fix(sqlparser): array of struct of struct

[xiangjinwu]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

When parsing >>[], whose token stream contains [ShiftRight, LBracket, RBracket], the parser first consumes ShiftRight and sets internal state remaining_close = true. At this stage, it shall consider the next token as effectively Gt rather than peeking at LBracket from the stream. In the latter case it would be wrongly parsed as >[]>

dot

Limitations:

• Ideally all places peeking from stream need to check remaining_close first as if it was the next token. But I cannot find a good way to ensure such a virtual stream is followed (mutating input stream is not an option as it breaks backtrack and location reporting). The current fix only applies to array syntax [] as suffix. Hopefully no future syntax extensions hit the same problem.

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

If this PR includes changes that directly affect users or other significant modifications relevant to the community, kindly draft a release note to provide a concise summary of these changes. Please prioritize highlighting the impact these changes will have on users.

[xxchan]

This test case seems to work in current version. Did I miss something?

dev=> CREATE TABLE t (params STRUCT<a STRUCT<b int>>[])
dev-> ;
CREATE_TABLE
dev=> describe t;
┌───────────────────┬─────────────────────┬───────────┬─────────────┐
│       Name        │        Type         │ Is Hidden │ Description │
├───────────────────┼─────────────────────┼───────────┼─────────────┤
│ params            │                     │ false     │ ∅           │
│ params.a          │ struct<b integer>[] │ false     │ ∅           │
│ _row_id           │ serial              │ true      │ ∅           │
│ primary key       │ _row_id             │ ∅         │ ∅           │
│ distribution key  │ _row_id             │ ∅         │ ∅           │
│ table description │ t                   │ ∅         │ ∅           │
└───────────────────┴─────────────────────┴───────────┴─────────────┘
(6 rows)

[xxchan]

Oh, it's wrong

dev=> show create table t;
┌──────────┬───────────────────────────────────────────────────┐
│   Name   │                    Create Sql                     │
├──────────┼───────────────────────────────────────────────────┤
│ public.t │ CREATE TABLE t (params STRUCT<a STRUCT<b INT>[]>) │
└──────────┴───────────────────────────────────────────────────┘

[xxchan]

How did you find the bug? Is it related with other bugs?

[xiangjinwu]

See the linked issue. It was causing a schema mismatch panic when sink into table.

[xxchan]

Oh, I didn't notice there's linked issue since there's no fix #xx in body 😄

[TennyZhuang]

LGTM

[xxchan]

LGTM for the fix. But agree that not sure whether it's universal.

[xxchan]

dot

Curious how did you generated this graph and what did you use it for? 👀

[xiangjinwu]

dot

Curious how did you generated this graph and what did you use it for? 👀

• Generated manually ...
• Used it to understand the indirect recursion, specially how parsing of >>[] became >[]>.