aboutsummaryrefslogtreecommitdiffhomepage
path: root/docs
diff options
context:
space:
mode:
authornsfisis <nsfisis@gmail.com>2026-06-14 15:02:44 +0900
committernsfisis <nsfisis@gmail.com>2026-06-14 15:03:24 +0900
commite71e24b5b5cacb93aab3a62e809414fbe9c8a438 (patch)
tree8d78b598f3f6ab9e766350ebb415bb752454f8c0 /docs
parent430f59c1938b9e5da381365172ab788b54895ffc (diff)
downloadphp-shirabe-e71e24b5b5cacb93aab3a62e809414fbe9c8a438.tar.gz
php-shirabe-e71e24b5b5cacb93aab3a62e809414fbe9c8a438.tar.zst
php-shirabe-e71e24b5b5cacb93aab3a62e809414fbe9c8a438.zip
docs(regex): add regex porting rules
Document the conventions for porting PCRE patterns to the regex crate (no PCRE crate, panic on compile failure, dropping performance-only possessive quantifiers, ad-hoc compatibility comments). Apply the rule to Platform::expand_path by rewriting its conditional subpattern as an explicit alternation, which the regex crate can compile. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'docs')
-rw-r--r--docs/dev/regex-porting.md39
1 files changed, 39 insertions, 0 deletions
diff --git a/docs/dev/regex-porting.md b/docs/dev/regex-porting.md
new file mode 100644
index 0000000..4503304
--- /dev/null
+++ b/docs/dev/regex-porting.md
@@ -0,0 +1,39 @@
+# Regex porting
+
+Composer uses PHP's `preg_*` functions (backed by PCRE). Shirabe ports these to the Rust
+[`regex`](https://docs.rs/regex) crate. Because PCRE and the `regex` crate have different feature
+sets, follow the rules below when porting.
+
+## Do not introduce a PCRE crate
+
+The `regex` crate is not based on PCRE library and does not support some features that PCRE does.
+
+Even when such a feature is needed, do not introduce a PCRE-family crate such as `pcre2`. Either
+rewrite the pattern into a form the `regex` crate can express, or decompose it into hand-written
+logic that does not use a regular expression.
+
+## Panic on pattern compile failure
+
+When a pattern fails to compile, `panic!`. Do not wrap it in a `Result` and propagate it to the
+caller.
+
+This reflects a design assumption that Composer-derived patterns always compile successfully at
+runtime: a compile failure is not a recoverable error but a programming mistake (or a porting bug).
+
+## Do not port performance-only possessive quantifiers
+
+When a PCRE possessive quantifier (`++`, `*+`, `?+`, etc.) is used solely to suppress backtracking
+for performance, replace it with the plain quantifier (`+`, `*`, `?`).
+
+The `regex` crate never backtracks, so it has no possessive quantifiers and has no need for them.
+
+## Port other unsupported features ad hoc
+
+For any other PCRE feature the `regex` crate does not support (conditional subpatterns,
+backreferences, etc.), port it case by case. When you do, make the transformation explicit with a
+comment in the following form:
+
+```
+// Regex pattern compatibility:
+// <description>
+```