Implement string contains for lpython (#2787)

Author: wolfv

integration_tests/CMakeLists.txt

@@ -532,6 +532,7 @@ RUN(NAME test_str_02         LABELS cpython llvm llvm_jit c)
 RUN(NAME test_str_03         LABELS cpython llvm llvm_jit c)
 RUN(NAME test_str_04         LABELS cpython llvm llvm_jit c wasm)
 RUN(NAME test_str_05         LABELS cpython llvm llvm_jit c)
+RUN(NAME test_str_06         LABELS cpython llvm llvm_jit c)
 RUN(NAME test_list_01        LABELS cpython llvm llvm_jit c)
 RUN(NAME test_list_02        LABELS cpython llvm llvm_jit c)
 RUN(NAME test_list_03        LABELS cpython llvm llvm_jit c NOFAST)

integration_tests/test_str_06.py

@@ -0,0 +1,11 @@
+def main0():
+    x: str
+    x = "Hello, World"
+
+    assert "Hello" in x
+    assert "," in x
+    assert "rld" in x
+
+    assert "Hello" not in "World"
+
+main0()

src/libasr/codegen/asr_to_c_cpp.h

@@ -1244,6 +1244,15 @@ PyMODINIT_FUNC PyInit_lpython_module_)" + fn_name + R"((void) {
         src = "_lfortran_strrepeat_c(" + s + ", " + n + ")";
     }
 
+    void visit_StringContains(const ASR::StringContains_t &x) {
+        CHECK_FAST_C_CPP(compiler_options, x)
+        self().visit_expr(*x.m_left);
+        std::string substr = src;
+        self().visit_expr(*x.m_right);
+        std::string str = src;
+        src = "_lfortran_str_contains(" + str + ", " + substr + ")";
+    }
+
     void visit_Assignment(const ASR::Assignment_t &x) {
         std::string target;
         ASR::ttype_t* m_target_type = ASRUtils::expr_type(x.m_target);

src/libasr/codegen/asr_to_llvm.cpp

@@ -723,6 +723,21 @@ class ASRToLLVMVisitor : public ASR::BaseVisitor<ASRToLLVMVisitor>
         return builder->CreateCall(fn, {str, idx1});
     }
 
+    llvm::Value* lfortran_str_contains(llvm::Value* str, llvm::Value* substr)
+    {
+        std::string runtime_func_name = "_lfortran_str_contains";
+        llvm::Function *fn = module->getFunction(runtime_func_name);
+        if (!fn) {
+            llvm::FunctionType *function_type = llvm::FunctionType::get(
+                    llvm::Type::getInt1Ty(context), {
+                        character_type, character_type
+                    }, false);
+            fn = llvm::Function::Create(function_type,
+                    llvm::Function::ExternalLinkage, runtime_func_name, *module);
+        }
+        return builder->CreateCall(fn, {str, substr});
+    }
+
     llvm::Value* lfortran_str_copy(llvm::Value* str, llvm::Value* idx1, llvm::Value* idx2)
     {
         std::string runtime_func_name = "_lfortran_str_copy";
@@ -6416,6 +6431,22 @@ class ASRToLLVMVisitor : public ASR::BaseVisitor<ASRToLLVMVisitor>
         }
     }
 
+    void visit_StringContains(const ASR::StringContains_t& x) {
+        if (x.m_value) {
+            this->visit_expr_wrapper(x.m_value, true);
+            return;
+        }
+
+        this->visit_expr_wrapper(x.m_left, true);
+        llvm::Value *substr = tmp;
+
+        this->visit_expr_wrapper(x.m_right, true);
+        llvm::Value *right = tmp;
+
+        tmp = lfortran_str_contains(right, substr);
+        strings_to_be_deallocated.push_back(al, tmp);
+    }
+
     void visit_StringSection(const ASR::StringSection_t& x) {
         if (x.m_value) {
             this->visit_expr_wrapper(x.m_value, true);

src/libasr/runtime/lfortran_intrinsics.c

@@ -2197,6 +2197,12 @@ LFORTRAN_API char* _lfortran_str_item(char* s, int32_t idx) {
     return res;
 }
 
+/// Find a substring in a string
+LFORTRAN_API bool _lfortran_str_contains(char* str, char* substr) {
+    char* res = strstr(str, substr);
+    return res != NULL;
+}
+
 // idx1 and idx2 both start from 1
 LFORTRAN_API char* _lfortran_str_copy(char* s, int32_t idx1, int32_t idx2) {
 

src/libasr/runtime/lfortran_intrinsics.h

@@ -209,6 +209,7 @@ LFORTRAN_API void _lfortran_free(char* ptr);
 LFORTRAN_API void _lfortran_alloc(char** ptr, int32_t len);
 LFORTRAN_API void _lfortran_string_init(int size_plus_one, char *s);
 LFORTRAN_API char* _lfortran_str_item(char* s, int32_t idx);
+LFORTRAN_API bool _lfortran_str_contains(char* str, char* substr);
 LFORTRAN_API char* _lfortran_str_copy(char* s, int32_t idx1, int32_t idx2); // idx1 and idx2 both start from 1
 LFORTRAN_API char* _lfortran_str_slice(char* s, int32_t idx1, int32_t idx2, int32_t step,
                         bool idx1_present, bool idx2_present);