MATCH (A) MATCH (A {prop: 'val'} ) ... Fails when it should not.

[dehowef]

Describe the bug

There is unintended behavior when a user tries to MATCH a node with property constraints after already MATCHing a node with the same variable name.

How are you accessing AGE (Command line, driver, etc.)?

Command Line

What data setup do we need to do?

Have AGE running.

What is the necessary configuration info needed?

• N/A

What is the command that caused the error?

SELECT * FROM cypher('my_graph', $$ MATCH (a) MATCH (a {prop:'val'}) RETURN a $$ ) AS (a agtype);

postgres=# SELECT * FROM cypher('my_graph', $$ MATCH (a) MATCH (a {prop:'val'}) RETURN a $$ ) AS (a agtype);
2023-02-22 01:29:05.230 PST [59866] ERROR:  missing FROM-clause entry for table "a" at character 34
2023-02-22 01:29:05.230 PST [59866] STATEMENT:  SELECT * FROM cypher('pilots', $$ MATCH (a) MATCH (a {s:'as'}) RETURN a $$ ) AS (a agtype);
ERROR:  missing FROM-clause entry for table "a"
LINE 1: SELECT * FROM cypher('my_graph', $$ MATCH (a) MATCH (a {prop:'val'...

Expected behavior

The program should detect the variable and return the according result. The RTE is not being detected and the program is throwing an error.

Environment (please complete the following information):

• Version: PG11 Version 1.1.X, Fedora OS 36

Additional context
Add any other context about the problem here.

[dehowef]

The issue seems to lie in transform_cypher_node() in cypher_clause.c

Because of how previously transformed match clauses are stored in an RTE, they are undetectable by create_property_constraints(). All of previous clauses are pooled into an RTE with the alias "_', and the find_rte() logic responsible detecting previously declared variables cannot produce property_constraints.

[jrgemignani]

@dehowef I checked and this issue goes back to, at least, 1.0.0, that was as far back as I went. It is also in PG12 and PG13 versions. So, it isn't anything new and was likely just overlooked as a case.