From 0e911d64e71a85d3958689debff0ccbee4c3891a Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Mon, 28 Oct 2024 12:52:11 +0200
Subject: Make .*search() functions not to match empty substrings in non empty
 strings

---
 libbuild2/functions-regex.cxx   | 31 ++++++++++++++--
 tests/function/regex/testscript | 81 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 108 insertions(+), 4 deletions(-)

diff --git a/libbuild2/functions-regex.cxx b/libbuild2/functions-regex.cxx
index c46f6f5..cac4e54 100644
--- a/libbuild2/functions-regex.cxx
+++ b/libbuild2/functions-regex.cxx
@@ -138,12 +138,19 @@ namespace build2
     //
     string s (to_string (move (v)));
 
+    // Match flags.
+    //
+    regex_constants::match_flag_type mf (regex_constants::match_default);
+
+    if (!s.empty ())
+      mf |= regex_constants::match_not_null;
+
     if (!match && !subs)
-      return value (regex_search (s, rge)); // Return boolean value.
+      return value (regex_search (s, rge, mf)); // Return boolean value.
 
     match_results<string::const_iterator> m;
 
-    if (regex_search (s, m, rge))
+    if (regex_search (s, m, rge, mf))
     {
       assert (!m.empty ());
 
@@ -483,7 +490,16 @@ namespace build2
 
     for (auto& n: ns)
     {
-      if (regex_search (convert<string> (move (n)), rge))
+      string s (convert<string> (move (n)));
+
+      // Match flags.
+      //
+      regex_constants::match_flag_type mf (regex_constants::match_default);
+
+      if (!s.empty ())
+        mf |= regex_constants::match_not_null;
+
+      if (regex_search (s, rge, mf))
         return true;
     }
 
@@ -516,7 +532,14 @@ namespace build2
       bool s (n.simple ());
       string v (convert<string> (s ? move (n) : name (n)));
 
-      if (regex_search (v, rge) == matching)
+      // Match flags.
+      //
+      regex_constants::match_flag_type mf (regex_constants::match_default);
+
+      if (!v.empty ())
+        mf |= regex_constants::match_not_null;
+
+      if (regex_search (v, rge, mf) == matching)
         r.emplace_back (s ? name (move (v)) : move (n));
     }
 
diff --git a/tests/function/regex/testscript b/tests/function/regex/testscript
index 538bdab..5564bbd 100644
--- a/tests/function/regex/testscript
+++ b/tests/function/regex/testscript
@@ -366,6 +366,33 @@
       EOI
     }
   }
+
+  : empty-substring
+  :
+  : Note that the regex search-based replacement with the match_not_null flag
+  : is broken for older versions of libstdc++ and libc++.
+  :
+  if (($cxx.id != 'gcc'   || $cxx.version.major >= 7) && \
+      ($cxx.id != 'clang' || $cxx.version.major >= 6))
+  {
+    : empty
+    :
+    $* <<EOI >'true'
+    print $regex.search('', '.*')
+    EOI
+
+    : match
+    :
+    $* <<EOI >'true'
+    print $regex.search('a', 'a*')
+    EOI
+
+    : no-match
+    :
+    $* <<EOI >'false'
+    print $regex.search('aa', 'b*')
+    EOI
+  }
 }
 
 : split
@@ -576,6 +603,33 @@
     print $regex.find_search(Foo.cxx, 'f', icase)
     EOI
   }
+
+  : empty-substring
+  :
+  : Note that the regex search-based replacement with the match_not_null flag
+  : is broken for older versions of libstdc++ and libc++.
+  :
+  if (($cxx.id != 'gcc'   || $cxx.version.major >= 7) && \
+      ($cxx.id != 'clang' || $cxx.version.major >= 6))
+  {
+    : empty
+    :
+    $* <<EOI >'true'
+    print $regex.find_search('', '.*')
+    EOI
+
+    : match
+    :
+    $* <<EOI >'true'
+    print $regex.find_search('a', 'a*')
+    EOI
+
+    : no-match
+    :
+    $* <<EOI >'false'
+    print $regex.find_search('aa', 'b*')
+    EOI
+  }
 }
 
 : filter-search
@@ -607,6 +661,33 @@
     $* <<EOI >''
     print $regex.filter_search(-g, '-O')
     EOI
+
+    : empty-substring
+    :
+    : Note that the regex search-based replacement with the match_not_null flag
+    : is broken for older versions of libstdc++ and libc++.
+    :
+    if (($cxx.id != 'gcc'   || $cxx.version.major >= 7) && \
+        ($cxx.id != 'clang' || $cxx.version.major >= 6))
+    {
+      : empty
+      :
+      $* <<EOI >'{}'
+      print $regex.filter_search('', '.*')
+      EOI
+
+      : match
+      :
+      $* <<EOI >'a'
+      print $regex.filter_search('a', 'a*')
+      EOI
+
+      : no-match
+      :
+      $* <<EOI >''
+      print $regex.filter_search('aa', 'b*')
+      EOI
+    }
   }
 
   : filter-out
-- 
cgit v1.1