From 5709f540811f80fc138fb0e084a3e5dd633b1873 Mon Sep 17 00:00:00 2001
From: SFG545 <khyrenjoseph253@gmail.com>
Date: Fri, 10 Jul 2026 18:13:27 -0500
Subject: [PATCH] Validate regular expressions before reading input

---
 meson.build                         |  9 ++++++++-
 scripts/test-pipe-regex-fallback.sh | 22 ++++++++++++++++++++++
 src/Compiler.cpp                    | 23 +++++++++++++++++++++++
 src/Options.cpp                     |  2 +-
 4 files changed, 54 insertions(+), 2 deletions(-)
 create mode 100755 scripts/test-pipe-regex-fallback.sh

diff --git a/meson.build b/meson.build
index be418c8..8552b5b 100644
--- a/meson.build
+++ b/meson.build
@@ -1,7 +1,7 @@
 project(
   'sedpp',
   'cpp',
-  version: '1.0.0',
+  version: '1.0.1',
   default_options: [
     'cpp_std=c++20',
     'warning_level=3',
@@ -24,3 +24,10 @@ test(
   depends: sed_exe,
   timeout: 240,
 )
+
+test(
+  'pipe-regex-fallback',
+  find_program('scripts/test-pipe-regex-fallback.sh'),
+  args: [sed_exe.full_path()],
+  depends: sed_exe,
+)
diff --git a/scripts/test-pipe-regex-fallback.sh b/scripts/test-pipe-regex-fallback.sh
new file mode 100755
index 0000000..2205d0f
--- /dev/null
+++ b/scripts/test-pipe-regex-fallback.sh
@@ -0,0 +1,22 @@
+#!/bin/sh
+
+set -eu
+
+sed_bin=$1
+
+# Limine uses this portability probe in gensyms.sh.  The first sed must reject
+# the unsupported word-boundary character classes without consuming the pipe,
+# leaving the input available to the GNU-compatible fallback expression.
+actual="$({
+    printf '%s\n' \
+        '00000000 l d .text discarded' \
+        '00000010 g F .text retained'
+} | ("$sed_bin" '/[[:<:]]d[[:>:]]/d' 2>/dev/null ||
+     "$sed_bin" '/\bd\b/d'))"
+
+expected='00000010 g F .text retained'
+
+if [ "$actual" != "$expected" ]; then
+    printf 'expected: %s\nactual:   %s\n' "$expected" "$actual" >&2
+    exit 1
+fi
diff --git a/src/Compiler.cpp b/src/Compiler.cpp
index 884b8c2..2c3f76c 100644
--- a/src/Compiler.cpp
+++ b/src/Compiler.cpp
@@ -1,4 +1,5 @@
 #include "sedpp/Compiler.h"
+#include "sedpp/Regex.h"
 
 #include <cctype>
 #include <iostream>
@@ -103,6 +104,28 @@ Program Compiler::compile(const std::vector<std::string> &scripts) {
   if (braceDepth_ > 0) {
     throw std::runtime_error("-e expression #1, char 0: unmatched '{'");
   }
+
+  // Validate every literal regular expression before Runner opens or reads an
+  // input stream.  Besides matching GNU sed's diagnostic timing, this matters
+  // for portability probes such as `(sed non_gnu_re || sed fallback_re)`: the
+  // rejected first expression must leave the shared pipe untouched so the
+  // fallback process can consume it.  Empty expressions are runtime references
+  // to the previously used regex and therefore cannot be compiled here.
+  auto validateAddress = [this](const std::optional<Address> &address) {
+    if (address && address->kind == AddressKind::Regex &&
+        !address->text.empty()) {
+      Regex regex(address->text, options_);
+    }
+  };
+  for (const Command &command : program_.commands) {
+    validateAddress(command.firstAddress);
+    validateAddress(command.secondAddress);
+    if (command.substitute && !command.substitute->pattern.empty()) {
+      Regex regex(command.substitute->pattern, options_,
+                  command.substitute->ignoreCase,
+                  command.substitute->multiline);
+    }
+  }
   return program_;
 }
 
diff --git a/src/Options.cpp b/src/Options.cpp
index ba00cbc..d662999 100644
--- a/src/Options.cpp
+++ b/src/Options.cpp
@@ -7,7 +7,7 @@
 #include <cstdlib>
 
 namespace sedpp {
-int SEDPP_VERSION = 1;
+constexpr const char *SEDPP_VERSION = "1.0.1";
 
 void OptionParser::printVersion() {
   // Version output intentionally names this implementation, not GNU sed.