Question about computeContentAssist()

[appocalypseltd]

This is more of a question than a bug or an error...not quite understanding the meaning of occurrenceStack returned as part of ISyntacticContentAssistPath[] from computeContentAssist().

The source code says:

The occurrence index (SUBRULE1/2/3/5/…) of each Grammar rule invoked and still unterminated. Used to distinguish between different invocations of the same subrule at the same top level rule.

So I don't understand how it can be an array like this:

[1, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0]

If the rule at index 4 is unterminated, then why are the preceding rules also not indicated as unterminated?

For context, here is one of the elements of the array:

{
  "nextTokenType": {
    "name": "OpenBracket",
    "PATTERN": {},
    "tokenTypeIdx": 12,
    "CATEGORIES": [],
    "categoryMatches": [],
    "categoryMatchesMap": {},
    "isParent": false
  },
  "nextTokenOccurrence": 0,
  "ruleStack": [
    "booleanExpression",
    "booleanTerm",
    "booleanFactor",
    "comparison",
    "additionExpression",
    "multiplicationExpression",
    "atomicExpression",
    "parenthesisExpression",
    "additionExpression",
    "multiplicationExpression",
    "atomicExpression",
    "parenthesisExpression"
  ],
  "occurrenceStack": [1, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0]
}

1 corresponds to booleanExpression, and 2 to additionExpression, defined like this:

public booleanExpression = this.RULE("booleanExpression", () => {
  this.SUBRULE(this.booleanTerm);
  this.MANY(() => {
    this.CONSUME(RuleLexer.Or);
    this.SUBRULE2(this.booleanTerm);
  });
});

public additionExpression = this.RULE("additionExpression", () => {
  this.SUBRULE(this.multiplicationExpression, { LABEL: "lhs" });
  this.MANY(() => {
    this.CONSUME(AdditionOperator);
    this.SUBRULE2(this.multiplicationExpression, { LABEL: "rhs" });
  });
});

(actually think that rule has a bug in it, just noticed 🤣)

So if additionExpression is unterminated, why is the preceding comparison also not unterminated?

public comparison = this.RULE("comparison", () => {
  this.SUBRULE(this.additionExpression);
  this.SUBRULE(this.comparisonOperator);
  this.SUBRULE2(this.additionExpression);
});

I'm just trying out lots of partial source snippets to see if I can get my head around what this object is telling me.

[msujew]

Hey @appocalypseltd,

If the rule at index 4 is unterminated, then why are the preceding rules also not indicated as unterminated?

Every item in the ruleStack/occurrenceStack array is not terminated. Them being in the array is the indicator.

So if additionExpression is unterminated, why is the preceding comparison also not unterminated?

I believe you're under a misconception there. The value in the occurenceStack indicates the SUBRULE<number> call. Nothing else. In fact, comparison is not terminated yet, given that it's in the ruleStack array.

[appocalypseltd]

Actually, that can't be right, or there would be no zeros...

[msujew]

So the 2 pertaining to the additionExpression indicates that the preceding rule (in my case comparison) invokes this rule, additionExpresison, twice?

Well, no, not really. It a consequence of invoking it twice, but not the real reason. The 2 indicates that it is being parsed by the SUBRULE2 call. If you were to use SUBRULE5 instead, it'd be 5. Just using SUBRULE leads to a 0 in the occurenceStack.

[appocalypseltd]

Ahahaha, I see. Then one question outstanding...

I can understand the zeros, since they are the only occurrence of that subrule invocation, but why the 1 at the beginning? Is that an edge case since that's where the parser was kicked off with ruleParser.booleanExpression()?

[msujew]

Yeah, that might be. I'm not too familiar with the content assist API of Chevrotain, since we wrote our own instead for Langium. Chevrotain's API has a few issues, the most problematic being that it doesn't work in the presence of parser errors.

[appocalypseltd]

Oo I shall have a good read around Langium. Another distraction!

Thank you for your help. This concludes the interrogation, I don't think that API fits my purpose.