int16_t german_context_process( cell* C )
{
int16_t return_code;
cell * wc;
uchar i, wp;
int16_t nv_c, nv_h, nv_b;
return_code = NO_ACTION;
if ( ( n_ltr == 1 ) && ( C->vers[0].let == ss_deaf_sound ) )
{
C->vers[0].prob = MAX( MIN_PROB, C->vers[0].prob - GER_DOUBLE_S_MONUS );
sort_vers( C );
return_code = CONTINUE_ACTION;
goto GCP_Ret;
}
if ( n_ltr > 1 )
{
if ( C->vers[0].let == ss_deaf_sound )
{
C->vers[0].prob =
MAX( MIN_PROB, C->vers[0].prob - GER_DOUBLE_S_MONUS );
sort_vers( C );
}
for ( i = 0, wc = C; i < n_ltr - 1; wc = wc->nextl, i++ )
{
nv_c = get_nvers( wc, 'c' );
if ( nv_c < 0 ) continue;
if ( ( nv_c > 1 ) || ( wc->vers[nv_c].prob < ADM_PROB ) )
continue;
nv_h = get_nvers( wc->nextl, 'h' );
if ( nv_h < 0 ) continue;
nv_b = get_nvers( wc->nextl, 'b' );
if ( nv_b < 0 ) continue;
if ( ( nv_h > nv_b ) &&
( wc->nextl->vers[nv_b].prob -
wc->nextl->vers[nv_h].prob < REFUSE_BOUND ) )
{
wp = wc->nextl->vers[nv_h].prob;
wc->nextl->vers[nv_h].prob = wc->nextl->vers[nv_b].prob;
wc->nextl->vers[nv_b].prob = wp;
sort_vers( wc->nextl );
}
}
}
GCP_Ret:
return return_code;
}
static int16_t final_slash_l(cell *c)
{
int16_t i;
uchar fnt;
get_b_lines(c,&bl);
if (c->row+c->h>=bl.b3+MIN(3,MAX(1,c->h/10)))
return 0;
if ((c->flg & (c_f_let|c_f_bad)) == 0)
return 0; // AL 940318
if (c->flg & c_f_dust)
return 0; // AL 940318
if ((fnt=c->prevl->font|c->nextl->font)&c_fp_ser && !(fnt&c_fp_gelv))
return 0;
c->nvers-=MAX(0,c->nvers-(VERS_IN_CELL-4));
c->vers[c->nvers].let='l';
c->vers[c->nvers+1].let='I';
c->vers[c->nvers+2].let='1';
c->vers[c->nvers].prob=c->vers[c->nvers+1].prob=c->vers[c->nvers+2].prob=
c->vers[0].prob;
for (i=0; i<c->nvers; i++)
if (memchr("lI1",c->vers[i].let,3))
c->vers[i].prob=0;
c->nvers+=3;
c->vers[c->nvers].let=0;
if (c->env!=NULL && !is_slash(c))
c->vers[0].prob=MAX(2,c->vers[0].prob-50);
sort_vers(c);
return 1;
}
static int16_t LiliI(cell *NC)
//
// This procedure gives bonus to 'I' in one letter word.
//
{
int16_t i;
version *v;
bon_to_dig(NC); // monus to non- "1"
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
{
if (v->let == 'I')
goto monus;
}
return 0;
monus:;
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
{
if (v->let == 'I')
continue;
v->prob=MAX(v->prob-MONUS,10); // didn't expect number
}
sort_vers(NC);
return 1;
}
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);
}
*/
}
}
int16_t short_recog_cell (cell *c)
{
// was && но если нет c->evn - что распознавать?
if( (c->cg_flag & c_cg_comp) ||
(!c->env) )
return 0;
comp_from_kit(c);
{
int16_t n,i;
uchar res[20];
CCOM_comp cc;
//запись в структуру CCOM_comp данных для EVNRecog_lp
cc.h = c->env->h;
cc.w = c->env->w;
cc.rw = c->env->rw;
cc.nl = c->env->nl;
cc.begs = c->env->begs;
cc.ends = c->env->ends;
cc.scale = c->env->scale;
// n = (int16_t)EVNRecog_lp(c->env,lpool,lpool_lth,&res[0]);
n = (int16_t)EVNRecog_lp(&cc,lpool,lpool_lth,&res[0]);
//восстановление из структуры CCOM_comp результатов EVNRecog_lp
c->env->type = cc.type;
c->env->cs = cc.cs;
c->env->pidx = cc.pidx;
c->env->reasno = cc.reasno;
if( n )
{
for(i=0;i<n;i++)
{
c->vers[i].let=res[i];
c->vers[i].prob=254;
}
c->vers[n].let=0;
c->flg = c_f_let;
c->recsource |= c_rs_ev;
c->history |= c_rs_ev;
}
else
{
c->vers[0].let=176;
c->vers[1].let=0;
c->flg = c_f_bad;
}
c->nvers=n;
}
//recog_letter();
v2_pidx_crit(c);
sort_vers(c);
//c->reasno = wcomp.reasno;
//return vers_to_cell(c);
return c->nvers;
}
void discrim_un( cell *GC, s_glue *GL, int16_t cod_let )
{
int16_t wi;
memset ( l_tab_shift, 0, sizeof (l_tab_shift) );
memset ( r_tab_shift, 0, sizeof (r_tab_shift) );
pen_u = 0; pen_n = 0; un_code = (uchar)cod_let;
comp_row_col( GL ); // compute coordinates of the composed box
// check italic incline of the component
wi = letincl(GC);
if ( ((GC->font|GC->font_new) & c_fp_it) && (GL->ncell == 1) ) return;
make_straight_abris( GL, l_tab_shift, r_tab_shift );
if (GL->ncell > 1)
{
if ( is_italic() )
{ compare_corners_mass(); goto ApplyPenalty;}
else return;
}
if ( cod_let & Let_n )
{
pen_n = n_around_ur_corner(GL); // study right upper corner against arounding
pen_n += left_right_dist();
}
if ( cod_let & Let_u )
pen_u = u_around_bl_corner(GL); // study left bottom corner against arounding
ApplyPenalty:
if ( pen_u )
{
cell_bonus_let(GC, 'u', (int16_t)(-(pen_u)) );
sort_vers( GC );
}
if ( pen_n )
{
cell_bonus_let(GC, 'n', (int16_t)(-(pen_n)) );
sort_vers( GC );
}
}
static void final_bh(cell *c)
{
int16_t i;
for (i=1; i<c->nvers; i++)
if (memchr("bh",c->vers[i].let,2))
{
c->vers[i].prob=MAX(c->vers[i].prob,c->vers[0].prob-10);
sort_vers(c);
return;
}
if (c->nvers==VERS_IN_CELL-1)
c->nvers--;
c->vers[c->nvers].let=(c->vers[0].let=='b')?'h':'b';
c->vers[c->nvers].prob=c->vers[0].prob-10;
c->nvers++;
c->vers[c->nvers].let=0;
sort_vers(c);
}
static void final_eg(cell *c)
{
if (c->vers[0].prob<=DEG) return;
if (c->nvers==VERS_IN_CELL-1) c->nvers--;
c->vers[c->nvers].let='g';
c->vers[c->nvers].prob=c->vers[0].prob-DEG;
c->nvers++;
c->vers[c->nvers].let=0;
sort_vers(c);
}
void add_vers(cell *bc, version *wv)
{
int16_t nv;
nv=bc->nvers;
if (nv == VERS_IN_CELL-1)
{
nv--;
if (bc->vers[nv].prob >= wv->prob) return;
bc->vers[nv].let=wv->let;
bc->vers[nv].prob=wv->prob;
sort_vers(bc);
}
else
{
bc->vers[bc->nvers].let=wv->let;
bc->vers[bc->nvers].prob=wv->prob;
bc->nvers++;
sort_vers(bc);
}
}
static void final_OQ(cell *c)
{
if (c->vers[0].prob<=DOQ) return;
if (c->nvers==VERS_IN_CELL-1) c->nvers--;
c->vers[c->nvers].let='Q';
c->vers[c->nvers].prob=c->vers[0].prob-DOQ;
c->nvers++;
c->vers[c->nvers].let=0;
sort_vers(c);
}
int16_t swedish_context_process ( cell* C )
{
int16_t return_code;
swed_lju_flag = FALSE;
return_code = NO_ACTION;
if (n_ltr == 1 && !(C->next->flg&c_f_punct && C->next->vers[0].let=='.'))
/** single letter case: **/
{
LiliI(C);
return_code = CONTINUE_ACTION;
goto SwCP_Ret;
}
if ( n_ltr > 2 ) // Process 'lju' at start of word
{
int16_t nv_l, nv_j, nv_u;
nv_l = get_nvers( C, 'l');
if ( nv_l < 0 ) goto SwCP_Ret;
if ( C->vers[ nv_l ].prob < ADM_PROB ) goto SwCP_Ret;
nv_j = get_nvers( C->nextl, 'j');
if ( nv_j < 0 ) goto SwCP_Ret;
if ( C->nextl->vers[ nv_j ].prob < ADM_PROB ) goto SwCP_Ret;
nv_u = get_nvers( C->nextl->nextl, 'u');
if ( nv_u < 0 ) goto SwCP_Ret;
if ( C->nextl->nextl->vers[ nv_u ].prob < ADM_PROB ) goto SwCP_Ret;
cell_bonus( C, &(C->vers[nv_l]), SWED_LJU_BONUS );
sort_vers( C );
cell_bonus( C->nextl, &(C->nextl->vers[nv_j]), SWED_LJU_BONUS );
sort_vers( C->nextl );
cell_bonus( C->nextl->nextl, &(C->nextl->nextl->vers[nv_u]), SWED_LJU_BONUS );
sort_vers( C->nextl->nextl );
swed_lju_flag = TRUE;
}
SwCP_Ret:
return return_code;
}
static Bool apostr_ll(cell *NC)
//
// This procedure gives bonus to ll in two letters word
// if there is after apostrophe. Example: he'll
// ON ENTRY: NC - address of cell for first letter.
//
{
int16_t i;
version *v;
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
if (v->let == 'l')
goto second;
return FALSE;
second:;
for (i=0,v=NC->nextl->vers; i < NC->nextl->nvers; i++,v++)
if (v->let == 'l')
goto monus;
return FALSE;
monus:;
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
{
if (v->let=='l')
continue;
v->prob=MAX(v->prob-MONUS,10); // didn't expect other than l
}
sort_vers(NC);
for (i=0,v=NC->nextl->vers; i < NC->nextl->nvers; i++,v++)
{
if (v->let == 'l')
continue;
v->prob=MAX(v->prob-MONUS,10); // didn't expect other than l
}
sort_vers(NC->nextl);
return TRUE;
}
void dif_II( void )
{
c_comp *cp;
lnhead *lp;
int16_t h=workI.c->h/2,ii=0;
if( !workI.twoDot && workI.ld == NULL && workI.rd && workI.rh > 0 ){
workI.c->vers[1].prob -= 20;
goto ret_dif;
}
if( workI.ld && workI.rd && /*workI.c->prevl->vers[0].let != UKR_i && */
abs(workI.lh-workI.rh) >= 2 &&
(workI.lw > 2*workI.rw || workI.rw > 2*workI.lw) ){
workI.c->vers[1].prob -= 20;
goto ret_dif;
}
if( workI.ld && workI.rd && workI.lh > 0 && workI.rh > 0 ){
workI.c->vers[0].prob -= 20;
goto ret_dif;
}
cp = workI.c->env;
for(;cp!=NULL;cp=cp->next){
lp = (lnhead *)((puchar)cp + cp->lines + sizeof(uint16_t));
while( lp->lth != 0 ){
if( lp->row <= workI.c->h/8 ){
ii++;
if( h > lp->h )
h = lp->h;
}
lp = (lnhead *)((puchar)lp + lp->lth);
}
}
workI.c->vers[0].prob -= 24*workI.twoDot;
if( ii >= 2 && h >= 2 ){
if( !workI.twoDot )
workI.c->vers[0].prob -= 20;
}
else
workI.c->vers[1].prob -= 20;
ret_dif:
sort_vers(workI.c);
return;
}/*dif_II*/
static int16_t bon_to_dig(cell *NC)
//
// This procedure gives bonus to digits in one letter word.
//
{
int16_t i;
version *v;
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
if ((v->let >= '0') && (v->let <= '9'))
goto monus;
return 0;
monus:;
for (i=0,v=NC->vers; i < NC->nvers; i++,v++)
if ((v->let < '0') || (v->let > '9'))
v->prob=MAX(v->prob-MONUS+MONUS1I,10); // didn't expect other than 1
sort_vers(NC);
return 1;
}
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);
}
}
static void inc_or_dec_slash(cell *wc, int16_t numv)
//
// This procedure decreases probability '/' if its neighbour letters are
// italic ( to prefer 'l' or 'I' ) and
// increases probability '/' if its heighbour letters are straight
// ( to avoid 'l' and 'I' appearence );
//
{
cell *cprev, *cnext;
Bool no_Il = (get_nvers(wc, 'I')==-1) && (get_nvers(wc, 'l')==-1 && wc->vers[numv].let=='/');
if ( wc->vers[0].prob - wc->vers[numv].prob > REFUSE_BOUND )
return;
cprev = wc->prev;
cnext = wc->next;
if ( !((cprev->flg & c_f_let) && (cnext->flg & c_f_let)) ) return;
if ( ((cprev->font|cprev->font_new) & c_fp_it) && ((cnext->font|cnext->font_new) & c_fp_it) )
{ cell_bonus(wc, &(wc->vers[numv]), (int16_t)(no_Il?2:-BONUS_SLASH)); goto IDS_Sort; }
if ( ((cprev->font|cprev->font_new) & c_fp_str) && ((cnext->font|cnext->font_new) & c_fp_str) )
cell_bonus(wc, &(wc->vers[numv]), (int16_t)(no_Il?2:BONUS_SLASH));
IDS_Sort:
sort_vers(wc);
}
static int16_t final_no_slash(cell *c)
{
int16_t i;
uchar fnt;
if (c->env==NULL)
return 0;
for (i=1; i<c->nvers; i++)
if (c->vers[i].let=='/') break;
if (i==c->nvers)
return 0;
if ((fnt=c->prevl->font|c->nextl->font)&c_fp_ser && !(fnt&c_fp_gelv) &&
!(c->font&c_fp_ser))
{
c->vers[i].let=c->vers[0].let;
c->vers[0].let='/';
return 1;
}
if (is_slash(c))
return 0;
c->vers[i].prob=0;
sort_vers(c);
return 1;
}
////////////////////
/////////////////
// Tools.c
int16_t short_recog_cell(CSTR_rast c, int line_scale) {
int16_t n, i;
uchar res[20];
CSTR_rast_attr attr;
CCOM_comp *comp;
// ExtComponent ec;
int16_t *pint16;
uchar * lpool;
UniVersions vers = { 0 };
CSTR_GetAttr(c, &attr);
comp = CSTR_GetComp(c);
// исходно была ошибка Tools.c -
// если c->env == 0 в lpool лежало неизвестно что!
// if( (c->cg_flag & CSTR_cg_comp) && !c->env )
// return 0;
if ((attr.cg_flag & CSTR_cg_comp) || !comp)
return 0;
// comp_from_kit(c);
// CCOM_comp -> to ExtComponent
/* memset(&ec,0,sizeof(ExtComponent));
if( !Ccomp2ExtComponenet(comp,&ec,&attr,line_scale) )
return 0;
*/
// n = (int16_t)EVNRecog_lp(c->env,lpool,lpool_lth,&res[0]);
pint16 = (int16_t *) comp->linerep;
// *pint16 == comp->size_linerep ?????
lpool = comp->linerep + 2;
// n = (int16_t)EVNRecog_lp(&ec,lpool,*pint16,&res[0]);
n = (int16_t) EVNRecog_lp(comp, lpool, *pint16, &res[0]);
vers.lnAltMax = REC_MAX_VERS;
if (n) {
for (i = 0; i < n; i++) {
//c->vers[i].let=res[i];
//c->vers[i].prob=254;
vers.Alt[i].Liga = res[i];
vers.Alt[i].Prob = 254;
vers.Alt[i].Method = REC_METHOD_EVN;
}
attr.flg = CSTR_f_let;
attr.recsource |= CSTR_rs_ev;
attr.RecogHistory |= CSTR_rs_ev;
} else {
attr.flg = CSTR_f_bad;
}
vers.lnAltCnt = n;
CSTR_StoreCollectionUni(c, &vers);
CSTR_SetAttr(c, &attr);
v2_pidx_crit(c);
sort_vers(c);
CSTR_GetCollectionUni(c, &vers);
return (int16_t) vers.lnAltCnt;
}
int16_t franch_context_process( cell* C )
{
int16_t return_code;
int16_t i, nvI, nvl, nvs;
return_code = NO_ACTION;
if (n_ltr == 1 && (C->next->flg&c_f_punct) &&
(C->next->vers[0].let== APOSTROPH ) )
/**** shorted article case: *****/
{
for ( i = 0; i < C->nvers; i++ )
{
switch( C->vers[i].let )
{
case 'l':
case 'd':
case 'L':
case 'D':
{
C->vers[i].prob =
MIN( MAX_PROB, C->vers[i].prob + FRANCH_ART_BONUS );
return_code = CONTINUE_ACTION;
break;
}
default :
{
C->vers[i].prob =
MAX( MIN_PROB, C->vers[i].prob - FRANCH_ART_BONUS );
}
}
}
if ( return_code == CONTINUE_ACTION )
{
sort_vers( C );
goto FCP_Ret;
}
}
if ( ( (n_ltr == 2) || (n_ltr == 3) ) &&
( ( word_flag & AFTERDOTCOM ) ||
( word_flag & AFTERDOT ) ||
( C->prev->flg & c_f_fict ) ) )
/*** "Il" or "Ils" case ***/
{
nvI = get_nvers( C, 'I' );
if ( nvI < 0 )
goto FCP_Ret;
if ( C->vers[ nvI ].prob < ADM_PROB )
goto FCP_Ret;
nvl = get_nvers( C->nextl, 'l' );
if ( nvl < 0 )
goto FCP_Ret;
if ( C->nextl->vers[ nvl ].prob < ADM_PROB )
goto FCP_Ret;
if ( n_ltr == 2 )
goto FCP_Il_case;
nvs = get_nvers( C->nextl->nextl, 's' );
if ( nvs < 0 )
goto FCP_Ret;
if ( C->nextl->nextl->vers[ nvs ].prob < ADM_PROB )
goto FCP_Ret;
vers_to_first_place( C->nextl->nextl, nvs );
FCP_Il_case:
vers_to_first_place( C, nvI );
vers_to_first_place( C->nextl, nvl );
return_code = CONTINUE_ACTION;
}
FCP_Ret:
return return_code;
}