add CharAt to theory_str and basic rewrite rule for constant CharAt exprs

src/ast/rewriter/str_rewriter.cpp

@@ -23,12 +23,56 @@ Notes:
 #include"ast_util.h"
 #include"well_sorted.h"
 
+br_status str_rewriter::mk_str_CharAt(expr * arg0, expr * arg1, expr_ref & result) {
+    TRACE("t_str_rw", tout << "rewrite (CharAt " << mk_pp(arg0, m()) << " " << mk_pp(arg1, m()) << ")" << std::endl;);
+    // if arg0 is a string constant and arg1 is an integer constant,
+    // we can rewrite this by evaluating the expression
+    rational arg1Int;
+    if (m_strutil.is_string(arg0) && m_autil.is_numeral(arg1, arg1Int)) {
+        TRACE("t_str_rw", tout << "evaluating constant CharAt expression" << std::endl;);
+        std::string arg0Str = m_strutil.get_string_constant_value(arg0);
+        std::string resultStr;
+        if (arg1Int >= rational(0) && arg1Int <= rational((unsigned)arg0Str.length())) {
+            resultStr = arg0Str.at(arg1Int.get_unsigned());
+            TRACE("t_str_rw", tout << "result is '" << resultStr << "'" << std::endl;);
+        } else {
+            resultStr = "";
+            TRACE("t_str_rw", tout << "bogus length argument, result is empty string" << std::endl;);
+        }
+        result = m_strutil.mk_string(resultStr);
+        return BR_DONE;
+    } else {
+        // TODO NEXT
+        NOT_IMPLEMENTED_YET();
+        /*
+        Z3_ast ts0 = my_mk_internal_string_var(t);
+        Z3_ast ts1 = my_mk_internal_string_var(t);
+        Z3_ast ts2 = my_mk_internal_string_var(t);
+
+        Z3_ast cond = mk_2_and(t, Z3_mk_ge(ctx, args[1], mk_int(ctx, 0)), Z3_mk_lt(ctx, args[1], mk_length(t, args[0])));
+
+        Z3_ast and_item[3];
+        and_item[0] = Z3_mk_eq(ctx, args[0], mk_concat(t, ts0, mk_concat(t, ts1, ts2)));
+        and_item[1] = Z3_mk_eq(ctx, args[1], mk_length(t, ts0));
+        and_item[2] = Z3_mk_eq(ctx, mk_length(t, ts1), mk_int(ctx, 1));
+        Z3_ast thenBranch = Z3_mk_and(ctx, 3, and_item);
+        Z3_ast elseBranch = Z3_mk_eq(ctx, ts1, my_mk_str_value(t, ""));
+        breakdownAssert = Z3_mk_ite(ctx, cond, thenBranch, elseBranch);
+        return ts1;
+        */
+    }
+}
+
 br_status str_rewriter::mk_app_core(func_decl * f, unsigned num_args, expr * const * args, expr_ref & result) {
     SASSERT(f->get_family_id() == get_fid());
 
     TRACE("t_str_rw", tout << "rewrite app: " << f->get_name() << std::endl;);
 
+    // TODO more rewrites for really easy cases, e.g. (Concat "abc" "def")...
     switch(f->get_decl_kind()) {
+    case OP_STR_CHARAT:
+        SASSERT(num_args == 2);
+        return mk_str_CharAt(args[0], args[1], result);
     default:
         return BR_FAILED;
     }

src/ast/rewriter/str_rewriter.h

@@ -40,6 +40,8 @@ public:
     br_status mk_app_core(func_decl * f, unsigned num_args, expr * const * args, expr_ref & result);
     br_status mk_eq_core(expr * lhs, expr * rhs, expr_ref & result);
 
+    br_status mk_str_CharAt(expr * arg0, expr * arg1, expr_ref & result);
+
     bool reduce_eq(expr * l, expr * r, expr_ref_vector & lhs, expr_ref_vector & rhs, bool & change);
     bool reduce_eq(expr_ref_vector& ls, expr_ref_vector& rs, expr_ref_vector& lhs, expr_ref_vector& rhs, bool& change);
 

src/ast/str_decl_plugin.cpp

@@ -26,6 +26,7 @@ str_decl_plugin::str_decl_plugin():
     m_str_decl(0),
     m_concat_decl(0),
     m_length_decl(0),
+    m_charat_decl(0),
     m_arith_plugin(0),
     m_arith_fid(0),
     m_int_sort(0){
@@ -39,6 +40,7 @@ void str_decl_plugin::finalize(void) {
     DEC_REF(m_str_decl);
     DEC_REF(m_concat_decl);
     DEC_REF(m_length_decl);
+    DEC_REF(m_charat_decl);
     DEC_REF(m_int_sort);
 }
 
@@ -64,7 +66,11 @@ void str_decl_plugin::set_manager(ast_manager * m, family_id id) {
 
     MK_OP(m_concat_decl, "Concat", OP_STRCAT, s);
 
-    m_length_decl = m->mk_func_decl(symbol("Length"), s, i, func_decl_info(id, OP_STRLEN)); m_manager->inc_ref(m_length_decl);
+    m_length_decl = m->mk_func_decl(symbol("Length"), s, i, func_decl_info(id, OP_STRLEN));
+    m_manager->inc_ref(m_length_decl);
+
+    m_charat_decl = m->mk_func_decl(symbol("CharAt"), s, i, s, func_decl_info(id, OP_STR_CHARAT));
+    m_manager->inc_ref(m_charat_decl);
 }
 
 decl_plugin * str_decl_plugin::mk_fresh() {
@@ -82,6 +88,7 @@ func_decl * str_decl_plugin::mk_func_decl(decl_kind k) {
     switch(k) {
     case OP_STRCAT: return m_concat_decl;
     case OP_STRLEN: return m_length_decl;
+    case OP_STR_CHARAT: return m_charat_decl;
     default: return 0;
     }
 }
@@ -138,6 +145,7 @@ app * str_decl_plugin::mk_fresh_string() {
 void str_decl_plugin::get_op_names(svector<builtin_name> & op_names, symbol const & logic) {
     op_names.push_back(builtin_name("Concat", OP_STRCAT));
     op_names.push_back(builtin_name("Length", OP_STRLEN));
+    op_names.push_back(builtin_name("CharAt", OP_STR_CHARAT));
 }
 
 void str_decl_plugin::get_sort_names(svector<builtin_name> & sort_names, symbol const & logic) {

src/ast/str_decl_plugin.h

@@ -27,9 +27,12 @@ enum str_sort_kind {
 
 enum str_op_kind {
     OP_STR, /* string constants */
-    //
+    // basic string operators
     OP_STRCAT,
     OP_STRLEN,
+    // higher-level string functions -- these are reduced to basic operations
+    OP_STR_CHARAT,
+    // end
     LAST_STR_OP
 };
 
@@ -40,6 +43,7 @@ protected:
 
     func_decl * m_concat_decl;
     func_decl * m_length_decl;
+    func_decl * m_charat_decl;
 
     arith_decl_plugin * m_arith_plugin;
     family_id           m_arith_fid;