C++ Codec::name方法代码示例

void ContentType::parse( const EString &s )
{
    EmailParser p( s );
    p.whitespace();
    while ( p.present( ":" ) )
        p.whitespace();

    bool mustGuess = false;

    if ( p.atEnd() ) {
        t = "text";
        st = "plain";
    }
    else {
        uint x = p.mark();
        if ( p.nextChar() == '/' )
            mustGuess = true;
        else
            t = p.mimeToken().lower();
        if ( p.atEnd() ) {
            if ( s == "text" ) {
                t = "text"; // elm? mailtool? someone does this, anyway.
                st = "plain";
            }
            // the remainder is from RFC 1049
            else if ( s == "postscript" ) {
                t = "application";
                st = "postscript";
            }
            else if ( s == "postscript" ) {
                t = "application";
                st = "postscript";
            }
            else if ( s == "sgml" ) {
                t = "text";
                st = "sgml";
            }
            else if ( s == "tex" ) {
                t = "application";
                st = "x-tex";
            }
            else if ( s == "troff" ) {
                t = "application";
                st = "x-troff";
            }
            else if ( s == "dvi" ) {
                t = "application";
                st = "x-dvi";
            }
            else if ( s.startsWith( "x-" ) ) {
                st = "x-rfc1049-" + s;
                t = "application";
            }
            else {
                // scribe and undefined types
                setError( "Invalid Content-Type: " + s.quoted() );
            }
        }
        else {
            if ( p.nextChar() == '/' ) {
                p.step();
                if ( !p.atEnd() || p.nextChar() != ';' )
                    st = p.mimeToken().lower();
                if ( st.isEmpty() )
                    mustGuess = true;
            }
            else if ( p.nextChar() == '=' ) {
                // oh no. someone skipped the content-type and
                // supplied only some parameters. we'll assume it's
                // text/plain and parse the parameters.
                t = "text";
                st = "plain";
                p.restore( x );
                mustGuess = true;
            }
            else {
                addParameter( "original-type", t + "/" + st );
                t = "application";
                st = "octet-stream";
                mustGuess = true;
            }
            parseParameters( &p );
        }
    }

    if ( mustGuess ) {
        EString fn = parameter( "name" );
        if ( fn.isEmpty() )
            fn = parameter( "filename" );
        while ( fn.endsWith( "." ) )
            fn.truncate( fn.length() - 1 );
        fn = fn.lower();
        if ( fn.endsWith( "jpg" ) || fn.endsWith( "jpeg" ) ) {
            t = "image";
            st = "jpeg";
        }
        else if ( fn.endsWith( "htm" ) || fn.endsWith( "html" ) ) {
            t = "text";
            st = "html";
        }
//.........这里部分代码省略.........

Bodypart * Bodypart::parseBodypart( uint start, uint end,
                                    const EString & rfc2822,
                                    Header * h, Multipart * parent )
{
    if ( rfc2822[start] == 13 )
        start++;
    if ( rfc2822[start] == 10 )
        start++;

    Bodypart * bp = new Bodypart;
    bp->setParent( parent );
    bp->setHeader( h );

    EString body;
    if ( end > start )
        body = rfc2822.mid( start, end-start );
    if ( !body.contains( '=' ) ) {
        // sometimes people send c-t-e: q-p _and_ c-t-e: 7bit or 8bit.
        // if they are equivalent we can accept it.
        uint i = 0;
        bool any = false;
        HeaderField * f = 0;
        while ( (f=h->field(HeaderField::ContentTransferEncoding,i)) != 0 ) {
            if ( ((ContentTransferEncoding*)f)->encoding() == EString::QP )
                any = true;
            i++;
        }
        if ( any && i > 1 )
            h->removeField( HeaderField::ContentTransferEncoding );
    }

    EString::Encoding e = EString::Binary;
    ContentTransferEncoding * cte = h->contentTransferEncoding();
    if ( cte )
        e = cte->encoding();
    if ( !body.isEmpty() ) {
        if ( e == EString::Base64 || e == EString::Uuencode )
            body = body.decoded( e );
        else
            body = body.crlf().decoded( e );
    }

    ContentType * ct = h->contentType();
    if ( !ct ) {
        switch ( h->defaultType() ) {
        case Header::TextPlain:
            h->add( "Content-Type", "text/plain" );
            break;
        case Header::MessageRfc822:
            h->add( "Content-Type", "message/rfc822" );
            break;
        }
        ct = h->contentType();
    }
    if ( ct->type() == "text" ) {
        bool specified = false;
        bool unknown = false;
        Codec * c = 0;

        if ( ct ) {
            EString csn = ct->parameter( "charset" );
            if ( csn.lower() == "default" )
                csn = "";
            if ( !csn.isEmpty() )
                specified = true;
            c = Codec::byName( csn );
            if ( !c )
                unknown = true;
            if ( c && c->name().lower() == "us-ascii" ) {
                // Some MTAs appear to say this in case there is no
                // Content-Type field - without checking whether the
                // body actually is ASCII. If it isn't, we'd better
                // call our charset guesser.
                (void)c->toUnicode( body );
                if ( !c->valid() )
                    specified = false;
                // Not pretty.
            }
        }

        if ( !c )
            c = new AsciiCodec;

        bp->d->hasText = true;
        bp->d->text = c->toUnicode( body.crlf() );

        if ( c->name() == "GB2312" || c->name() == "ISO-2022-JP" ||
             c->name() == "KS_C_5601-1987" ) {
            // undefined code point usage in GB2312 spam is much too
            // common. (GB2312 spam is much too common, but that's
            // another matter.) Gb2312Codec turns all undefined code
            // points into U+FFFD, so here, we can take the unicode
            // form and say it's the canonical form. when a client
            // later reads the message, it gets the text in unicode,
            // including U+FFFD.

            bool bad = !c->valid();

            // the header may contain some unencoded gb2312. we bang
            // it by hand, ignoring errors.
//.........这里部分代码省略.........

int main(int argc, char **argv)
{
    if (argc < 3)
        return 1;

    av::init();
    av::setFFmpegLoggingLevel(AV_LOG_TRACE);

    string uri (argv[1]);
    string out (argv[2]);

    ssize_t      audioStream = -1;
    AudioDecoderContext adec;
    Stream       ast;
    error_code   ec;

    int count = 0;

    {

        //
        // INPUT
        //
        FormatContext ictx;

        ictx.openInput(uri, ec);
        if (ec) {
            cerr << "Can't open input\n";
            return 1;
        }

        ictx.findStreamInfo();

        for (size_t i = 0; i < ictx.streamsCount(); ++i) {
            auto st = ictx.stream(i);
            if (st.isAudio()) {
                audioStream = i;
                ast = st;
                break;
            }
        }

        cerr << audioStream << endl;

        if (ast.isNull()) {
            cerr << "Audio stream not found\n";
            return 1;
        }

        if (ast.isValid()) {
            adec = AudioDecoderContext(ast);

            //Codec codec = findDecodingCodec(adec.raw()->codec_id);

            //adec.setCodec(codec);
            //adec.setRefCountedFrames(true);

            adec.open(ec);

            if (ec) {
                cerr << "Can't open codec\n";
                return 1;
            }
        }

        //
        // OUTPUT
        //
        OutputFormat  ofmt;
        FormatContext octx;

        ofmt = av::guessOutputFormat(out, out);
        clog << "Output format: " << ofmt.name() << " / " << ofmt.longName() << '\n';
        octx.setFormat(ofmt);

        Codec        ocodec = av::findEncodingCodec(ofmt, false);
        Stream      ost    = octx.addStream(ocodec);
        AudioEncoderContext enc (ost);

        clog << ocodec.name() << " / " << ocodec.longName() << ", audio: " << (ocodec.type()==AVMEDIA_TYPE_AUDIO) << '\n';

        auto sampleFmts  = ocodec.supportedSampleFormats();
        auto sampleRates = ocodec.supportedSamplerates();
        auto layouts     = ocodec.supportedChannelLayouts();

        clog << "Supported sample formats:\n";
        for (const auto &fmt : sampleFmts) {
            clog << "  " << av_get_sample_fmt_name(fmt) << '\n';
        }

        clog << "Supported sample rates:\n";
        for (const auto &rate : sampleRates) {
            clog << "  " << rate << '\n';
        }

        clog << "Supported sample layouts:\n";
        for (const auto &lay : layouts) {
            char buf[128] = {0};
            av_get_channel_layout_string(buf,
                                         sizeof(buf),
//.........这里部分代码省略.........

static Codec * guessHtmlCodec( const EString & body )
{
    // Let's see if the general function has something for us.
    Codec * guess = guessTextCodec( body );

    // HTML prescribes that 8859-1 is the default. Let's see if 8859-1
    // works.
    if ( !guess ) {
        guess = new Iso88591Codec;
        (void)guess->toUnicode( body );
        if ( !guess->valid() )
            guess = 0;
    }

    if ( !guess ||
         ( !guess->wellformed() &&
           ( guess->name() == "ISO-8859-1" ||
             guess->name() == "ISO-8859-15" ) ) ) {
        // Some people believe that Windows codepage 1252 is
        // ISO-8859-1. Let's see if that works.
        Codec * windoze = new Cp1252Codec;
        (void)windoze->toUnicode( body );
        if ( windoze->wellformed() )
            guess = windoze;
    }


    // Some user-agents add a <meta http-equiv="content-type"> instead
    // of the Content-Type field. Maybe that exists? And if it exists,
    // is it more likely to be correct than our guess above?

    EString b = body.lower().simplified();
    int i = 0;
    while ( i >= 0 ) {
        EString tag( "<meta http-equiv=\"content-type\" content=\"" );
        i = b.find( tag, i );
        if ( i >= 0 ) {
            i = i + tag.length();
            int j = i;
            while ( j < (int)b.length() && b[j] != '"' )
                j++;
            HeaderField * hf
                = HeaderField::create( "Content-Type",
                                       b.mid( i, j-i ) );
            EString cs = ((MimeField*)hf)->parameter( "charset" );
            Codec * meta = 0;
            if ( !cs.isEmpty() )
                meta = Codec::byName( cs );
            UString m;
            if ( meta )
                m = meta->toUnicode( body );
            UString g;
            if ( guess )
                g = guess->toUnicode( body );
            if ( meta &&
                 ( ( !m.isEmpty() && m == g ) ||
                   ( meta->wellformed() &&
                     ( !guess || !guess->wellformed() ) ) ||
                   ( meta->valid() && !guess ) ||
                   ( meta->valid() && guess &&
                     guess->name() == "ISO-8859-1" ) ||
                   ( meta->valid() && guess && !guess->valid() ) ) &&
                 meta->toUnicode( b ).ascii().contains( tag ) ) {
                guess = meta;
            }
        }
    }

    return guess;
}