static obj_t add_who_irritants(obj_t cont, obj_t values, obj_t ex)
{
//oprintf("add_who_irritants: cont_proc = %p\n", cont_proc(cont));
//oprintf("add_who_irritants: cont4_arg = %O\n", cont4_arg(cont));
//oprintf("add_who_irritants: values = %O\n", values);
assert(is_cont(cont));
cont_proc_t proc = cont_proc(cont);
if (proc == c_apply_proc) {
assert(is_cont5(cont));
obj_t op = cont5_arg1(cont);
obj_t who_sym = procedure_is_C(op) ? procedure_name(op) : FALSE_OBJ;
obj_t who_ex = MAKE_RECORD(who, who_sym);
obj_t irr_ex = MAKE_RECORD(irritants, reverse_list(values));
return MAKE_COMPOUND_CONDITION(who_ex, irr_ex, ex);
}
if (proc == c_eval_operator) {
assert(is_cont4(cont));
obj_t arg = cont4_arg(cont);
obj_t who_ex = MAKE_RECORD(who, CAR(arg));
obj_t irr_ex = MAKE_RECORD(irritants, CDR(arg));
return MAKE_COMPOUND_CONDITION(who_ex, irr_ex, ex);
}
if (proc == c_eval) {
assert(is_cont4(cont));
obj_t expr = cont4_arg(cont);
obj_t who_ex = MAKE_RECORD(who, expr);
return MAKE_COMPOUND_CONDITION(who_ex, ex);
}
return ex;
}
static cv_t default_handler(obj_t cont, obj_t values)
{
/*
* For all non-message-irritants-who conditions,
* print the condition's class.
* For all message conditions, print the condition's message.
* If who or irritants given, print (cons who irritants).
*/
assert(is_cont(cont));
EVAL_LOG("cont=%O values=%O", cont, values);
obj_t ex = CAR(values);
obj_t parts = record_get_field(ex, 0);
const char *psn = program_short_name();
size_t psnl = strlen(psn);
size_t i, size = vector_len(parts);
for (i = 0; i < size; i++) {
obj_t rtd = record_rtd(vector_ref(parts, i));
if (rtd != message && rtd != irritants && rtd != who) {
ofprintf(stderr, "%*s: %O\n", (int)psnl, psn, rtd_name(rtd));
psn = "";
}
}
obj_t who_p = FALSE_OBJ;
obj_t irr_p = make_uninitialized();
for (i = 0; i < size; i++) {
obj_t p = vector_ref(parts, i);
obj_t rtd = record_rtd(p);
if (rtd == message) {
obj_t msg = record_get_field(p, 0);
const wchar_t *chars = string_value(msg);
fprintf(stderr, "%*s %ls\n", (int)psnl, psn, chars);
psn = "";
} else if (rtd == who)
who_p = record_get_field(p, 0);
else if (rtd == irritants)
irr_p = record_get_field(p, 0);
}
if (who_p != FALSE_OBJ && !is_uninitialized(irr_p)) {
ofprintf(stderr, "%*s %O\n", (int)psnl, psn, CONS(who_p, irr_p));
psn = "";
} else if (who_p != FALSE_OBJ) {
ofprintf(stderr, "%*s %O\n", (int)psnl, psn, who_p);
psn = "";
} else if (!is_uninitialized(irr_p)) {
ofprintf(stderr, "%*s %O\n", (int)psnl, psn, irr_p);
psn = "";
}
if (*psn)
fprintf(stderr, "%s: unknown exception\n", psn);
ofprintf(stderr, "\n");
return cv(EMPTY_LIST, CONS(make_uninitialized(), EMPTY_LIST));
}
static cv_t push_exception(obj_t cont, obj_t values)
{
assert(is_cont(cont));
obj_t ex = add_who_irritants(cont, values, eval_exception);
obj_t handler = record_get_field(eval_dyn_env, DE_HANDLER);
obj_t second = make_cont4(c_exception_returned,
EMPTY_LIST,
cont_env(cont),
EMPTY_LIST);
obj_t first = make_cont5(c_apply_proc,
second,
cont_env(cont),
handler,
values);
return cv(first, CONS(ex, values));
}
/* An explicit control evaluator, taken almost directly from SICP, section
* 5.2. list is a flat list of expressions to evaluate. where is a label to
* begin at. Return value depends on where.
*/
NODE *evaluator(NODE *list, enum labels where) {
/* registers */
NODE *exp = NIL, /* the current expression */
*val = NIL, /* the value of the last expression */
*proc = NIL, /* the procedure definition */
*argl = NIL, /* evaluated argument list */
*unev = NIL, /* list of unevaluated expressions */
*stack = NIL, /* register stack */
*parm = NIL, /* the current formal */
*catch_tag = NIL,
*arg = NIL; /* the current actual */
/* registers that don't get reference counted, so we pretend they're ints */
FIXNUM vsp = 0, /* temp ptr into var_stack */
cont = 0, /* where to go next */
formals = (FIXNUM)NIL; /* list of formal parameters */
int i, nargs;
BOOLEAN tracing; /* are we tracing the current procedure? */
FIXNUM oldtailcall; /* in case of reentrant use of evaluator */
FIXNUM repcount; /* count for repeat */
FIXNUM old_ift_iff;
oldtailcall = tailcall;
old_ift_iff = ift_iff_flag;
save2(var,this_line);
assign(var, var_stack);
save2(fun,ufun);
cont = (FIXNUM)all_done;
numsave((FIXNUM)cont);
newcont(where);
goto fetch_cont;
begin_line:
ref(list);
assign(this_line, list);
newcont(end_line);
begin_seq:
make_tree(list);
if (!is_tree(list)) {
assign(val, UNBOUND);
goto fetch_cont;
}
assign(unev, tree__tree(list));
assign(val, UNBOUND);
goto eval_sequence;
end_line:
if (val != UNBOUND) {
if (NOT_THROWING) err_logo(DK_WHAT, val);
deref(val);
}
val = NIL;
deref(list);
goto fetch_cont;
/* ----------------- EVAL ---------------------------------- */
tail_eval_dispatch:
tailcall = 1;
eval_dispatch:
switch (nodetype(exp)) {
case QUOTE: /* quoted literal */
assign(val, node__quote(exp));
goto fetch_cont;
case COLON: /* variable */
assign(val, valnode__colon(exp));
while (val == UNBOUND && NOT_THROWING)
assign(val, err_logo(NO_VALUE, node__colon(exp)));
goto fetch_cont;
case CONS: /* procedure application */
if (tailcall == 1 && is_macro(car(exp)) &&
is_list(procnode__caseobj(car(exp)))) {
/* tail call to user-defined macro must be treated as non-tail
* because the expression returned by the macro
* remains to be evaluated in the caller's context */
assign(unev, NIL);
goto non_tail_eval;
}
assign(fun, car(exp));
if (cdr(exp) != NIL)
goto ev_application;
else
goto ev_no_args;
default:
assign(val, exp); /* self-evaluating */
goto fetch_cont;
}
ev_no_args:
/* Evaluate an application of a procedure with no arguments. */
assign(argl, NIL);
goto apply_dispatch; /* apply the procedure */
ev_application:
/* Evaluate an application of a procedure with arguments. */
assign(unev, cdr(exp));
assign(argl, NIL);
mixsave(tailcall,var);
num2save(val_status,ift_iff_flag);
save2(didnt_get_output,didnt_output_name);
eval_arg_loop:
if (unev == NIL) goto eval_args_done;
assign(exp, car(unev));
if (exp == Not_Enough_Node) {
if (NOT_THROWING)
err_logo(NOT_ENOUGH, NIL);
goto eval_args_done;
}
save(argl);
save2(unev,fun);
save2(ufun,last_ufun);
save2(this_line,last_line);
assign(var, var_stack);
tailcall = -1;
val_status = 1;
assign(didnt_get_output,
cons_list(0,fun,ufun,this_line,END_OF_LIST));
assign(didnt_output_name, NIL);
newcont(accumulate_arg);
goto eval_dispatch; /* evaluate the current argument */
accumulate_arg:
/* Put the evaluated argument into the argl list. */
reset_args(var);
restore2(this_line,last_line);
restore2(ufun,last_ufun);
assign(last_call, fun);
restore2(unev,fun);
restore(argl);
while (NOT_THROWING && val == UNBOUND) {
assign(val, err_logo(DIDNT_OUTPUT, NIL));
}
push(val, argl);
pop(unev);
goto eval_arg_loop;
eval_args_done:
restore2(didnt_get_output,didnt_output_name);
num2restore(val_status,ift_iff_flag);
mixrestore(tailcall,var);
if (stopping_flag == THROWING) {
assign(val, UNBOUND);
goto fetch_cont;
}
assign(argl, reverse(argl));
/* --------------------- APPLY ---------------------------- */
apply_dispatch:
/* Load in the procedure's definition and decide whether it's a compound
* procedure or a primitive procedure.
*/
proc = procnode__caseobj(fun);
if (is_macro(fun)) {
num2save(val_status,tailcall);
val_status = 1;
newcont(macro_return);
}
if (proc == UNDEFINED) {
if (ufun != NIL) {
untreeify_proc(ufun);
}
if (NOT_THROWING)
assign(val, err_logo(DK_HOW, fun));
else
assign(val, UNBOUND);
goto fetch_cont;
}
if (is_list(proc)) goto compound_apply;
/* primitive_apply */
if (NOT_THROWING)
assign(val, (*getprimfun(proc))(argl));
else
assign(val, UNBOUND);
#define do_case(x) case x: goto x;
fetch_cont:
{
enum labels x = (enum labels)cont;
cont = (FIXNUM)car(stack);
numpop(&stack);
switch (x) {
do_list(do_case)
default: abort();
}
}
compound_apply:
#ifdef mac
check_mac_stop();
#endif
#ifdef ibm
check_ibm_stop();
#endif
if (tracing = flag__caseobj(fun, PROC_TRACED)) {
for (i = 0; i < trace_level; i++) print_space(writestream);
trace_level++;
ndprintf(writestream, "( %s ", fun);
}
/* Bind the actuals to the formals */
vsp = (FIXNUM)var_stack; /* remember where we came in */
for (formals = (FIXNUM)formals__procnode(proc);
formals != (FIXNUM)NIL;
formals = (FIXNUM)cdr((NODE *)formals)) {
parm = car((NODE *)formals);
if (nodetype(parm) == INT) break; /* default # args */
if (argl != NIL) {
arg = car(argl);
if (tracing) {
print_node(writestream, maybe_quote(arg));
print_space(writestream);
}
} else
arg = UNBOUND;
if (nodetype(parm) == CASEOBJ) {
if (not_local(parm,(NODE *)vsp)) {
push(parm, var_stack);
setobject(var_stack, valnode__caseobj(parm));
}
tell_shadow(parm);
setvalnode__caseobj(parm, arg);
} else if (nodetype(parm) == CONS) {
/* parm is optional or rest */
if (not_local(car(parm),(NODE *)vsp)) {
push(car(parm), var_stack);
setobject(var_stack, valnode__caseobj(car(parm)));
}
tell_shadow(car(parm));
if (cdr(parm) == NIL) { /* parm is rest */
setvalnode__caseobj(car(parm), argl);
break;
}
if (arg == UNBOUND) { /* use default */
save2(fun,var);
save2(ufun,last_ufun);
save2(this_line,last_line);
save2(didnt_output_name,didnt_get_output);
num2save(ift_iff_flag,val_status);
assign(var, var_stack);
tailcall = -1;
val_status = 1;
mixsave(formals,argl);
numsave(vsp);
assign(list, cdr(parm));
if (NOT_THROWING)
make_tree(list);
else
assign(list, NIL);
if (!is_tree(list)) {
assign(val, UNBOUND);
goto set_args_continue;
}
assign(unev, tree__tree(list));
assign(val, UNBOUND);
newcont(set_args_continue);
goto eval_sequence;
set_args_continue:
numrestore(vsp);
mixrestore(formals,argl);
parm = car((NODE *)formals);
reset_args(var);
num2restore(ift_iff_flag,val_status);
restore2(didnt_output_name,didnt_get_output);
restore2(this_line,last_line);
restore2(ufun,last_ufun);
restore2(fun,var);
arg = val;
}
setvalnode__caseobj(car(parm), arg);
}
if (argl != NIL) pop(argl);
}
if (check_throwing) {
assign(val, UNBOUND);
goto fetch_cont;
}
vsp = 0;
if (tracing = flag__caseobj(fun, PROC_TRACED)) {
if (NOT_THROWING) print_char(writestream, ')');
new_line(writestream);
save(fun);
newcont(compound_apply_continue);
}
assign(val, UNBOUND);
assign(last_ufun, ufun);
assign(ufun, fun);
assign(last_line, this_line);
assign(this_line, NIL);
proc = procnode__caseobj(fun);
assign(list, bodylist__procnode(proc)); /* get the body ... */
make_tree_from_body(list);
if (!is_tree(list)) {
goto fetch_cont;
}
assign(unev, tree__tree(list));
if (NOT_THROWING) stopping_flag = RUN;
assign(output_node, UNBOUND);
if (val_status == 1) val_status = 2;
else if (val_status == 5) val_status = 3;
else val_status = 0;
eval_sequence:
/* Evaluate each expression in the sequence. Stop as soon as
* val != UNBOUND.
*/
if (!RUNNING || val != UNBOUND) {
goto fetch_cont;
}
if (nodetype(unev) == LINE) {
assign(this_line, unparsed__line(unev));
if (flag__caseobj(ufun, PROC_STEPPED)) {
char junk[20];
if (tracing) {
int i = 1;
while (i++ < trace_level) print_space(stdout);
}
print_node(stdout, this_line);
ndprintf(stdout, " >>> ");
input_blocking++;
#ifndef TIOCSTI
if (!setjmp(iblk_buf))
#endif
#ifdef __ZTC__
ztc_getcr();
#else
fgets(junk, 19, stdin);
#endif
input_blocking = 0;
update_coords('\n');
}
}
assign(exp, car(unev));
pop(unev);
if (is_list(exp) && (is_tailform(procnode__caseobj(car(exp))))) {
if (nameis(car(exp),Output) || nameis(car(exp),Op)) {
assign(didnt_get_output,
cons_list(0,car(exp),ufun,this_line,END_OF_LIST));
assign(didnt_output_name, NIL);
if (val_status == 2 || val_status == 3) {
val_status = 1;
assign(exp, cadr(exp));
goto tail_eval_dispatch;
} else if (ufun == NIL) {
err_logo(AT_TOPLEVEL,car(exp));
assign(val, UNBOUND);
goto fetch_cont;
} else if (val_status < 4) {
val_status = 1;
assign(exp, cadr(exp));
assign(unev, NIL);
goto non_tail_eval; /* compute value then give error */
}
} else if (nameis(car(exp),Stop)) {
if (ufun == NIL) {
err_logo(AT_TOPLEVEL,car(exp));
assign(val, UNBOUND);
goto fetch_cont;
} else if (val_status == 0 || val_status == 3) {
assign(val, UNBOUND);
goto fetch_cont;
} else if (val_status < 4) {
assign(didnt_output_name, fun);
assign(val, UNBOUND);
goto fetch_cont;
}
} else { /* maybeoutput */
assign(exp, cadr(exp));
val_status = 5;
goto tail_eval_dispatch;
}
}
if (unev == NIL) {
if (val_status == 2 || val_status == 4) {
assign(didnt_output_name, fun);
assign(unev, UNBOUND);
goto non_tail_eval;
} else {
goto tail_eval_dispatch;
}
}
if (is_list(car(unev)) && nameis(car(car(unev)),Stop)) {
if ((val_status == 0 || val_status == 3) && ufun != NIL) {
goto tail_eval_dispatch;
} else if (val_status < 4) {
assign(didnt_output_name, fun);
goto tail_eval_dispatch;
}
}
non_tail_eval:
save2(unev,fun);
num2save(ift_iff_flag,val_status);
save2(ufun,last_ufun);
save2(this_line,last_line);
save(var);
assign(var, var_stack);
tailcall = 0;
newcont(eval_sequence_continue);
goto eval_dispatch;
eval_sequence_continue:
reset_args(var);
restore(var);
restore2(this_line,last_line);
restore2(ufun,last_ufun);
if (dont_fix_ift) {
num2restore(dont_fix_ift,val_status);
dont_fix_ift = 0;
} else
num2restore(ift_iff_flag,val_status);
restore2(unev,fun);
if (stopping_flag == MACRO_RETURN) {
if (unev == UNBOUND) assign(unev, NIL);
assign(unev, append(val, unev));
assign(val, UNBOUND);
stopping_flag = RUN;
if (unev == NIL) goto fetch_cont;
} else if (val_status < 4) {
if (STOPPING || RUNNING) assign(output_node, UNBOUND);
if (stopping_flag == OUTPUT || STOPPING) {
stopping_flag = RUN;
assign(val, output_node);
if (val != UNBOUND && val_status < 2 && NOT_THROWING) {
assign(didnt_output_name,Output);
err_logo(DIDNT_OUTPUT,Output);
}
if (val == UNBOUND && val_status == 1 && NOT_THROWING) {
assign(didnt_output_name,Stop);
err_logo(DIDNT_OUTPUT,Output);
}
goto fetch_cont;
}
}
if (val != UNBOUND) {
err_logo((unev == NIL ? DK_WHAT_UP : DK_WHAT), val);
assign(val, UNBOUND);
}
if (NOT_THROWING && (unev == NIL || unev == UNBOUND)) {
if (val_status != 4) err_logo(DIDNT_OUTPUT,NIL);
goto fetch_cont;
}
goto eval_sequence;
compound_apply_continue:
/* Only get here if tracing */
restore(fun);
--trace_level;
if (NOT_THROWING) {
for (i = 0; i < trace_level; i++) print_space(writestream);
print_node(writestream, fun);
if (val == UNBOUND)
ndprintf(writestream, " stops\n");
else {
ref(val);
ndprintf(writestream, " outputs %s\n", maybe_quote(val));
deref(val);
}
}
goto fetch_cont;
/* --------------------- MACROS ---------------------------- */
macro_return:
num2restore(val_status,tailcall);
while (!is_list(val) && NOT_THROWING) {
assign(val,err_logo(ERR_MACRO,val));
}
if (NOT_THROWING) {
if (is_cont(val)) {
newcont(cont__cont(val));
val->n_car = NIL;
assign(val, val__cont(val));
goto fetch_cont;
}
macro_reval:
if (tailcall == 0) {
make_tree(val);
stopping_flag = MACRO_RETURN;
if (!is_tree(val)) assign(val, NIL);
else assign(val, tree__tree(val));
goto fetch_cont;
}
assign(list,val);
goto begin_seq;
}
assign(val, UNBOUND);
goto fetch_cont;
runresult_continuation:
assign(list, val);
newcont(runresult_followup);
val_status = 5;
goto begin_seq;
runresult_followup:
if (val == UNBOUND) {
assign(val, NIL);
} else {
assign(val, cons(val, NIL));
}
goto fetch_cont;
repeat_continuation:
assign(list, cdr(val));
repcount = getint(car(val));
repeat_again:
assign(val, UNBOUND);
if (repcount == 0) goto fetch_cont;
mixsave(repcount,list);
num2save(val_status,tailcall);
val_status = 4;
newcont(repeat_followup);
goto begin_seq;
repeat_followup:
if (val != UNBOUND && NOT_THROWING) {
ref(val);
err_logo(DK_WHAT, val);
unref(val);
}
num2restore(val_status,tailcall);
mixrestore(repcount,list);
if (val_status < 4 && tailcall != 0) {
if (STOPPING || RUNNING) assign(output_node, UNBOUND);
if (stopping_flag == OUTPUT || STOPPING) {
stopping_flag = RUN;
assign(val, output_node);
if (val != UNBOUND && val_status < 2) {
err_logo(DK_WHAT_UP,val);
}
goto fetch_cont;
}
}
if (repcount > 0) /* negative means forever */
--repcount;
#ifdef mac
check_mac_stop();
#endif
#ifdef ibm
check_ibm_stop();
#endif
if (RUNNING) goto repeat_again;
assign(val, UNBOUND);
goto fetch_cont;
catch_continuation:
assign(list, cdr(val));
assign(catch_tag, car(val));
if (compare_node(catch_tag,Error,TRUE) == 0) {
push(Erract, var_stack);
setobject(var_stack, valnode__caseobj(Erract));
setvalnode__caseobj(Erract, UNBOUND);
}
save(catch_tag);
save2(didnt_output_name,didnt_get_output);
num2save(val_status,tailcall);
newcont(catch_followup);
val_status = 5;
goto begin_seq;
catch_followup:
num2restore(val_status,tailcall);
restore2(didnt_output_name,didnt_get_output);
restore(catch_tag);
if (val_status < 4 && tailcall != 0) {
if (STOPPING || RUNNING) assign(output_node, UNBOUND);
if (stopping_flag == OUTPUT || STOPPING) {
stopping_flag = RUN;
assign(val, output_node);
if (val != UNBOUND && val_status < 2) {
err_logo(DK_WHAT_UP,val);
}
}
}
if (stopping_flag == THROWING &&
compare_node(throw_node, catch_tag, TRUE) == 0) {
throw_node = reref(throw_node, UNBOUND);
stopping_flag = RUN;
assign(val, output_node);
}
goto fetch_cont;
begin_apply:
/* This is for lapply. */
assign(fun, car(val));
while (nodetype(fun) == ARRAY && NOT_THROWING)
assign(fun, err_logo(APPLY_BAD_DATA, fun));
assign(argl, cadr(val));
assign(val, UNBOUND);
while (!is_list(argl) && NOT_THROWING)
assign(argl, err_logo(APPLY_BAD_DATA, argl));
if (NOT_THROWING && fun != NIL) {
if (is_list(fun)) { /* template */
if (is_list(car(fun)) && cdr(fun) != NIL) {
/* lambda form */
formals = (FIXNUM)car(fun);
numsave(tailcall);
tailcall = 0;
llocal((NODE *)formals); /* bind the formals locally */
numrestore(tailcall);
for ( ;
formals && argl && NOT_THROWING;
formals = (FIXNUM)cdr((NODE *)formals),
assign(argl, cdr(argl)))
setvalnode__caseobj(car((NODE *)formals), car(argl));
assign(val, cdr(fun));
goto macro_reval;
} else { /* question-mark form */
save(qm_list);
assign(qm_list, argl);
assign(list, fun);
make_tree(list);
if (list == NIL || !is_tree(list)) {
goto qm_failed;
}
assign(unev, tree__tree(list));
save2(didnt_output_name,didnt_get_output);
num2save(val_status,tailcall);
newcont(qm_continue);
val_status = 5;
goto eval_sequence;
qm_continue:
num2restore(val_status,tailcall);
restore2(didnt_output_name,didnt_get_output);
if (val_status < 4 && tailcall != 0) {
if (STOPPING || RUNNING) assign(output_node, UNBOUND);
if (stopping_flag == OUTPUT || STOPPING) {
stopping_flag = RUN;
assign(val, output_node);
if (val != UNBOUND && val_status < 2) {
err_logo(DK_WHAT_UP,val);
}
}
}
qm_failed:
restore(qm_list);
goto fetch_cont;
}
} else { /* name of procedure to apply */
int min, max, n;
NODE *arg;
assign(fun, intern(fun));
if (procnode__caseobj(fun) == UNDEFINED && NOT_THROWING &&
fun != Null_Word)
silent_load(fun, NULL); /* try ./<fun>.lg */
if (procnode__caseobj(fun) == UNDEFINED && NOT_THROWING &&
fun != Null_Word)
silent_load(fun, logolib); /* try <logolib>/<fun> */
proc = procnode__caseobj(fun);
while (proc == UNDEFINED && NOT_THROWING) {
assign(val, err_logo(DK_HOW_UNREC, fun));
}
if (NOT_THROWING) {
if (nodetype(proc) == CONS) {
min = getint(minargs__procnode(proc));
max = getint(maxargs__procnode(proc));
} else {
if (getprimdflt(proc) < 0) { /* special form */
err_logo(DK_HOW_UNREC, fun); /* can't apply */
goto fetch_cont;
} else {
min = getprimmin(proc);
max = getprimmax(proc);
}
}
for (n = 0, arg = argl; arg != NIL; n++, arg = cdr(arg));
if (n < min) {
err_logo(NOT_ENOUGH, NIL);
} else if (n > max && max >= 0) {
err_logo(TOO_MUCH, NIL);
} else {
goto apply_dispatch;
}
}
}
}
goto fetch_cont;
all_done:
tailcall = oldtailcall;
ift_iff_flag = old_ift_iff;
restore2(fun,ufun);
reset_args(var);
restore2(var,this_line);
deref(argl);deref(unev);deref(stack);deref(catch_tag);deref(exp);
return(val);
}
static VALUE protected_tidy_bytes(VALUE arg) {
struct protected_tidy_bytes_arg* args = (struct protected_tidy_bytes_arg *)arg;
VALUE string = args->string;
uint8_t *arr = args->arr;
int conts_expected = 0;
int i, j;
uint8_t byte;
uint8_t *curr = arr;
uint8_t *last_lead = curr;
int did_write = 0;
for (i = 0; i < RSTRING_LEN(string) ; i++) {
byte = RSTRING_PTR(string)[i];
curr[0] = byte;
did_write = 0;
if (is_unused(byte) || is_restricted(byte)) {
curr = tidy_byte2(byte, curr);
did_write = 1;
} else if (is_cont(byte)) {
if (conts_expected == 0) {
curr = tidy_byte2(byte, curr);
did_write = 1;
} else {
conts_expected--;
}
} else {
if (conts_expected > 0) {
int start_index = last_lead-arr;
int end_index = curr-arr;
int len = end_index - start_index;
uint8_t temp[len];
for (j = 0; j < len; j++) {
temp[j] = arr[start_index + j];
}
curr = arr + start_index;
for (j = 0; j < len; j++) {
curr = tidy_byte2(temp[j], curr);
}
conts_expected = 0;
}
if (is_lead(byte)) {
if (i == RSTRING_LEN(string) - 1) {
curr = tidy_byte2(byte, curr);
did_write = 1;
} else {
if (byte < 224) {
conts_expected = 1;
} else if (byte < 240) {
conts_expected = 2;
} else {
conts_expected = 3;
}
last_lead = curr;
}
}
}
if (!did_write) {
*curr++ = byte;
}
}
VALUE str = rb_str_new((const char *)arr, curr-arr);
return rb_funcall(str, idForceEncoding, 1, rb_str_new2("UTF-8"));
}