How to obtain character range of null value in map

[easbar]

YAML.parseDocument(`name:      \nage: 26`).contents.items

is a list of objects of type 'PAIR' that have a key and value property.
Each key and value provide a range field that is very useful to figure out the exact location in the original text. Unfortunately, for name in this example the values is simply null so it is not possible to obtain the range. How can we get this range?

I tried

YAML.parseDocument(`name:      \nage: 26`, {keepCstNodes: true}).contents.cstNode.items

which seems to be an alternating list of objects of type PLAIN and MAP_VALUE which provide a range field even for the 'null' value of name. The range we get for the MAP_VALUE corresponding to name is exactly what I am looking for. But unfortunately if I write the same YAML as

const doc = YAML.parseDocument(`{name:     , age: 26}`, {keepCstNodes: true});

doc.contents.cstNode.items will be an array of some objects with type PLAIN (for name, age and 26) and some nodes that simply provide a character and offset (for {, :, , and }). So this does not seem to be a reliable way to obtain the character range of the name value (in case it is null).

My use-case is that I want to provide auto-completion for the name field and for this I need to detect the position of the name value in the YAML document. My current workaround is basing the value range on the range of the key, but that is not ideal either.

[eemeli]

You've managed to hit one of the issues (the handling of empty nodes) that lead me to rewrite the parser for yaml@2. Admittedly this particular case could (and probably should) be handled by the v1 AST composer as well, but, well, it clearly doesn't.

The new parser has been merged into the master branch, but I haven't yet cut a beta release from it due to a few missing features & incomplete documentation. It does report a range for your input, albeit placing it just after all the whitespace, including the newline:

const doc = YAML.parseDocument(`name:      \nage: 26`)
doc.contents.items[0].value
// Scalar { value: null, range: [ 12, 12 ], source: '', type: 'PLAIN' }

That range ought to be [ 11, 11 ].

[easbar]

Thanks a lot for your quick reply (and this nice parser of course) :) So this means the new parser changes doc.contents.items[0].value from null to the Scalar object you wrote as a comment? That's great to hear and certainly seems more consistent. But why is the character range [11, 11] and does not start right after name:? And if not would it be reliable to assume that everything between the name key` and this range was whitespace?

Btw I ran into another, similar (but different) case (using the version 2.0.0-3):

const doc = YAML.parseDocument(`people: [{age: 16, name:`)
console.log(doc.contents.items[0].value.items[0].items[1].value);
// Scalar {value: null, sourceStr: "", range: Array(2), cstNode: PlainValue, type: "PLAIN"}

Here the value is not null. I guess, because the 'empty' value of name is at the very end of the input string.

[easbar]

I see you adjusted the range in ed3bce7, thanks a lot for this. So for my initial example we now get Scalar { value: null, range: [ 11, 11 ], source: '', type: 'PLAIN' }. But can you explain why the range is 'empty'? Should it not rather be [11, 12]? In this case it would be the range of the new line character. But why does it not just span all the whitespace between the colon after name and the next non-empty node (age)?

[eemeli]

That would be because a string consisting of one or more spaces is not a valid representation of an empty node. Technically, the YAML spec-mandated position for the empty node would be immediately after the : character, but I figured that it'd be more useful to place it at the end of the whitespace, as there it's possible to actually enter a non-empty value.

It is possible for there to be a tag or an alias between the : and the position of the empty string. And possibly comments as well, if they're followed by a tag or alias.