You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
tdelibs/khtml/xml/dom_stringimpl.cpp

461 lines
11 KiB

/**
* This file is part of the DOM implementation for KDE.
*
* Copyright (C) 1999-2003 Lars Knoll (knoll@kde.org)
* (C) 1999 Antti Koivisto (koivisto@kde.org)
* (C) 2001-2003 Dirk Mueller ( mueller@kde.org )
* (C) 2002 Apple Computer, Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public License
* along with this library; see the file COPYING.LIB. If not, write to
* the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
* Boston, MA 02110-1301, USA.
*
*/
#include "dom_stringimpl.h"
#include <kdebug.h>
#include <string.h>
#include <tqstringlist.h>
using namespace DOM;
using namespace khtml;
DOMStringImpl::DOMStringImpl(const char *str)
{
if(str && *str)
{
l = strlen(str);
s = QT_ALLOC_QCHAR_VEC( l );
int i = l;
TQChar* ptr = s;
while( i-- )
*ptr++ = *str++;
}
else
{
s = QT_ALLOC_QCHAR_VEC( 1 ); // crash protection
s[0] = 0x0; // == TQChar::null;
l = 0;
}
}
// FIXME: should be a cached flag maybe.
bool DOMStringImpl::containsOnlyWhitespace() const
{
if (!s)
return true;
for (uint i = 0; i < l; i++) {
TQChar c = s[i];
if (c.unicode() <= 0x7F) {
if (c.unicode() > ' ')
return false;
} else {
if (c.direction() != TQChar::DirWS)
return false;
}
}
return true;
}
void DOMStringImpl::append(DOMStringImpl *str)
{
if(str && str->l != 0)
{
int newlen = l+str->l;
TQChar *c = QT_ALLOC_QCHAR_VEC(newlen);
memcpy(c, s, l*sizeof(TQChar));
memcpy(c+l, str->s, str->l*sizeof(TQChar));
if(s) QT_DELETE_QCHAR_VEC(s);
s = c;
l = newlen;
}
}
void DOMStringImpl::insert(DOMStringImpl *str, unsigned int pos)
{
if(pos > l)
{
append(str);
return;
}
if(str && str->l != 0)
{
int newlen = l+str->l;
TQChar *c = QT_ALLOC_QCHAR_VEC(newlen);
memcpy(c, s, pos*sizeof(TQChar));
memcpy(c+pos, str->s, str->l*sizeof(TQChar));
memcpy(c+pos+str->l, s+pos, (l-pos)*sizeof(TQChar));
if(s) QT_DELETE_QCHAR_VEC(s);
s = c;
l = newlen;
}
}
void DOMStringImpl::truncate(int len)
{
if(len > (int)l) return;
int nl = len < 1 ? 1 : len;
TQChar *c = QT_ALLOC_QCHAR_VEC(nl);
memcpy(c, s, nl*sizeof(TQChar));
if(s) QT_DELETE_QCHAR_VEC(s);
s = c;
l = len;
}
void DOMStringImpl::remove(unsigned int pos, int len)
{
if(pos >= l ) return;
if(pos+len > l)
len = l - pos;
uint newLen = l-len;
TQChar *c = QT_ALLOC_QCHAR_VEC(newLen);
memcpy(c, s, pos*sizeof(TQChar));
memcpy(c+pos, s+pos+len, (l-len-pos)*sizeof(TQChar));
if(s) QT_DELETE_QCHAR_VEC(s);
s = c;
l = newLen;
}
DOMStringImpl *DOMStringImpl::split(unsigned int pos)
{
if( pos >=l ) return new DOMStringImpl();
uint newLen = l-pos;
DOMStringImpl *str = new DOMStringImpl(s + pos, newLen);
truncate(pos);
return str;
}
DOMStringImpl *DOMStringImpl::substring(unsigned int pos, unsigned int len)
{
if( pos >=l ) return new DOMStringImpl();
if(pos+len > l)
len = l - pos;
return new DOMStringImpl(s + pos, len);
}
// Collapses white-space according to CSS 2.1 rules
DOMStringImpl *DOMStringImpl::collapseWhiteSpace(bool preserveLF, bool preserveWS)
{
if (preserveLF && preserveWS) return this;
// Notice we are likely allocating more space than needed (worst case)
TQChar *n = QT_ALLOC_QCHAR_VEC(l);
unsigned int pos = 0;
bool collapsing = false; // collapsing white-space
bool collapsingLF = false; // collapsing around linefeed
bool changedLF = false;
for(unsigned int i=0; i<l; i++) {
TQChar ch = s[i];
// We act on \r as we would on \n because CSS uses it to indicate new-line
if (ch == '\r') ch = '\n';
else
// ### The XML parser lets \t through, for now treat them as spaces
if (ch == '\t') ch = ' ';
if (!preserveLF && ch == '\n') {
// ### Not strictly correct according to CSS3 text-module.
// - In ideographic languages linefeed should be ignored
// - and in Thai and Khmer it should be treated as a zero-width space
ch = ' '; // Treat as space
changedLF = true;
}
if (collapsing) {
if (ch == ' ')
continue;
if (ch == '\n') {
collapsingLF = true;
continue;
}
n[pos++] = (collapsingLF) ? '\n' : ' ';
collapsing = false;
collapsingLF = false;
}
else
if (!preserveWS && ch == ' ') {
collapsing = true;
continue;
}
else
if (!preserveWS && ch == '\n') {
collapsing = true;
collapsingLF = true;
continue;
}
n[pos++] = ch;
}
if (collapsing)
n[pos++] = ((collapsingLF) ? '\n' : ' ');
if (pos == l && !changedLF) {
QT_DELETE_QCHAR_VEC(n);
return this;
}
else {
DOMStringImpl* out = new DOMStringImpl();
out->s = n;
out->l = pos;
return out;
}
}
static Length parseLength(const TQChar *s, unsigned int l)
{
if (l == 0) {
return Length(1, Relative);
}
unsigned i = 0;
while (i < l && s[i].isSpace())
++i;
if (i < l && (s[i] == '+' || s[i] == '-'))
++i;
while (i < l && s[i].isDigit())
++i;
bool ok;
int r = TQConstString(s, i).string().toInt(&ok);
/* Skip over any remaining digits, we are not that accurate (5.5% => 5%) */
while (i < l && (s[i].isDigit() || s[i] == '.'))
++i;
/* IE Quirk: Skip any whitespace (20 % => 20%) */
while (i < l && s[i].isSpace())
++i;
if (ok) {
if (i == l) {
return Length(r, Fixed);
} else {
const TQChar* next = s+i;
if (*next == '%')
return Length(r, Percent);
if (*next == '*')
return Length(r, Relative);
}
return Length(r, Fixed);
} else {
if (i < l) {
const TQChar* next = s+i;
if (*next == '*')
return Length(1, Relative);
if (*next == '%')
return Length(1, Relative);
}
}
return Length(0, Relative);
}
khtml::Length* DOMStringImpl::toCoordsArray(int& len) const
{
TQString str(s, l);
for(unsigned int i=0; i < l; i++) {
TQChar cc = s[i];
if (cc > '9' || (cc < '0' && cc != '-' && cc != '*' && cc != '.'))
str[i] = ' ';
}
str = str.simplifyWhiteSpace();
len = str.contains(' ') + 1;
khtml::Length* r = new khtml::Length[len];
int i = 0;
int pos = 0;
int pos2;
while((pos2 = str.find(' ', pos)) != -1) {
r[i++] = parseLength((TQChar *) str.unicode()+pos, pos2-pos);
pos = pos2+1;
}
r[i] = parseLength((TQChar *) str.unicode()+pos, str.length()-pos);
return r;
}
khtml::Length* DOMStringImpl::toLengthArray(int& len) const
{
TQString str(s, l);
str = str.simplifyWhiteSpace();
len = str.contains(',') + 1;
// If we have no commas, we have no array.
if( len == 1 )
return 0L;
khtml::Length* r = new khtml::Length[len];
int i = 0;
int pos = 0;
int pos2;
while((pos2 = str.find(',', pos)) != -1) {
r[i++] = parseLength((TQChar *) str.unicode()+pos, pos2-pos);
pos = pos2+1;
}
/* IE Quirk: If the last comma is the last char skip it and reduce len by one */
if (str.length()-pos > 0)
r[i] = parseLength((TQChar *) str.unicode()+pos, str.length()-pos);
else
len--;
return r;
}
bool DOMStringImpl::isLower() const
{
unsigned int i;
for (i = 0; i < l; i++)
if (s[i].lower() != s[i])
return false;
return true;
}
DOMStringImpl *DOMStringImpl::lower() const
{
DOMStringImpl *c = new DOMStringImpl;
if(!l) return c;
c->s = QT_ALLOC_QCHAR_VEC(l);
c->l = l;
for (unsigned int i = 0; i < l; i++)
c->s[i] = s[i].lower();
return c;
}
DOMStringImpl *DOMStringImpl::upper() const
{
DOMStringImpl *c = new DOMStringImpl;
if(!l) return c;
c->s = QT_ALLOC_QCHAR_VEC(l);
c->l = l;
for (unsigned int i = 0; i < l; i++)
c->s[i] = s[i].upper();
return c;
}
DOMStringImpl *DOMStringImpl::capitalize(bool noFirstCap) const
{
bool canCapitalize= !noFirstCap;
DOMStringImpl *c = new DOMStringImpl;
if(!l) return c;
c->s = QT_ALLOC_QCHAR_VEC(l);
c->l = l;
for (unsigned int i=0; i<l; i++)
{
if (s[i].isLetterOrNumber() && canCapitalize)
{
c->s[i]=s[i].upper();
canCapitalize=false;
}
else
{
c->s[i]=s[i];
if (s[i].isSpace())
canCapitalize=true;
}
}
return c;
}
TQString DOMStringImpl::string() const
{
return TQString(s, l);
}
int DOMStringImpl::toInt(bool* ok) const
{
// match \s*[+-]?\d*
unsigned i = 0;
while (i < l && s[i].isSpace())
++i;
if (i < l && (s[i] == '+' || s[i] == '-'))
++i;
while (i < l && s[i].isDigit())
++i;
return TQConstString(s, i).string().toInt(ok);
}
static const unsigned short amp[] = {'&', 'a', 'm', 'p', ';'};
static const unsigned short lt[] = {'&', 'l', 't', ';'};
static const unsigned short gt[] = {'&', 'g', 't', ';'};
DOMStringImpl *DOMStringImpl::escapeHTML()
{
unsigned outL = 0;
for (unsigned int i = 0; i < l; ++i ) {
if ( s[i] == '&' )
outL += 5; //&amp;
else if (s[i] == '<' || s[i] == '>')
outL += 4; //&gt;/&lt;
else
++outL;
}
if (outL == l)
return this;
DOMStringImpl* toRet = new DOMStringImpl();
toRet->s = QT_ALLOC_QCHAR_VEC(outL);
toRet->l = outL;
unsigned outP = 0;
for (unsigned int i = 0; i < l; ++i ) {
if ( s[i] == '&' ) {
memcpy(&toRet->s[outP], amp, sizeof(amp));
outP += 5;
} else if (s[i] == '<') {
memcpy(&toRet->s[outP], lt, sizeof(lt));
outP += 4;
} else if (s[i] == '>') {
memcpy(&toRet->s[outP], gt, sizeof(gt));
outP += 4;
} else {
toRet->s[outP] = s[i];
++outP;
}
}
return toRet;
}