Index out of range during parsing

[danpalmer]

I've found a markdown document that causes a fatal error when parsing. While the document itself is fairly useless, I feel like the library should attempt to never fatal error, as this makes detecting and recovering from errors impossible in code that uses it.

Example code:

let text = try String(contentsOfFile: "example.md")
_ = Document(parsing: text, options: [.parseBlockDirectives, .parseSymbolLinks])

This crashes. It looks like it's something to do with CRLF line endings because this does not crash:

var text = try String(contentsOfFile: "example.md")
text = text.replacingOccurrences(of: "\r\n", with: "\n")
_ = Document(parsing: text, options: [.parseBlockDirectives, .parseSymbolLinks])

The example file that triggers this is attached. I am not pasting it in because it's line ending specific.

example.md

[d-ronnqvist]

I thought that there was only an Issue for this but I could only find rdar://114291306 tracking this.

Copying over the extra information from that:

This crash happens in RangeAdjuster.defaultVisit(_:) on line 30:

column: range.lowerBound.column + (trimmedIndentationPerLine[range.lowerBound.line - 1] ),

This seems to do with splitting content into newlines. "\r\n" is a single character—because it's defined like that in Unicode—so splitting on just "\n" won't always produce the correct results.

There are a handful of places that should probably use Swift/Character/isNewline instead of "\n".

[Roshankumar350]

I want to give it a try.
Thanks :)

[d-ronnqvist]

Great. I've assigned this issue to you so that other people who find it can see that someone is already working on it.