Added ws skipping. Non ws skipping versions of base parsers. Added comment skipping

CMakeLists.txt

@@ -13,4 +13,5 @@ include_directories(dispatchkit langkit)
 add_subdirectory(langkit bin)
 add_subdirectory(dispatchkit bin)
 add_subdirectory(chaiscript bin)
+add_subdirectory(chaioop bin)
  

chaioop/CMakeLists.txt

@@ -3,8 +3,8 @@ cmake_minimum_required(VERSION 2.6)
 project(chaioop)
 
 SET (CMAKE_BUILD_TYPE gdb)
-SET (CMAKE_C_FLAGS_GDB " -Wall -ggdb -O3")
-SET (CMAKE_CXX_FLAGS_GDB " -Wall -ggdb -O3")
+SET (CMAKE_C_FLAGS_GDB " -Wall -ggdb")
+SET (CMAKE_CXX_FLAGS_GDB " -Wall -ggdb")
 
 include_directories(../langkit ../dispatchkit)
 

chaioop/main.cpp

@@ -7,6 +7,27 @@
 #include <string.h>
 
 namespace langkit {
+    struct File_Position {
+        int line;
+        int column;
+        const char *filename;
+
+        File_Position(int file_line, int file_column, const char *file_name)
+            : line(file_line), column(file_column), filename(file_name) { }
+
+        File_Position() : line(0), column(0), filename(NULL) { }
+    };
+
+    struct Parse_Error {
+        std::string reason;
+        File_Position position;
+
+        Parse_Error(const std::string &why, const File_Position &where) :
+            reason(why), position(where) { }
+
+        virtual ~Parse_Error() throw() {}
+    };
+
     class Token {
         std::string text;
 
@@ -15,14 +36,85 @@ namespace langkit {
 
     class Parser {
         std::string::iterator input_pos, input_end;
+        int line, col;
+        std::string multiline_comment_begin, multiline_comment_end;
+        std::string singleline_comment;
 
     public:
-        bool Int() {
+        bool SkipComment() {
+            bool retval = false;
+
+            if (Str_(multiline_comment_begin.c_str())) {
+                while (input_pos != input_end) {
+                    if (Str_(multiline_comment_end.c_str())) {
+                        break;
+                    }
+                    else if (!Eol_()) {
+                        ++col;
+                        ++input_pos;
+                    }
+                }
+                retval = true;
+            }
+            else if (Str_(singleline_comment.c_str())) {
+                while (input_pos != input_end) {
+                    if (Eol_()) {
+                        break;
+                    }
+                    else {
+                        ++col;
+                        ++input_pos;
+                    }
+                }
+                retval = true;
+            }
+            return retval;
+        }
+
+        bool SkipWS() {
+            bool retval = false;
+            while (input_pos != input_end) {
+                if ((*input_pos == ' ') || (*input_pos == '\t')) {
+                    ++input_pos;
+                    ++col;
+                    retval = true;
+                }
+                else if (SkipComment()) {
+                    retval = true;
+                }
+                else {
+                    break;
+                }
+            }
+            return retval;
+        }
+
+        bool Int_() {
             bool retval = false;
             if ((input_pos != input_end) && (*input_pos >= '0') && (*input_pos <= '9')) {
                 retval = true;
                 while ((input_pos != input_end) && (*input_pos >= '0') && (*input_pos <= '9')) {
                     ++input_pos;
+                    ++col;
+                }
+            }
+
+            return retval;
+        }
+
+        bool Int() {
+            SkipWS();
+
+            return Int_();
+        }
+
+        bool Id_() {
+            bool retval = false;
+            if ((input_pos != input_end) && (((*input_pos >= 'A') && (*input_pos <= 'Z')) || ((*input_pos >= 'a') && (*input_pos <= 'z')))) {
+                retval = true;
+                while ((input_pos != input_end) && (((*input_pos >= 'A') && (*input_pos <= 'Z')) || ((*input_pos >= 'a') && (*input_pos <= 'z')))) {
+                    ++input_pos;
+                    ++col;
                 }
             }
 
@@ -30,28 +122,29 @@ namespace langkit {
         }
 
         bool Id() {
+            SkipWS();
+
+            return Id_();
+        }
+
+        bool Char_(char c) {
             bool retval = false;
-            if ((input_pos != input_end) && (((*input_pos >= 'A') && (*input_pos <= 'Z')) || ((*input_pos >= 'a') && (*input_pos <= 'z')))) {
+            if ((input_pos != input_end) && (*input_pos == c)) {
+                ++input_pos;
+                ++col;
                 retval = true;
-                while ((input_pos != input_end) && (((*input_pos >= 'A') && (*input_pos <= 'Z')) || ((*input_pos >= 'a') && (*input_pos <= 'z')))) {
-                    ++input_pos;
-                }
             }
 
             return retval;
         }
 
         bool Char(char c) {
-            bool retval = false;
-            if ((input_pos != input_end) && (*input_pos == c)) {
-                ++input_pos;
-                retval = true;
-            }
+            SkipWS();
 
-            return retval;
+            return Char_(c);
         }
 
-        bool Str(const char *s) {
+        bool Str_(const char *s) {
             bool retval = false;
             int len = strlen(s);
 
@@ -61,18 +154,46 @@ namespace langkit {
                     if (*tmp != s[i]) {
                         return false;
                     }
+                    ++tmp;
                 }
                 retval = true;
                 input_pos = tmp;
+                col += len;
             }
 
             return retval;
         }
 
+        bool Str(const char *s) {
+            SkipWS();
+            return Str_(s);
+        }
+
+        bool Eol_() {
+            bool retval = false;
+
+            if ((input_pos != input_end) && (Str_("\r\n") || Char_('\n'))) {
+                retval = true;
+                ++line;
+                col = 1;
+            }
+
+            return retval;
+        }
+
+        bool Eol() {
+            SkipWS();
+
+            return Eol_();
+        }
+
         bool Arg_List() {
             bool retval = Id() || Int();
             while (retval && Char(',')) {
                 retval = Id() || Int();
+                if (!retval) {
+                    throw Parse_Error("Unexpected value in parameter list", File_Position(line, col, "_INPUT_"));
+                }
             }
             return retval;
         }
@@ -81,6 +202,8 @@ namespace langkit {
             bool retval = false;
 
             std::string::iterator prev = input_pos;
+            int prev_line = line;
+            int prev_col = col;
 
             if (Id() && Char('(')) {
 
@@ -88,16 +211,10 @@ namespace langkit {
                 retval = Char(')');
             }
 
-            if (!retval) { input_pos = prev; }
-
-            return retval;
-        }
-
-        bool Eol() {
-            bool retval = false;
-
-            if ((input_pos != input_end) && (Str("\r\n") || Char('\n'))) {
-                retval = true;
+            if (!retval) {
+                input_pos = prev;
+                prev_line = line;
+                prev_col = col;
             }
 
             return retval;
@@ -117,6 +234,10 @@ namespace langkit {
         bool parse(std::string input) {
             input_pos = input.begin();
             input_end = input.end();
+            line = 1; col = 1;
+            multiline_comment_begin = "/*";
+            multiline_comment_end = "*/";
+            singleline_comment = "//";
 
             return Fun_Calls();
         }
@@ -144,9 +265,29 @@ std::string load_file(const char *filename) {
 
 int main(int argc, char *argv[]) {
     langkit::Parser parser;
+    std::string input;
 
     if (argc > 1) {
-        std::cout << parser.parse(load_file(argv[1])) << std::endl;
+        try {
+            std::cout << parser.parse(load_file(argv[1])) << std::endl;
+        }
+        catch (langkit::Parse_Error &pe) {
+            std::cout << pe.reason << " at " << pe.position.line << ", " << pe.position.column << std::endl;
+        }
+    }
+    else {
+        std::cout << "eval> ";
+        std::getline(std::cin, input);
+        while (input != "quit") {
+            try {
+                std::cout << parser.parse(input) << std::endl;
+            }
+            catch (langkit::Parse_Error &pe) {
+                std::cout << pe.reason << " at " << pe.position.line << ", " << pe.position.column << std::endl;
+            }
+            std::cout << "eval> ";
+            std::getline(std::cin, input);
+        }
     }
 
     return 0;

chaiscript/CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 2.6)
 enable_testing()
 
-project(wesley)
+project(chaiscript)
 
 SET (CMAKE_BUILD_TYPE gdb)
 SET (CMAKE_C_FLAGS_GDB " -Wall -ggdb")
@@ -15,11 +15,11 @@ if(Boost_FOUND)
    add_executable(chaiscript_test main.cpp)
    target_link_libraries(chaiscript_test ${Boost_LIBRARIES})
 
-   add_executable(chaiscript_callbacktest callbacktest.cpp)
-   target_link_libraries(chaiscript_callbacktest ${Boost_LIBRARIES})
+	 add_executable(chaiscript_callbacktest callbacktest.cpp)
+	 target_link_libraries(chaiscript_callbacktest ${Boost_LIBRARIES})
 
-   add_executable(sensors sensors.cpp)
-   target_link_libraries(sensors ${Boost_LIBRARIES})
+	 add_executable(sensors sensors.cpp)
+	 target_link_libraries(sensors ${Boost_LIBRARIES})
 
 
 endif()

chaiscript/main.cpp

@@ -18,11 +18,11 @@ int main(int argc, char *argv[]) {
 
             if (val.get_type_info().m_bare_type_info && *(val.get_type_info().m_bare_type_info) != typeid(void)) {
                 try {
-                    dispatchkit::Boxed_Value printeval 
-                      = dispatchkit::dispatch(chai.get_eval_engine().get_function("to_string"), 
+                    dispatchkit::Boxed_Value printeval
+                      = dispatchkit::dispatch(chai.get_eval_engine().get_function("to_string"),
                           dispatchkit::Param_List_Builder() << val);
                     std::cout << "result: ";
-                    dispatchkit::dispatch(chai.get_eval_engine().get_function("print"), 
+                    dispatchkit::dispatch(chai.get_eval_engine().get_function("print"),
                         dispatchkit::Param_List_Builder() << printeval);
                 } catch (const std::runtime_error &e) {
                     std::cout << "result: object #" << &val << std::endl;

dispatchkit/CMakeLists.txt

@@ -12,11 +12,11 @@ if(Boost_FOUND)
    include_directories(${Boost_INCLUDE_DIRS})
    include_directories(.)
    
-   add_executable(dispatchkit_test test.cpp)
+	 add_executable(dispatchkit_test test.cpp)
 
-   add_executable(dispatchkit_unittest unittest.cpp)
+	 add_executable(dispatchkit_unittest unittest.cpp)
 
-   target_link_libraries(dispatchkit_unittest ${Boost_LIBRARIES})
+	 target_link_libraries(dispatchkit_unittest ${Boost_LIBRARIES})
 
 endif()
 

langkit/CMakeLists.txt

@@ -11,11 +11,11 @@ find_package( Boost 1.36.0 COMPONENTS regex unit_test_framework)
 if(Boost_FOUND)
    include_directories(${Boost_INCLUDE_DIRS})
    
-   add_executable(langkit_test main.cpp)
-   target_link_libraries(langkit_test ${Boost_LIBRARIES})
+	 add_executable(langkit_test main.cpp)
+	 target_link_libraries(langkit_test ${Boost_LIBRARIES})
 
-   add_executable(langkit_unittest unittest.cpp)
-   target_link_libraries(langkit_unittest ${Boost_LIBRARIES})
+	 add_executable(langkit_unittest unittest.cpp)
+	 target_link_libraries(langkit_unittest ${Boost_LIBRARIES})
 endif()
 
 add_test(langkit_unittest ${EXECUTABLE_OUTPUT_PATH}/langkit_unittest)