diff options
author | tpearson <tpearson@283d02a7-25f6-0310-bc7c-ecb5cbfe19da> | 2011-01-07 03:45:53 +0000 |
---|---|---|
committer | tpearson <tpearson@283d02a7-25f6-0310-bc7c-ecb5cbfe19da> | 2011-01-07 03:45:53 +0000 |
commit | 10308be19ef7fa44699562cc75946e7ea1fdf6b9 (patch) | |
tree | 4bc444c00a79e88105f2cfce5b6209994c413ca0 /kdecore/kcharsets.cpp | |
parent | 307136d8eef0ba133b78ceee8e901138d4c996a1 (diff) | |
download | tdelibs-10308be19ef7fa44699562cc75946e7ea1fdf6b9.tar.gz tdelibs-10308be19ef7fa44699562cc75946e7ea1fdf6b9.zip |
Revert automated changes
Sorry guys, they are just not ready for prime time
Work will continue as always
git-svn-id: svn://anonsvn.kde.org/home/kde/branches/trinity/kdelibs@1212479 283d02a7-25f6-0310-bc7c-ecb5cbfe19da
Diffstat (limited to 'kdecore/kcharsets.cpp')
-rw-r--r-- | kdecore/kcharsets.cpp | 46 |
1 files changed, 23 insertions, 23 deletions
diff --git a/kdecore/kcharsets.cpp b/kdecore/kcharsets.cpp index d55060d49..52cf87849 100644 --- a/kdecore/kcharsets.cpp +++ b/kdecore/kcharsets.cpp @@ -202,7 +202,7 @@ static struct Builtin { "ascii", "iso 8859-1" }, { "x-utf-8", "utf-8" }, { "x-utf-7", "utf-7" }, // ### FIXME: UTF-7 is not in Qt - { "tqunicode-1-1-utf-7", "utf-7" }, // ### FIXME: UTF-7 is not in Qt + { "unicode-1-1-utf-7", "utf-7" }, // ### FIXME: UTF-7 is not in Qt { "utf-16", "iso-10646-ucs-2" }, { "utf16", "iso-10646-ucs-2" }, { "ucs2", "iso-10646-ucs-2" }, @@ -323,7 +323,7 @@ static struct ConversionHints { 0, 0 }}; -// search an array of items index/data, index is const char*, data is T, tqfind first matching index +// search an array of items index/data, index is const char*, data is T, find first matching index // and return data, or return 0 template< typename T, typename Data > static Data kcharsets_array_search( const T* start, const char* entry ) @@ -381,17 +381,17 @@ TQChar KCharsets::fromEntity(const TQString &str) if (str[pos] == (QChar)'x' || str[pos] == (QChar)'X') { pos++; // '�', hexadeciaml character reference - TQString tmp(str.tqunicode()+pos, str.length()-pos); + TQString tmp(str.unicode()+pos, str.length()-pos); res = tmp.toInt(&ok, 16); } else { // '�', decimal character reference - TQString tmp(str.tqunicode()+pos, str.length()-pos); + TQString tmp(str.unicode()+pos, str.length()-pos); res = tmp.toInt(&ok, 10); } return res; } - const entity *e = kde_tqfindEntity(str.ascii(), str.length()); + const entity *e = kde_findEntity(str.ascii(), str.length()); if(!e) { @@ -422,14 +422,14 @@ TQChar KCharsets::fromEntity(const TQString &str, int &len) TQString KCharsets::toEntity(const TQChar &ch) { TQString ent; - ent.sprintf("�x%x;", ch.tqunicode()); + ent.sprintf("�x%x;", ch.unicode()); return ent; } TQString KCharsets::resolveEntities( const TQString &input ) { TQString text = input; - const TQChar *p = text.tqunicode(); + const TQChar *p = text.unicode(); const TQChar *end = p + text.length(); const TQChar *ampersand = 0; bool scanForSemicolon = false; @@ -460,12 +460,12 @@ TQString KCharsets::resolveEntities( const TQString &input ) if ( entityValue.isNull() ) continue; - const uint ampersandPos = ampersand - text.tqunicode(); + const uint ampersandPos = ampersand - text.unicode(); text[ (int)ampersandPos ] = entityValue; text.remove( ampersandPos + 1, entityLength + 1 ); - p = text.tqunicode() + ampersandPos; - end = text.tqunicode() + text.length(); + p = text.unicode() + ampersandPos; + end = text.unicode() + text.length(); ampersand = 0; } @@ -477,7 +477,7 @@ TQStringList KCharsets::availableEncodingNames() TQStringList available; for ( const char* const* pos = charsets_for_encoding; *pos; ++pos ) { //kdDebug(0) << *charsets << " available" << endl; - available.append( TQString::tqfromLatin1( *pos )); + available.append( TQString::fromLatin1( *pos )); } return available; } @@ -491,14 +491,14 @@ TQString KCharsets::languageForEncoding( const TQString &encoding ) TQString KCharsets::encodingForName( const TQString &descriptiveName ) { - const int left = descriptiveName.tqfindRev( '(' ); + const int left = descriptiveName.findRev( '(' ); if (left<0) // No parenthesis, so assume it is a normal encoding name return descriptiveName.stripWhiteSpace(); TQString name(descriptiveName.mid(left+1)); - const int right = name.tqfindRev( ')' ); + const int right = name.findRev( ')' ); if (right<0) return name; @@ -511,7 +511,7 @@ TQStringList KCharsets::descriptiveEncodingNames() // As we are sorting, we can directly read the array language_for_encoding TQStringList encodings; for ( const LanguageForEncoding* pos = language_for_encoding; pos->index; ++pos ) { - const TQString name = TQString::tqfromLatin1( pos->index ); + const TQString name = TQString::fromLatin1( pos->index ); const TQString description = i18n( language_names[ pos->data ] ); encodings.append( i18n("Descriptive Encoding Name", "%1 ( %2 )"). arg ( description ). arg( name ) ); } @@ -536,7 +536,7 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const if (n.isEmpty()) { codec = KGlobal::locale()->codecForEncoding(); - d->codecForNameDict.tqreplace("->locale<-", codec); + d->codecForNameDict.replace("->locale<-", codec); return codec; } @@ -553,7 +553,7 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const codec = TQTextCodec::codecForName(name); if(codec) { - d->codecForNameDict.tqreplace(key, codec); + d->codecForNameDict.replace(key, codec); return codec; } @@ -566,14 +566,14 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const if(codec) { - d->codecForNameDict.tqreplace(key, codec); + d->codecForNameDict.replace(key, codec); return codec; } TQString dir; { KConfigGroupSaver cfgsav( KGlobal::config(), "i18n" ); - dir = KGlobal::config()->readPathEntry("i18ndir", TQString::tqfromLatin1("/usr/share/i18n/charmaps")); + dir = KGlobal::config()->readPathEntry("i18ndir", TQString::fromLatin1("/usr/share/i18n/charmaps")); } // these are codecs not included in Qt. They can be build up if the corresponding charmap @@ -584,8 +584,8 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const cname = name; cname = cname.upper(); - const TQString basicName = TQString::tqfromLatin1(cname); - kdDebug() << k_funcinfo << endl << " Trying to tqfind " << cname << " in " << dir << endl; + const TQString basicName = TQString::fromLatin1(cname); + kdDebug() << k_funcinfo << endl << " Trying to find " << cname << " in " << dir << endl; TQString charMapFileName; bool gzipped = false; @@ -603,7 +603,7 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const else { // Check if we are asking a code page // If yes, then check "CP99999" and "IBM99999" - // First we need to tqfind the number of the codepage + // First we need to find the number of the codepage TQRegExp regexp("^(X-)?(CP|IBM)(-| )?(0-9)+"); if ( regexp.search(basicName) != -1) { const TQString num = regexp.cap(4); @@ -642,7 +642,7 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const } if(codec) { - d->codecForNameDict.tqreplace(key, codec); + d->codecForNameDict.replace(key, codec); return codec; } @@ -654,7 +654,7 @@ TQTextCodec *KCharsets::codecForName(const TQString &n, bool &ok) const codec = TQTextCodec::codecForName(cname); if(codec) { - d->codecForNameDict.tqreplace(key, codec); + d->codecForNameDict.replace(key, codec); return codec; } |