lotobo/regex-tools
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 6 Wiki Activity

feat: add multi-engine regex flavour support

Browse Source
This commit is contained in:
Albrecht Degering
2026-07-27 11:43:51 +02:00
parent 7079cde15f
commit 7f3a91ad37
340 changed files with 643286 additions and 483 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
docs/PARSER_PROFILES.md
View File

@@ -6,20 +6,38 @@ Regex Tools has one public profile:
community
```
It uses only reviewed open-source dependencies and builds without private
registry credentials. ECMAScript uses regexpp 4.12.2. PCRE2 uses an
application-owned lexical provider for partial explanations and the actual
bundled PCRE2 compiler for authoritative acceptance. Python and Java use
application-owned lexical providers for partial explanations and
flavour-native replacement tokens; bundled CPython `re` and TeaVM
`java.util.regex` respectively remain authoritative for compilation.
The Java provider follows the bundled TeaVM 0.15.0 replacement subset:
single-digit `$n` is executable, `$12` is `$1` followed by literal `2`, and
Java SE `${name}` references are surfaced as a compatibility limitation.
It uses reviewed open-source dependencies and builds without private registry
credentials. No licensed parser is installed, referenced or required.
JavaScript/ECMAScript uses regexpp 4.12.2 for ECMAScript 2025 syntax. That is
the only current complete grammar-backed normalized syntax tree. All other
profiles use application-owned, deliberately partial lexical providers:
| Profile | Authoritative compiler | Lexical-provider boundary |
| ----------------------- | --------------------------------------------- | -------------------------------------------------------------------------------------------------------------------- |
| PCRE2 | Standalone PCRE2 10.47 | Common captures and PCRE2 constructs; no complete structural grammar or branch-reset numbering |
| PHP | PHP 8.5.8 `preg` / PCRE2 10.44 | Common PCRE2/PHP capture forms and numbered replacement references; no delimiter or complete `preg` template model |
| Perl | Perl 5.28.1 via legacy WebPerl 0.09-beta | Common captures and fixed bridge references only; embedded code and complete Perl interpolation are unavailable |
| Python | CPython 3.14.2 `re` | Common Python captures and replacement references |
| Ruby | CRuby 4.0.0 `Regexp` | Common Onigmo capture forms and `String#gsub` references |
| Java | TeaVM 0.15.0 `java.util.regex` | Common Java captures and TeaVM replacement subset; not OpenJDK |
| C++ | Emscripten 6.0.4 libc++ `std::wregex` | Capture discovery only for the default modified ECMAScript grammar; alternate grammars remain compiler-authoritative |
| Go | Go 1.26.5 standard-library `regexp` | Common RE2 capture forms and `regexp.Expand` references |
| Rust | Rust `regex` crate 1.13.1 | Common capture forms and `Captures::expand` references |
| .NET | .NET 10.0.10 `System.Text.RegularExpressions` | Common captures/substitutions; no balancing-group or capture-history structure |
| Scala/JVM compatibility | Same TeaVM 0.15.0 Java engine | Java-compatible syntax only; no Scala runtime, wrapper or extractor model |
The Java and Scala providers follow the bundled TeaVM 0.15.0 replacement
subset: single-digit `$n` is executable, `$12` is `$1` followed by literal
`2`, and `${name}` is surfaced as a compatibility limitation. The Scala
provider deliberately shares that implementation and says so.
Every partial provider exposes its gaps through coverage metadata. Actual
acceptance, capture numbering and replacement output always come from the
selected runtime. ECMAScript analysis, case generation and formatting are not
applied to these partial trees.
The product deliberately does not implement or reference an `@r101/parser`
profile. No commercial package alias, import, tarball, credential, licence key
or private CI path exists. Additional flavour syntax is implemented
incrementally as open-source providers behind the same application-owned
interface. Partial coverage is exposed as metadata and never presented as a
complete parser.
or private CI path exists. Further syntax coverage can grow behind the same
application-owned interface without changing the runtime identity.