/////////////////////////////////////////////////////////////////////////////
-// Name: tokenzr.cpp
+// Name: src/common/tokenzr.cpp
// Purpose: String tokenizer
// Author: Guilhem Lavaux
// Modified by: Vadim Zeitlin (almost full rewrite)
#endif
#include "wx/tokenzr.h"
-#include "wx/arrstr.h"
+
+#ifndef WX_PRECOMP
+ #include "wx/arrstr.h"
+ #include "wx/crt.h"
+#endif
// Required for wxIs... functions
#include <ctype.h>
// implementation
// ============================================================================
+// ----------------------------------------------------------------------------
+// helpers
+// ----------------------------------------------------------------------------
+
+static wxString::const_iterator
+find_first_of(const wxChar *delims, size_t len,
+ const wxString::const_iterator& from,
+ const wxString::const_iterator& end)
+{
+ wxASSERT_MSG( from <= end, _T("invalid index") );
+
+ for ( wxString::const_iterator i = from; i != end; ++i )
+ {
+ if ( wxTmemchr(delims, *i, len) )
+ return i;
+ }
+
+ return end;
+}
+
+static wxString::const_iterator
+find_first_not_of(const wxChar *delims, size_t len,
+ const wxString::const_iterator& from,
+ const wxString::const_iterator& end)
+{
+ wxASSERT_MSG( from <= end, _T("invalid index") );
+
+ for ( wxString::const_iterator i = from; i != end; ++i )
+ {
+ if ( !wxTmemchr(delims, *i, len) )
+ return i;
+ }
+
+ return end;
+}
+
// ----------------------------------------------------------------------------
// wxStringTokenizer construction
// ----------------------------------------------------------------------------
// whitespace characters and as wxTOKEN_RET_EMPTY otherwise (for
// whitespace delimiters, strtok() behaviour is better because we want
// to count consecutive spaces as one delimiter)
- const wxChar *p;
- for ( p = delims.c_str(); *p; p++ )
+ wxString::const_iterator p;
+ for ( p = delims.begin(); p != delims.end(); ++p )
{
if ( !wxIsspace(*p) )
break;
}
- if ( *p )
+ if ( p != delims.end() )
{
// not whitespace char in delims
mode = wxTOKEN_RET_EMPTY;
}
}
- m_delims = delims;
+#if wxUSE_UNICODE // FIXME-UTF8: only wc_str()
+ m_delims = delims.wc_str();
+#else
+ m_delims = delims.mb_str();
+#endif
+ m_delimsLen = delims.length();
+
m_mode = mode;
Reinit(str);
wxASSERT_MSG( IsOk(), _T("you should call SetString() first") );
m_string = str;
- m_pos = 0;
-
- // empty string doesn't have any tokens
- m_hasMore = !m_string.empty();
+ m_stringEnd = m_string.end();
+ m_pos = m_string.begin();
+ m_lastDelim = _T('\0');
+ m_hasMoreTokens = MoreTokens_Unknown;
}
// ----------------------------------------------------------------------------
// do we have more of them?
bool wxStringTokenizer::HasMoreTokens() const
{
- wxCHECK_MSG( IsOk(), false, _T("you should call SetString() first") );
-
- if ( m_string.find_first_not_of(m_delims) == wxString::npos )
+ // GetNextToken() calls HasMoreTokens() and so HasMoreTokens() is called
+ // twice in every interation in the following common usage patten:
+ // while ( HasMoreTokens() )
+ // GetNextToken();
+ // We optimize this case by caching HasMoreTokens() return value here:
+ if ( m_hasMoreTokens == MoreTokens_Unknown )
{
- // no non empty tokens left, but in 2 cases we still may return true if
- // GetNextToken() wasn't called yet for this empty token:
- //
- // a) in wxTOKEN_RET_EMPTY_ALL mode we always do it
- // b) in wxTOKEN_RET_EMPTY mode we do it in the special case of a
- // string containing only the delimiter: then there is an empty
- // token just before it
- return (m_mode == wxTOKEN_RET_EMPTY_ALL) ||
- (m_mode == wxTOKEN_RET_EMPTY && m_pos == 0)
- ? m_hasMore : false;
+ bool r = DoHasMoreTokens();
+ wxConstCast(this, wxStringTokenizer)->m_hasMoreTokens =
+ r ? MoreTokens_Yes : MoreTokens_No;
+ return r;
}
else
+ return m_hasMoreTokens == MoreTokens_Yes;
+}
+
+bool wxStringTokenizer::DoHasMoreTokens() const
+{
+ wxCHECK_MSG( IsOk(), false, _T("you should call SetString() first") );
+
+ if ( find_first_not_of(m_delims, m_delimsLen, m_pos, m_stringEnd)
+ != m_stringEnd )
{
- // there are non delimiter characters left, hence we do have more
- // tokens
+ // there are non delimiter characters left, so we do have more tokens
return true;
}
+
+ switch ( m_mode )
+ {
+ case wxTOKEN_RET_EMPTY:
+ case wxTOKEN_RET_DELIMS:
+ // special hack for wxTOKEN_RET_EMPTY: we should return the initial
+ // empty token even if there are only delimiters after it
+ return !m_string.empty() && m_pos == m_string.begin();
+
+ case wxTOKEN_RET_EMPTY_ALL:
+ // special hack for wxTOKEN_RET_EMPTY_ALL: we can know if we had
+ // already returned the trailing empty token after the last
+ // delimiter by examining m_lastDelim: it is set to NUL if we run
+ // up to the end of the string in GetNextToken(), but if it is not
+ // NUL yet we still have this last token to return even if m_pos is
+ // already at m_string.length()
+ return m_pos < m_stringEnd || m_lastDelim != _T('\0');
+
+ case wxTOKEN_INVALID:
+ case wxTOKEN_DEFAULT:
+ wxFAIL_MSG( _T("unexpected tokenizer mode") );
+ // fall through
+
+ case wxTOKEN_STRTOK:
+ // never return empty delimiters
+ break;
+ }
+
+ return false;
}
-// count the number of tokens in the string
+// count the number of (remaining) tokens in the string
size_t wxStringTokenizer::CountTokens() const
{
wxCHECK_MSG( IsOk(), 0, _T("you should call SetString() first") );
// VZ: this function is IMHO not very useful, so it's probably not very
- // important if it's implementation here is not as efficient as it
- // could be - but OTOH like this we're sure to get the correct answer
+ // important if its implementation here is not as efficient as it
+ // could be -- but OTOH like this we're sure to get the correct answer
// in all modes
- wxStringTokenizer *self = (wxStringTokenizer *)this; // const_cast
- wxString stringInitial = m_string;
+ wxStringTokenizer tkz(wxString(m_pos, m_stringEnd), m_delims, m_mode);
size_t count = 0;
- while ( self->HasMoreTokens() )
+ while ( tkz.HasMoreTokens() )
{
count++;
- (void)self->GetNextToken();
+ (void)tkz.GetNextToken();
}
- self->Reinit(stringInitial);
-
return count;
}
wxString wxStringTokenizer::GetNextToken()
{
- // strtok() doesn't return empty tokens, all other modes do
- bool allowEmpty = m_mode != wxTOKEN_STRTOK;
-
wxString token;
do
{
{
break;
}
+
+ m_hasMoreTokens = MoreTokens_Unknown;
+
// find the end of this token
- size_t pos = m_string.find_first_of(m_delims);
+ wxString::const_iterator pos =
+ find_first_of(m_delims, m_delimsLen, m_pos, m_stringEnd);
// and the start of the next one
- if ( pos == wxString::npos )
+ if ( pos == m_stringEnd )
{
// no more delimiters, the token is everything till the end of
// string
- token = m_string;
+ token.assign(m_pos, m_stringEnd);
- m_pos += m_string.length();
- m_string.clear();
+ // skip the token
+ m_pos = m_stringEnd;
- // no more tokens in this string, even in wxTOKEN_RET_EMPTY_ALL
- // mode (we will return the trailing one right now in this case)
- m_hasMore = false;
+ // it wasn't terminated
+ m_lastDelim = _T('\0');
}
- else
+ else // we found a delimiter at pos
{
- size_t pos2 = pos + 1;
-
// in wxTOKEN_RET_DELIMS mode we return the delimiter character
- // with token
- token = wxString(m_string, m_mode == wxTOKEN_RET_DELIMS ? pos2
- : pos);
+ // with token, otherwise leave it out
+ wxString::const_iterator tokenEnd(pos);
+ if ( m_mode == wxTOKEN_RET_DELIMS )
+ ++tokenEnd;
+
+ token.assign(m_pos, tokenEnd);
- // remove token with the following it delimiter from string
- m_string.erase(0, pos2);
+ // skip the token and the trailing delimiter
+ m_pos = pos + 1;
- // keep track of the position in the original string too
- m_pos += pos2;
+ m_lastDelim = (pos == m_stringEnd) ? _T('\0') : (wxChar)*pos;
}
}
- while ( !allowEmpty && token.empty() );
+ while ( !AllowEmpty() && token.empty() );
return token;
}
projects / wxWidgets.git / blobdiff