Undefined behaviour in re2c lexers caused by unspecified default rule.

Description

Hi! [re2c author here] It came to my attention in this nixos thread that re2c emits a some serious warnings on the current code for the lexers, e.g. this one. When running re2c with -W (or version >= 4.0) you should see this:

sapi/phpdbg/phpdbg_lexer.l:80:20: warning: escape has no effect: '\.' [-Wuseless-escape]
sapi/phpdbg/phpdbg_lexer.l:64:0: warning: control flow in condition 'NORMAL' is undefined for strings that match
        '\x22 [\x0\xA]'
        '\x27 [\x0\xA]'
        '\x22 \x22 [\x0\x9-\xA\xD\x20\x23]'
...

In particular, -Wundefined-control-flow warning indicates really serious issues; it is documented here. The fix should be simple: add default rule <*> * { /* error handling / abort / etc. */ }. Ideally also simplify some too-complex constructs like GENERIC_ID which make it hard to understand what's going on.

Some other warnings report unreachable rules, which is also not good (there's some rule that you think is doing something, but it's not).

I'm happy to help with further investigation and fixing these issues. I do recommend enabling the warnings from now on for all lexers - new bugs keep crawling in as the lexer code changes.

PHP Version

HEAD

Operating System

NixOS

Commencez par sapi/phpdbg/phpdbg_lexer.l et exécutez re2c avec -W pour reproduire les avertissements undefined-control-flow, useless-escape et unreachable-rule. Inspectez les règles du lexer ainsi que les autres lexers concernés, puis vérifiez que la gestion par défaut est définie et que les avertissements pertinents sont résolus ou traités intentionnellement.