Do not add an empty cell when an entire Markdown cell gets extracted into a virtual document

[krassowski]

Description

JupyterLab 4 added support for Markdown documents by using extractors mechanisms; it seems that markdown cells are erroneously included as empty cells in the resulting Python document.

It seems that the logic in TextForeignCodeExtractor and in the extractForeignCode() or appendCodeBlock() need to be modified so that when an entire cell gets extracted no cell would be added (rather than adding an empty code cell, padded by two empty lines on each side).

Reproduce

See jupyterlab-lsp issues:

• pycodestyle false positive jupyter-lsp/jupyterlab-lsp#1054
• What is "python-markdown" and how to install it? jupyter-lsp/jupyterlab-lsp#1040

Expected behavior

When an entire cell is extracted do not add new lines before and after it.

This could be solved by special casing the logic in:

jupyterlab/packages/lsp/src/virtual/document.ts

Lines 614 to 673 in 2d68127

	appendCodeBlock(
	block: Document.ICodeBlockOptions,
	editorShift: CodeEditor.IPosition = { line: 0, column: 0 },
	virtualShift?: CodeEditor.IPosition
	): void {
	let cellCode = block.value;
	let ceEditor = block.ceEditor;
	if (this.isDisposed) {
	console.warn('Cannot append code block: document disposed');
	return;
	}
	let sourceCellLines = cellCode.split('\n');
	let { lines, foreignDocumentsMap } = this.prepareCodeBlock(
	block,
	editorShift
	);
	for (let i = 0; i < lines.length; i++) {
	this.virtualLines.set(this.lastVirtualLine + i, {
	skipInspect: [],
	editor: ceEditor,
	// TODO this is incorrect, wont work if something was extracted
	sourceLine: this.lastSourceLine + i
	});
	}
	for (let i = 0; i < sourceCellLines.length; i++) {
	this.sourceLines.set(this.lastSourceLine + i, {
	editorLine: i,
	editorShift: {
	line: editorShift.line - (virtualShift?.line || 0),
	column: i === 0 ? editorShift.column - (virtualShift?.column || 0) : 0
	},
	// TODO: move those to a new abstraction layer (DocumentBlock class)
	editor: ceEditor,
	foreignDocumentsMap,
	// TODO this is incorrect, wont work if something was extracted
	virtualLine: this.lastVirtualLine + i
	});
	}
	this.lastVirtualLine += lines.length;
	// one empty line is necessary to separate code blocks, next 'n' lines are to silence linters;
	// the final cell does not get the additional lines (thanks to the use of join, see below)
	this.lineBlocks.push(lines.join('\n') + '\n');
	// adding the virtual lines for the blank lines
	for (let i = 0; i < this.blankLinesBetweenCells; i++) {
	this.virtualLines.set(this.lastVirtualLine + i, {
	skipInspect: [this.idPath],
	editor: ceEditor,
	sourceLine: null
	});
	}
	this.lastVirtualLine += this.blankLinesBetweenCells;
	this.lastSourceLine += sourceCellLines.length;
	}

For example if lines is just one empty line we could skip adding this.blankLinesBetweenCells. This need to be conditional on foreignDocumentsMap being non-empty so that it does not lead to false negatives (e.g. undetected empty code cells),

Context

• JupyterLab version: 4.1.2