void setup_string()
{
int32_t number_of_cells=((intptr_t)CellsPageEnd_rstr-(intptr_t)CellsPage_rstr)/sizeof (cell);
empty_cell=(cell *) (CellsPage_rstr);
cell_boundary = empty_cell + number_of_cells;//NUMBER_OF_CELLS;
free_cell_chain=NULL;
//Paul 04-04-2002
cell_f()->next =cell_l();
cell_f()->nextl =cell_l();
cell_l()->prev =cell_f();
cell_l()->prevl =cell_f();
//
}
void serbian_J2j(void)
{
B_LINES my_bases;
cell *c,*e;
int16_t n;
char bb[90];
get_b_lines(NULL,&my_bases);
for(c=cell_f()->nextl,e=cell_l();c!=e;c=c->nextl)
if( (c->flg&c_f_let) && c->nvers && c->w*3<=c->h &&
(c->vers[0].let=='J' ||
memchr("[]!",c->vers[0].let,3)&&
c->row<(my_bases.b2 + my_bases.b1)/2&&
c->row+c->h>(my_bases.b3 + my_bases.b4)/2) )
{
if( (n=geom_neck((int16_t)(my_bases.b2 - my_bases.b1),c))!=0 )
{
if( db_status && snap_activity('c') )
{
sprintf (bb,"serbian reverse J-->>j. bs= %d %d (%d). find neck on %d",
my_bases.b1, my_bases.b2, my_bases.b2 - my_bases.b1, n);
snap_show_text(bb);
snap_monitor();
}
c->vers[0].let='j';
}
if( c->nvers && c->vers[0].let=='j' && check_let(c, '/') )
del_version( c,'/');
}
return;
}
void add_digital_versions(void)
{
cell *curr;
uchar c,p;
curr = cell_f();
while( (curr=curr->next)->next )
{
if( !(curr->flg&(c_f_let|c_f_bad)) ) continue;
c = curr->vers[0].let;
p = curr->vers[0].prob;
if( c=='|' && curr->nvers==2 )
{
c = curr->vers[1].let;
p = curr->vers[1].prob;
}
if( (curr->nvers==1 || c==(uchar)'Т' && curr->nvers==2)&& p>220 )
switch( c )
{
case (uchar)'Т' : add_stick_vers (curr, '1', p);break;
case (uchar)'°' : add_stick_vers (curr, '6', p);break;
default : break;
}
}
return;
}
void AKClearVers()
{
cell * Count;
char Let;
//uchar Prob;
int i;
// проверяем цепочку на случай зацикливания или обрыва
AKCheckChain();
for (Count = cell_f()->next; Count != cell_l(); Count=Count->next)
{
for ( i = 0; /*Count->vers[i].let != 0x0 &&*/ i < VERS_IN_CELL/*Count->nvers*/ ; i++ )
{
Let = Count->vers[i].let;
if ( !proplet(Let) /*&& Let != bad_char*/ )
{ // если символ не от сюда
Count->vers[i].let = 0; //bad_char;
Count->vers[i].prob = 0;
}
}
sort_vers(Count);
/*
if (Count->vers[0].let == bad_char)
{
Count = del_cell(Count);
}
*/
}
}
void criteries()
{
cell *c;
if (!line_number || current_fragment!=curr_frag)
{
dist_point_of_i_1=dist_point_of_i_2=dist_point_of_i_3=
dist_point_of_i_b=0;
curr_frag=current_fragment;
}
c=cell_f();
while ((c=c->nextl)->nextl != NULL)
{
criteria(c);
if(language == PUMA_LANG_RUSSIAN )
{
r_criteria(c,NULL);
if( c->nvers>0 && memchr("’вѓЈ",c->vers[0].let,4) &&
!is_russian_baltic_conflict(c->vers[0].let)&& // 17.07.2001 E.P.
!is_russian_turkish_conflict(c->vers[0].let) // 21.05.2002 E.P.
)
stick_center_study(c,NULL,1);
}
}
}
// finding begin of word
cell * erect_next_word(cell *cs)
{
cell *c=cs ;
if( c==cell_l() ) return NULL;
if( c==cell_f() ) c=c->next;
return (c->flg&c_f_fict)?NULL:c;
}
/********************************************************
* *
* Special programs for Й *
* *
********************************************************/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include "func.h"
#include "struct.h"
#include "linear.h"
#include "linutil.h"
#include "ligas.h"
#include "func.h"
#include "minmax.h"
extern uchar p2_active;
// Discrim й by base lines *
static int16_t rec_ii(cell*c,cell*cap);
static int16_t rec_ii_halo(cell * c);
static uchar iot_pen_lc[]= { 120,60,10,0,0 };
static uchar iot_pen_uc[]= { 140,10,0 ,0,0 };
int16_t cut_by_pos_ii(s_glue * const gl,uchar let)
{
B_LINES bl;
int16_t pen=0,upper=32000,dis,i;
get_b_lines(gl->celist[0],&bl);
for(i=0; i < gl->ncell; i++)
upper = MIN(upper,gl->celist[i]->row);
if(let==(uchar)'\xa9' /* й */ &&
!is_russian_turkish_conflict(let) // 21.05.2002 E.P.
) {
if((dis=upper-bl.b2) <= 0) { // letter upper than bbs2
dis = abs(dis);
if(dis < 5) pen = iot_pen_lc[dis];
}
else pen = 160; // letter lower than bbs2
if( gl->ncell==1 && (gl->celist[0]->recsource == c_rs_ev ||
gl->celist[0]->recsource == (c_rs_ev|c_rs_deskr) ))// events brought vers
if((Ns1+Ns2)>0 && bl.b2-bl.b1>6) {
dis=upper-bl.b1;
pen += dis<3 ? 60 : 0;
}
}
if(let==(uchar)'\x89' /* Й */) { // Capital iot
if((dis=upper-bl.b1) <= 0 ) { // letter upper than bbs1
dis = abs(dis);
if(dis < 5) pen = iot_pen_uc[dis];
}
else pen = 160; // letter lower than bbs1
}
return pen;
}
// Go by string and recog '\xa9' /* й */
void proc_ii(void)
{
cell * c,*cap;
uchar let;
int16_t ndust;
c = cell_f();
while((c=c->nextLetter()) != NULL ) {
if( !c->hasCellFlag(c_f_let + c_f_bad))
continue;
let = c->vers[0].let;
ndust=0;
//if( !memchr("иИнНпыў",let,7) ||
if( !memchr("\xA8\x88\xAD\x8D\xAF\xEB\xF7",let,7) ||
is_russian_baltic_conflict(let) // 17.07.2001 E.P.
)
continue;
cap = c;
while((cap=cap->next)!=NULL && cap != c->nextLetter()) {
if(cap->isDust()) {
ndust++;
switch(rec_ii(c,cap)) {
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
cap = c;
while((cap=cap->prev)!=NULL && cap != c->previousLetter()) {
if(cap->isDust()) {
ndust++;
//if(rec_ii(c,cap))goto next_let;
switch( rec_ii(c,cap))
{
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
if( ndust > 1)
rec_ii_halo(c); // many dusts, try all together
next_let:
;
} // while by letters
}
void proc_Ukr( void )
{
cell *c;
if (db_status)
snap_newpass('b');
for (c = cell_f(); c != cell_l(); c = c -> next){
proc_UKR_I(c);
}
}/*proc_Ukr*/
int16_t GetColumn(int16_t *VertLines,int16_t VertNum )
{
int16_t start_col, end_col;
int16_t col;
start_col = cell_f()->next->col;
end_col = cell_l()->prev->col;
if( start_col<VertLines[0] || start_col>=VertLines[VertNum-1] ) return -1;
col = dichotomy_array(start_col,VertLines,0,VertNum);
if( end_col>VertLines[col+1] ) return -1;
return col;
}
// Go by string and recog 'ч'
void cf::proc_shortu()
{
cell * c, * cap;
uchar let;
int ndust;
c = cell_f();
while((c = c->nextLetter()) != NULL) {
if(! c->hasCellFlag(c_f_let + c_f_bad))
continue;
let = c->vers[0].let;
ndust = 0;
if(!memchr("\xE3\x93", let, 2))
continue;
cap = c;
while((cap = cap->next) != NULL && cap != c->nextLetter()) {
if(cap->isDust()) {
ndust++;
switch(rec_shortu(c,cap)){
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
cap = c;
while((cap = cap->prev) != NULL && cap != c->previousLetter()) {
if(cap->isDust()) {
ndust++;
//if(rec_shortu(c,cap))goto next_let;
switch(rec_shortu(c,cap)) {
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
if(ndust > 1)
rec_shortu_halo(c); // many dusts, try all together
next_let: ;
} // while by letters
}
// Go by string and recog 'ч'
void proc_shortu(void)
{
cell * c,*cap;
uchar let;
int16_t ndust;
c = cell_f();
while((c=c->nextl) != NULL ){
if( !(c->flg & (c_f_let+c_f_bad)) ) continue;
let = c->vers[0].let; ndust=0;
if( !memchr("гУ",let,2)
)
continue;
cap = c;
while((cap=cap->next)!=NULL && cap != c->nextl )
{
if( cap->flg & c_f_dust)
{
ndust++;
switch( rec_shortu(c,cap))
{
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
cap = c;
while((cap=cap->prev)!=NULL && cap != c->prevl )
{
if( cap->flg & c_f_dust)
{
ndust++;
//if(rec_shortu(c,cap))goto next_let;
switch( rec_shortu(c,cap))
{
case 1:
goto next_let;
case -1:
return;
case 0:
break;
}
}
}
if( ndust > 1) rec_shortu_halo(c); // many dusts, try all together
next_let: ;
} // while by letters
}
void analysis_words(void)
{
int16_t left,right,flag=0,flag_qual;
cell *WB,*WE,*cl;
char snap[380],*buf=snap;
left=right=0;
WB=clbeg;
WE=clend;
//AK! add crashguard
while( (WB->flg & c_f_dust) || (WB->flg & c_f_fict) ||
(WB->flg & c_f_punct) )
{
if ((WB->next == cell_l() || WB->next == NULL) )
return;
WB=WB->next;
}
//AK! add c/g
while( (WE->flg & c_f_dust) || (WB->flg & c_f_fict) ||
(WB->flg & c_f_punct))
{
if ((WE->prev == cell_f() || WE->prev == NULL))
return;
WE=WE->prev;
}
if(WB==WE) return;
/* ---------------- LEFT -----------------------------*/
if(WB->vers[0].let==left_quocket) left=100;
else if((WB->flg & c_f_bad) ||
(memchr("¢э™",WB->vers[0].let,6) &&
!is_russian_baltic_conflict(WB->vers[0].let) && // 17.07.2001 E.P.
!is_russian_turkish_conflict(WB->vers[0].let)&& // 21.05.2002 E.P.
WB->vers[0].prob < MAX_PROB) ||
(memchr("Cc",WB->vers[0].let,2) && WB->vers[0].prob < MIN_PROB) ||
(memchr("u",WB->vers[0].let,1) && WB->vers[0].prob <=SMAX_PROB &&
WB->h <= ok_K_hgt) ||
(WB->vers[0].let== r_cu_u && WB->vers[0].prob < MID_PROB) ||
(memchr("ae",WB->vers[0].let,2) && WB->vers[0].prob < MAX_PROB)
)
{
if(WB->vers[0].prob<MIN_PROB && (WB->h>>1)*3 <= (WB->nextl)->h)
flag_qual=1;
else flag_qual=0;
if(WB->vers[0].let=='1') flag_qual=0;
left=check_shevron(WB,flag_qual);
}
void AKTryChainRecover(cell ** c) {
AKCheckChain();
// если правка цепочки не принесла результата,
// проверяем, находится ли c в цепочке
if ((*c)->next == NULL) {
// если нет, попробуем вернуться к предыдущей, иначе к началу
if (!AKCellInChain(*c)) {
if (AKCellInChain((*c)->prev))
*c = (*c)->prev;
else
*c = cell_f()->next;
}
}
}
////////////////////////////////////////////////////////////////////////////////////////////////////////////
// смотрим. есть ли такой cell в цепочке
int16_t AKCellInChain(cell * Cell) {
cell * Count;
// проверяем цепочку на случай зацикливания или обрыва
AKCheckChain();
for (Count = cell_f(); Count != cell_l(); Count = Count->next) {
if (Count == Cell)
return TRUE;
}
if (Cell == cell_l())
return TRUE;
return FALSE;
}
void insert_cell(cell *c,cell *ci)
{
int16_t col=c->col;
if(!(ci->flg & c_f_space)) // Valdemar 02-15-96 00:17am
// Paul 10-11-96
/*
if (col <= ci->col) while ((ci->prev)->col > col) ci=ci->prev;
else while ((ci=ci->next)->col <= col);
*/
//AK! Crashed line when ci->prev pointed to first_cell or ci
// pointed to last_cell
if (col <= ci->col)
while ( (ci->prev && ci->prev != cell_f()) && ((ci->prev)->flg & c_f_space || (ci->prev)->col > col))
{
if ( ci->prev )
ci=ci->prev;
else
AKCheckChain();
}
else
while ((ci != cell_l() && ci->next) && (ci->flg & c_f_space || ci->col <= col))
{
if ( ci->next )
ci=ci->next;
else
AKCheckChain();
}
//
(ci->prev)->next=c;
c->prev=ci->prev;
c->next=ci;
ci->prev=c;
if (c->flg & (c_f_let+c_f_bad) && !(c->flg & c_f_dust) )
{
ci=c->next;
while (!(ci->flg & (c_f_fict+c_f_let+c_f_bad)) || ci->flg & c_f_dust)
ci=ci->next;
(ci->prevl)->nextl=c;
c->prevl=ci->prevl;
c->nextl=ci;
ci->prevl=c;
}
else // AL 940318
err_pnlet (c);
}
void make_simples_diff(int16_t lang)
{
extern char db_pass;
cell *c,*e=cell_l();
int16_t dbp = db_pass;
db_pass=0;
for(c=cell_f()->next;c!=e;c=c->next)
if( !c->env->scale )
{
criteria(c);
if( lang==PUMA_LANG_RUSSIAN )
r_criteria(c,NULL);
}
db_pass = (uchar)dbp;
return;
}
void test_O_and_slash()
{
cell *C;
int16_t nv;
C = cell_f();
while ((C=C->nextl)->nextl)
{
nv = get_nvers(C, '0');
if ( nv >= 0 )
// Minimum decrease for 'O' to prefer '0' in the cases without
// obvious decision
{
#ifndef UFA
if( language!=PUMA_LANG_RUSSIAN )
cell_bonus(C, &(C->vers[nv]), 2);
#else
if( language!=PUMA_LANG_RUSSIAN )
cell_bonus(C, &(C->vers[nv]), 2);
else if( language==PUMA_LANG_RUSSIAN )
{
if( get_nvers(C,'9')<0 && get_nvers(C,'6')<0 && get_nvers(C,'8')<0 )
cell_bonus(C, &(C->vers[nv]), 2);
// Oleg : 06-09-1994 : not exist '698' - bonus 0
else
{ // Oleg : 06-09-1994 : exist any '698'-version
nv=get_nvers(C,'Ћ');
if( nv>=0 )
cell_bonus(C, &(C->vers[nv]), -2); // monus O
}
}
#endif
sort_vers(C);
}
nv = get_nvers(C, '/');
if ( nv == -1 ) continue;
if( language == PUMA_LANG_RUSSIAN && langSer && nv &&
C->nvers && C->vers[0].let==SERB_j )
continue;
inc_or_dec_slash(C, nv);
}
}
void erection_delete(void)
{
cell *c=cell_f()->next, *e=cell_l();
erection_delete_seq(c,e);
return;
}
//
// main erection function :
// find and erected incline words
//
uchar erection_incline_words(int16_t pass)
{
cell *c, *e, *tmp, *sc ;
uchar buf[MAX_LEN_WORD+40] ;
uchar wrd[MAX_LEN_WORD+40], word_len ;
Bool inc, incline, no_res ;
B_LINES bl ;
#ifdef STEND_INC
static char oldstr[256]="c:\\";
char str[256] ;
static int16_t oldline=-1 ;
FILE *fp ;
static int16_t first=1 ;
#endif
int16_t av_inc=0, n_inc=0;
local_pass=pass;
if( !setup_incline_word(cell_f()->nextl,cell_l(), TRUE) )
{
if( db_status && snap_activity('i') )
{
snap_show_text("ERECT : No erection images");
snap_monitor();
}
clear_incline_word(cell_f()->nextl,cell_l());
return 0;
}
if( db_status && snap_activity('i') )
{
snap_show_text("ERECT : Begin erection words");
snap_monitor();
}
c = cell_f() ;
incline = FALSE ;
while(1)
{
if( (c = erect_next_word(c))==NULL )break;
no_res=1;
inc =0;
strcpy(buf,"ERECT : Word : ");
get_b_lines(c,&bl);
e = erect_end_word(c, wrd, &word_len, (int16_t)(bl.ps/2) );
strcpy(&buf[14],wrd);
erect_page_words++;
sc = snap_stopcell(c,e)?snap_get_stopcell():NULL;
tmp = c->prev;
inc = erection_incline_word(c, e, bl.b3, 0);
if( inc && !local_pass )
{
incline = 1;
av_inc+=inc;
n_inc++;
erect_page_incline[inc/16]++;
}
#ifdef STEND_INC
if( inc && !(db_status && snap_activity('i')) )
{
set_page_name(str);
if( strcmp(str, oldstr) )
{
strcpy(oldstr, str);
fp = fopen("erection.txt",first?"wt":"at");
if( fp==NULL )
fp = fopen("erection.txt","wt");
if( first) first=0;
fprintf(fp,"\nfile : %s", str);
fclose(fp);
}
if( oldline!=line_number)
{
oldline=line_number;
fp = fopen("erection.txt","at");
fprintf(fp,"\nline %d : ", line_number+1);
fclose(fp);
}
sprintf(wrd,"(%d)",inc);
no_res=0;
strcat(buf,wrd) ;
fp = fopen("erection.txt","at");
fprintf(fp," %s ", &buf[14]);
fclose(fp);
}
#endif
if( db_status && snap_activity('i') )
{
if( inc || sc )
{
if( inc )
sprintf(wrd,"(%d)",inc);
else
sprintf(wrd,"(no erection)");
if( no_res )
strcat(buf,wrd) ;
snap_newcell(sc?sc:tmp->next) ;
snap_show_text(buf) ;
snap_monitor() ;
}
}
if( e==NULL )
break;
c = e;
}
if( n_inc )
erection_inc = av_inc / n_inc;
clear_incline_word(cell_f()->nextl,cell_l());
if( db_status && snap_activity('i') )
{
snap_show_text("ERECT : End erection words ");
snap_monitor();
}
return(uchar) incline ;
}
//////////////////////////////////////////////////////////////////////////////////////////////////////////////
//AK debug
// Проверяем цепочку и пытаемся ее поправить
void AKCheckChain()
{
cell * ForwardChain;
cell * BackwardChain;
cell * Back;
cell * BackTwo;
cell * Appendix = NULL;
cell * LossCell;
cell * Cicle;
// проверяем цепочку вперед
for ( Back = cell_f(), ForwardChain = cell_f()->next;
ForwardChain != cell_l() && ForwardChain->next != NULL;
ForwardChain = ForwardChain->next )
{
// проверяем на зацикленность
for ( Cicle = cell_f();
Cicle != ForwardChain;
Cicle = Cicle->next )
if (ForwardChain->next == Cicle)
{ //если зацикливается...
ForwardChain = NULL;
break;
}
// проверка обратной связи
if ( ForwardChain->prev != Back )
{
LossCell = ForwardChain->prev;
ForwardChain->prev = Back;
}
Back = ForwardChain;
}
// если прошли до конца...
if ( ForwardChain == cell_l() )
return;
// или не прошли...
if ( ForwardChain->prev != Back )
ForwardChain->prev = Back;
// проверяем цепочку надзад
for ( Back = cell_l(), BackwardChain = cell_l()->prev;
BackwardChain != cell_f() && BackwardChain->prev != NULL;
BackwardChain = BackwardChain->prev )
{
// проверка обратной связи
if ( BackwardChain->next != Back )
{
// смотрим, куда идет апендикс
for ( BackTwo = BackwardChain, Appendix = BackwardChain->next;
Appendix != cell_l() && Appendix->next != NULL;
Appendix = Appendix->next )
{
if ( Appendix->prev != BackTwo )
{
LossCell = Appendix->prev;
Appendix->prev = BackTwo;
}
}
// если по апендиксу дошли до обрыва от начала, то соединяем
if ( Appendix == ForwardChain )
{
ForwardChain->next = Back;
Back->prev = ForwardChain;
return;
}
// если конец апендикса не обрыв от начала...
if ( Appendix != cell_l() )
{
Appendix->next = Back;
Back->prev = Appendix;
BackwardChain = Back;
}
}
// проверяем на потерянный next
if ( BackwardChain->prev == ForwardChain )
{
ForwardChain->next = BackwardChain;
return;
}
Back = BackwardChain;
}
// Если не прошли до начала...
if ( BackwardChain != cell_f() )
{
ForwardChain->next = BackwardChain;
BackwardChain->prev = ForwardChain;
return;
}
// Если назад прошли до начала, а вперед только до обрыва,
// то есть аппендикс от начала, пристраеваем его в начало
// (на всякий случай проверим, остался ли апендикс целым
if ( BackwardChain == cell_f() && ForwardChain != cell_l() )
{
ForwardChain->next = Back;
Back->prev = ForwardChain;
}
}
void final_crit()
{
cell *c;
uchar let,let1;
int16_t cnt;
for (cnt=0,c=cell_f()->next; c->next!=NULL; c=c->next)
{
if (c->flg&c_f_space) continue;
let=c->vers[0].let;
if(final_crit_russian(c)) cnt++;
if (c->font&c_fp_undrln)
{
if (let=='v')
final_vy(c);
if (let=='e' || let=='u')
final_eg(c);
if (let=='o' || let=='a' && c->nvers>=2 && c->vers[1].let=='o')
final_ao_gpq(c);
if (let=='O')
final_OQ(c);
if (let=='.')
final_dotcom(c);
}
if (let=='/')
cnt+=final_slash_l(c);
else
{
if ((c->font|c->font_new)&c_fp_it && memchr("Il1",let,3))
cnt+=final_no_slash(c);
if (c->flg&c_f_bad || c->vers[0].prob<PROBL)
cnt+=final_back_slash(c);
}
if ((let=='c' || let=='C') && c->vers[1].let=='<')
{
c->nvers=1;
c->vers[1].let=0;
}
if ((let=='b' || let=='h') && (c->font|c->font_new)&c_fp_it)
final_bh(c);
if ( (language==LANG_GERMAN ||
language==LANG_SPANISH ||
language==LANG_FRENCH ||
language==LANG_ESTONIAN || // 19.07.2001 E.P.
language==LANG_TURKISH // 30.05.2002 E.P.
) &&
((c->font|c->font_new)&c_fp_it || let=='i' ||
let==liga_i ||
language == LANG_TURKISH && // 30.05.2002 E.P.
(let==i_sans_accent||let==II_dot_accent)
) &&
(let=='i' || let==liga_i ||
language == LANG_TURKISH && // 30.05.2002 E.P.
(let==i_sans_accent||let==II_dot_accent) ||
let==liga_fi || let==liga_ffi
) &&
( (let1=c->nextl->vers[0].let)=='i' ||
let1==liga_i ||
language == LANG_TURKISH && // 30.05.2002 E.P.
(let1==i_sans_accent||let1==II_dot_accent)
) &&
(c->cg_flag&c_cg_cutr && c->nextl->cg_flag&c_cg_cutl ||
((c->font|c->font_new)&c_fp_str) && c->nextl->col-(c->col+c->w)<=1
)
)
final_ii_u(c);
if (language==LANG_GERMAN && memchr("0AOU",let,4) ||
language==LANG_SWEDISH && memchr("0AO",let,3) ||
language==LANG_SPANISH && let=='U' ||
language==LANG_ESTONIAN && memchr("0AOU",let,4) // 19.07.2001 E.P.
)
{
get_b_lines(c,&bl);
if (bl.b1>c->row)
final_AOU_2dot(c);
}
if (language==LANG_SWEDISH && c->vers[0].let=='A')
{
get_b_lines(c,&bl);
if (bl.b1>c->row+MAX(4,bl.ps/4))
final_A_circle(c);
}
if (let=='-' && 3*c->h<c->w &&
(c->prev->flg&(c_f_fict|c_f_space) && c->prev->vers[0].let!=0x1e ||
memchr(".,:;",c->prev->vers[0].let,4)) &&
(c->next->flg&(c_f_fict|c_f_space) && c->next->vers[0].let!=0x1e ||
memchr(".,:;",c->next->vers[0].let,4)))
c->vers[0].let='_';
if (let=='F' && c->cg_flag&c_cg_cutr &&
memchr(".,",c->next->vers[0].let,2) && c->next->cg_flag&c_cg_cutl &&
c->col+c->w>c->next->col)
{
clist[0]=c;
clist[1]=c->next;
compose_cell(2,clist,c);
c->vers[0].let='E';
c->left=c->col;
c->right=c->col+c->w;
}
if (memchr("Vv",let,2) && c->cg_flag&c_cg_cutr &&
c->next->vers[0].let==let && c->next->cg_flag&c_cg_cutl)
final_vv_w(c);
if (let=='d' && c->vers[0].prob>=PROBD &&
memchr("cistx",c->next->vers[0].let,5) &&
c->next->vers[0].prob<=PROBB &&
c->next->col-(c->col+c->w)<=1)
final_d_chck(c);
if (let==0x27 && c->cg_flag&c_cg_cutr &&
memchr("Il",c->next->vers[0].let,2) && c->next->cg_flag&c_cg_cutl)
{
c=c->next; // AL 940321
final_Il_1(c);
cnt++;
}
if (let=='7' && c->cg_flag&c_cg_cutr &&
c->next->vers[0].let==0x27 && c->next->cg_flag&c_cg_cutl)
{final_7_T(c); cnt++;}
}
if (cnt)
{
delspace();
space_size((int16_t)get_size());
space_cell();
context_proc_line();
cont_space();
}
}
Bool digital_last_context(void)
{
cell *curr,*tmp;
char punct_list[]="'\"=:";
char sign_list[]="+-_";
int16_t l = strlen(punct_list),num,num_dig,num_bad,num_broken;
uchar c,p;
if( db_status && snap_activity('c') )
{
snap_show_text("DIG : Digital last context");
snap_monitor();
}
tmp = cell_f();
while( (tmp=tmp->next)->next )
{
if( tmp->nvers && tmp->vers[0].prob==2 )
set_bad_cell(tmp);
}
num=num_dig=num_bad=num_broken=0;
tmp = cell_f();
while( (tmp=tmp->next)->next && !(tmp->flg&(c_f_let|c_f_bad)) ); /* find let */
curr = cell_f();
if( tmp->next )
{
while( (curr=curr->next)!=tmp )
if( (curr->flg&c_f_punct) &&
!(curr->nvers>0 && memchr(sign_list,curr->vers[0].let,sizeof(sign_list))) )
{
curr = del_cell(curr); /* kill first punctuation-cell */
if( curr->next==NULL )break;
}
}
curr = cell_l();
while( (curr=curr->prev)->prev && !(curr->flg&(c_f_let|c_f_bad)) ); /* find let */
if( curr->next )
{
while( (curr=curr->next)->next )
if( (curr->flg&c_f_punct) )
{
curr = del_cell(curr); /* kill last punctuation-cell */
if( curr->next==NULL )break;
}
}
curr = cell_f();
while( (curr=curr->next)->next )
{
if( curr->flg&c_f_bad ) num_bad++;
if( !(curr->flg&(c_f_let|c_f_punct))) continue;
c = curr->vers[0].let;
if( memchr(punct_list,c,l) ||
c==left_quocket || c==right_quocket ||
c==low_quotes // macros E.P.
)
{
curr = del_cell(curr);
if( curr->next==NULL )break;
continue;
}
/*
if( !enable_setup_alphabet && !test_alphabet_elem(c) )
{
curr = del_cell(curr);
if( curr->next==NULL )break;
continue;
}
*/
if( (curr->flg&c_f_let) )
{
//if( memchr("0123456789ОЃ°ЗІТВшl",c,19) &&
if( isLikeDigit(c) && // 21.05.2002 E.P.
curr->vers[0].prob > 150 // Nick 26.02.2001
)
num_dig++;
if( memchr("С",c,1) )num_broken++;
if( curr->flg&c_f_let )num++;
}
}
if( digital_mode==2 ||
num_dig*4>=num*3 ||
num_dig==2 && num==3 ||
num_dig==1 && num==1 && num_bad==1 )
{ /* digital string */
curr = cell_f();
if( digital_mode==2 )
while( (curr=curr->next)->next )
{
if( !(curr->flg&(c_f_let|c_f_punct))) continue;
c = curr->vers[0].let;
if ( c==liga_exm || c=='!' )
curr=del_cell(curr);
}
curr = cell_f();
while( (curr=curr->next)->next )
{
if( curr->nvers<1 ) continue;
c = curr->vers[0].let;
p = curr->vers[0].prob;
if( c==(uchar)'_' )
curr->vers[0].let='-' ;
if( c==(uchar)'ш' )
curr->vers[0].let='2' ;
if( c==(uchar)'О' ||
// 21.05.2002 E.P.
( c==(uchar)'Ѓ')&& !is_russian_turkish_conflict(c)
)
curr->vers[0].let='0' ;
if( c==(uchar)'°' )
curr->vers[0].let='6' ;
if( c==(uchar)'З' || c==(uchar)'І')
curr->vers[0].let='3' ;
if( c==(uchar)'В' )
curr->vers[0].let='8' ;
if( c==(uchar)'С' && (p<190 || num_broken+num_dig==num ) )
curr->vers[0].let='0' ;
if( c==(uchar)'Т' || c==liga_exm || c=='!' || c=='|' )
curr->vers[0].let='1' ;
if( c==(uchar)'l' )
curr->vers[0].let='1' ;
if( c=='$' && curr->nvers==1 )
{
curr->vers[0].let='5';
curr->vers[1].let='$';
curr->vers[1].prob=curr->vers[0].prob-10;
curr->nvers=2;
}
if( c=='5' && curr->nvers>2 && curr->vers[1].let==(uchar)'°' &&
p==curr->vers[1].prob && curr->vers[2].let=='6' &&
curr->vers[2].prob>150 )
{ /* alternates : °56 -->>CONTEXT-->> 5°6 */
curr->vers[0].let='6';
curr->vers[2].let='5';
}
}
}
digital_mode=0;
set_all_alphabet();
for(curr=cell_f()->nextl;curr!=cell_l();curr=curr->nextl)
{
if( curr->flg&c_f_bad )
{
s_glue GL={0};
short_recog_cell(curr);
if( curr->nvers<1 )
{
GL.celist[0]=curr;
GL.maxlist[0]=curr;
GL.complist[0]=curr->env;
GL.maxlist[1]=GL.celist[1]=NULL;
GL.complist[1]=NULL;
GL.ncell=1;
GL.row = curr->row; GL.col = curr->col;
GL.width = curr->w; GL.height = curr->h;
if( crecell(curr,&GL,3)>=0)
dmBOX(curr,&GL);
}
criteria(curr);
levcut(curr,1);
if( curr->nvers<1 || curr->vers[0].prob<220 )
{ // restore bad sybmol
set_bad_cell(curr);
}
if( curr->nvers && curr->vers[0].let=='|' )
{
curr->vers[0].let='1';
// Nick add 26.02.2001 - test artificial '1'
levcut(curr,1);
if( curr->nvers<1 || curr->nvers==1 && curr->vers[0].let!='1' || curr->vers[0].prob<220 )
{ // restore bad sybmol
set_bad_cell(curr);
}
}
}
}
for(curr=cell_f()->nextl;curr!=cell_l();curr=curr->nextl)
{
if( curr->nvers && curr->vers[0].let=='(' &&
curr->next &&
curr->next->nvers && curr->next->vers[0].let==')' )
{
clist[0]=curr;
clist[1]=curr->next;
compose_cell(2,clist,curr);
curr->vers[0].let='0';
curr->vers[1].prob=220;
curr->nvers=1;
}
}
if( db_status && snap_activity('c') )
{
snap_show_text("DIG : Digital last context end");
snap_monitor();
}
num_dig = 0; // Nick 26.02.2001
num = 0;
for(curr=cell_f()->nextl;curr!=cell_l();curr=curr->nextl,num++)
{
// 21.05.2002 E.P.
Bool likeDigit = isLikeDigit(curr->vers[0].let);
if( curr->nvers<1 || curr->vers[0].prob<150 && likeDigit)
return 0; // bad symbols or small propability
if( likeDigit )
num_dig++;
}
if( num_dig*2 <= num ) // Nick 26.02.2001
return 0;
return 1;
}
void setup_digital_mode_pass2(str_info *str, int16_t first_pass)
{
cell *curr;
int16_t num_dig,num_let,num_bad,num_all, num_dig_let;
///char dig_list[]="|1234567890°ОЃЗІВш¢", 21.05.2002 E.P.
char dig_let_list[]="036",let_dig_list[]="В¢ш";
// letters_ini пунктуацию переводит в letter !!!
char punct_list[]="'\"=:!Љ?\\.,-"; // remove (){}[]/
uchar c,p;
digital_mode=0;
set_all_alphabet();
curr = cell_f();
num_all=num_dig=num_let=num_bad=num_dig_let=0;
while( (curr=curr->next)->next )
{
if( !(curr->flg&c_f_let) ) //!(curr->flg&(c_f_let|c_f_bad)) )
continue;
num_all++;
if( (curr->flg&c_f_bad) )
num_bad++;
else
{
c = curr->vers[0].let;
p = curr->vers[0].prob;
if( memchr(punct_list,c,sizeof(punct_list)) )
{
num_all--;
continue;
}
if( !isLikeDigit(c) )
{
if( p>220 )
num_let++;
if( p > 240 ) // Nick 11.05.2001
{
//num_let++;
if( curr->nvers < 2 ||
!isLikeDigit(curr->vers[1].let) ||
curr->vers[1].prob < 220
)
num_dig--; // штраф на цифровой
}
}
if( isLikeDigit(c) )
{
num_dig++;
if( memchr(let_dig_list,c,sizeof(let_dig_list)) && p>220 )
num_let++;
}
if( memchr(dig_let_list,c,sizeof(dig_let_list)) )
num_dig_let++;
}
}
str->all += num_all;
//if( num_dig < 0 )
// num_dig = 0;
str->dig += num_dig;
str->fuzzy_dig += num_dig_let;
return;
}
void setup_digital_mode(str_info *str, int16_t first_pass)
{
cell *curr;
int16_t num_dig,num_let,num_bad,num_all, num_dig_let;
int16_t num_asOne;
//char dig_list[]="|1234567890°ОЃЗІВш¢",
char dig_let_list[]="036",let_dig_list[]="В¢ш";
//char punct_list[]="'\"=:!(){}[]!Љ?\\/";
char punct_list[]="'\"=:;";
char asOne_list[]="1[]/!|";
uchar c,p;
digital_mode=0;
set_all_alphabet();
curr = cell_f();
num_all=num_dig=num_let=num_bad=num_dig_let=0;
num_asOne = 0;
while( (curr=curr->next)->next )
{
if( !(curr->flg&(c_f_let|c_f_bad)) )
continue;
num_all++;
if( (curr->flg&c_f_bad) )
num_bad++;
else
{
c = curr->vers[0].let;
p = curr->vers[0].prob;
if( memchr(punct_list,c,sizeof(punct_list)) )
{
num_all--;
continue;
}
if( memchr(asOne_list,c,sizeof(asOne_list)) )
{
num_asOne++;
}
if( !isLikeDigit(c) && p>220 ) // 21.05.2002 E.P.
num_let++;
if( isLikeDigit(c) )
{
num_dig++;
if( memchr(let_dig_list,c,sizeof(let_dig_list)) && p>220 )
num_let++;
}
if( memchr(dig_let_list,c,sizeof(dig_let_list)) )
num_dig_let++;
}
}
if( str->alphabet==ALPHA_ALL )
{
int16_t dig=0;
switch( num_all )
{
case 0 : break;
case 1 :
if( num_dig>0 && num_dig > num_asOne )
dig = 2;
break;
case 2 :
if( num_dig>0 && num_dig > num_asOne && num_let==0 )
dig = 2;
break;
case 3 :
if( num_dig>1 && num_dig > num_asOne && num_let==0 )
dig = 2;
break;
case 4 :
if( num_dig>2 && num_let==0 && num_dig > num_asOne )
dig = 2;
break;
default :
if( num_dig>MAX(2,(3*num_all)/4) && num_let==0 ||
num_dig>0 && num_dig > num_asOne && num_bad==0 && num_let==0
)
dig = 2;
else if( num_dig*10>num_all*6 && num_all>5 /*&& num_bad*/)
dig = 1;
break;
}
if( dig==1 && num_dig*10>num_all*6 && num_all>5 /*&& num_bad*/)
{
digital_mode=1;
}
if( dig==2 &&
(num_dig>num_all/2&&num_all>5 || str->dig*4>str->all || first_pass) )
{
digital_mode=2;
set_digital_alphabet();
if( str->dollars )
add_alphabet_elem('$');
}
}
else if (str->alphabet==ALPHA_DIGITAL || str->alphabet==ALPHA_DIGITAL_TRUE)
{
if( num_dig>=MAX(1,num_all/2) && num_let<MAX(1,num_all/4) ||
num_dig>0 && num_bad==0 && num_let==0)
{
digital_mode=2;
set_digital_alphabet();
if( str->dollars )
add_alphabet_elem('$');
}
}
if( digital_mode==2 )
{
digital_string_penalty = (num_dig_let!=num_dig)?254:20;
for(curr = cell_f();curr!=cell_l();curr=curr->nextl)
{
if( !(curr->flg&c_f_let) )
continue;
if( curr->nvers>0 &&
(
memchr(punct_list,curr->vers[0].let,sizeof(punct_list)) ||
curr->vers[0].let != '1' && memchr(asOne_list,curr->vers[0].let,sizeof(asOne_list))
)
)
{
if( curr->vers[0].prob>220 && !memchr("()",curr->vers[0].let,2) )
add_alphabet_elem(curr->vers[0].let);
continue;
}
if( !is_digit(curr->vers[0].let) )
down_all_versions(curr,100);
}
}
return;
}
void save_alphabet_information(str_info *str,int16_t column,CSTR_line ln)
{
uchar buf[256]={0};
cell * c;
RecObject ro={0};
CSTR_rast rst;
RecRaster rrst;
RecVersions ver;
LeoFieldSetup fs={0};
CSTR_attr lattr;
CSTR_rast_attr attr;
UniVersions uvs;
if( db_status && snap_activity('n') )
{
snap_show_text("DIG : start selection");
snap_monitor();
}
str[column].all++;
fs.nStyle = LS_PRINT;
buf['0']=buf['1']=buf['2']=buf['3']=buf['4']=
buf['5']=buf['6']=buf['7']=buf['8']=buf['9']=1;
memcpy(fs.AlphaTable,buf,256);
#ifdef _USE_LEO_
LEOSetupField(&fs);
for(c=cell_f()->nextl;c!=cell_l();c=c->nextl)
{
if( c->nvers==0 && c->w*2>c->h && c->h*2>c->w )
{
puchar wr;
// to work_rst
wr=save_raster_align8(c);
// rst -->> rrst
rrst.lnPixHeight = c->h;
rrst.lnPixWidth = c->w;
rrst.lnRasterBufSize = REC_MAX_RASTER_SIZE;
memcpy(rrst.Raster,wr,((c->w+63)/64)*8*c->h);
memcpy(&ro.recData.recRaster,&rrst,REC_MAX_RASTER_SIZE);
LEORecogPrnMethod( &ro , REC_METHOD_3X5, 1);
memcpy(&ver,&ro.recResults,sizeof(ver));
if( ver.lnAltCnt && ver.Alt[0].Prob>220 )
{
c->nvers=1;
c->vers[0].let=ver.Alt[0].Code;
c->vers[0].prob=ver.Alt[0].Prob;
c->vers[1].let=0;
c->flg=c_f_let;
}
}
}
#endif
setup_digital_mode(&str[column], 1);
for(c=cell_f()->next;c!=cell_l();c=c->next)
if( c->nvers>0 && c->vers[0].let=='$' && c->vers[0].prob>220 )
{
str[column].dollars++;
break;
}
if( digital_mode==2 )
{
str[column].dig++;
if( db_status && snap_activity('n') )
{
sprintf(buf,"DIG : digital str, col %d, all %d, dig %d, plus_minus %d" ,
column, str[column].all,str[column].dig,str[column].pm);
snap_show_text(buf);
snap_monitor();
}
}
else if( digital_mode==1 )
{
str[column].fuzzy_dig++;
if( db_status && snap_activity('n') )
{
sprintf(buf,"DIG : fuzzy digital str, col %d, all %d, dig %d, plus_minus %d" ,
column, str[column].all,str[column].dig,str[column].pm);
snap_show_text(buf);
snap_monitor();
}
}
else
{
plusminus_mode=0;
if( cell_f()->next && cell_f()->next==cell_l()->prev )
{
c=cell_f()->next;
rst=CSTR_GetNextRaster(CSTR_GetFirstRaster(ln),CSTR_f_bad|CSTR_f_let);
if( rst && CSTR_GetImage(rst,(uchar *)&rrst,CSTR_TYPE_IMAGE_RS))
{
if( c->w>c->h*5
|| c->w>c->h*3 && c->env && c->env->nl == 1 // Nick 18.01.2002
)
{
c->nvers=1;
c->vers[0].let='-';
c->vers[0].prob=254;
ver.Alt[0].Code='-';
ver.Alt[0].Prob=255;
ver.lnAltCnt =1;
CSTR_StoreCollection(rst,&ver);
CSTR_GetAttr(rst,&attr);
//attr.keg=Ps<<line_scale;
CSTR_SetAttr(rst,&attr);
CSTR_GetLineAttr(ln,&lattr);
lattr.Flags |= CSTR_STR_ReadyResults ;
CSTR_SetLineAttr(ln,&lattr);
plusminus_mode=1;
}
else if( abs(c->w-c->h)<5 )
{
#ifdef _USE_LEO_
fs.nStyle = LS_PRINT;
buf['+']=1;
memcpy(fs.AlphaTable,buf,256);
LEOSetupField(&fs);
memcpy(&ro.recData.recRaster,&rrst,REC_MAX_RASTER_SIZE);
LEORecogPrnMethod( &ro , REC_METHOD_3X5, 1);
memcpy(&ver,&ro.recResults,sizeof(ver));
if( ver.lnAltCnt && ver.Alt[0].Code=='+' && ver.Alt[0].Prob>220 )
{
c->nvers=1;
c->vers[0].let='+';
c->vers[0].prob=254;
ver.lnAltCnt =1;
CSTR_StoreCollection(rst,&ver);
CSTR_GetAttr(rst,&attr);
//attr.keg=Ps<<line_scale;
CSTR_SetAttr(rst,&attr);
CSTR_GetLineAttr(ln,&lattr);
lattr.Flags |= CSTR_STR_ReadyResults ;
CSTR_SetLineAttr(ln,&lattr);
plusminus_mode=1;
}
#endif
}
}
}
if( plusminus_mode )
{
str[column].pm++;
rst=CSTR_GetNextRaster(CSTR_GetFirstRaster(ln),CSTR_f_bad|CSTR_f_let);
CSTR_GetCollectionUni(rst,&uvs);
if( uvs.lnAltCnt )
{
uvs.Alt[0].Charset=CodePages[language];
CSTR_StoreCollectionUni(rst,&uvs);
}
if( db_status && snap_activity('n') )
{
sprintf(buf,"DIG : plus_minus str , col %d, all %d, dig %d, plus_min %d" ,
column, str[column].all,str[column].dig,str[column].pm);
snap_show_text(buf);
snap_monitor();
}
}
else
{
if( db_status && snap_activity('n') )
{
sprintf(buf,"DIG : unknown alphabet , col %d, all %d, dig %d, plus_min %d" ,
column, str[column].all,str[column].dig,str[column].pm);
snap_show_text(buf);
snap_monitor();
}
}
}
return;
}