/* ML type: bdd -> bdd -> bdd */
EXTERNML value mlbdd_bdd_restrict(value r, value var) /* ML */
{
return mlbdd_make(bdd_restrict(Bdd_val(r),Bdd_val(var)));
}
int main(int argc,char **argv) {
char ambiguous_treatment='A';
char linebuff[1024];
char *parseptr;
PARSE_STATE ps;
uint32_t low_code,high_code;
int width,i,j,vi,vj;
FILE *unicode_db;
BDD x,y,child[8];
BDD *queue;
int queue_low,queue_high,queue_max;
puts("/*\n"
" * GENERATED CODE - DO NOT EDIT!\n"
" * Edit mkwcw.c, which generates this, or the input to that\n"
" * program, instead. Distributions of IDSgrep will nonetheless\n"
" * usually include a ready-made copy of this file because\n"
" * compiling and running mkwcw.c requires a library and data\n"
" * file that, although free, not everyone is expected to have.\n"
" */\n\n"
"#include \"_stdint.h\"\n"
);
if (argc>1)
ambiguous_treatment=argv[1][0]&~32;
bdd_init(1000000,15625);
bdd_setcacheratio(64);
bdd_setvarnum(32);
bdd_gbc_hook(NULL);
defined_codes=bddfalse;
zero_codes=bddfalse;
wide_codes=bddfalse;
/* yes, unfortunately UnicodeData.txt and EastAsianWidth.txt are just
* different enough to need separate parsers, at least if the parsers
* are as stupid as I'd like these ones to be */
if (argc>2) {
unicode_db=fopen(argv[2],"rt");
while (1) {
fgets(linebuff,sizeof(linebuff),unicode_db);
if (feof(unicode_db))
break;
ps=psLOW;
linebuff[sizeof(linebuff)-1]='\0';
low_code=0;
width=-1;
for (parseptr=linebuff;(*parseptr) && (ps!=psSTOP);parseptr++)
switch (ps) {
case psLOW:
if ((*parseptr>='0') && (*parseptr<='9'))
low_code=(low_code<<4)+(*parseptr-'0');
else if ((*parseptr>='a') && (*parseptr<='f'))
low_code=(low_code<<4)+(*parseptr-'a'+10);
else if ((*parseptr>='A') && (*parseptr<='F'))
low_code=(low_code<<4)+(*parseptr-'A'+10);
else if (*parseptr==';')
ps=psSEMI;
else if ((*parseptr==' ') || (*parseptr=='\t'))
{ /* skip spaces and tabs */ }
else
ps=psSTOP; /* this catches comment lines */
break;
case psSEMI:
if (*parseptr==';')
ps=psWIDTH;
break;
case psWIDTH:
if (((parseptr[0]=='M') && ((parseptr[1]=='e') ||
(parseptr[1]=='n'))) ||
((parseptr[0]=='C') && (parseptr[1]=='f')))
width=0;
/* FALL THROUGH */
default:
ps=psSTOP;
break;
}
if (width==0)
set_range_width(low_code,low_code,0);
}
fclose(unicode_db);
}
while (1) {
fgets(linebuff,sizeof(linebuff),stdin);
if (feof(stdin))
break;
ps=psLOW;
linebuff[sizeof(linebuff)-1]='\0';
low_code=0;
high_code=0;
width=-1;
for (parseptr=linebuff;(*parseptr) && (ps!=psSTOP);parseptr++)
switch (ps) {
case psLOW:
if ((*parseptr>='0') && (*parseptr<='9'))
low_code=(low_code<<4)+(*parseptr-'0');
else if ((*parseptr>='a') && (*parseptr<='f'))
low_code=(low_code<<4)+(*parseptr-'a'+10);
else if ((*parseptr>='A') && (*parseptr<='F'))
low_code=(low_code<<4)+(*parseptr-'A'+10);
else if (*parseptr=='.')
ps=psHIGH;
else if (*parseptr==';') {
high_code=low_code;
ps=psWIDTH;
} else if ((*parseptr==' ') || (*parseptr=='\t'))
{ /* skip spaces and tabs */ }
else
ps=psSTOP; /* this catches comment lines */
break;
case psHIGH:
if ((*parseptr>='0') && (*parseptr<='9'))
high_code=(high_code<<4)+(*parseptr-'0');
else if ((*parseptr>='a') && (*parseptr<='f'))
high_code=(high_code<<4)+(*parseptr-'a'+10);
else if ((*parseptr>='A') && (*parseptr<='F'))
high_code=(high_code<<4)+(*parseptr-'A'+10);
else if ((*parseptr=='.') || (*parseptr==' ') || (*parseptr=='\t'))
{ /* skip spaces, tabs, and dots */ }
else if (*parseptr==';')
ps=psWIDTH;
else
ps=psSTOP;
break;
case psWIDTH:
if (*parseptr=='A')
*parseptr=ambiguous_treatment;
switch (*parseptr) {
case 'F': /* full-width treated as wide */
case 'W': /* wide */
width=2;
break;
case 'H': /* half-width treated as narrow */
case 'N': /* narrow or neutral */
width=1;
break;
case '0': /* zero-width - should only appear in user database */
width=0;
break;
default:
/* ignore all others */
break;
}
/* FALL THROUGH */
default:
ps=psSTOP;
break;
}
if (width>=0)
set_range_width(low_code,high_code,width);
}
printf("/* node counts before simplification: %d %d %d */\n",
bdd_nodecount(defined_codes),
bdd_nodecount(zero_codes),
bdd_nodecount(wide_codes));
x=bdd_addref(bdd_simplify(wide_codes,defined_codes));
bdd_delref(wide_codes);
wide_codes=x;
x=bdd_addref(bdd_apply(defined_codes,wide_codes,bddop_diff));
bdd_delref(defined_codes);
defined_codes=x;
x=bdd_addref(bdd_simplify(zero_codes,defined_codes));
bdd_delref(zero_codes);
zero_codes=x;
printf("/* node counts after simplification: %d %d %d */\n\n",
bdd_nodecount(defined_codes),
bdd_nodecount(zero_codes),
bdd_nodecount(wide_codes));
bdd_varblockall();
bdd_intaddvarblock(0,7,0);
bdd_intaddvarblock(8,15,0);
bdd_intaddvarblock(16,23,0);
bdd_intaddvarblock(24,31,0);
bdd_intaddvarblock(0,31,1);
bdd_reorder_probe(&reordering_size_callback);
puts("typedef struct _WIDTH_BBD_ENT {\n"
" int16_t child[8];\n"
" char byte,shift;\n"
"} WIDTH_BDD_ENT;\n\n"
"static WIDTH_BDD_ENT width_bdd[]={");
queue=(BDD *)malloc(sizeof(BDD)*1000);
queue_max=1000;
queue_low=2;
queue_high=4;
queue[0]=bddfalse;
queue[1]=bddtrue;
queue[2]=wide_codes;
queue[3]=zero_codes;
while (queue_low<queue_high) {
if (queue_high+8>queue_max) {
queue_max/=3;
queue_max*=4;
queue=(BDD *)realloc(queue,sizeof(BDD)*queue_max);
}
reorder_focus=queue[queue_low];
bdd_reorder(BDD_REORDER_WIN2ITE);
vj=bdd_var(queue[queue_low]);
vi=(vj/8)*8;
vj=((vj-vi+1)/3)*3-1;
if (vj<0) vj=0;
x=bdd_addref(bdd_restrict(queue[queue_low],bdd_nithvar(vi+vj)));
y=bdd_addref(bdd_restrict(x,bdd_nithvar(vi+vj+1)));
child[0]=bdd_addref(bdd_restrict(y,bdd_nithvar(vi+vj+2)));
child[1]=bdd_addref(bdd_restrict(y,bdd_ithvar(vi+vj+2)));
bdd_delref(y);
y=bdd_addref(bdd_restrict(x,bdd_ithvar(vi+vj+1)));
child[2]=bdd_addref(bdd_restrict(y,bdd_nithvar(vi+vj+2)));
child[3]=bdd_addref(bdd_restrict(y,bdd_ithvar(vi+vj+2)));
bdd_delref(y);
bdd_delref(x);
x=bdd_addref(bdd_restrict(queue[queue_low],bdd_ithvar(vi+vj)));
y=bdd_addref(bdd_restrict(x,bdd_nithvar(vi+vj+1)));
child[4]=bdd_addref(bdd_restrict(y,bdd_nithvar(vi+vj+2)));
child[5]=bdd_addref(bdd_restrict(y,bdd_ithvar(vi+vj+2)));
bdd_delref(y);
y=bdd_addref(bdd_restrict(x,bdd_ithvar(vi+vj+1)));
child[6]=bdd_addref(bdd_restrict(y,bdd_nithvar(vi+vj+2)));
child[7]=bdd_addref(bdd_restrict(y,bdd_ithvar(vi+vj+2)));
bdd_delref(y);
bdd_delref(x);
fputs(" {{",stdout);
for (i=0;i<8;i++) {
queue[queue_high]=child[i];
for (j=0;queue[j]!=child[i];j++);
if (j==queue_high)
queue_high++;
else
bdd_delref(child[i]);
printf("%d",j-2);
if (i<7) putchar(',');
}
printf("},%d,%d},\n",vi/8,5-vj);
queue_low++;
}
puts("};\n\n"
"int idsgrep_utf8cw(char *);\n"
"\n"
"#define WBS width_bdd[search]\n"
"\n"
"int idsgrep_utf8cw(char *cp) {\n"
" int search;\n"
"\n"
" for (search=0;search>=0;)\n"
" search=WBS.child[(cp[WBS.byte]>>WBS.shift)&7];\n"
" if (search==-1)\n"
" return 2;\n"
" for (search=1;search>=0;)\n"
" search=WBS.child[(cp[WBS.byte]>>WBS.shift)&7];\n"
" return ((-1)-search);\n"
"}\n");
bdd_done();
exit(0);
}