/****************************************************************************

** 

**

** Implementation of QTextCodec class

**

** Created : 981015

**

** Copyright (C)1998-2000 Trolltech AS.  All rights reserved.

**

** This file is part of the tools module of the Qt GUI Toolkit.

**

** This file may be distributed under the terms of the Q Public License

** as defined by Trolltech AS of Norway and appearing in the file

** LICENSE.QPL included in the packaging of this file.

**

** This file may be distributed and/or modified under the terms of the

** GNU General Public License version 2 as published by the Free Software

** Foundation and appearing in the file LICENSE.GPL included in the

** packaging of this file.

**

** Licensees holding valid Qt Enterprise Edition or Qt Professional Edition

** licenses may use this file in accordance with the Qt Commercial License

** Agreement provided with the Software.

**

** This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE

** WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

**

** See http://www.trolltech.com/pricing.html or email sales@trolltech.com for

**   information about Qt Commercial License Agreements.

** See http://www.trolltech.com/qpl/ for QPL licensing information.

** See http://www.trolltech.com/gpl/ for GPL licensing information.

**

** Contact info@trolltech.com if any conditions of this licensing are

** not clear to you.

**

**********************************************************************/

#include "qtextcodec.h"

#ifndef QT_NO_TEXTCODEC

#include "qlist.h"

#ifndef QT_NO_CODECS

#include "qutfcodec.h"

#include "qgbkcodec.h"

#include "qeucjpcodec.h"

#include "qjiscodec.h"

#include "qsjiscodec.h"

#include "qeuckrcodec.h"

#include "qbig5codec.h"

#include "qrtlcodec.h"

#include "qtsciicodec.h"

#endif

#include "qfile.h"

#include "qstrlist.h"

#include "qstring.h"

#include <stdlib.h>

#include <ctype.h>

#include <locale.h>

static QList<QTextCodec> * all = 0;

static bool destroying_is_ok; // starts out as 0

/*!  Deletes all the created codecs.

  \warning Do not call this function.

  QApplication calls this just before exiting, to delete any

  QTextCodec objects that may be lying around.  Since various other

  classes hold pointers to QTextCodec objects, it is not safe to call

  this function earlier.

  If you are using the utility classes (like QString) but not using

  QApplication, calling this function at the very end of your

  application can be helpful to chasing down memory leaks, as

  QTextCodec objects will not show up.

*/

void QTextCodec::deleteAllCodecs()

{

    if ( !all )

        return;

    destroying_is_ok = TRUE;

    QList<QTextCodec> * ball = all;

    all = 0;

    ball->clear();

    delete ball;

    destroying_is_ok = FALSE;

}

static void setupBuiltinCodecs();

static void realSetup()

{

#if defined(CHECK_STATE)

    if ( destroying_is_ok )

        qWarning( "creating new codec during codec cleanup" );

#endif

    all = new QList<QTextCodec>;

    all->setAutoDelete( TRUE );

    setupBuiltinCodecs();

}

static inline void setup()

{

    if ( !all )

        realSetup();

}

class QTextStatelessEncoder: public QTextEncoder {

    const QTextCodec* codec;

public:

    QTextStatelessEncoder(const QTextCodec*);

    QCString fromUnicode(const QString& uc, int& lenInOut);

};

class QTextStatelessDecoder : public QTextDecoder {

    const QTextCodec* codec;

public:

    QTextStatelessDecoder(const QTextCodec*);

    QString toUnicode(const char* chars, int len);

};

QTextStatelessEncoder::QTextStatelessEncoder(const QTextCodec* c) :

    codec(c)

{

}

QCString QTextStatelessEncoder::fromUnicode(const QString& uc, int& lenInOut)

{

    return codec->fromUnicode(uc,lenInOut);

}

QTextStatelessDecoder::QTextStatelessDecoder(const QTextCodec* c) :

    codec(c)

{

}

QString QTextStatelessDecoder::toUnicode(const char* chars, int len)

{

    return codec->toUnicode(chars,len);

}

// NOT REVISED

/*!

  \class QTextCodec qtextcodec.h

  \brief Provides conversion between text encodings.

  By making objects of subclasses of QTextCodec, support for

  new text encodings can be added to Qt.

  The abstract virtual functions describe the encoder to the

  system and the coder is used as required in the different

  text file formats supported QTextStream and, under X11 for the

  locale-specific character input and output (under Windows NT

  codecs are not needed for GUI I/O since the system works

  with Unicode already, and Windows 95/98 has built-in convertors

  for the 8-bit local encoding).

  More recently created QTextCodec objects take precedence

  over earlier ones.

  To add support for another 8-bit encoding to Qt, make a subclass

  or QTextCodec and implement at least the following methods:

  <dl>

   <dt>\c const char* name() const

    <dd>Return the official name for the encoding.

   <dt>\c int mibEnum() const

    <dd>Return the MIB enum for the encoding if it is listed in the

      <a href=ftp://ftp.isi.edu/in-notes/iana/assignments/character-sets>

      IANA character-sets encoding file</a>.

  </dl>

  If the encoding is multi-byte then it will have "state"; that is,

  the interpretation of some bytes will be dependent on some preceding

  bytes.  For such an encoding, you will need to implement

  <dl>

   <dt> \c QTextDecoder* makeDecoder() const

    <dd>Return a QTextDecoder that remembers incomplete multibyte

        sequence prefixes or other required state.

  </dl>

  If the encoding does \e not require state, you should implement:

  <dl>

   <dt> \c QString toUnicode(const char* chars, int len) const

    <dd>Converts \e len characters from \e chars to Unicode.

  </dl>

  The base QTextCodec class has default implementations of the above

  two functions, <i>but they are mutually recursive</i>, so you must

  re-implement at least one of them, or both for improved efficiency.

  For conversion from Unicode to 8-bit encodings, it is rarely necessary

  to maintain state.  However, two functions similar to the two above

  are used for encoding:

  <dl>

   <dt> \c QTextEncoder* makeEncoder() const

    <dd>Return a QTextDecoder.

   <dt> \c QCString fromUnicode(const QString& uc, int& lenInOut ) const;

    <dd>Converts \e lenInOut characters (of type QChar) from the start

        of the string \a uc, returning a QCString result, and also returning

        the \link QCString::length() length\endlink

        of the result in lenInOut.

  </dl>

  Again, these are mutually recursive so only one needs to be implemented,

  or both if better efficiency is possible.

  Finally, you must implement:

  <dl>

   <dt> \c int heuristicContentMatch(const char* chars, int len) const

    <dd>Gives a value indicating how likely it is that \e len characters

        from \e chars are in the encoding.

  </dl>

  A good model for this function is the

  QWindowsLocalCodec::heuristicContentMatch function found in the Qt sources.

  A QTextCodec subclass might have improved performance if you also

  re-implement:

  <dl>

   <dt> \c bool canEncode( QChar ) const

    <dd>Test if a Unicode character can be encoded.

   <dt> \c bool canEncode( const QString& ) const

    <dd>Test if a string of Unicode characters can be encoded.

   <dt> \c int heuristicNameMatch(const char* hint) const

    <dd>Test if a possibly non-standard name is referring to the codec.

  </dl>

*/

/*!

  Constructs a QTextCodec, making it of highest precedence.

  The QTextCodec should always be constructed on the heap

  (with new), and once constructed it becomes the responsibility

  of Qt to delete it (which is done at QApplication destruction).

*/

QTextCodec::QTextCodec()

{

    setup();

    all->insert(0,this);

}

/*!

  Destructs the QTextCodec.  Note that you should not delete

  codecs yourself - once created they become the responsibility

  of Qt to delete.

*/

QTextCodec::~QTextCodec()

{

    if ( !destroying_is_ok )

        qWarning("QTextCodec::~QTextCodec() called by application");

    if ( all )

        all->remove( this );

}

/*!

  Returns a value indicating how likely this decoder is

  for decoding some format that has the given name.

  A good match returns a positive number around

  the length of the string.  A bad match is negative.

  The default implementation calls simpleHeuristicNameMatch()

  with the name of the codec.

*/

int QTextCodec::heuristicNameMatch(const char* hint) const

{

    return simpleHeuristicNameMatch(name(),hint);

}

// returns a string cotnaining the letters and numbers from input,

// with a space separating run of a character class.  e.g. "iso8859-1"

// becomes "iso 8859 1"

static QString lettersAndNumbers( const char * input )

{

    QString result;

    QChar c;

    while( input && *input ) {

        c = *input;

        if ( c.isLetter() || c.isNumber() )

            result += c.lower();

        if ( input[1] ) {

            // add space at character class transition, except

            // transition from upper-case to lower-case letter

            QChar n( input[1] );

            if ( c.isLetter() && n.isLetter() ) {

                if ( c == c.lower() && n == n.upper() )

                    result += ' ';

            } else if ( c.category() != n.category() ) {

                result += ' ';

            }

        }

        input++;

    }

    return result.simplifyWhiteSpace();

}

/*!

  A simple utility function for heuristicNameMatch() - it

  does some very minor character-skipping

  so that almost-exact matches score high.

*/

int QTextCodec::simpleHeuristicNameMatch(const char* name, const char* hint)

{

    // if they're the same, return a perfect score.

    if ( name && hint && qstrcmp( name, hint ) == 0 )

        return qstrlen( hint );

    // if the letters and numbers are the same, we have an "almost"

    // perfect match.

    QString h( lettersAndNumbers( hint ) );

    QString n( lettersAndNumbers( name ) );

    if ( h == n )

        return qstrlen( hint )-1;

    if ( h.stripWhiteSpace() == n.stripWhiteSpace() )

        return qstrlen( hint )-2;

    // could do some more here, but I don't think it's worth it

    return 0;

}

/*!

  Returns the QTextCodec \a i places from the more recently

  inserted, or NULL if there is no such QTextCodec.  Thus,

  codecForIndex(0) returns the most recently created QTextCodec.

*/

QTextCodec* QTextCodec::codecForIndex(int i)

{

    setup();

    return (uint)i >= all->count() ? 0 : all->at(i);

}

/*!

  Returns the QTextCodec which matches the

  \link QTextCodec::mibEnum() MIBenum\endlink \a mib.

*/

QTextCodec* QTextCodec::codecForMib(int mib)

{

    setup();

    QListIterator<QTextCodec> i(*all);

    QTextCodec* result;

    for ( ; (result=i); ++i ) {

        if ( result->mibEnum()==mib )

            break;

    }

    return result;

}

#ifdef _OS_WIN32_

class QWindowsLocalCodec: public QTextCodec

{

public:

    QWindowsLocalCodec();

    ~QWindowsLocalCodec();

    QString toUnicode(const char* chars, int len) const;

    QCString fromUnicode(const QString& uc, int& lenInOut ) const;

    const char* name() const;

    int mibEnum() const;

    int heuristicContentMatch(const char* chars, int len) const;

};

QWindowsLocalCodec::QWindowsLocalCodec()

{

}

QWindowsLocalCodec::~QWindowsLocalCodec()

{

}

QString QWindowsLocalCodec::toUnicode(const char* chars, int len) const

{

    if ( len == 1 && chars ) {          // Optimization; avoids allocation

        char c[2];

        c[0] = *chars;

        c[1] = 0;

        return qt_winMB2QString( c, 2 );

    }

    if ( len < 0 )

        return qt_winMB2QString( chars );

    QCString s(chars,len+1);

    return qt_winMB2QString(s);

}

QCString QWindowsLocalCodec::fromUnicode(const QString& uc, int& lenInOut ) const

{

    QCString r = qt_winQString2MB( uc, lenInOut );

    lenInOut = r.length();

    return r;

}

const char* QWindowsLocalCodec::name() const

{

    return "System";

}

int QWindowsLocalCodec::mibEnum() const

{

    return 0;

}

int QWindowsLocalCodec::heuristicContentMatch(const char* chars, int len) const

{

    // ### Not a bad default implementation?

    QString t = toUnicode(chars,len);

    int l = t.length();

    QCString mb = fromUnicode(t,l);

    int i=0;

    while ( i < len )

        if ( chars[i] == mb[i] )

            i++;

    return i;

}

#else

/* locale names mostly copied from XFree86 */

static const char * const iso8859_2locales[] = {

    "croatian", "cs", "cs_CS", "cs_CZ","cz", "cz_CZ", "czech", "hr",

    "hr_HR", "hu", "hu_HU", "hungarian", "pl", "pl_PL", "polish", "ro",

    "ro_RO", "rumanian", "serbocroatian", "sh", "sh_SP", "sh_YU", "sk",

    "sk_SK", "sl", "sl_CS", "sl_SI", "slovak", "slovene", "sr_SP", 0 };

static const char * const iso8859_3locales[] = {

    "eo", 0 };

static const char * const iso8859_4locales[] = {

    "ee", "ee_EE", "lt", "lt_LT", "lv", "lv_LV", 0 };

static const char * const iso8859_5locales[] = {

    "bg", "bg_BG", "bulgarian", "mk", "mk_MK",

    "sp", "sp_YU", 0 };

static const char * const iso8859_6locales[] = {

    "ar_AA", "ar_SA", "arabic", 0 };

static const char * const iso8859_7locales[] = {

    "el", "el_GR", "greek", 0 };

static const char * const iso8859_8locales[] = {

    "hebrew", "he", "he_IL", "iw", "iw_IL", 0 };

static const char * const iso8859_9locales[] = {

    "tr", "tr_TR", "turkish", 0 };

static const char * const iso8859_15locales[] = {

    "fr", "fi", "french", "finnish", "et", "et_EE", 0 };

static const char * const koi8_ulocales[] = {

    "uk", "uk_UA", "ru_UA", "ukrainian", 0 };

static const char * const tis_620locales[] = {

    "th", "th_TH", "thai", 0 };

static bool try_locale_list( const char * const locale[], const char * lang )

{

    int i;

    for( i=0; locale[i] && qstrcmp(locale[i], lang); i++ )

    { }

    return locale[i] != 0;

}

// For the probably_koi8_locales we have to look. the standard says

// these are 8859-5, but almsot all Russion users uses KOI8-R and

// incorrectly set $LANG to ru_RU. We'll check tolower() to see what

// tolower() thinks ru_RU means.

// If you read the history, it seems that many Russians blame ISO and

// Peristroika for the confusion.

//

// The real bug is that some programs break if the user specifies

// ru_RU.KOI8-R.

static const char * const probably_koi8_rlocales[] = {

    "ru", "ru_SU", "ru_RU", "russian", 0 };

// this means ANY of these locale aliases. if they're aliases for

// different locales, the code breaks.

static QTextCodec * ru_RU_codec = 0;

static QTextCodec * ru_RU_hack( const char * i ) {

    if ( ! ru_RU_codec ) {

        QCString origlocale = setlocale( LC_CTYPE, i );

        // unicode   koi8r   latin5   name

        // 0x044E    0xC0    0xEE     CYRILLIC SMALL LETTER YU

        // 0x042E    0xE0    0xCE     CYRILLIC CAPITAL LETTER YU

        int latin5 = tolower( 0xCE );

        int koi8r = tolower( 0xE0 );

        if ( koi8r == 0xC0 && latin5 != 0xEE ) {

            ru_RU_codec = QTextCodec::codecForName( "KOI8-R" );

        } else if ( koi8r != 0xC0 && latin5 == 0xEE ) {

            ru_RU_codec = QTextCodec::codecForName( "ISO 8859-5" );

        } else {

            // something else again... let's assume... *throws dice*

            ru_RU_codec = QTextCodec::codecForName( "KOI8-R" );

            qWarning( "QTextCodec: using KOI8-R, probe failed (%02x %02x %s)",

                      koi8r, latin5, i );

        }

        setlocale( LC_CTYPE, origlocale.data() );

    }

    return ru_RU_codec;