Added support for psql meta commands to Postgres

[WittierDinosaur]

Brief summary of the change made

This PR adds support for psql meta commands. Psql commands here start with , followed by at least one alphabet character, and are then terminated by newline or \. The File and Statement Segments had to be edited to accommodate for this new functionality.

Fixes #1327

Are there any other side effects of this change that we should be aware of?

No, Postgres-specific, and all dialect tests are passing

Pull Request checklist

Test cases added for psql meta commands, and sql commands in the same file (or same line for \)
Documented regex in file

[codecov]

Codecov Report

Merging #1423 (6676f12) into main (ce4e5a3) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main     #1423   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files          127       127           
  Lines         8603      8604    +1     
=========================================
+ Hits          8603      8604    +1     

Impacted Files	Coverage Δ
src/sqlfluff/dialects/dialect_postgres.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ce4e5a3...6676f12. Read the comment docs.

[tunetheweb]

I was curious how comments were handled as seems like a similar use case.

From a quick look it looks like it basically trims the details as part of the lexing, so those symbols aren’t sent to the parser and are effectively ignored.

The downside would be that you can’t check the grammar of meta commands (as they will effectively not exist to the parser), but doesn’t look like you are doing that here anyway so for now that wouldn’t matter. But your solution at least opens that possibility later.

Any thoughts on whether following the comments process would be the better solution than what’s proposed here?

[WittierDinosaur]

Yeah, that might be easier. The only thing I think could cause an issue, is:
\echo "thing" \\ SELECT 1;
Comments are easy to ignore because they can't be terminated mid-line, psql meta commands can. We would just need to make sure the whitespace is preserved during linting, if the command gets ignored in parsing. I could add an optional \s after the \\?

[tunetheweb]

Comments are easy to ignore because they can't be terminated mid-line, psql meta commands can.

Sure they can, with /*...*/ syntax:

SELECT *
/* this is our main table */ FROM table 1

[WittierDinosaur]

Good point. I've updated meta commands to function like comments.

[tunetheweb]

Good point. I've updated meta commands to function like comments.

Wow that seems almost too simple now!!! I took ages trying to hunt for the code to figure out where the actual magic was happening - but guess it's all handled just by calling it a CommentSegment?

[tunetheweb]

LGTM! One small suggestion.