static CSTR_rast ed_end_word(CSTR_rast cs,uchar *str,uchar *word_len)
{
CSTR_rast c=cs;
CSTR_rast_attr attr;
int16_t i;
RecVersions vers;
CSTR_GetCollection(c,&vers);
*str++ = vers.Alt[0].Code;
i=1;
do{
c=CSTR_GetNext(c);
CSTR_GetCollection(c,&vers);
*str++ = vers.Alt[0].Code;
CSTR_GetAttr(c,&attr);
i++;
}while( c && !memchr(right_limit_word,vers.Alt[0].Code,sizeof(right_limit_word)) &&
!(attr.flg&CSTR_f_fict) && i<MAX_LEN_WORD-1);
c=c ? c : CSTR_GetLastRaster(CSTR_GetRasterLine(cs));
if( i )
{
i--;
str--;
}
*str='\0';
*word_len = (uchar)i;
return c;
}
/////////////////////////////////////////////////////////////////////////////
// CheckLineForFilling
Bool CheckLineForFilling( CSTR_line* Comingline )
{
CSTR_line line;
CSTR_rast rast;
RecVersions vers;
CSTR_rast_attr rast_attr;
line = *Comingline;
rast = CSTR_GetFirstRaster( line ); // I фикт. растр
rast = CSTR_GetNextRaster( rast , CSTR_f_all );
while( rast )
{
CSTR_GetCollection( rast,&vers );
CSTR_GetAttr( rast,&rast_attr );
if((rast_attr.flg&CSTR_f_dust) || (rast_attr.flg&CSTR_f_fict ) ||//Дусты и
((rast_attr.flg&CSTR_f_space) &&
(!vers.Alt[0].Code || vers.Alt[0].Code == 0x1e ||
vers.Alt[0].Code == 0x1f))) //полупробелы выкидываем.
goto next_raster;
return TRUE;
next_raster:
rast = CSTR_GetNextRaster( rast , CSTR_f_all );
}
return FALSE;
}
Bool32 p2_msk_inc(CSTR_rast rst)
{
#ifdef _USE_MSK_
uchar prob=0, name, lim, code;
RecRaster rc;
RecVersions ver;
int32_t i;
if(!HndTabInc )
return 0;
if(language!=PUMA_LANG_RUSSIAN && language!=PUMA_LANG_RUSENG )
return 0;
CSTR_GetImage(rst,(uchar*)&rc,CSTR_TYPE_IMAGE_RS);
if( !rc.lnPixHeight || !rc.lnPixWidth )
return 0;
MSKSetAlphabet ( alphabet );
CSTR_GetCollection(rst,&ver);
ver.lnAltCnt=2;
ver.lnAltMax=REC_MAX_VERS;
name = ver.Alt[0].Code;
ver.Alt[0].Code=(uchar)to_upper((uchar)name);
ver.Alt[0].Prob=255;
ver.Alt[1].Code=(uchar)to_lower((uchar)name);
ver.Alt[1].Prob=255;
if( name=='0' )
{
ver.Alt[1].Code=(uchar)'О';
ver.Alt[2].Code=(uchar)'Ѓ';
ver.Alt[2].Prob=255;
ver.lnAltCnt=3;
}
if( name=='3' )
{
ver.Alt[1].Code=(uchar)'З';
ver.Alt[2].Code=(uchar)'І';
ver.Alt[2].Prob=255;
ver.lnAltCnt=3;
}
if( !MSKRecogCharExpPuma( HndTabInc, &rc, &ver,1 ) )
return 0;
if( !ver.lnAltCnt )
return 0;
for(prob=0, i=0;i<ver.lnAltCnt;i++)
{
if( prob<ver.Alt[i].Prob)
{
prob=ver.Alt[i].Prob ;
code=ver.Alt[i].Code;
}
}
lim = MIN_MSK_PROB;
if( strchr(similar_letters,code) )
lim = MIN_MSK_PROB_SIM;
return (prob>lim);
#else
return FALSE;
#endif
}
static CSTR_rast ed_next_word(CSTR_rast cs)
{
CSTR_rast c=cs;
CSTR_rast_attr attr;
RecVersions vers;
CSTR_GetAttr(c,&attr);
if( attr.flg==CSTR_f_fict && attr.col==0x7fff )
return (CSTR_rast)0; // last rast
do{
c=CSTR_GetNext(c);
CSTR_GetAttr(c,&attr);
CSTR_GetCollection(c,&vers);
}while( c && memchr(left_limit_word,vers.Alt[0].Code,sizeof(left_limit_word)) &&
!(attr.flg&CSTR_f_fict) );
return (attr.flg&CSTR_f_fict)?(CSTR_rast)0:c;
}
/////////////////////////////////////////////////////////////////////////////
// CFString::ExtractWordsFromString
void CFString::ExtractWordsFromString( CSTR_line* Comingline , PageElementCount* Count)
{
CSTR_rast_attr rast_attr;
CSTR_rast rast;
RecVersions vers;
CSTR_line line;
CSTR_attr line_attr;
CWord* CurrentWord;
const char* SeparatorsWord=" ";
const char* result=NULL;
int16_t FlagWord;
Bool FlagString,FlagCapDrop;
line = *Comingline;
FlagWord = 0;
FlagString = FALSE;
FlagCapDrop = FALSE;
CSTR_GetLineAttr( line , &line_attr );
if(line_attr.Flags & CSTR_STR_CapDrop) //буквица - см. первая буква в сказках
FlagCapDrop = TRUE;
SetRect(&m_rectBaseLine, line_attr.bs1, line_attr.bs2,
line_attr.bs3, line_attr.bs4); //don't used now
SetRect(&m_rectString, line_attr.col - TemplateOffset.x,
line_attr.row - TemplateOffset.y,
line_attr.col - TemplateOffset.x + line_attr.wid,
line_attr.row - TemplateOffset.y + line_attr.hei);
#ifdef alDebug //obsolete option
{
RECT rect;
SetRect(&rect,line_attr.col, line_attr.row,
line_attr.col + line_attr.wid,
line_attr.row + line_attr.hei);
pInputArray->push_back(rect);
}
#endif
rast = CSTR_GetFirstRaster( line ); // I фикт. растр
rast = CSTR_GetNextRaster( rast , CSTR_f_all );
while( rast )
{
CSTR_GetCollection( rast,&vers );
CSTR_GetAttr( rast,&rast_attr );
//#define NEG_HALF_SPACE 0x1e
//#define POS_HALF_SPACE 0x1f
//#define REGULAR_SPACE 0x20
if((rast_attr.flg&CSTR_f_dust) || (rast_attr.flg&CSTR_f_fict ) ||//Дусты и
((rast_attr.flg&CSTR_f_space) &&
(!vers.Alt[0].Code || vers.Alt[0].Code == 0x1e ||
vers.Alt[0].Code == 0x1f)))//полупробелы выкидываем.
goto next_raster;
FlagString = TRUE;
result = strchr( SeparatorsWord, vers.Alt[0].Code );
if( FlagWord ){ // word openly
if(result) // word is broken
FlagWord =0;
else{ // word lasts
assert( CurrentWord );
CurrentWord->AddLetter2Word( &rast , Count ,&FlagCapDrop );
}
}
else{ // expect new word
if(!result){ // beginning word
++m_wWordsCount;
++Count->Words;
FlagWord = 1;
m_arWords.push_back( new CWord() );
CurrentWord = m_arWords[m_wWordsCount-1];
assert( CurrentWord );
if(rast_attr.font_spec == CSTR_fs_courier)
CurrentWord->m_wFontNumber = rast_attr.font & 0xFC;
else
CurrentWord->m_wFontNumber = rast_attr.font;
CurrentWord->m_wFontPointSize = rast_attr.keg;
CurrentWord->AddLetter2Word( &rast , Count ,&FlagCapDrop );
if(FlagCapDrop){
CurrentWord->m_wFontPointSize = 14;
FlagCapDrop = FALSE;
FlagWord = 0;
}
}
}
next_raster:
rast = CSTR_GetNextRaster( rast , CSTR_f_all );
}
if(FlagString == FALSE)
Count->Strings--;
}
