#include <regex.h>
#include "wx/regex.h"
+// WXREGEX_USING_BUILTIN defined when using the built-in regex lib
+// WXREGEX_BUILTIN_ONLY() wrap a parameter only used with the built-in regex
+// WXREGEX_CONVERT_TO_MB indicates when the regex lib is using chars and
+// wxChar is wide, so conversion must be done
+#ifdef __REG_NOFRONT
+# define WXREGEX_USING_BUILTIN
+# define WXREGEX_BUILTIN_ONLY(x) ,x
+#else
+# define WXREGEX_BUILTIN_ONLY(x)
+# if wxUSE_UNICODE
+# define WXREGEX_CONVERT_TO_MB
+# endif
+#endif
+
// ----------------------------------------------------------------------------
// private classes
// ----------------------------------------------------------------------------
+// the character type used by the regular expression engine
+#ifndef WXREGEX_CONVERT_TO_MB
+typedef wxChar wxRegChar;
+#else
+typedef char wxRegChar;
+#endif
+
// the real implementation of wxRegEx
class wxRegExImpl
{
// RE operations
bool Compile(const wxString& expr, int flags = 0);
- bool Matches(const wxChar *str, int flags = 0) const;
+ bool Matches(const wxRegChar *str, int flags
+ WXREGEX_BUILTIN_ONLY(size_t len)) const;
bool GetMatch(size_t *start, size_t *len, size_t index = 0) const;
size_t GetMatchCount() const;
int Replace(wxString *pattern, const wxString& replacement,
wxString wxRegExImpl::GetErrorMsg(int errorcode, bool badconv) const
{
-#if wxUSE_UNICODE && !defined(__REG_NOFRONT)
+#ifdef WXREGEX_CONVERT_TO_MB
// currently only needed when using system library in Unicode mode
if ( badconv )
{
flagsRE |= REG_NEWLINE;
// compile it
-#ifdef __REG_NOFRONT
+#ifdef WXREGEX_USING_BUILTIN
bool conv = true;
int errorcode = wx_re_comp(&m_RegEx, expr, expr.length(), flagsRE);
#else
return IsValid();
}
-bool wxRegExImpl::Matches(const wxChar *str, int flags) const
+bool wxRegExImpl::Matches(const wxRegChar *str,
+ int flags
+ WXREGEX_BUILTIN_ONLY(size_t len)) const
{
wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
}
// do match it
-#ifdef __REG_NOFRONT
- bool conv = true;
- int rc = wx_re_exec(&self->m_RegEx, str, wxStrlen(str), NULL, m_nMatches, m_Matches, flagsRE);
+#ifdef WXREGEX_USING_BUILTIN
+ int rc = wx_re_exec(&self->m_RegEx, str, len, NULL, m_nMatches, m_Matches, flagsRE);
#else
- const wxWX2MBbuf conv = wxConvertWX2MB(str);
- int rc = conv ? regexec(&self->m_RegEx, conv, m_nMatches, m_Matches, flagsRE) : REG_BADPAT;
+ int rc = str ? regexec(&self->m_RegEx, str, m_nMatches, m_Matches, flagsRE) : REG_BADPAT;
#endif
switch ( rc )
default:
// an error occurred
- wxLogError(_("Failed to match '%s' in regular expression: %s"),
- str, GetErrorMsg(rc, !conv).c_str());
+ wxLogError(_("Failed to find match for regular expression: %s"),
+ GetErrorMsg(rc, !str).c_str());
// fall through
case REG_NOMATCH:
wxCHECK_MSG( text, wxNOT_FOUND, _T("NULL text in wxRegEx::Replace") );
wxCHECK_MSG( IsValid(), wxNOT_FOUND, _T("must successfully Compile() first") );
+ // the input string
+#ifndef WXREGEX_CONVERT_TO_MB
+ const wxChar *textstr = text->c_str();
+ size_t textlen = text->length();
+#else
+ const wxWX2MBbuf textstr = wxConvertWX2MB(*text);
+ if (!textstr)
+ {
+ wxLogError(_("Failed to find match for regular expression: %s"),
+ GetErrorMsg(0, true).c_str());
+ return 0;
+ }
+ size_t textlen = strlen(textstr);
+ text->clear();
+#endif
+
// the replacement text
wxString textNew;
+ // the result, allow 25% extra
+ wxString result;
+ result.reserve(5 * textlen / 4);
+
// attempt at optimization: don't iterate over the string if it doesn't
// contain back references at all
bool mayHaveBackrefs =
}
// the position where we start looking for the match
- //
- // NB: initial version had a nasty bug because it used a wxChar* instead of
- // an index but the problem is that replace() in the loop invalidates
- // all pointers into the string so we have to use indices instead
size_t matchStart = 0;
// number of replacement made: we won't make more than maxMatches of them
// note that "^" shouldn't match after the first call to Matches() so we
// use wxRE_NOTBOL to prevent it from happening
while ( (!maxMatches || countRepl < maxMatches) &&
- Matches(text->c_str() + matchStart, countRepl ? wxRE_NOTBOL : 0) )
+ Matches(textstr + matchStart,
+ countRepl ? wxRE_NOTBOL : 0
+ WXREGEX_BUILTIN_ONLY(textlen - matchStart)) )
{
// the string possibly contains back references: we need to calculate
// the replacement text anew after each match
}
else
{
- textNew += wxString(text->c_str() + matchStart + start,
- len);
+ textNew += wxString(textstr + matchStart + start,
+ *wxConvCurrent, len);
mayHaveBackrefs = true;
}
return wxNOT_FOUND;
}
+ // an insurance against implementations that don't grow exponentially
+ // to ensure building the result takes linear time
+ if (result.capacity() < result.length() + start + textNew.length())
+ result.reserve(2 * result.length());
+
+#ifndef WXREGEX_CONVERT_TO_MB
+ result.append(*text, matchStart, start);
+#else
+ result.append(wxString(textstr + matchStart, *wxConvCurrent, start));
+#endif
matchStart += start;
- text->replace(matchStart, len, textNew);
+ result.append(textNew);
countRepl++;
- matchStart += textNew.length();
+ matchStart += len;
}
+#ifndef WXREGEX_CONVERT_TO_MB
+ result.append(*text, matchStart, wxString::npos);
+#else
+ result.append(wxString(textstr + matchStart, *wxConvCurrent));
+#endif
+ *text = result;
+
return countRepl;
}
m_impl = NULL;
}
-
wxRegEx::~wxRegEx()
{
delete m_impl;
return true;
}
+bool wxRegEx::Matches(const wxChar *str, int flags, size_t len) const
+{
+ wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
+ (void)len;
+
+#ifdef WXREGEX_CONVERT_TO_MB
+ return m_impl->Matches(wxConvertWX2MB(str), flags);
+#else
+ return m_impl->Matches(str, flags WXREGEX_BUILTIN_ONLY(len));
+#endif
+}
+
bool wxRegEx::Matches(const wxChar *str, int flags) const
{
wxCHECK_MSG( IsValid(), false, _T("must successfully Compile() first") );
- return m_impl->Matches(str, flags);
+#ifdef WXREGEX_CONVERT_TO_MB
+ return m_impl->Matches(wxConvertWX2MB(str), flags);
+#else
+ return m_impl->Matches(str, flags WXREGEX_BUILTIN_ONLY(wxStrlen(str)));
+#endif
}
bool wxRegEx::GetMatch(size_t *start, size_t *len, size_t index) const