Regex Across Languages and Tools

Regular expression syntax looks similar across languages, but important differences exist. A pattern that works in one language may behave differently — or fail entirely — in another.

**Flavor families:** - **PCRE** (Perl Compatible): PHP and many command-line tools (grep -P). The most feature-rich flavor. Ruby uses the Oniguruma engine, which is PCRE-like but distinct. - **JavaScript/ECMAScript**: Supported in all browsers and Node.js. Has added features steadily (named groups in ES2018, lookbehind in ES2018, \p{...} with u flag). - **Python re**: Similar to PCRE but with distinct syntax for named groups ((?P<name>...)) and no support for \p{...} property escapes (use the third-party regex module). - **Java java.util.regex**: PCRE-like but with some quirks — backslashes must be double-escaped in string literals (\\d instead of \d). - **Go regexp**: Uses RE2 engine, which guarantees linear-time matching. Does NOT support backreferences or lookaheads/lookbehinds. This is a deliberate trade-off for performance safety. - **.NET**: The most feature-rich built-in engine — supports balancing groups, variable-length lookbehinds, and more.

**Key differences to watch for:**

| Feature | JS | Python | Java | Go | PCRE | |---|---|---|---|---|---| | Named groups | (?<n>...) | (?P<n>...) | (?<n>...) | (?P<n>...) | Both | | Lookbehind | Fixed + variable | Fixed-length only | Variable (caveats in Java 9+) | Not supported | Fixed + variable | | \p{L} Unicode | With u flag | regex module only | Yes | Yes | Yes | | Backreferences | \1 or \k<n> | \1 or (?P=n) | \1 | Not supported | \1 or \k<n> | | Possessive x++ | Not supported | Yes (3.11+) | Yes | Not supported | Yes | | Atomic groups | Not supported | Yes (3.11+) | Yes | Not supported | Yes |

**Command-line tools:** - grep — basic regex by default; use grep -E for extended regex (ERE) or grep -P for PCRE (on GNU grep) - sed — basic regex by default; sed -E for ERE. No lookaheads or named groups. - **VS Code Find/Replace** — uses JavaScript regex syntax with the g and m flags implicitly enabled

**Practical tip:** If your pattern must work across multiple languages, stick to these universally supported features: character classes ([...], \d, \w, \s), quantifiers (+, *, ?, {n,m}), anchors (^, $), alternation (|), and basic groups (...). Avoid backreferences and lookarounds if Go support is required.