Add positive field query syntax

[RedCMD]

Fixes #3956

possible Fixes/Workaround for:

• Node contains field | query syntax  #3956
• Caching or Serializing a TSQuery #1942
• Slow query parsing #1903
• Loading of query files much slower with 0.19.2 #973
• Slow query parsing, not sure why #1853
• TSQuery serialization and deserialization #2374

I've just copy pasted the negative fields syntax !field
changing:

• negative to positive
• ! to &
• adding a negation when checking the field nodes
  https://github.com/tree-sitter/tree-sitter/pull/4456/files#diff-cfcaac4236741ab3da24a7768f96bba49ae7382f054c0e7757d78862c44ec9d5R4013

using #983 reference

[ribru17]

How is this a fix/workaround for query serialization or the slow query parsing issues?

[RedCMD]

the slow query

(parent (childA) (childB) (childC))
(parent (childA) (childC) (childB))
(parent (childB) (childA) (childC))
(parent (childB) (childC) (childA))
(parent (childC) (childA) (childB))
(parent (childC) (childB) (childA))

can be replaced with the much faster

(parent &childA &childB &childC)

query serialization isn't needed as badly if all queries are fast

---

wow github...
the checks failed cause it took 20min 8sec
which is longer than the allowed 20min

[ribru17]

I guess, but I personally wouldn't consider this a workaround because this specific case isn't the cause for the truly horrendously slow query parses (which are covered by #4042)

[ribru17]

Also is this for fields or children? I thought the ! syntax applied to field names but it looks like in your example it is meant to test for the existence of nodes, rather than fields

[RedCMD]

(parent &fieldA &fieldB &fieldC)

[ribru17]

? ok you can change the names of the hypothetical fields but I'm just saying that this:

(parent &childA &childB &childC)

is not equivalent to this:

(parent (childA) (childB) (childC))
(parent (childA) (childC) (childB))
(parent (childB) (childA) (childC))
(parent (childB) (childC) (childA))
(parent (childC) (childA) (childB))
(parent (childC) (childB) (childA))

if & tests for the presence of fields. Recall that:

this_is_a_field: (this_is_a_node)

That said, I'm not sure this PR works as intended. Would you mind adding some tests for the behavior (not just for syntax errors)?

[RedCMD]

I am now able to build and test the PR
everything seems to be working as expected

is not equivalent to this:

ik
it works in the way I would like
being both orderless and performant
having to use fields instead of node names is kinda a bummer
but that doesn't really affect me much

---

I replaced this old query: (https://github.com/RedCMD/TmLanguage-Syntax-Highlighter/blob/015b41533fbfda48e881f0af0101e0b51036bde4/src/DiagnosticCollection.ts#L547-L581)

(repo
	[(include) (patterns)] (repository
		(repo
			(key) @nestRepo))
	!match !begin)
(repo
	(repository
		(repo
			(key) @nestRepo)) [(include) (patterns)]
	!match !begin)
(pattern
	(patterns) (repository
		(repo
			(key) @nestRepo))
	!match !begin !include)
(pattern
	(repository
		(repo
			(key) @nestRepo)) (patterns)
	!match !begin !include)
(capture
	(patterns) (repository
		(repo
			(key) @nestRepo))
	!match !begin)
(capture
	(repository
		(repo
			(key) @nestRepo)) (patterns)
	!match !begin)

with this new one:

(repo
	&include (repository
		(repo
			(key) @nestRepo))
	!match !begin)
(repo
	&patterns (repository
		(repo
			(key) @nestRepo))
	!match !begin !include)
(pattern
	&patterns (repository
		(repo
			(key) @nestRepo))
	!match !begin !include)
(capture
	&patterns (repository
		(repo
			(key) @nestRepo))
	!match !begin)

the old one took 4200ms to compile
returned duplicate captures if multiple of the same node existed (patterns)
and is just bigger

while the new one compiles in just 22ms, 190x times faster
doesn't return duplicate captures
and is smaller

and I was able to improve this query as well (https://github.com/RedCMD/TmLanguage-Syntax-Highlighter/blob/015b41533fbfda48e881f0af0101e0b51036bde4/src/DiagnosticCollection.ts#L754-L794)
from 17600ms to 530ms. only a 33x speed up

---

I will look into creating tests soon

[RedCMD]

I've added a working positive field query test

[RedCMD]

Fixed merge conflict

even tho there was no conflict..... github??

[ribru17]

there was a conflict (#4496)

[clason]

git can resolve conflicts that Github can't.