Three lots of changes:

1) Changes post code review by Jason
2) Fixing bug in end of line processing on Windows with cr-lf
3) Adding command-line options to chai

include/chaiscript/language/chaiscript_engine.hpp

@@ -492,7 +492,7 @@ namespace chaiscript
      * Helper function for loading a file
      */
     std::string load_file(const std::string &t_filename) {
-      std::ifstream infile(t_filename.c_str(), std::ios::in | std::ios::ate);
+      std::ifstream infile(t_filename.c_str(), std::ios::in | std::ios::ate | std::ios::binary );
 
       if (!infile.is_open()) {
         throw File_Not_Found_Error(t_filename);

include/chaiscript/language/chaiscript_parser.hpp

@@ -7,7 +7,6 @@
 #ifndef CHAISCRIPT_PARSER_HPP_
 #define CHAISCRIPT_PARSER_HPP_
 
-
 #include <exception>
 #include <fstream>
 #include <sstream>
@@ -15,9 +14,6 @@
 #include "chaiscript_prelude.hpp"
 #include "chaiscript_common.hpp"
 
-#define lengthof(x) (((int)sizeof(x))/((int)sizeof(x[0])))
-
-
 namespace chaiscript
 {
   enum Alphabet
@@ -32,6 +28,7 @@ namespace chaiscript
   ,   id_alphabet
   ,   white_alphabet
   ,   max_alphabet
+  ,   lengthof_alphabet = 256
   };
 
   class ChaiScript_Parser {
@@ -43,7 +40,7 @@ namespace chaiscript
     std::string m_singleline_comment;
     boost::shared_ptr<std::string> m_filename;
     std::vector<AST_NodePtr> m_match_stack;
-	bool alphabet[max_alphabet][256];
+    bool alphabet[max_alphabet][lengthof_alphabet];
 
     std::vector<std::vector<std::string> > m_operator_matches;
     std::vector<AST_Node_Type::Type> m_operators;
@@ -60,7 +57,7 @@ namespace chaiscript
     ChaiScript_Parser(const ChaiScript_Parser &); // explicitly unimplemented copy constructor
     ChaiScript_Parser &operator=(const ChaiScript_Parser &); // explicitly unimplemented assignment operator
 
-	void setup_operators() 
+    void setup_operators()
     {
       m_operators.push_back(AST_Node_Type::Logical_Or);
       std::vector<std::string> logical_or;
@@ -125,28 +122,28 @@ namespace chaiscript
       dot_access.push_back(".");
       m_operator_matches.push_back(dot_access);
 
-	  int c;
+      int c;
-	  for ( c = 0 ; c < lengthof(alphabet[0]) ; c++ ) {
+      for ( c = 0 ; c < lengthof_alphabet ; c++ ) {
         for ( int a = 0 ; a < max_alphabet ; a ++ ) {
           alphabet[a][c]=false;
         }
-	  }
+      }
-	  alphabet[symbol_alphabet]['+']=true;
+      alphabet[symbol_alphabet]['+']=true;
-	  alphabet[symbol_alphabet]['-']=true; 
+      alphabet[symbol_alphabet]['-']=true;
-	  alphabet[symbol_alphabet]['*']=true; 
+      alphabet[symbol_alphabet]['*']=true;
-	  alphabet[symbol_alphabet]['/']=true;
+      alphabet[symbol_alphabet]['/']=true;
-	  alphabet[symbol_alphabet]['|']=true; 
+      alphabet[symbol_alphabet]['|']=true;
-	  alphabet[symbol_alphabet]['&']=true; 
+      alphabet[symbol_alphabet]['&']=true;
-	  alphabet[symbol_alphabet]['^']=true; 
+      alphabet[symbol_alphabet]['^']=true;
-	  alphabet[symbol_alphabet]['=']=true;
+      alphabet[symbol_alphabet]['=']=true;
-	  alphabet[symbol_alphabet]['.']=true;
+      alphabet[symbol_alphabet]['.']=true;
-	  alphabet[symbol_alphabet]['<']=true; 
+      alphabet[symbol_alphabet]['<']=true;
-	  alphabet[symbol_alphabet]['>']=true;
+      alphabet[symbol_alphabet]['>']=true;
 
       for ( c = 'a' ; c <= 'z' ; c++ ) alphabet[keyword_alphabet][c]=true;
-	  for ( c = 'A' ; c <= 'Z' ; c++ ) alphabet[keyword_alphabet][c]=true;
+      for ( c = 'A' ; c <= 'Z' ; c++ ) alphabet[keyword_alphabet][c]=true;
-	  for ( c = '0' ; c <= '9' ; c++ ) alphabet[keyword_alphabet][c]=true;
+      for ( c = '0' ; c <= '9' ; c++ ) alphabet[keyword_alphabet][c]=true;
-	  alphabet[keyword_alphabet]['_']=true;
+      alphabet[keyword_alphabet]['_']=true;
 
       for ( c = '0' ; c <= '9' ; c++ ) alphabet[int_alphabet][c]=true;
       for ( c = '0' ; c <= '9' ; c++ ) alphabet[float_alphabet][c]=true;
@@ -171,10 +168,10 @@ namespace chaiscript
       alphabet[white_alphabet]['\t']=true;
     }
 
-	/**
+    /**
-	 * test a char in an alphabet
+     * test a char in an alphabet
-	 */
+     */
-	bool char_in_alphabet(unsigned char c,Alphabet a) { return alphabet[a][c]; } 
+    bool char_in_alphabet(unsigned char c,Alphabet a) { return alphabet[a][c]; }
 
     /**
      * Prints the parsed ast_nodes as a tree
@@ -251,19 +248,6 @@ namespace chaiscript
       }
     }
 
-    /**
-     * Does ranged char check
-     */
-/*
-    inline bool char_between(char t_start, char t_end) {
-      if ((*m_input_pos >= t_start) && (*m_input_pos <= t_end)) {
-        return true;
-      }
-      else {
-        return false;
-      }
-    }
-*/
     /**
      * Check to see if there is more text parse
      */
@@ -315,7 +299,7 @@ namespace chaiscript
     bool SkipWS() {
       bool retval = false;
       while (has_more_input()) {
-        if ( char_in_alphabet(*m_input_pos,white_alphabet) ) { // (*m_input_pos == ' ') || (*m_input_pos == '\t')) {
+        if ( char_in_alphabet(*m_input_pos,white_alphabet) ) {
           ++m_input_pos;
           ++m_col;
           retval = true;
@@ -337,17 +321,17 @@ namespace chaiscript
       bool retval = false;
       std::string::const_iterator start = m_input_pos;
 
-      if (has_more_input() && char_in_alphabet(*m_input_pos,float_alphabet) ) { // (char_between('0', '9') || (*m_input_pos == '.'))) {
+      if (has_more_input() && char_in_alphabet(*m_input_pos,float_alphabet) ) {
-        while (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet) ) { // char_between('0', '9')) {
+        while (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet) ) {
           ++m_input_pos;
           ++m_col;
         }
         if (has_more_input() && (*m_input_pos == '.')) {
           ++m_input_pos;
           ++m_col;
-          if (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet)) { // char_between('0', '9')) {
+          if (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet)) {
             retval = true;
-            while (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet) ) { //char_between('0', '9')) {
+            while (has_more_input() && char_in_alphabet(*m_input_pos,int_alphabet) ) {
               ++m_input_pos;
               ++m_col;
             }
@@ -370,16 +354,12 @@ namespace chaiscript
         ++m_input_pos;
         ++m_col;
 
-        if (has_more_input() && char_in_alphabet(*m_input_pos,x_alphabet) ) { // ((*m_input_pos == 'x') || (*m_input_pos == 'X'))) {
+        if (has_more_input() && char_in_alphabet(*m_input_pos,x_alphabet) ) {
           ++m_input_pos;
           ++m_col;
-          if (has_more_input() && char_in_alphabet(*m_input_pos,hex_alphabet)) { // (char_between('0', '9') ||
+          if (has_more_input() && char_in_alphabet(*m_input_pos,hex_alphabet)) {
-                                 //  char_between('a', 'f') ||
-                                 //  char_between('A', 'F'))) {
             retval = true;
-            while (has_more_input() && char_in_alphabet(*m_input_pos,hex_alphabet) ) { // (char_between('0', '9') ||
+            while (has_more_input() && char_in_alphabet(*m_input_pos,hex_alphabet) ) {
-                                     //   char_between('a', 'f') ||
-                                     //   char_between('A', 'F'))) {
               ++m_input_pos;
               ++m_col;
             }
@@ -407,12 +387,12 @@ namespace chaiscript
         ++m_input_pos;
         ++m_col;
 
-        if (has_more_input() && char_in_alphabet(*m_input_pos,b_alphabet) ) { //  ((*m_input_pos == 'b') || (*m_input_pos == 'B'))) {
+        if (has_more_input() && char_in_alphabet(*m_input_pos,b_alphabet) ) {
           ++m_input_pos;
           ++m_col;
-          if (has_more_input() && char_in_alphabet(*m_input_pos,bin_alphabet) ) { // char_between('0', '1')) {
+          if (has_more_input() && char_in_alphabet(*m_input_pos,bin_alphabet) ) {
             retval = true;
-            while (has_more_input() && char_in_alphabet(*m_input_pos,bin_alphabet) ) { // char_between('0', '1')) {
+            while (has_more_input() && char_in_alphabet(*m_input_pos,bin_alphabet) ) {
               ++m_input_pos;
               ++m_col;
             }
@@ -444,7 +424,7 @@ namespace chaiscript
         std::string::const_iterator start = m_input_pos;
         int prev_col = m_col;
         int prev_line = m_line;
-        if (has_more_input() && char_in_alphabet(*m_input_pos,float_alphabet) ) { // (char_between('0', '9') || (*m_input_pos == '.')) ) {
+        if (has_more_input() && char_in_alphabet(*m_input_pos,float_alphabet) ) {
           if (Hex_()) {
             std::string match(start, m_input_pos);
             std::stringstream ss(match);
@@ -512,11 +492,9 @@ namespace chaiscript
      */
     bool Id_() {
       bool retval = false;
-      if (has_more_input() && char_in_alphabet(*m_input_pos,id_alphabet)) { //  (char_between('A', 'Z') || (*m_input_pos == '_') || char_between('a', 'z'))) {
+      if (has_more_input() && char_in_alphabet(*m_input_pos,id_alphabet)) {
         retval = true;
         while (has_more_input() && char_in_alphabet(*m_input_pos,keyword_alphabet) ) {
-                                  // (char_between('A', 'Z') || (*m_input_pos == '_') ||
-                                  //  char_between('a', 'z') || char_between('0', '9'))) {
           ++m_input_pos;
           ++m_col;
         }
@@ -1972,6 +1950,7 @@ namespace chaiscript
         while ((m_input_pos != m_input_end) && (!Eol())) {
           ++m_input_pos;
         }
+        // TODO: respect // -*- coding: utf-8 -*- on line 1 or 2 see: http://evanjones.ca/python-utf8.html)
       }
 
       if (Statements()) {

src/main.cpp

@@ -24,14 +24,27 @@ static void add_history(const char*){}
 static void using_history(){}
 #endif
 
-void print_help() {
+static void help(int n) {
-  std::cout << "ChaiScript evaluator.  To evaluate an expression, type it and press <enter>." << std::endl;
+  if ( n >= 0 ) {
-  std::cout << "Additionally, you can inspect the runtime system using:" << std::endl;
+    std::cout << "ChaiScript evaluator.  To evaluate an expression, type it and press <enter>." << std::endl;
-  std::cout << "  dump_system() - outputs all functions registered to the system" << std::endl;
+    std::cout << "Additionally, you can inspect the runtime system using:" << std::endl;
-  std::cout << "  dump_object(x) - dumps information about the given symbol" << std::endl;
+    std::cout << "  dump_system() - outputs all functions registered to the system" << std::endl;
+    std::cout << "  dump_object(x) - dumps information about the given symbol" << std::endl;
+  } else {
+    std::cout << "usage: chai [option]+" << std::endl;
+    std::cout << "  -h | --help"         << std::endl;
+    std::cout << "  -i | --interactive"  << std::endl;
+    std::cout << "  -c | --command cmd"  << std::endl;
+    std::cout << "  -v | --version"      << std::endl;
+    std::cout << "  -    --stdin"        << std::endl;
+  }
 }
 
-bool throws_exception(const chaiscript::Proxy_Function &f)
+static void version(int){
+  std::cout << "chai: compiled " << __TIME__ << " " << __DATE__ << std::endl;
+}
+
+static bool throws_exception(const chaiscript::Proxy_Function &f)
 {
   try {
     chaiscript::functor<void ()>(f)();
@@ -42,34 +55,41 @@ bool throws_exception(const chaiscript::Proxy_Function &f)
   return false;
 }
 
-std::string get_next_command() {
+static std::string trim(std::string source,const std::string& t)
+{
+  std::string result = source ;
+  result.erase(result.find_last_not_of(t)+1);
+  result.erase(0, result.find_first_not_of(t));
+  return result ;
+}
+
+static std::string get_next_command() {
   std::string retval("quit");
   if ( ! std::cin.eof() ) {
     char *input_raw = readline("eval> ");
     if ( input_raw ) {
       add_history(input_raw);
-      retval = input_raw;
+      retval = trim(std::string(input_raw),std::string(" \t"));
       ::free(input_raw);
-      if ( retval == "help" ) {
-         print_help();
-         retval="";
-      }
-      if ( retval == "quit" || retval == "exit" ) {
-        retval="exit(0)";
-      }
     }
   }
+  if(retval == "quit"
+  || retval == "exit"
+  || retval == "help"
+  || retval == "version"
+  ){
+    retval += "(0)";
+  }
   return retval;
 }
 
 // We have to wrap exit with our own because Clang has a hard time with
 // function pointers to functions with special attributes (system exit being marked NORETURN)
-void myexit(int return_val) {
+static void myexit(int return_val) {
-  std::cout << "thanks for using ChaiScript" << std::endl ;
   exit(return_val);
 }
 
-void interactive(chaiscript::ChaiScript& chai)
+static void interactive(chaiscript::ChaiScript& chai)
 {
   using_history();
 
@@ -103,23 +123,21 @@ void interactive(chaiscript::ChaiScript& chai)
 
 int main(int argc, char *argv[])
 {
-  int error = EXIT_SUCCESS;
-
   std::vector<std::string> usepaths;
   std::vector<std::string> modulepaths;
 
   // Disable deprecation warning for getenv call.
-  #ifdef BOOST_MSVC
+#ifdef BOOST_MSVC
-  #pragma warning(push)
+#pragma warning(push)
-  #pragma warning(disable : 4996)
+#pragma warning(disable : 4996)
-  #endif
+#endif
 
   const char *usepath = getenv("CHAI_USE_PATH");
   const char *modulepath = getenv("CHAI_MODULE_PATH");
 
-  #ifdef BOOST_MSVC
+#ifdef BOOST_MSVC
-  #pragma warning(pop)
+#pragma warning(pop)
-  #endif
+#endif
 
   usepaths.push_back("");
   if (usepath)
@@ -136,34 +154,70 @@ int main(int argc, char *argv[])
   chaiscript::ChaiScript chai(modulepaths,usepaths);
 
   chai.add(chaiscript::fun(&myexit), "exit");
+  chai.add(chaiscript::fun(&myexit), "quit");
+  chai.add(chaiscript::fun(&help), "help");
+  chai.add(chaiscript::fun(&version), "version");
   chai.add(chaiscript::fun(&throws_exception), "throws_exception");
 
-  if (argc < 2) {
+  for (int i = 0; i < argc; ++i) {
-    interactive(chai);
+    if ( i == 0 && argc > 1 ) i++ ;
-  } else {
+    std::string arg( i ? argv[i] : "--interactive" );
-    for (int i = 1; !error && (i < argc); ++i) {
+    enum 
-      try {
+    { eInteractive
-        chaiscript::Boxed_Value val = chai.eval_file(argv[i]);
+    , eCommand
+    , eFile
+    } mode = eCommand ;
+    if  ( arg == "-c" || arg == "--command" ) {
+      if ( (i+1) >= argc ) {
+        std::cout << "insufficient input following " << arg << std::endl;
+        return EXIT_FAILURE;
+      } else {
+        arg = argv[++i];
       }
-      catch (chaiscript::Eval_Error &ee) {
+    } else if ( arg == "-" || arg == "--stdin" ) {
-        std::cout << ee.what();
+      arg = "" ;
-        if (ee.call_stack.size() > 0) {
+      std::string line;
-          std::cout << "during evaluation at (" << *(ee.call_stack[0]->filename) << " " << ee.call_stack[0]->start.line << ", " << ee.call_stack[0]->start.column << ")";
+      while ( std::getline(std::cin, line) ) {
-          for (unsigned int j = 1; j < ee.call_stack.size(); ++j) {
+        arg += line ;
-            std::cout << std::endl;
+      }
-            std::cout << "  from " << *(ee.call_stack[j]->filename) << " (" << ee.call_stack[j]->start.line << ", " << ee.call_stack[j]->start.column << ")";
+    } else if ( arg == "-v" || arg == "--version" ) {
-          }
+      arg = "version(0)" ;
+    } else if ( arg == "-h" || arg == "--help" ) {
+      arg = "help(-1)";
+    } else if ( arg == "-i" || arg == "--interactive" ) {
+      mode = eInteractive ;
+    } else if ( (arg.length() ? arg[0] : ' ') == '-' ) {
+      std::cout << "unrecognised argument " << arg << std::endl;
+      return EXIT_FAILURE;
+    } else {
+      mode = eFile;
+    }
+
+    chaiscript::Boxed_Value val ;
+    try {
+      switch ( mode ) {
+        case eInteractive : interactive(chai); break;
+        case eCommand     : val = chai.eval(arg); break;
+        case eFile        : val = chai.eval_file(arg); break;
+      }
+    }
+    catch (chaiscript::Eval_Error &ee) {
+      std::cout << ee.what();
+      if (ee.call_stack.size() > 0) {
+        std::cout << "during evaluation at (" << *(ee.call_stack[0]->filename) << " " << ee.call_stack[0]->start.line << ", " << ee.call_stack[0]->start.column << ")";
+        for (unsigned int j = 1; j < ee.call_stack.size(); ++j) {
+          std::cout << std::endl;
+          std::cout << "  from " << *(ee.call_stack[j]->filename) << " (" << ee.call_stack[j]->start.line << ", " << ee.call_stack[j]->start.column << ")";
         }
-        std::cout << std::endl;
-        error = EXIT_FAILURE;
-      }
-      catch (std::exception &e) {
-        std::cout << e.what() << std::endl;
-        error = EXIT_FAILURE;
       }
+      std::cout << std::endl;
+      return EXIT_FAILURE;
+    }
+    catch (std::exception &e) {
+      std::cout << e.what() << std::endl;
+      return EXIT_FAILURE;
     }
   }
 
-  return error ;
+  return EXIT_SUCCESS;
 }
-