SCSUUTF8 Class Reference

#include <scsuutf8.h>

Inheritance diagram for SCSUUTF8:

Collaboration diagram for SCSUUTF8:

Public Member Functions
virtual const char *	getHeader () const
virtual char	processText (SWBuf &text, const SWKey *key=0, const SWModule *module=0)
	SCSUUTF8 ()
	~SCSUUTF8 ()

Private Member Functions
int	UTF8Output (unsigned long, SWBuf *utf8Buf)

Private Attributes
unsigned char	active
unsigned long	c
unsigned long	d
bool	mode

Static Private Attributes
static unsigned short	slide [8] = {0x0080,0x00C0,0x0400,0x0600,0x0900,0x3040,0x30A0,0xFF00}
static unsigned short	start [8] = {0x0000,0x0080,0x0100,0x0300,0x2000,0x2080,0x2100,0x3000}
static unsigned short	win [256]

Detailed Description

This filter converts SCSU compressed (encoded) text to UTF-8

Definition at line 38 of file scsuutf8.h.

Constructor & Destructor Documentation

SWORD_NAMESPACE_START SCSUUTF8::SCSUUTF8

(

)

Definition at line 43 of file scsuutf8.cpp.

 : err()
#endif
{
#ifdef _ICU_
    
    // initialize SCSU converter
    scsuConv = ucnv_open("SCSU", &err);
    // initialize UTF-8 converter
    utf8Conv = ucnv_open("UTF-8", &err);    
#else
    active = 0;
    mode = 0;
#endif
}

SCSUUTF8::~SCSUUTF8

(

)

Definition at line 60 of file scsuutf8.cpp.

                    {
#ifdef _ICU_
    ucnv_close(scsuConv);
    ucnv_close(utf8Conv);
#endif
}

Member Function Documentation

virtual const char* SWFilter::getHeader ( ) const

inlinevirtualinherited

This method can supply a header associated with the processing done with this filter. A typical example is a suggested CSS style block for classed containers.

Reimplemented in OSISLaTeX, OSISXHTML, ThMLLaTeX, ThMLXHTML, TEIXHTML, GBFLaTeX, and GBFXHTML.

Definition at line 62 of file swfilter.h.

{ return ""; }

char SCSUUTF8::processText ( SWBuf &  text, const SWKey *  key = 0, const SWModule *  module = 0  )

virtual

This method processes and appropriately modifies the text given it for a particular filter task

Parameters

text	The text to be filtered/converted
key	Current key That was used.
module	Current module.

Returns

0

Implements SWFilter.

Definition at line 141 of file scsuutf8.cpp.

                                                                                {
    if ((unsigned long)key < 2) // hack, we're en(1)/de(0)ciphering
        return -1;
    
#ifdef _ICU_
    // Try decoding with ICU if possible
    err = U_ZERO_ERROR;
    icu::UnicodeString utf16Text(text.getRawData(), text.length(), scsuConv, err);
    err = U_ZERO_ERROR;
    int32_t len = utf16Text.extract(text.getRawData(), text.size(), utf8Conv, err);
    if (len > (int32_t)text.size()+1) {
        text.setSize(len+1);
        utf16Text.extract(text.getRawData(), text.size(), utf8Conv, err);
    }
#else
    // If ICU is unavailable, decode using Czyborra's decoder
    SWBuf utf8Buf = "";
    int len = text.length();
    const char* scsuString = text.c_str();

    for (int i = 0; i < len;) {
        
        if (i >= len) break;
        c = scsuString[i++];
        
        if (c >= 0x80)
        {
            UTF8Output(c - 0x80 + slide[active], &utf8Buf);
        }
        else if (c >= 0x20 && c <= 0x7F)
        {
            UTF8Output(c, &utf8Buf);
        }
        else if (c == 0x0 || c == 0x9 || c == 0xA || c == 0xC || c == 0xD)
        {
            UTF8Output(c, &utf8Buf);
        }
        else if (c >= 0x1 && c <= 0x8) // SQn
        {
            if (i >= len) break;
            d = scsuString[i++]; // single quote
            
            UTF8Output(d < 0x80 ? d + start[c - 0x1] :
                    d - 0x80 + slide[c - 0x1], &utf8Buf);
        }
        else if (c >= 0x10 && c <= 0x17) // SCn
        {
            active = c - 0x10; // change window
        }
        else if (c >= 0x18 && c <= 0x1F) // SDn
        {
            active = c - 0x18;  // define window
            if (i >= len) break;
            slide[active] = win[(unsigned char)scsuString[i++]];
        }
        else if (c == 0xB) // SDX
        {
            if (i >= len) break;
            c = scsuString[i++];
            
            if (i >= len) break;
            d = scsuString[i++];
            
            slide[active = c>>5] = 0x10000 + (((c & 0x1F) << 8 | d) << 7);
        }
        else if (c == 0xE) // SQU
        {
            if (i >= len) break;
            c = scsuString[i++]; // SQU
            
            if (i >= len) break;
            UTF8Output(c << 8 | scsuString[i++], &utf8Buf);
        }
        else if (c == 0xF) // SCU
        {
            mode = 1; // change to Unicode mode
            
            while (mode)
            {
                if (i >= len) break;
                c = scsuString[i++];
                
                if (c <= 0xDF || c >= 0xF3)
                {
                    if (i >= len) break;
                    UTF8Output(c << 8 | scsuString[i++], &utf8Buf);
                }
                else if (c == 0xF0) // UQU
                {
                    if (i >= len) break;
                    c = scsuString[i++];
                    
                    if (i >= len) break;
                    UTF8Output(c << 8 | scsuString[i++], &utf8Buf);
                }
                else if (c >= 0xE0 && c <= 0xE7) // UCn
                {
                    active = c - 0xE0;
                    mode = 0;
                }
                else if (c >= 0xE8 && c <= 0xEF) // UDn
                {
                    if (i >= len) break;
                    slide[active=c-0xE8] = win[(unsigned char)scsuString[i++]];
                    mode = 0;
                }
                else if (c == 0xF1) // UDX
                {
                    if (i >= len) break;
                    c = scsuString[i++];
                    
                    if (i >= len) break;
                    d = scsuString[i++];
                    
                    slide[active = c>>5] =
                        0x10000 + (((c & 0x1F) << 8 | d) << 7);
                    mode = 0;
                }
            }
        }
    }
#endif
    
    return 0;
}

int SCSUUTF8::UTF8Output ( unsigned long  uchar, SWBuf *  utf8Buf  )

private

Definition at line 105 of file scsuutf8.cpp.

{
    // join UTF-16 surrogates without any pairing sanity checks
    if (uchar >= 0xd800 && uchar <= 0xdbff) {
        d = uchar & 0x3ff;
        return 0;
    }
    if (uchar >= 0xdc00 && uchar <= 0xdfff) {
        uchar = uchar + 0x2400 + d * 0x400;
    }
    
    // output one character as UTF-8 multibyte sequence
    
    if (uchar < 0x80) {
        utf8Buf += uchar;
    }
    else if (uchar < 0x800) {
        utf8Buf += (0xc0 | (uchar>>6));
        utf8Buf += (0x80 | (uchar & 0x3f));
    }
    else if (uchar < 0x10000) {
        utf8Buf += (0xe0 | (uchar>>12));
        utf8Buf += (0x80 | (uchar>>6 & 0x3f));
        utf8Buf += (0x80 | (uchar & 0x3f));
    }
    else if (uchar < 0x200000) {
        utf8Buf += (0xf0 | (uchar>>18));
        utf8Buf += (0x80 | (uchar>>12 & 0x3f));
        utf8Buf += (0x80 | (uchar>>6 & 0x3f));
        utf8Buf += (0x80 | (uchar & 0x3f));
    }
    
    return 0;
}

Member Data Documentation

unsigned char SCSUUTF8::active

private

Definition at line 46 of file scsuutf8.h.

unsigned long SCSUUTF8::c

private

Definition at line 48 of file scsuutf8.h.

unsigned long SCSUUTF8::d

private

Definition at line 48 of file scsuutf8.h.

bool SCSUUTF8::mode

private

Definition at line 47 of file scsuutf8.h.

unsigned short SCSUUTF8::slide = {0x0080,0x00C0,0x0400,0x0600,0x0900,0x3040,0x30A0,0xFF00}

staticprivate

Definition at line 51 of file scsuutf8.h.

unsigned short SCSUUTF8::start = {0x0000,0x0080,0x0100,0x0300,0x2000,0x2080,0x2100,0x3000}

staticprivate

Definition at line 50 of file scsuutf8.h.

unsigned short SCSUUTF8::win

staticprivate

Definition at line 52 of file scsuutf8.h.

---

The documentation for this class was generated from the following files:

• include/scsuutf8.h
• src/modules/filters/scsuutf8.cpp