Compatibility with google/lit

[43081j]

Google currently use this in lit, specifically here.

There's some flexibility in changing interface i think as i can contribute back to lit to get those changes made.

Suggested migration/support:

function	summary	migration/action
filter	filters an iterable by a predicate	probably unnecessary, only used internally in parse5-utils
getAttr	gets the string value of an attribute by name	use getAttribute
getTextContent	gets the text contents of a node	use getTextContent
setAttr	sets an attribute by name	use setAttribute
insertBefore	inserts a node before another node	use tree adapter insertBefore
insertNode	inserts a node at a given position, optionally replacing	use spliceChildren
isElement	determines if a node is an element	use isElementNode
isCommentNode	determines if a node is a comment	same
isDocumentFragment	determines if a node is a document fragment	same
isTextNode	determines if a node is a text node	same
defaultChildNodes/GetChildNodes	helper passed to other functions to assist in retrieving child nodes	probably can be dropped
depthFirst	traverses the tree of a node depth-first, as an iterable	use queryAll
nodeWalkAll	traverses the tree of a node depth-first, as an array	use queryAll
removeFakeRootElements	removes parse5's fix-up elements (e.g. missing body tags)	TBD
removeNode	removes a given node	use tree adapter (detachNode)
removeNodeSaveChildren	replaces a node with its children	TBD
setTextContent	sets the text content of a node	use setTextContent
newTextNode	creates a new text node	use createTextNode
traverse	traverses the tree of a node depth-first using a visitor	use traverse

Gaps to discuss in compatibility

There are a couple of compatibility gaps we need to decide on:

removeFakeRootElements

This is used by lit to remove the missing root elements parse5 adds when producing an AST (e.g. missing body tags).

It works by finding all elements without a location and replacing them with their children.

• Option 1 - implement this or a very similar function (e.g. removeByQuery might be a little more generic, and the lit team specify a query for locationless nodes)
• Option 2 - don't implement this, just keep it in the lit repo

removeNodeSaveChildren

Only really used by lit during the previously explained removeFakeRootElements call.

It replaces a node with its children (removing a level in the hierarchy).

Could still be useful to people though as its own standalone function, so i think we should at least ship something similar (no clue of any better name though 😬 ).

[43081j]

@justinfagnani @rictic fyi this is probably the relevant issue for you guys.

i have a branch of lit where i made these migration changes and it seemed to build / test fine, though i left the two TBD methods inside lit for that.

[43081j]

Closing this as lit now uses this package as of lit/lit#3266