Click here to Skip to main content
Click here to Skip to main content

Command line parser

By , 17 Aug 2002
 

Introduction

Getting list of arguments from command line is a common task which is required by a lot of applications. However, there is no standard solution (as far as I know ;). So I wrote class CCmdLineParser, which can parse arguments from command line, if they are typed in folowing form: 

  • /Key 
  • /KeyWithValue:Value 
  • /KeyWithComplexValue:"Some really complex value of /KeyWithComplexValue"

Of course, multiple keys, Unicode and long (up to 32Kb) command lines are supported.

This implementation requires the MFC or ATL CString class, or some clone with similar interface as well as the STL class map.

Usage

First, you should construct the object and call the Parse function (from constructor or
CCmdLineParser parser(::GetCommandLine());
or
CCmdLineParser parser;
parser.Parse(_T("/Key /Key2:Val /Key3:\"Complex-Value\" -Key4"));

Then, there are two ways of working with results. You can check if some particular key was specified in the command line:

if(parser.HasKey(_T("Key")) {
	// Do some stuff
}
if(parser.HasKey(_T("Key2")) {
	LPCTSTR szKey2Value = parser.GetVal(_T("Key2"));
	// Do something with value of Key2
}

LPCTSTR szKey3Value = parser.GetVal(_T("Key3"));
if(szKey3Value) {
	// There was key "Key3" in input,  
} else {
	// No key "Key3" in input
}

LPCTSTR szKey4Value = parser.GetVal(_T("Key4"));
// Key4 was found in input, but since no value was specified, 
// szKey4Value points to empty string

Another way to use is to enumerate all keys in command line:

CString sKey, sValue;

CCmdLineParser::POSITION pos = parser.getFirst();
while(!realParser.isLast(pos)) {
	realParser.getNext(pos, sKey, sValue);
	// Do something with current key and value
}

Customization and "how it works"

Repeated keys

If several different values are specified with same key, only the first value is stored. So, if user passes command line /Add:One /Add:Two, /Add:Two will be ignored and will not be added to parsed list.

Case sensitive/insensitive

By default, keys are not case-sensitive. So, /KeyOne is equal to -keyONE. This is done by converting all keys to lowercase before storing them. If you want to change this behaviour, call setCaseSensitive(true) or call the constructor with the second argument set to true:
CCmdLineParser parser(::GetCommandLine(), true);
This will switch the parser to case-sensitive mode, and if the user passes -key, then GetKey(_T("Key")) will return false

Syntax

Formally, command line should be in following form:
CommandLine::=[<Key> [,<Key>...]]
<Key>::=<Delimeter>KeyName[<Separator><Value>]
<Value> ::= { KeyValue | <QuoteChar>Quoted Key Value<QuoteChar>} ][
<Delimeter>::= { - | / }
<Separator>::= { : }
<QuoteChar>::= { " }

Values for <Delimeter>, <Separator> and <QuoteChar> are stored in static variables m_sDelimeters, m_sValueSep and m_sQuotes respectively. If you want to change them (for instance, allow user to specify quoted values in apostrophes), you can do it in the beginning of CmdLineParser.cpp:

const TCHAR CCmdLineParser::m_sQuotes[] = _T("\"\'");
Note: If you want to change m_sDelimeters, space must be the first character of this string. Also, if you have your own CString class with other name than CString, you can change it in the beginning of CmdLineParser.h:
typedef MyOwnCString CCmdLineParser_String;
That's it! ;)

License

This article has no explicit license attached to it but may contain usage terms in the article text or the download files themselves. If in doubt please contact the author via the discussion board below.

A list of licenses authors might use can be found here

About the Author

Pavel Antonov
Web Developer
Russian Federation Russian Federation
Member
No Biography provided

Sign Up to vote   Poor Excellent
Add a reason or comment to your vote: x
Votes of 3 or less require a comment

Comments and Discussions

 
You must Sign In to use this message board.
Search this forum  
    Spacing  Noise  Layout  Per page   
QuestionThis wheel has been reinvented a number of times: TCLAP, boost program_options to name a fewmemberLarry S. Smith17 Dec '11 - 21:28 
http://tclap.sourceforge.net/[^]
 
http://www.boost.org/doc/libs/1_48_0/doc/html/program_options.html[^]
GeneralMy vote of 4memberhivhiv17 Dec '10 - 23:18 
you integrated easily with GetCommandLine(). A very easy to use
QuestionGood!But a question.membersuxiaojack26 Apr '08 - 1:28 
Could we easily find out some filepath string in cmdline?
It's easy if we used /FileNameKey:"THE PATH OF FILE",like the sample in your article.
But in normally, filepath string be used don't have the marked part /FileNameKey . and then there is a easy way to find out it?
GeneralBug report and my implementation of CCmdLineParser::CCmdLineParser() [modified]memberdoroboy27 Nov '07 - 21:11 
Good job! thanks very much.
But when the execute file is in one folder which include the Delimeter character "-", the Parse() function will take it as a key mistakenly. I think this is a bug.
And why don't you use the argc & argv? Below is my implementation of CCmdLineParser::CCmdLineParser(). I think this is simpler and easier to understand than yours.
const TCHAR CCmdLineParser::m_sDelimeters[] = _T("/");  // Can be _T("-/"),  for instance
const TCHAR CCmdLineParser::m_sValueSep[] = _T(":");    // Can be _T("=:"),  for instance. Space need NOT be in set.
//const TCHAR CCmdLineParser::m_sQuotes[] = _T("\"");   // No need any more! so I comment out this line.

CCmdLineParser::CCmdLineParser(int argc, TCHAR* argv[], bool bCaseSensitive) : m_bCaseSensitive(bCaseSensitive)
{
    const CString sEmpty;
    m_ValsMap.clear();
 
    for (int i=1; i<argc; i++) {
        LPCTSTR sArg = _tcspbrk(argv[i], m_sDelimeters);
        if ( (NULL == sArg) || (sArg != argv[i]) || ('\0' == sArg[1]) ) {  // The m_sDelimeters chacracter not found || The first character is not in m_sDelimeters set || cmdline ends with /
            continue;
        }
        
        sArg = _tcsinc(sArg);
        LPCTSTR sVal = _tcspbrk(sArg, m_sValueSep);
 
        if( NULL == sVal ) {          // cmdline ends with /Key
            CString csKey(sArg);
            if(!m_bCaseSensitive) {
                csKey.MakeLower();
            }
            m_ValsMap.insert(CValsMap::value_type(csKey, sEmpty));
        } else {                    // cmdline ends with /Key:*
            CString csKey(sArg, sVal - sArg);
            if(!csKey.IsEmpty()) {      // Prevent /: case
                if(!m_bCaseSensitive) {
                    csKey.MakeLower();
                }
                if ( 1 == _tcslen(sVal) ) {     // cmdline ends with /Key:
                    sVal = sEmpty;
                } else {                    // cmdline ends with /Key:Val
                    sVal = _tcsinc(sVal);
                }
                m_ValsMap.insert(CValsMap::value_type(csKey, sVal));
            }
        }
    }
}

GeneralRe: Bug report and my implementation of CCmdLineParser::CCmdLineParser()memberdoroboy13 Dec '07 - 13:56 
Sorry, there was a bug in my code, too. When the "Val" is enclosed in double-quotation, the last charater of argv[i] became '\"'.
Below is the revised version.
 

CCmdLineParser::CCmdLineParser(int argc, TCHAR* argv[], bool bCaseSensitive) : m_bCaseSensitive(bCaseSensitive)
{
m_ValsMap.clear();
 
for (int i=1; i<argc; i++) {
LPCTSTR sArg = _tcspbrk(argv[i], m_sDelimeters);
if ( (NULL == sArg) || (sArg != argv[i]) || ('\0' == sArg[1]) ) { // The m_sDelimeters chacracter not found || The first character is not in m_sDelimeters set || cmdline ends with /
continue;
}

sArg = _tcsinc(sArg);
LPTSTR sVal = _tcspbrk(sArg, m_sValueSep);
 
if(sVal == NULL) { // cmdline ends with /Key
CString csKey(sArg);
if( !m_bCaseSensitive ) {
csKey.MakeLower();
}
m_ValsMap.insert(CValsMap::value_type(csKey, ""));
} else { // cmdline ends with /Key:*
CString csKey(sArg, sVal - sArg);
if( !csKey.IsEmpty() ) { // Prevent /: case
if( !m_bCaseSensitive ) {
csKey.MakeLower();
}

size_t stLen = _tcslen(sVal);
if ( 1 == stLen ) { // cmdline ends with /Key:
sVal = _T("");
} else { // cmdline ends with /Key:Val
if ('\"' == sVal[stLen-1]) {
sVal[stLen-1] = '\\';
}
sVal = _tcsinc(sVal);
}
m_ValsMap.insert(CValsMap::value_type(csKey, sVal));
}
}
}
}

GeneralThank you, I'm just looking for thismemberPeter Liu26 Aug '07 - 3:46 
Big Grin | :-D
GeneralExcellentmemberkpinkert19 Aug '05 - 9:06 
A very easy to use and simple interface. I like the fact that you integrated easily with GetCommandLine().
 
Thanks!
Kevin
GeneralSupporting - - parameters and map.memberdB.29 Jun '05 - 6:52 
I have been using this class with modifications in my project, so thought I'd contribute something back.
 
Added support for parameters with two dashes (like --foobar) or actually any number of delimiter combinations (-/-foobar):
 
LPCTSTR pszArg = _tcspbrk(pszCurrent, m_pszDelimeters);
if (! pszArg) 
{
 // No delimeters found
 break; 
}
 

LPCTSTR pszInArg;
do
{
 pszArg = _tcsinc(pszArg);
 pszInArg = _tcspbrk(pszArg, m_pszDelimeters);
} while (pszInArg == pszArg);

...
 
I also derived from CMapStringToString as many have suggested already and removed all the find and iterator functions to use those of the base class. Made code a lot simpler.
 
In addition, for those that use CppUnit for unit testing, here's a unit test:
 
typedef struct
{
    LPCTSTR pszName;
    LPCTSTR pszValue;
} CCmdLineTestDataPair;
 
typedef struct 
{
    LPCTSTR pszCmdLine;
    BOOL fCaseSensitive;
    ULONG ulParameters;
    CCmdLineTestDataPair testDataPairs[3];
} CCmdLineTestData;
 
void CCmdLineUnitTest::testParseCmdLine(void)
{
    CCmdLineTestData testdata[] = 
    {
       { _T(""), FALSE, 0, { { NULL, NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("--test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/-/test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("----test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/p1 /p2"), TRUE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/P1 /p2"), FALSE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/P1 /p2 /p2"), FALSE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/p1=p1value /p2=p2value /p3=p3value"), FALSE, 3, { { _T("p1"), _T("p1value") }, { _T("p2"), _T("p2value") }, { _T("p3"), _T("p3value") } } },
       { _T("/p1=\"p1value\" /p2=\'p2value\' /p3:`p3value`"), FALSE, 3, { { _T("p1"), _T("p1value") }, { _T("p2"), _T("p2value") }, { _T("p3"), _T("p3value") } } }      
    };
 
    for (int i = 0; i < sizeof(testdata) / sizeof(CCmdLineTestData); i++)
    {
        // std::wcout << testdata[i].pszCmdLine << std::endl;
        CCmdLine c(testdata[i].pszCmdLine, testdata[i].fCaseSensitive);
        CPPUNIT_ASSERT_MESSAGE("Invalid command line parsed.", c.GetCmdLine().Compare(testdata[i].pszCmdLine) == 0);
        CPPUNIT_ASSERT_MESSAGE("Invalid case sensitive parameter.", c.GetCaseSensitive() == testdata[i].fCaseSensitive);
        CPPUNIT_ASSERT_MESSAGE("Invalid parameter count.", c.GetCount() == testdata[i].ulParameters);
 
        for (int j = 0; j < sizeof(testdata[i].testDataPairs) / sizeof(CCmdLineTestDataPair); j++)
        {
            if (testdata[i].testDataPairs[j].pszName == NULL && testdata[i].testDataPairs[j].pszValue == NULL)
                break;
 
            CString strValue;
            CPPUNIT_ASSERT_MESSAGE("Expected parameter not found.", c.Lookup(testdata[i].testDataPairs[j].pszName, strValue));
            CPPUNIT_ASSERT_MESSAGE("Parameter value does not match.", 
                ((strValue.GetLength() == 0) && (testdata[i].testDataPairs[j].pszValue == NULL)) 
                || strValue.Compare(testdata[i].testDataPairs[j].pszValue) == 0);
        }
    }
 
}
 
cheers
dB.
GeneralRe: Supporting - - parameters and map.memberdB.19 Dec '05 - 11:05 
Someone asked me for full code.
 
CmdLine.h:
 
/*++
 
    Notes:
        Based on CCmdLine
        Copyright (c) Pavel Antonov, 2002
        http://www.codeproject.com/cpp/cmdlineparser.asp
        Changes by dB. (www.dblock.org)
 
    Formats:
        CommandLine::=[<Key> [,<Key>...]]
        <Key>::=<Delimeter>KeyName[<Separator><Value>]
        <Value> ::= { KeyValue | <QuoteChar>Quoted Key Value<QuoteChar>} ][
        <Delimeter>::= { - | / }
        <Separator>::= { : }
        <QuoteChar>::= { " }
--*/
 
#pragma once
 
class CCmdLine 
    : public CMapStringToString
{
    DECLARE_DYNAMIC(CCmdLine)
public:
    CCmdLine(LPCTSTR sCmdLine = NULL, BOOL bCaseSensitive = FALSE);
    virtual ~CCmdLine(void);
    void Parse(LPCTSTR pszCmdLine);
    inline const CString& GetCmdLine(void) const { return m_strCmdLine; }
    inline void SetCaseSensitive(BOOL bSensitive) { m_bCaseSensitive = bSensitive; }
    inline BOOL GetCaseSensitive(void) const { return m_bCaseSensitive; }
    inline BOOL HasKey(LPCTSTR pszKey) const { CString strValue; return CMapStringToString::Lookup(pszKey, strValue); }
private:
    CString m_strCmdLine;
    BOOL m_bCaseSensitive;
    static const TCHAR m_pszDelimeters[];
    static const TCHAR m_pszValueSep[];
    static const TCHAR m_pszQuotes[];
};
 
#ifdef CPPUNIT_API
class CCmdLineUnitTest : public CppUnit::TestCase 
{
public:
    void testParseCmdLine(void);
public:
    CPPUNIT_TEST_SUITE(CCmdLineUnitTest);
    CPPUNIT_TEST(testParseCmdLine);
    CPPUNIT_TEST_SUITE_END();
};
 
CPPUNIT_TEST_SUITE_REGISTRATION(CCmdLineUnitTest); 
#endif
 
CmdLine.cpp:
 
#include "stdafx.h"
 
IMPLEMENT_DYNAMIC(CCmdLine, CMapStringToString)
 
const TCHAR CCmdLine::m_pszDelimeters[] = _T("-/");
const TCHAR CCmdLine::m_pszQuotes[] = _T("\"\'`");
const TCHAR CCmdLine::m_pszValueSep[] = _T(" :="); // Space MUST be in set, but is not a name/value delimiter.

CCmdLine::CCmdLine(LPCTSTR pszCmdLine, BOOL bCaseSensitive)
    : m_bCaseSensitive(bCaseSensitive)
{
    if (pszCmdLine != NULL) 
    {
        Parse(pszCmdLine);
    }
}
 
CCmdLine::~CCmdLine()
{
 
}
 
void CCmdLine::Parse(LPCTSTR pszCmdLine) 
{
    if (! pszCmdLine)
        return;
    
    m_strCmdLine = pszCmdLine;
    RemoveAll();
 
    int nArgs = 0;
    LPCTSTR pszCurrent = pszCmdLine;
    while(TRUE) 
    {
        // /Key:"arg"
        if (_tcslen(pszCurrent) == 0) 
        { 
            // No data left
            break; 
        } 
 
        LPCTSTR pszArg = _tcspbrk(pszCurrent, m_pszDelimeters);
        if (! pszArg) 
        {
            // No delimeters found
            break; 
        }
 
        LPCTSTR pszInArg;
        do
        {
            pszArg = _tcsinc(pszArg);
            pszInArg = _tcspbrk(pszArg, m_pszDelimeters);
        } while (pszInArg == pszArg);
    
        // Key:"arg"
        if (_tcslen(pszArg) == 0) 
        {
            // String ends with delimeter
            break; 
        }
 
        LPCTSTR pszVal = _tcspbrk(pszArg, m_pszValueSep);
        if (pszVal == NULL) 
        { 
            // Key ends command line
            CString strKey(pszArg);
            if(! m_bCaseSensitive) 
            {
                strKey.MakeLower();
            }
            SetAt(strKey, _T(""));
            break;
        } 
        else if (pszVal[0] == _T(' ') || _tcslen(pszVal) == 1 ) 
        { 
            // Key with no value or cmdline ends with /Key:
            CString strKey(pszArg, (int) (pszVal - pszArg));
            if(! strKey.IsEmpty()) 
            { 
                // Prevent /: case
                if(!m_bCaseSensitive) 
                {
                    strKey.MakeLower();
                }
                SetAt(strKey, _T(""));
            }
 
            pszCurrent = _tcsinc(pszVal);
            continue;
        } 
        else 
        { 
            // Key with value
            CString strKey(pszArg, (int) (pszVal - pszArg));
            if (! m_bCaseSensitive) 
            {
                strKey.MakeLower();
            }
 
            pszVal = _tcsinc(pszVal);
            // "arg"
            LPCTSTR pszQuote = _tcspbrk(pszVal, m_pszQuotes), pszEndQuote(NULL);
            
            if (pszQuote == pszVal) 
            { 
                // Quoted String
                pszQuote = _tcsinc(pszVal);
                pszEndQuote = _tcspbrk(pszQuote, m_pszQuotes);
            } 
            else 
            {
                pszQuote = pszVal;
                pszEndQuote = _tcschr(pszQuote, _T(' '));
            }
 
            if (pszEndQuote == NULL) 
            { 
                // No end quotes or terminating space, take rest of string
                CString strValue(pszQuote);
                if(! strKey.IsEmpty()) 
                { 
                    // Prevent /:val case
                    SetAt(strKey, strValue);
                }
                break;
            } 
            else 
            { 
                // End quote or space present
                if(! strKey.IsEmpty()) 
                {    
                    // Prevent /:"val" case
                    CString strValue(pszQuote, (int) (pszEndQuote - pszQuote));
                    SetAt(strKey, strValue);
                }
                pszCurrent = _tcsinc(pszEndQuote);
                continue;
            }
        }
    }
}
 
#ifdef CPPUNIT_API
typedef struct
{
    LPCTSTR pszName;
    LPCTSTR pszValue;
} CCmdLineTestDataPair;
 
typedef struct 
{
    LPCTSTR pszCmdLine;
    BOOL fCaseSensitive;
    ULONG ulParameters;
    CCmdLineTestDataPair testDataPairs[3];
} CCmdLineTestData;
 
void CCmdLineUnitTest::testParseCmdLine(void)
{
    CCmdLineTestData testdata[] = 
    {
       { _T(""), FALSE, 0, { { NULL, NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("--test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/-/test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("----test"), FALSE, 1, { { _T("test"), NULL }, { NULL, NULL }, { NULL, NULL } } },
       { _T("/p1 /p2"), TRUE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/P1 /p2"), FALSE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/P1 /p2 /p2"), FALSE, 2, { { _T("p1"), NULL }, { _T("p2"), NULL }, { NULL, NULL } } },
       { _T("/p1=p1value /p2=p2value /p3=p3value"), FALSE, 3, { { _T("p1"), _T("p1value") }, { _T("p2"), _T("p2value") }, { _T("p3"), _T("p3value") } } },
       { _T("/p1=\"p1value\" /p2=\'p2value\' /p3:`p3value`"), FALSE, 3, { { _T("p1"), _T("p1value") }, { _T("p2"), _T("p2value") }, { _T("p3"), _T("p3value") } } }      
    };
 
    for (int i = 0; i < sizeof(testdata) / sizeof(CCmdLineTestData); i++)
    {
        // std::wcout << testdata[i].pszCmdLine << std::endl;
        CCmdLine c(testdata[i].pszCmdLine, testdata[i].fCaseSensitive);
        CPPUNIT_ASSERT_MESSAGE("Invalid command line parsed.", c.GetCmdLine().Compare(testdata[i].pszCmdLine) == 0);
        CPPUNIT_ASSERT_MESSAGE("Invalid case sensitive parameter.", c.GetCaseSensitive() == testdata[i].fCaseSensitive);
        CPPUNIT_ASSERT_MESSAGE("Invalid parameter count.", c.GetCount() == testdata[i].ulParameters);
 
        for (int j = 0; j < sizeof(testdata[i].testDataPairs) / sizeof(CCmdLineTestDataPair); j++)
        {
            if (testdata[i].testDataPairs[j].pszName == NULL && testdata[i].testDataPairs[j].pszValue == NULL)
                break;
 
            CString strValue;
            CPPUNIT_ASSERT_MESSAGE("Expected parameter not found.", c.Lookup(testdata[i].testDataPairs[j].pszName, strValue));
            CPPUNIT_ASSERT_MESSAGE("Parameter value does not match.", 
                ((strValue.GetLength() == 0) && (testdata[i].testDataPairs[j].pszValue == NULL)) 
                || strValue.Compare(testdata[i].testDataPairs[j].pszValue) == 0);
        }
    }
 
}
#endif

GeneralToo much blah....memberdouglash128 Oct '04 - 6:29 
Your coding style is too verbose.
 
You have created wrapper functions for every member variable. There are even member functions that simply call the appropriate stl::map method.
 
You have mixed CString with stl code - not good.
 
And your coding style reminds me of a C coder who once told me he could write an editor in 3 lines.
 
Go figure.
 
Still thanks for the code, I have chopped most of it but the algortihm works.

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Rant Rant    Admin Admin   

Permalink | Advertise | Privacy | Mobile
Web02 | 2.6.130523.1 | Last Updated 18 Aug 2002
Article Copyright 2002 by Pavel Antonov
Everything else Copyright © CodeProject, 1999-2013
Terms of Use
Layout: fixed | fluid