[Boost-commit] svn:boost r58133 - in trunk: boost/program_options libs/program_options/src libs/program_options/test

Subject: [Boost-commit] svn:boost r58133 - in trunk: boost/program_options libs/program_options/src libs/program_options/test
From: s.ochsenknecht_at_[hidden]
Date: 2009-12-04 03:09:44

Author: s_ochsenknecht
Date: 2009-12-04 03:09:43 EST (Fri, 04 Dec 2009)
New Revision: 58133
URL: http://svn.boost.org/trac/boost/changeset/58133

Log:
enhance split_unix() to allow unix style splitting of command line string
Text files modified: 
   trunk/boost/program_options/parsers.hpp        |    10 +++-                                    
   trunk/libs/program_options/src/split.cpp       |    65 ++++++++++++++++------------------      
   trunk/libs/program_options/test/split_test.cpp |    74 +++++++++++++++++++++++++++++++++++++-- 
   3 files changed, 107 insertions(+), 42 deletions(-)

Modified: trunk/boost/program_options/parsers.hpp
==============================================================================
--- trunk/boost/program_options/parsers.hpp	(original)
+++ trunk/boost/program_options/parsers.hpp	2009-12-04 03:09:43 EST (Fri, 04 Dec 2009)
@@ -206,14 +206,18 @@
         can be passed to command_line_parser. The second parameter is
         used to specify a collection of possible seperator chars used
         for splitting. The seperator is defaulted to space " ".
+        Splitting is done in a unix style way, with respect to quotes '"'
+        and escape characters '\'
     */
     BOOST_PROGRAM_OPTIONS_DECL std::vector<std::string>
-    split(const std::string& cmdline, const std::string& sep = " ");
-    
+    split_unix(const std::string& cmdline, const std::string& seperator = " ", 
+         const std::string& quote = "\"", const std::string& escape = "\\");
+         
 #ifndef BOOST_NO_STD_WSTRING
     /** @overload */
     BOOST_PROGRAM_OPTIONS_DECL std::vector<std::wstring>
-    split(const std::wstring& cmdline, const std::wstring& sep = L" ");
+    split_unix(const std::wstring& cmdline, const std::wstring& seperator = L" ", 
+         const std::wstring& quote = L"\"", const std::wstring& escape = L"\\");
 #endif
 
     #ifdef _WIN32

Modified: trunk/libs/program_options/src/split.cpp
==============================================================================
--- trunk/libs/program_options/src/split.cpp	(original)
+++ trunk/libs/program_options/src/split.cpp	2009-12-04 03:09:43 EST (Fri, 04 Dec 2009)
@@ -4,62 +4,59 @@
 // or copy at http://www.boost.org/LICENSE_1_0.txt)
 
 #define BOOST_PROGRAM_OPTIONS_SOURCE
+
 #include <boost/program_options/parsers.hpp>
+#include <boost/tokenizer.hpp>
+
 #include <string>
 #include <vector>
 
 namespace boost { namespace program_options { namespace detail {
 
-
-   template<class charT>
+   template< class charT >
    std::vector<std::basic_string<charT> > 
-   split(const std::basic_string<charT>& cmdline, const std::basic_string<charT>& sep)
-   {
-      std::vector<std::basic_string<charT> > result;
-      if (!cmdline.empty()) 
-      { 
-         std::basic_string<charT> sub(cmdline), val;
-         std::size_t pos; 
-
-         while (sub.size() > 0) 
-         {                                                     
-            if ((pos = sub.find_first_of(sep)) != sub.npos) 
-            {                 
-               val = sub.substr(0,pos);                                               
-               sub = sub.substr(pos+1);                                               
-            } 
-            else 
-            {                                                                  
-               val = sub;                                                             
-               sub.erase();                                                           
-            }                                                                         
-            if (!val.empty()) 
-            {
-               result.push_back(val);
-            }
-         }      
+   split_unix(
+         const std::basic_string<charT>& cmdline, 
+         const std::basic_string<charT>& seperator, 
+         const std::basic_string<charT>& quote, 
+         const std::basic_string<charT>& escape)
+   {   
+      typedef boost::tokenizer< boost::escaped_list_separator<charT>, 
+            typename std::basic_string<charT>::const_iterator, 
+            std::basic_string<charT> >  tokenizerT;
+         
+      tokenizerT tok(cmdline.begin(), cmdline.end(), 
+                boost::escaped_list_separator< charT >(escape, seperator, quote));
+         
+      std::vector< std::basic_string<charT> > result;
+      for (typename tokenizerT::iterator cur_token(tok.begin()), end_token(tok.end()); cur_token != end_token; ++cur_token) {
+         if (!cur_token->empty())
+            result.push_back(*cur_token);
       }
-      return result;   
+      return result;
    }
    
-}}}
+}}}   // namespace
 
 namespace boost { namespace program_options {
 
    // Take a command line string and splits in into tokens, according
    // to the given collection of seperators chars.
    BOOST_PROGRAM_OPTIONS_DECL std::vector<std::string> 
-   split(const std::string& cmdline, const std::string& sep)
+   split_unix(const std::string& cmdline, const std::string& seperator, 
+         const std::string& quote, const std::string& escape)
    {
-      return detail::split(cmdline, sep);
+      return detail::split_unix< char >(cmdline, seperator, quote, escape);
    }
 
 #ifndef BOOST_NO_STD_WSTRING
    BOOST_PROGRAM_OPTIONS_DECL std::vector<std::wstring>
-   split(const std::wstring& cmdline, const std::wstring& sep)
+   split_unix(const std::wstring& cmdline, const std::wstring& seperator, 
+         const std::wstring& quote, const std::wstring& escape)
    {
-      return detail::split(cmdline, sep);
+      return detail::split_unix< wchar_t >(cmdline, seperator, quote, escape);
    }
 #endif
 
-}}
+}}  // namespace
+

Modified: trunk/libs/program_options/test/split_test.cpp
==============================================================================
--- trunk/libs/program_options/test/split_test.cpp	(original)
+++ trunk/libs/program_options/test/split_test.cpp	2009-12-04 03:09:43 EST (Fri, 04 Dec 2009)
@@ -26,11 +26,12 @@
 void split_whitespace(const options_description& description)
 {
 
-   const char* cmdline = "prg --input input.txt \t --optimization 4 \t\n --opt    option";
+   const char* cmdline = "prg --input input.txt \r --optimization 4  \t  --opt \n  option";
    
-   vector< string > tokens =  split(cmdline, " \t\n");
+   vector< string > tokens =  split_unix(cmdline, " \t\n\r");
    
    BOOST_REQUIRE(tokens.size() == 7);
+   
    check_value(tokens[0], "prg");
    check_value(tokens[1], "--input");
    check_value(tokens[2], "input.txt");
@@ -49,8 +50,8 @@
 
    const char* cmdline = "prg --input=input.txt  --optimization=4 --opt=option";
    
-   vector< string > tokens =  split(cmdline, "= ");
-   
+   vector< string > tokens =  split_unix(cmdline, "= ");
+ 
    BOOST_REQUIRE(tokens.size() == 7);
    check_value(tokens[0], "prg");
    check_value(tokens[1], "--input");
@@ -70,7 +71,7 @@
 
    const char* cmdline = "prg;--input input.txt;--optimization 4;--opt option";
    
-   vector< string > tokens =  split(cmdline, "; ");
+   vector< string > tokens =  split_unix(cmdline, "; ");
    
    BOOST_REQUIRE(tokens.size() == 7);
    check_value(tokens[0], "prg");
@@ -86,6 +87,66 @@
    notify(vm);
 }
 
+void split_quotes(const options_description& description)
+{
+   const char* cmdline = "prg --input \"input.txt input.txt\" --optimization 4 --opt \"option1 option2\"";
+   
+   vector< string > tokens =  split_unix(cmdline, " ");
+ 
+   BOOST_REQUIRE(tokens.size() == 7);
+   check_value(tokens[0], "prg");
+   check_value(tokens[1], "--input");
+   check_value(tokens[2], "input.txt input.txt");
+   check_value(tokens[3], "--optimization");
+   check_value(tokens[4], "4");
+   check_value(tokens[5], "--opt");
+   check_value(tokens[6], "option1 option2");
+
+   variables_map vm;  
+   store(command_line_parser(tokens).options(description).run(), vm);
+   notify(vm);
+}
+
+void split_escape(const options_description& description)
+{
+   const char* cmdline = "prg --input \\\"input.txt\\\" --optimization 4 --opt \\\"option1\\ option2\\\"";
+   
+   vector< string > tokens =  split_unix(cmdline, " ");
+  
+   BOOST_REQUIRE(tokens.size() == 7);
+   check_value(tokens[0], "prg");
+   check_value(tokens[1], "--input");
+   check_value(tokens[2], "\"input.txt\"");
+   check_value(tokens[3], "--optimization");
+   check_value(tokens[4], "4");
+   check_value(tokens[5], "--opt");
+   check_value(tokens[6], "\"option1 option2\"");
+
+   variables_map vm;  
+   store(command_line_parser(tokens).options(description).run(), vm);
+   notify(vm);
+}
+
+
+void split_single_quote(const options_description& description)
+{
+   const char* cmdline = "prg --input 'input.txt input.txt' --optimization 4 --opt 'option1 option2'";
+   
+   vector< string > tokens =  split_unix(cmdline, " ", "'");
+ 
+   BOOST_REQUIRE(tokens.size() == 7);
+   check_value(tokens[0], "prg");
+   check_value(tokens[1], "--input");
+   check_value(tokens[2], "input.txt input.txt");
+   check_value(tokens[3], "--optimization");
+   check_value(tokens[4], "4");
+   check_value(tokens[5], "--opt");
+   check_value(tokens[6], "option1 option2");
+
+   variables_map vm;  
+   store(command_line_parser(tokens).options(description).run(), vm);
+   notify(vm);
+}
 
 int main(int /*ac*/, char** /*av*/)
 {
@@ -99,6 +160,9 @@
    split_whitespace(desc);
    split_equalsign(desc);
    split_semi(desc);
+   split_quotes(desc);
+   split_escape(desc);
+   split_single_quote(desc);
    
    return 0;
 }