/////////////////////////////////////////////////////////////////////////////
-// Name: tokenzr.h
-// Purpose: String tokenizer
+// Name: wx/tokenzr.h
+// Purpose: String tokenizer - a C++ replacement for strtok(3)
// Author: Guilhem Lavaux
-// Modified by: Vadim Zeitlin
+// Modified by: (or rather rewritten by) Vadim Zeitlin
// Created: 04/22/98
// RCS-ID: $Id$
// Copyright: (c) Guilhem Lavaux
#ifndef _WX_TOKENZRH
#define _WX_TOKENZRH
-#ifdef __GNUG__
- #pragma interface "tokenzr.h"
-#endif
-
#include "wx/object.h"
#include "wx/string.h"
+#include "wx/arrstr.h"
+
+// ----------------------------------------------------------------------------
+// constants
+// ----------------------------------------------------------------------------
// default: delimiters are usual white space characters
#define wxDEFAULT_DELIMITERS (_T(" \t\r\n"))
-class WXDLLEXPORT wxStringTokenizer : public wxObject
+// wxStringTokenizer mode flags which determine its behaviour
+enum wxStringTokenizerMode
+{
+ wxTOKEN_INVALID = -1, // set by def ctor until SetString() is called
+ wxTOKEN_DEFAULT, // strtok() for whitespace delims, RET_EMPTY else
+ wxTOKEN_RET_EMPTY, // return empty token in the middle of the string
+ wxTOKEN_RET_EMPTY_ALL, // return trailing empty tokens too
+ wxTOKEN_RET_DELIMS, // return the delim with token (implies RET_EMPTY)
+ wxTOKEN_STRTOK // behave exactly like strtok(3)
+};
+
+// ----------------------------------------------------------------------------
+// wxStringTokenizer: replaces infamous strtok() and has some other features
+// ----------------------------------------------------------------------------
+
+class WXDLLIMPEXP_BASE wxStringTokenizer : public wxObject
{
public:
- // ctors and such
- wxStringTokenizer() { m_retdelims = FALSE; m_pos = 0; }
- wxStringTokenizer(const wxString& to_tokenize,
+ // ctors and initializers
+ // default ctor, call SetString() later
+ wxStringTokenizer() { m_mode = wxTOKEN_INVALID; }
+ // ctor which gives us the string
+ wxStringTokenizer(const wxString& str,
const wxString& delims = wxDEFAULT_DELIMITERS,
- bool ret_delim = FALSE);
- void SetString(const wxString& to_tokenize,
+ wxStringTokenizerMode mode = wxTOKEN_DEFAULT);
+
+ // args are same as for the non default ctor above
+ void SetString(const wxString& str,
const wxString& delims = wxDEFAULT_DELIMITERS,
- bool ret_delim = FALSE);
- virtual ~wxStringTokenizer();
+ wxStringTokenizerMode mode = wxTOKEN_DEFAULT);
- // count tokens/get next token
+ // reinitialize the tokenizer with the same delimiters/mode
+ void Reinit(const wxString& str);
+
+ // tokens access
+ // return the number of remaining tokens
size_t CountTokens() const;
- bool HasMoreTokens() { return m_hasMore; }
+ // did we reach the end of the string?
+ bool HasMoreTokens() const;
+ // get the next token, will return empty string if !HasMoreTokens()
wxString GetNextToken();
+ // get the delimiter which terminated the token last retrieved by
+ // GetNextToken() or NUL if there had been no tokens yet or the last
+ // one wasn't terminated (but ran to the end of the string)
+ wxChar GetLastDelimiter() const { return m_lastDelim; }
+
+ // get current tokenizer state
+ // returns the part of the string which remains to tokenize (*not* the
+ // initial string)
+ wxString GetString() const { return wxString(m_pos, m_string.end()); }
+
+ // returns the current position (i.e. one index after the last
+ // returned token or 0 if GetNextToken() has never been called) in the
+ // original string
+ size_t GetPosition() const { return m_pos - m_string.begin(); }
- // One note about GetString -- it returns the string
- // remaining after the previous tokens have been removed,
- // not the original string
- wxString GetString() const { return m_string; }
+ // misc
+ // get the current mode - can be different from the one passed to the
+ // ctor if it was wxTOKEN_DEFAULT
+ wxStringTokenizerMode GetMode() const { return m_mode; }
+ // do we return empty tokens?
+ bool AllowEmpty() const { return m_mode != wxTOKEN_STRTOK; }
- // returns the current position (i.e. one index after the last returned
- // token or 0 if GetNextToken() has never been called) in the original
- // string
- size_t GetPosition() const { return m_pos; }
+
+ // backwards compatibility section from now on
+ // -------------------------------------------
// for compatibility only, use GetNextToken() instead
wxString NextToken() { return GetNextToken(); }
+ // compatibility only, don't use
+ void SetString(const wxString& to_tokenize,
+ const wxString& delims,
+ bool WXUNUSED(ret_delim))
+ {
+ SetString(to_tokenize, delims, wxTOKEN_RET_DELIMS);
+ }
+
+ wxStringTokenizer(const wxString& to_tokenize,
+ const wxString& delims,
+ bool ret_delim)
+ {
+ SetString(to_tokenize, delims, ret_delim);
+ }
+
protected:
- wxString m_string, // the (rest of) string to tokenize
- m_delims; // all delimiters
+ bool IsOk() const { return m_mode != wxTOKEN_INVALID; }
+
+ bool DoHasMoreTokens() const;
- size_t m_pos; // the position in the original string
+ enum MoreTokensState
+ {
+ MoreTokens_Unknown,
+ MoreTokens_Yes,
+ MoreTokens_No
+ };
- bool m_retdelims; // if TRUE, return delims with tokens
- bool m_hasMore; // do we have more tokens?
+ MoreTokensState m_hasMoreTokens;
+
+ wxString m_string; // the string we tokenize
+ wxString::const_iterator m_stringEnd;
+ // FIXME-UTF8: use wxWcharBuffer
+ wxWxCharBuffer m_delims; // all possible delimiters
+ size_t m_delimsLen;
+
+ wxString::const_iterator m_pos; // the current position in m_string
+
+ wxStringTokenizerMode m_mode; // see wxTOKEN_XXX values
+
+ wxChar m_lastDelim; // delimiter after last token or '\0'
};
+// ----------------------------------------------------------------------------
+// convenience function which returns all tokens at once
+// ----------------------------------------------------------------------------
+
+// the function takes the same parameters as wxStringTokenizer ctor and returns
+// the array containing all tokens
+wxArrayString WXDLLIMPEXP_BASE
+wxStringTokenize(const wxString& str,
+ const wxString& delims = wxDEFAULT_DELIMITERS,
+ wxStringTokenizerMode mode = wxTOKEN_DEFAULT);
+
#endif // _WX_TOKENZRH
projects / wxWidgets.git / blobdiff