Relax parsing requirements to allow hyphens and periods

[ColtonPayne]

Summary

Relaxes identifier/entity validation across the fact and rule parsers, splits the rule parser's single identifier regex into separate predicate/component regexes, and aligns fact and rule component rules so any entity valid as a fact can also appear as a grounded atom in a rule.

Final regex layout

Role	Regex	Leading digit?	. / -	@
Fact predicate	[a-zA-Z_][a-zA-Z0-9_.\-]*	no	yes	no
Fact component	[a-zA-Z0-9_][a-zA-Z0-9_.@\-]*	yes	yes	yes
Rule predicate	[a-zA-Z_][a-zA-Z0-9_.\-]*	no	yes	no
Rule component	[a-zA-Z0-9_][a-zA-Z0-9_.@\-]*	yes	yes	yes

Fact and rule components share the same regex. Predicates are stricter than components (no leading digit, no @).

Changes

pyreason/scripts/utils/fact_parser.py

• Added _PREDICATE_RE and _COMPONENT_RE.
• Added _validate_predicate() and _validate_component() helpers.
• Replaces inline predicate checks and ad-hoc (, ), : bans on components — covered by the regex.
• Node components are now validated (previously only edge components had any validation, and only an empty-check).

pyreason/scripts/utils/rule_parser.py

• Split _IDENTIFIER_RE into _PREDICATE_RE (identifiers) and _COMPONENT_RE (entities — matches fact_parser._COMPONENT_RE).
• _validate_component_name now uses _COMPONENT_RE; removed the now-unreachable digit-start error branch.
• Error messages updated to reflect the new allowed sets.

tests/unit/dont_disable_jit/test_rule_parser.py

• Negative-case inputs switched from - (now valid) to ! (still invalid) in three tests.
• Removed test_head_variable_starts_with_digit — digit-leading rule components are now valid by design.

tests/api_tests/test_pyreason_reasoning.py

• 13 facts changed from person("A") to person(A) — quoted entity names are no longer valid under the stricter component allowlist.

Test plan

• Parser unit tests pass (168/168)
• All 11 previously-failing API reasoning tests pass after the quoted→unquoted fix
• Smoke tests confirm:
  • Facts with hyphens/periods/digits/@ parse: has-vuln(node-1), cve.2024.1234(host.a, host.b), person(123), user(alice@example.com)
  • Rule components accept the same entity shapes: p(1X) <- b(1X), p(a@b) <- q(a@b)
  • Invalid chars (!, ", leading @, ~, /) are still rejected

Note

The branch name and first commit mention "spaces" but spaces are not allowed — only -, ., and @ were added across the changes.

[jaikrishnap98]

Looks good to me. I ran the tests and aslo verified the changes