From b55e1a122e5e93c84f3712bb576ef179213231cd Mon Sep 17 00:00:00 2001
From: Arne Welzel <arne.welzel@corelight.com>
Date: Wed, 1 Nov 2023 11:51:25 +0100
Subject: [PATCH] Reuse CompileSet() instead of || string formatting

---
 src/RE.cc      | 19 +++++++++++++------
 src/RE.h       |  1 -
 src/re-parse.y | 18 +-----------------
 src/re-scan.l  |  4 ----
 4 files changed, 14 insertions(+), 28 deletions(-)

diff --git a/src/RE.cc b/src/RE.cc
index 87b69d29a2..70bb71b055 100644
--- a/src/RE.cc
+++ b/src/RE.cc
@@ -172,6 +172,10 @@ bool Specific_RE_Matcher::CompileSet(const string_list& set, const int_list& idx
     dfa = new DFA_Machine(nfa, EC());
     ecs = EC()->EquivClasses();
 
+    // dfa took ownership
+    Unref(nfa);
+    nfa = nullptr;
+
     return true;
 }
 
@@ -455,13 +459,16 @@ bool RE_Matcher::Compile(bool lazy) { return re_anywhere->Compile(lazy) && re_ex
 RE_DisjunctiveMatcher::RE_DisjunctiveMatcher(const std::vector<const RE_Matcher*>& REs) {
     matcher = std::make_unique<detail::Specific_RE_Matcher>(detail::MATCH_EXACTLY);
 
-    std::string disjunction;
-    for ( auto re : REs )
-        disjunction += std::string("||") + re->PatternText();
+    zeek::detail::string_list sl;
+    zeek::detail::int_list il;
 
-    matcher->SetPat(disjunction.c_str());
-    auto status = matcher->Compile();
-    ASSERT(status);
+    for ( const auto* re : REs ) {
+        sl.push_back(const_cast<char*>(re->PatternText()));
+        il.push_back(sl.size());
+    }
+
+    if ( ! matcher->CompileSet(sl, il) )
+        reporter->FatalError("failed compile set for disjunctive matcher");
 }
 
 void RE_DisjunctiveMatcher::Match(const String* s, std::vector<int>& matches) {
diff --git a/src/RE.h b/src/RE.h
index 8d7b28da30..52b446a306 100644
--- a/src/RE.h
+++ b/src/RE.h
@@ -36,7 +36,6 @@ extern CCL* curr_ccl;
 extern NFA_Machine* nfa;
 extern Specific_RE_Matcher* rem;
 extern const char* RE_parse_input;
-extern int RE_accept_num;
 
 extern int clower(int);
 extern void synerr(const char str[]);
diff --git a/src/re-parse.y b/src/re-parse.y
index 2ee6bec9e2..2d6672df8d 100644
--- a/src/re-parse.y
+++ b/src/re-parse.y
@@ -21,7 +21,6 @@ void yyerror(const char msg[]);
 %}
 
 %token TOK_CHAR TOK_NUMBER TOK_CCL TOK_CCE TOK_CASE_INSENSITIVE TOK_SINGLE_LINE
-%token TOK_DISJUNCTION
 
 %union {
 	int int_val;
@@ -33,7 +32,7 @@ void yyerror(const char msg[]);
 %type <int_val> TOK_CHAR TOK_NUMBER
 %type <cce_val> TOK_CCE
 %type <ccl_val> TOK_CCL ccl full_ccl
-%type <mach_val> re singleton series string disjunction
+%type <mach_val> re singleton series string
 
 %destructor { delete $$; } <mach_val>
 
@@ -41,9 +40,6 @@ void yyerror(const char msg[]);
 flexrule	:	re
 			{ $1->AddAccept(1); zeek::detail::nfa = $1; }
 
-		|  disjunction
-			{ zeek::detail::nfa = $1; }
-
 		|  error
 			{ return 1; }
 		;
@@ -55,18 +51,6 @@ re		:  re '|' series
 			{ $$ = new zeek::detail::NFA_Machine(new zeek::detail::EpsilonState()); }
 		;
 
-disjunction	:  disjunction TOK_DISJUNCTION re
-			{
-			$3->AddAccept(++zeek::detail::RE_accept_num);
-			$$ = zeek::detail::make_alternate($1, $3);
-			}
-		|  TOK_DISJUNCTION re
-			{
-			$2->AddAccept(++zeek::detail::RE_accept_num);
-			$$ = $2;
-			}
-		;
-
 series		:  series singleton
 			{ $1->AppendMachine($2); $$ = $1; }
 		|  singleton
diff --git a/src/re-scan.l b/src/re-scan.l
index 7df4665640..f382393477 100644
--- a/src/re-scan.l
+++ b/src/re-scan.l
@@ -23,7 +23,6 @@
 #include "re-parse.h"
 
 const char* zeek::detail::RE_parse_input = nullptr;
-int zeek::detail::RE_accept_num = 0;
 
 #define RET_CCE(func) \
 	BEGIN(SC_CCL); \
@@ -144,8 +143,6 @@ CCL_EXPR	("[:"[[:alpha:]]+":]")
 				}
 			}
 
-	"||"		return TOK_DISJUNCTION;
-
 	[|*+?.(){}]	return yytext[0];
 	.		yylval.int_val = yytext[0]; return TOK_CHAR;
 	\n		return 0;	// treat as end of pattern
@@ -240,7 +237,6 @@ YY_BUFFER_STATE RE_buf;
 void RE_set_input(const char* str)
 	{
 	zeek::detail::RE_parse_input = str;
-	zeek::detail::RE_accept_num = 0;
 	RE_buf = yy_scan_string(str);
 	}