Ensure nb cell source uses LF instead of CRLF #133762
Conversation
DonJayamanne
requested review from
joyceerhl,
rebornix,
DavidKutu,
IanMatthewHuff,
rchiodo and
roblourens
| @@ -91,6 +91,7 @@ export class NotebookSerializer implements vscode.NotebookSerializer { | |||
| const indentAmount = data.metadata && 'indentAmount' in data.metadata && typeof data.metadata.indentAmount === 'string' ? | |||
| data.metadata.indentAmount : | |||
| ' '; | |||
| return JSON.stringify(notebookContent, undefined, indentAmount); | |||
| // ipynb always ends with a trailing new line (we add this so that SCMs do not show unnecesary changes, resulting from a missing trailing new line). | |||
| return JSON.stringify(sortObjectPropertiesRecursively(notebookContent), undefined, indentAmount) + '\n'; | |||
There was a problem hiding this comment.
ipynb have a trailing \n
| * Jupyter notbeooks/labs sorts the JSON keys in alphabetical order. | ||
| * https://github.com/microsoft/vscode-python/issues/13155 | ||
| */ | ||
| export function sortObjectPropertiesRecursively(obj: any): any { |
There was a problem hiding this comment.
This was missed when moving the serializer into vscode, adding this back.
I think at one point we relialized this was missing, can't remember. but added comments with the link to the original issue so we have some history.
There was a problem hiding this comment.
I think there is an issue on the vscode side for it but I thought we figured out that it wasn't actually being done in the other serializer at the time we moved it
| function createCodeCellFromNotebookCell(cell: NotebookCellData): nbformat.ICodeCell { | ||
| const cellMetadata = cell.metadata?.custom as CellMetadata | undefined; | ||
| const codeCell: any = { | ||
| cell_type: 'code', | ||
| execution_count: cell.executionSummary?.executionOrder ?? null, | ||
| source: splitMultilineString(cell.value), | ||
| source: splitMultilineString(cell.value.replace(/\r\n/g, '\n')), |
There was a problem hiding this comment.
Use \n instead of \r\n
| @@ -81,7 +81,7 @@ export class NotebookSerializer implements vscode.NotebookSerializer { | |||
| const notebookContent: Partial<nbformat.INotebookContent> = data.metadata?.custom || {}; | |||
| notebookContent.cells = notebookContent.cells || []; | |||
| notebookContent.nbformat = notebookContent.nbformat || 4; | |||
| notebookContent.nbformat_minor = notebookContent.nbformat_minor || 2; | |||
| notebookContent.nbformat_minor = notebookContent.nbformat_minor ?? 2; | |||
There was a problem hiding this comment.
a Bug, if the value was 0 we were replacing with 2
There was a problem hiding this comment.
Same for the above? Maybe that can't happen
| /** | ||
| * Cell id for notebooks created with the new 4.5 version of nbformat. | ||
| */ | ||
| id?: string; |
There was a problem hiding this comment.
Brought in a PR that was in jupyter extnesion to support forwards compatibility with the new notebook format 4.5
microsoft/vscode-jupyter#5970
There was a problem hiding this comment.
This new property isn't required in older versions, and only required in the latest 4.5 version (since we support older versions we're not going to upgrade to 4.5, but ensure users can create notebooks in format 4.2 and open notebooks created using format 4.5)
Note: Assume user creates a notebook in vscode (or older version of jupyter) & then opens this in latest version of Jupyter, the notebook format is upgrade to version 4.5 & jupyter notebook/lab will add the new id attribute.
Unfortunately I don't have windows, will need to test this out on windows, give me some time to get this tested.
This PR fixes #125132