fix: preserve hard breaks when toggling editors
This commit is contained in:
@@ -47,6 +47,21 @@ describe('compactMarkdown', () => {
|
||||
expect(compactMarkdown(input)).toBe('```js\nconst a = 1\n\nconst b = 2\n```\n')
|
||||
})
|
||||
|
||||
it('removes stray hard-break-only lines after a hard line break', () => {
|
||||
const input = 'some text\\\\\n\\\\\nafter\n'
|
||||
expect(compactMarkdown(input)).toBe('some text\\\\\nafter\n')
|
||||
})
|
||||
|
||||
it('removes stray hard-break-only lines after formatted hard line breaks', () => {
|
||||
const input = '**text\\\\**\n\\\\\nafter\n'
|
||||
expect(compactMarkdown(input)).toBe('**text\\\\**\nafter\n')
|
||||
})
|
||||
|
||||
it('preserves intentional backslash-only lines when no hard line break precedes them', () => {
|
||||
const input = 'just text\n\\\\\nafter\n'
|
||||
expect(compactMarkdown(input)).toBe('just text\n\\\\\nafter\n')
|
||||
})
|
||||
|
||||
it('does not add trailing blank lines', () => {
|
||||
const input = '## Title\n\nText.\n\n\n'
|
||||
expect(compactMarkdown(input)).toBe('## Title\n\nText.\n')
|
||||
|
||||
@@ -4,6 +4,7 @@
|
||||
* - Tight lists (no blank lines between consecutive list items)
|
||||
* - Bullet list markers normalized to `-` (BlockNote outputs `*`)
|
||||
* - HTML entities like ` ` decoded back to spaces
|
||||
* - Stray hard-break-only lines removed after a markdown hard break
|
||||
* - No runs of 3+ blank lines (collapsed to one blank line)
|
||||
* - No trailing blank lines
|
||||
* - Code block content is never modified
|
||||
@@ -11,84 +12,129 @@
|
||||
export function compactMarkdown(md: string): string {
|
||||
if (!md) return md
|
||||
|
||||
const lines = md.split('\n')
|
||||
const result: string[] = []
|
||||
const source = { lines: md.split('\n') }
|
||||
const result = { lines: [] as string[] }
|
||||
let inCodeBlock = false
|
||||
|
||||
for (let i = 0; i < lines.length; i++) {
|
||||
let line = lines[i]
|
||||
|
||||
// Track fenced code blocks — never modify content inside them
|
||||
if (line.trimStart().startsWith('```')) {
|
||||
inCodeBlock = !inCodeBlock
|
||||
result.push(line)
|
||||
continue
|
||||
for (let i = 0; i < source.lines.length; i++) {
|
||||
const next = processMarkdownLine({ doc: source, idx: i, inCodeBlock })
|
||||
inCodeBlock = next.inCodeBlock
|
||||
if (next.line !== null) {
|
||||
result.lines.push(next.line)
|
||||
}
|
||||
|
||||
if (inCodeBlock) {
|
||||
result.push(line)
|
||||
continue
|
||||
}
|
||||
|
||||
// Normalize bullet markers: BlockNote uses `*`, convention is `-`
|
||||
line = normalizeBulletMarker(line)
|
||||
|
||||
// Decode HTML entities that BlockNote inserts (e.g.   for spaces)
|
||||
line = decodeHtmlEntities(line)
|
||||
|
||||
// Skip blank lines that sit between two list items (tight list rule)
|
||||
if (line.trim() === '') {
|
||||
if (isBlankBetweenListItems(lines, i)) continue
|
||||
// Collapse runs of 3+ blank lines to a single blank line
|
||||
if (isExcessiveBlankLine(lines, i)) continue
|
||||
result.push(line)
|
||||
continue
|
||||
}
|
||||
|
||||
result.push(line)
|
||||
}
|
||||
|
||||
// Trim trailing blank lines, keep exactly one trailing newline
|
||||
while (result.length > 0 && result[result.length - 1].trim() === '') {
|
||||
result.pop()
|
||||
}
|
||||
if (result.length > 0) result.push('')
|
||||
|
||||
return result.join('\n')
|
||||
return finalizeMarkdown(result)
|
||||
}
|
||||
|
||||
const LIST_RE = /^(\s*)([-*+]|\d+\.)\s/
|
||||
const HARD_BREAK_ONLY_RE = /^\\+$/
|
||||
const TRAILING_INLINE_CLOSERS_RE = /(?:[*_~`]+)$/
|
||||
|
||||
interface MarkdownDocument {
|
||||
lines: string[]
|
||||
}
|
||||
|
||||
interface LinePosition {
|
||||
doc: MarkdownDocument
|
||||
idx: number
|
||||
}
|
||||
|
||||
interface NormalizedLinePosition extends LinePosition {
|
||||
line: string
|
||||
}
|
||||
|
||||
interface ProcessMarkdownLineArgs extends LinePosition {
|
||||
inCodeBlock: boolean
|
||||
}
|
||||
|
||||
function processMarkdownLine(
|
||||
{ doc, idx, inCodeBlock }: ProcessMarkdownLineArgs,
|
||||
): { inCodeBlock: boolean; line: string | null } {
|
||||
const rawLine = doc.lines[idx]
|
||||
|
||||
if (isFenceDelimiter(rawLine)) {
|
||||
return { inCodeBlock: !inCodeBlock, line: rawLine }
|
||||
}
|
||||
|
||||
if (inCodeBlock) {
|
||||
return { inCodeBlock, line: rawLine }
|
||||
}
|
||||
|
||||
const line = normalizeMarkdownLine(rawLine)
|
||||
if (shouldSkipLine({ doc, idx, line })) {
|
||||
return { inCodeBlock, line: null }
|
||||
}
|
||||
|
||||
return { inCodeBlock, line }
|
||||
}
|
||||
|
||||
function isFenceDelimiter(line: string): boolean {
|
||||
return line.trimStart().startsWith('```')
|
||||
}
|
||||
|
||||
function normalizeMarkdownLine(line: string): string {
|
||||
return decodeHtmlEntities(normalizeBulletMarker(line))
|
||||
}
|
||||
|
||||
function shouldSkipLine({ doc, idx, line }: NormalizedLinePosition): boolean {
|
||||
if (line.trim() === '') {
|
||||
return isBlankBetweenListItems({ doc, idx }) || isExcessiveBlankLine({ doc, idx })
|
||||
}
|
||||
|
||||
return isRedundantHardBreakLine({ doc, idx, line })
|
||||
}
|
||||
|
||||
/** True if this blank line sits between two list items (including nested) */
|
||||
function isBlankBetweenListItems(lines: string[], idx: number): boolean {
|
||||
const prev = findPrevNonBlank(lines, idx)
|
||||
const next = findNextNonBlank(lines, idx)
|
||||
function isBlankBetweenListItems({ doc, idx }: LinePosition): boolean {
|
||||
const prev = findPrevNonBlank({ doc, idx })
|
||||
const next = findNextNonBlank({ doc, idx })
|
||||
if (prev === null || next === null) return false
|
||||
return LIST_RE.test(lines[prev]) && LIST_RE.test(lines[next])
|
||||
return LIST_RE.test(doc.lines[prev]) && LIST_RE.test(doc.lines[next])
|
||||
}
|
||||
|
||||
/** True if this blank line is part of a run of 2+ consecutive blank lines
|
||||
* (i.e. would create 3+ newlines in a row — collapse to just one blank line) */
|
||||
function isExcessiveBlankLine(lines: string[], idx: number): boolean {
|
||||
function isExcessiveBlankLine({ doc, idx }: LinePosition): boolean {
|
||||
// Keep the first blank line in a run, skip subsequent ones
|
||||
if (idx > 0 && lines[idx - 1].trim() === '') return true
|
||||
if (idx > 0 && doc.lines[idx - 1].trim() === '') return true
|
||||
return false
|
||||
}
|
||||
|
||||
function findPrevNonBlank(lines: string[], idx: number): number | null {
|
||||
function findPrevNonBlank({ doc, idx }: LinePosition): number | null {
|
||||
for (let i = idx - 1; i >= 0; i--) {
|
||||
if (lines[i].trim() !== '') return i
|
||||
if (doc.lines[i].trim() !== '') return i
|
||||
}
|
||||
return null
|
||||
}
|
||||
|
||||
function findNextNonBlank(lines: string[], idx: number): number | null {
|
||||
for (let i = idx + 1; i < lines.length; i++) {
|
||||
if (lines[i].trim() !== '') return i
|
||||
function findNextNonBlank({ doc, idx }: LinePosition): number | null {
|
||||
for (let i = idx + 1; i < doc.lines.length; i++) {
|
||||
if (doc.lines[i].trim() !== '') return i
|
||||
}
|
||||
return null
|
||||
}
|
||||
|
||||
function isRedundantHardBreakLine({ doc, idx, line }: NormalizedLinePosition): boolean {
|
||||
if (!isHardBreakOnlyLine(line)) return false
|
||||
|
||||
const prev = findPrevNonBlank({ doc, idx })
|
||||
if (prev === null) return false
|
||||
|
||||
const prevLine = normalizeMarkdownLine(doc.lines[prev])
|
||||
return isHardBreakOnlyLine(prevLine) || endsWithHardBreakMarker(prevLine)
|
||||
}
|
||||
|
||||
function isHardBreakOnlyLine(line: string): boolean {
|
||||
return HARD_BREAK_ONLY_RE.test(line.trim())
|
||||
}
|
||||
|
||||
function endsWithHardBreakMarker(line: string): boolean {
|
||||
const trimmed = line.trimEnd()
|
||||
if (trimmed.endsWith('\\\\')) return true
|
||||
return trimmed.replace(TRAILING_INLINE_CLOSERS_RE, '').endsWith('\\\\')
|
||||
}
|
||||
|
||||
const BULLET_RE = /^(\s*)\*(\s)/
|
||||
/** Normalize `*` bullet markers to `-` (BlockNote default → standard convention) */
|
||||
function normalizeBulletMarker(line: string): string {
|
||||
@@ -100,3 +146,13 @@ function decodeHtmlEntities(line: string): string {
|
||||
if (!line.includes('&#x')) return line
|
||||
return line.replace(/&#x([0-9a-fA-F]+);/g, (_, hex) => String.fromCharCode(parseInt(hex, 16)))
|
||||
}
|
||||
|
||||
function finalizeMarkdown(doc: MarkdownDocument): string {
|
||||
while (doc.lines.length > 0 && doc.lines[doc.lines.length - 1].trim() === '') {
|
||||
doc.lines.pop()
|
||||
}
|
||||
if (doc.lines.length > 0) {
|
||||
doc.lines.push('')
|
||||
}
|
||||
return doc.lines.join('\n')
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user