From eb1015cdfbf78837453727113f532a961b193925 Mon Sep 17 00:00:00 2001
From: Maxime Coste <mawww@kakoune.org>
Date: Tue, 26 Sep 2017 16:44:30 +0900
Subject: [PATCH] Regex: whenever Kakoune compiles a regex, pass it to the
 custom impl as well

That way we can see which features are missing.
---
 src/regex.cc      |  5 ++++-
 src/regex_impl.cc | 28 +++++++++++++++++++++-------
 src/regex_impl.hh |  9 +++++++++
 3 files changed, 34 insertions(+), 8 deletions(-)

diff --git a/src/regex.cc b/src/regex.cc
index 1658af76..77e01d89 100644
--- a/src/regex.cc
+++ b/src/regex.cc
@@ -1,6 +1,7 @@
 #include "regex.hh"
 
 #include "exception.hh"
+#include "regex_impl.hh"
 
 namespace Kakoune
 {
@@ -9,7 +10,9 @@ using Utf8It = RegexUtf8It<const char*>;
 
 Regex::Regex(StringView re, flag_type flags) try
     : RegexBase{Utf8It{re.begin(), re}, Utf8It{re.end(), re}, flags}, m_str{re.str()}
-{} catch (std::runtime_error& err) { throw regex_error(err.what()); }
+{
+    validate_regex(re);
+} catch (std::runtime_error& err) { throw regex_error(err.what()); }
 
 String option_to_string(const Regex& re)
 {
diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index ffc8edc8..8a61b760 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -8,6 +8,8 @@
 #include "exception.hh"
 #include "array_view.hh"
 
+#include "buffer_utils.hh"
+
 namespace Kakoune
 {
 
@@ -114,13 +116,6 @@ AstNodePtr make_ast_node(Op op, Codepoint value = -1,
 // standard, although the syntax is not fully compatible.
 struct Parser
 {
-    struct InvalidPolicy
-    {
-        Codepoint operator()(Codepoint cp) { throw runtime_error{"Invalid utf8 in regex"}; }
-    };
-
-    using Iterator = utf8::iterator<const char*, Codepoint, int, InvalidPolicy>;
-
     static ParsedRegex parse(StringView re)
     {
         ParsedRegex res;
@@ -131,6 +126,13 @@ struct Parser
     }
 
 private:
+    struct InvalidPolicy
+    {
+        Codepoint operator()(Codepoint cp) { throw runtime_error{"Invalid utf8 in regex"}; }
+    };
+
+    using Iterator = utf8::iterator<const char*, Codepoint, int, InvalidPolicy>;
+
     static AstNodePtr disjunction(ParsedRegex& parsed_regex, Iterator& pos, Iterator end, unsigned capture = -1)
     {
         AstNodePtr node = alternative(parsed_regex, pos, end);
@@ -790,6 +792,18 @@ struct ThreadedRegexVM
     Vector<const char*> m_captures;
 };
 
+void validate_regex(StringView re)
+{
+    try
+    {
+        RegexCompiler::Parser::parse(re);
+    }
+    catch (runtime_error& err)
+    {
+        write_to_debug_buffer(format("regex-impl: <<{}>> failed to parse: {}", re, err.what()));
+    }
+}
+
 auto test_regex = UnitTest{[]{
     {
         auto program = RegexCompiler::compile(R"(a*b)");
diff --git a/src/regex_impl.hh b/src/regex_impl.hh
index 24de6907..8fafcacd 100644
--- a/src/regex_impl.hh
+++ b/src/regex_impl.hh
@@ -1,4 +1,13 @@
 #ifndef regex_impl_hh_INCLUDED
 #define regex_impl_hh_INCLUDED
 
+namespace Kakoune
+{
+
+class StringView;
+
+void validate_regex(StringView re);
+
+}
+
 #endif // regex_impl_hh_INCLUDED