phpmyadmin / phpmyadmin/sql-parser
CTE parsing fails when name is a non-reserved keyword
Nobody has claimed this yet.
- Dominant language
- PHP
- Stars
- 485
- Forks
- 119
- PR merge metrics
- No merged PRs in 30d
Description
The parser rejects valid CTEs whose names are non-reserved keywords, throwing:
The name of the CTE was expected.
MySQL has a large set of keywords that are explicitly not reserved (no (R) marker in the official MySQL 9.7 keyword list^1). Non-reserved keywords are explicitly permitted as unquoted identifiers, including CTE names. MySQL executes these queries without error^2.
Reproducer
Unquoted:
vendor/bin/sql-parser --lint --query "WITH data AS (SELECT 1) SELECT * FROM data"
#1: The name of the CTE was expected. (near "data" at position 5)
#2: Unexpected end of the WITH CTE. (near "data" at position 5)
#3: Unrecognized statement type. (near "data" at position 5)
Backtick-quoted (which should unconditionally allow any keyword as an identifier):
vendor/bin/sql-parser --lint --query "WITH `data` AS (SELECT 1) SELECT * FROM `data`"
#1: The name of the CTE was expected. (near "`data`" at position 5)
#2: Unexpected end of the WITH CTE. (near "`data`" at position 5)
#3: Unexpected beginning of statement. (near "`data`" at position 5)
#4: Unrecognized statement type. (near "AS" at position 12)
Expected behaviour
The query parses successfully for any non-reserved keyword used as a CTE name, whether unquoted or backtick-quoted.
Actual behaviour
Parser throws errors in both cases. The root cause appears to be that the lexer classifies all keywords as keyword tokens unconditionally, rather than resolving them contextually. After WITH, the parser expects an identifier token; receiving a keyword token instead, it fails, even though MySQL allows non-reserved keywords in identifier position without quoting, and allows any keyword in identifier position when backtick-quoted.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by running the two vendor/bin/sql-parser --lint --query reproducers and trace how the lexer and parser handle the CTE name after WITH, including backtick-quoted names. Done means both queries parse successfully, including non-reserved keyword names and quoted keyword names, without the reported errors.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- mysql, php, sql
- Domain
- compilers, databases
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 65/100