Fix SWIG bindings SEGFAULT when .dl filename contains hyphens (souffle-lang#2573)

Author: qequ

src/include/souffle/SouffleInterface.h

@@ -21,6 +21,7 @@
 #include "souffle/SymbolTable.h"
 #include "souffle/datastructure/ConcurrentCache.h"
 #include "souffle/utility/MiscUtil.h"
+#include "souffle/utility/StringUtil.h"
 #include <algorithm>
 #include <cassert>
 #include <cstddef>
@@ -1134,7 +1135,10 @@ class ProgramFactory {
      * @return The new instance(SouffleProgram*), or null pointer if the instance not found
      */
     static SouffleProgram* newInstance(const std::string& name) {
-        ProgramFactory* factory = find(name);
+        // Normalize name to handle special characters (e.g., hyphens become underscores)
+        // This matches how factory names are generated from filenames in Synthesiser.cpp
+        std::string normalizedName = identifier(name);
+        ProgramFactory* factory = find(normalizedName);
         if (factory != nullptr) {
             return factory->newInstance();
         } else {

tests/swig/CMakeLists.txt

@@ -152,6 +152,7 @@ endfunction()
 souffle_positive_swig_test(dump_output COMPARE_STDOUT)
 souffle_positive_swig_test(family COMPARE_STDOUT)
 souffle_positive_swig_test(flights)
+souffle_positive_swig_test(hyphen-name)
 souffle_positive_swig_test(insert_for)
 souffle_positive_swig_test(movies)
 souffle_positive_swig_test(paths)

tests/swig/python/hyphen-name/driver.py

@@ -0,0 +1,32 @@
+"""
+Souffle - A Datalog Compiler
+Copyright (c) 2025, The Souffle Developers. All rights reserved
+Licensed under the Universal Permissive License v 1.0 as shown at:
+- https://opensource.org/licenses/UPL
+- <souffle root>/licenses/SOUFFLE-UPL.txt
+
+Test case for issue #2533: SWIG python bindings SEGFAULT if name of .dl file contains -
+
+The bug: when a .dl file contains a hyphen in its name (e.g., "hyphen-name.dl"),
+the identifier() function converts it to a valid C++ identifier by replacing
+hyphens with underscores ("hyphen_name"). However, users calling newInstance()
+expect to use the original filename with the hyphen.
+
+This test verifies that newInstance("hyphen-name") works correctly.
+"""
+
+import SwigInterface
+import sys
+
+# This should work with the original filename containing a hyphen
+# Before the fix, this would cause a SEGFAULT because:
+# 1. The program is registered in the factory as "hyphen_name" (with underscore)
+# 2. But newInstance("hyphen-name") looks for "hyphen-name" (with hyphen)
+# 3. The factory returns nullptr
+# 4. SWIGSouffleProgram tries to dereference nullptr -> SEGFAULT
+p = SwigInterface.newInstance('hyphen-name')
+p.loadAll(sys.argv[1])
+p.run()
+p.printAll('.')
+p.thisown = 1
+del p

tests/swig/python/hyphen-name/hyphen-name.dl

@@ -0,0 +1,19 @@
+// Souffle - A Datalog Compiler
+// Copyright (c) 2025, The Souffle Developers. All rights reserved
+// Licensed under the Universal Permissive License v 1.0 as shown at:
+// - https://opensource.org/licenses/UPL
+// - <souffle root>/licenses/SOUFFLE-UPL.txt
+
+// Test case for issue #2533: SWIG python bindings SEGFAULT if name of .dl file contains -
+
+.type Node <: symbol
+
+.decl edge(node1:Node, node2:Node)
+.decl path(node1:Node, node2:Node)
+.output path()
+
+edge("a","b").
+edge("b","c").
+
+path(X,Y) :- edge(X,Y).
+path(X,Y) :- path(X,Z), edge(Z,Y).

tests/swig/python/hyphen-name/hyphen-name.err

@@ -0,0 +1,3 @@
+a	b
+a	c
+b	c