.
"""
# Clear leading and trailing newlines.
for cell in data["cells"]:
cell_source = cell["source"]
while cell_source and cell_source[0] == "\n":
cell_source.pop(0)
while cell_source and cell_source[-1] == "\n":
cell_source.pop()
cell["source"] = cell_source
# Remove empty cells.
data["cells"] = [cell for cell in data["cells"] if any(cell["source"])]
# Clean cell metadata.
cell_count = 0
for cell in data["cells"]:
cell_count += 1
cell_metadata = cell.get("metadata", {})
if "id" not in cell_metadata:
cell_metadata["id"] = notebook_utils.generate_cell_id(
cell["source"], cell_count)
notebook_utils.del_entries_except(cell_metadata,
keep=["id", "cellView", "colab"])
_clean_metadata_colab(cell_metadata, remove_outputs)
cell["metadata"] = cell_metadata
# The presence of this field indicates that ouputs are already saved.
has_outputs = True if '"output_type"' in nb_source else False
for cell in data["cells"]:
if cell["cell_type"] == "code":
_clean_code_cell(cell, remove_outputs)
if has_outputs and remove_outputs:
notebook_utils.warn("Removed the existing output cells.")
def _clean_metadata_colab(cell_metadata: Dict[str, Any],
remove_outputs: bool) -> None:
"""Clean up a cell's `metadata.colab` field.
Remove all `metadata.colab` contents except for `metadata.colab.resources`, if
present. The Colab resources are used to embed data within the notebook and
can be treated like output cells (kept unless explictly removed).
Args:
cell_metadata: object representing the parsed JSON metadata from a cell.
remove_outputs: Boolean to clear cell output.
"""
colab = cell_metadata.pop("colab", {})
# If no outputs, just clear out `metadata.colab`.
if remove_outputs:
return
# Clear around `resources` if not empty. Otherwise, clear out `metata.colab`.
if colab.get("resources"):
notebook_utils.del_entries_except(colab, keep=["resources"])
cell_metadata["colab"] = colab
def clean_root(data: Dict[str, Any], filepath: pathlib.Path) -> None:
"""Deletes extra top-level notebook fields and metadata.
Jupyter format spec:
https://nbformat.readthedocs.io/en/latest/format_description.html
Args:
data: object representing a parsed JSON notebook.
filepath: String of notebook filepath passed to the command-line.
"""
# These top-level fields are required:
notebook_utils.del_entries_except(
data, keep=["cells", "metadata", "nbformat_minor", "nbformat"])
# All metadata is optional according to spec, but we use some of it.
notebook_utils.del_entries_except(
data["metadata"], keep=["accelerator", "colab", "kernelspec"])
metadata = data.get("metadata", {})
colab = metadata.get("colab", {})
# Set top-level notebook defaults.
data["nbformat"] = 4
data["nbformat_minor"] = 0
# Colab metadata
notebook_utils.del_entries_except(
colab, keep=["collapsed_sections", "name", "toc_visible"])
colab["name"] = os.path.basename(filepath)
colab["toc_visible"] = True
metadata["colab"] = colab
# Kernelspec metadata
kernelspec = metadata.get("kernelspec", {})
notebook_utils.del_entries_except(kernelspec,
keep=["display_name", "name"])
supported_kernels = {"python3": "Python 3", "swift": "Swift"}
kernel_name = kernelspec.get("name")
if kernel_name not in supported_kernels:
kernel_name = "python3" # Notebook defaults to Python3 (same as Colab).
kernelspec["name"] = kernel_name
kernelspec["display_name"] = supported_kernels[kernel_name]
metadata["kernelspec"] = kernelspec
data["metadata"] = metadata