438 lines
13 KiB
C
438 lines
13 KiB
C
#include "../stdio_p.h"
|
|
#include "../../stdlib/stdlib_p.h"
|
|
#include <stdio.h>
|
|
#include <ctype.h>
|
|
#include <string.h>
|
|
#include <stdbool.h>
|
|
|
|
void __scanf_start(struct __scanf_input *in)
|
|
{
|
|
if(in->fp)
|
|
in->buffer = fgetc(in->fp);
|
|
else {
|
|
in->buffer = (*in->str ? *in->str : EOF);
|
|
in->str += (in->buffer != EOF);
|
|
}
|
|
}
|
|
|
|
int __scanf_fetch(struct __scanf_input *in)
|
|
{
|
|
if(in->fp)
|
|
return fgetc(in->fp);
|
|
|
|
int c = *in->str;
|
|
if(c == 0)
|
|
return EOF;
|
|
in->str++;
|
|
return c;
|
|
}
|
|
|
|
void __scanf_end(struct __scanf_input *in)
|
|
{
|
|
if(in->buffer == EOF)
|
|
return;
|
|
|
|
if(in->fp)
|
|
ungetc(in->buffer, in->fp);
|
|
else
|
|
in->str--;
|
|
}
|
|
|
|
void __purge_space( struct __scanf_input * __restrict__ in )
|
|
{
|
|
while (isspace(__scanf_peek(in))) __scanf_in(in);
|
|
}
|
|
|
|
void __scanf_store_i(int64_t value, int size, va_list *args)
|
|
{
|
|
if(size == 1)
|
|
*va_arg(*args, int8_t *) = value;
|
|
else if(size == 2)
|
|
*va_arg(*args, int16_t *) = value;
|
|
else if(size == 4)
|
|
*va_arg(*args, int32_t *) = value;
|
|
else if(size == 8)
|
|
*va_arg(*args, int64_t *) = value;
|
|
}
|
|
|
|
void __scanf_store_d(long double value, int size, va_list *args)
|
|
{
|
|
if(size == sizeof(float))
|
|
*va_arg(*args, float *) = value;
|
|
else if(size == sizeof(double))
|
|
*va_arg(*args, double *) = value;
|
|
else if(size == sizeof(long double))
|
|
*va_arg(*args, long double *) = value;
|
|
}
|
|
|
|
/* STATUS OF __scanf DEVELOPMENT */
|
|
|
|
// XX = not done yet (but will be done)
|
|
// OK = OK, done and tested
|
|
// -- = not applicable
|
|
// NO = not supported (and will not be done) only for %lc as long char are not supported by gint
|
|
|
|
|
|
/*************************************************************************************************************/
|
|
/* Specifier * Explanation * num * hh * h *none* l * ll * j * z * t * L */
|
|
/*************************************************************************************************************/
|
|
/* % * Parse literal '%' * -- * -- * -- * OK * -- * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
/* c * match a char or several char * OK * -- * -- * OK * NO * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
/* s * match a string * OK * -- * -- * OK * NO * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
/* [set] * match a set of char * OK * -- * -- * OK * -- * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
/* d * match a decimal integer * OK * OK * OK * OK * OK * OK * OK * OK * OK * -- */
|
|
/*************************************************************************************************************/
|
|
/* i * match an integer * OK * OK * OK * OK * OK * OK * OK * OK * OK * -- */
|
|
/*************************************************************************************************************/
|
|
/* d * match an unsigned decimal integer * OK * OK * OK * OK * OK * OK * OK * OK * OK * -- */
|
|
/*************************************************************************************************************/
|
|
/* o * match a unsigned octal integer * OK * OK * OK * OK * OK * OK * OK * OK * OK * -- */
|
|
/*************************************************************************************************************/
|
|
/* x,X * match a unsigned hexadecimal integer * OK * OK * OK * OK * OK * OK * OK * OK * OK * -- */
|
|
/*************************************************************************************************************/
|
|
/* n * return the nb of chars read so far * -- * -- * -- * OK * -- * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
/* a,A * match a floating point number * OK * -- * -- * OK * OK * -- * -- * -- * -- * OK */
|
|
/* e,E * match a floating point number * OK * -- * -- * OK * OK * -- * -- * -- * -- * OK */
|
|
/* f,F * match a floating point number * OK * -- * -- * OK * OK * -- * -- * -- * -- * OK */
|
|
/* g,G * match a floating point number * OK * -- * -- * OK * OK * -- * -- * -- * -- * OK */
|
|
/*************************************************************************************************************/
|
|
/* p * match a pointer * -- * -- * -- * OK * -- * -- * -- * -- * -- * -- */
|
|
/*************************************************************************************************************/
|
|
|
|
// %ms and %m[set] are not implemented (with memory allocation while parsing a chain or a set of characters)
|
|
|
|
|
|
/* Set of bytes allowed in a given set %[]. */
|
|
static uint8_t bracket_set[32];
|
|
|
|
/* Allow/disallow the entire set */
|
|
static void bracket_set_init(bool allow)
|
|
{
|
|
memset(bracket_set, allow ? 0xff : 0x00, sizeof bracket_set);
|
|
}
|
|
|
|
/* Allow/disallow a range of characters. Both ends are included. */
|
|
static void bracket_set_range(uint8_t start, uint8_t end, bool allow)
|
|
{
|
|
for(int u = start; u <= end; u++) {
|
|
int byte = u >> 3;
|
|
int bit = 1 << (u & 7);
|
|
if(allow)
|
|
bracket_set[byte] |= bit;
|
|
else
|
|
bracket_set[byte] &= ~bit;
|
|
}
|
|
}
|
|
|
|
/* Check whether a byte is allowed by the bracket set. */
|
|
static bool bracket_set_test(int c)
|
|
{
|
|
int byte = (c >> 3);
|
|
int bit = 1 << (c & 7);
|
|
return (c != EOF) && (bracket_set[byte] & bit);
|
|
}
|
|
|
|
/* return 0 if Ok or -1 if syntax err in the set format */
|
|
static int bracket_set_parse(char const * __restrict__ format, int *pos )
|
|
{
|
|
int __sor = 0;
|
|
int __eor = 0;
|
|
bool __neg = false;
|
|
|
|
bracket_set_init(false);
|
|
|
|
(*pos)++;
|
|
|
|
// next will be a "negation" set
|
|
if (format[*pos] == '^' ) {
|
|
__neg = true;
|
|
(*pos)++;
|
|
bracket_set_init(true);
|
|
|
|
// the char ']' is part of the set
|
|
if (format[*pos] == ']' ) {
|
|
bracket_set_range(']', ']', !__neg);
|
|
(*pos)++;
|
|
}
|
|
}
|
|
// the char ']' is included in the allowed set
|
|
else if (format[*pos] == ']' ) {
|
|
__neg = false;
|
|
// the char ']' is part of the set
|
|
if (format[*pos] == ']' ) {
|
|
bracket_set_range(']', ']', !__neg);
|
|
(*pos)++;
|
|
}
|
|
}
|
|
|
|
while(1) {
|
|
// we find a '-' so need to check if we are considering a range or the char '-' only
|
|
if (format[*pos]=='-') {
|
|
// the char '-' is included in the allowed set
|
|
if (format[*pos+1]==']') {
|
|
bracket_set_range('-', '-', !__neg);
|
|
(*pos)++;
|
|
// we have now finished the reading of the set cause the following char is ']'
|
|
return 0;
|
|
}
|
|
// the char '-' indicates a range of char to be included into the set
|
|
else {
|
|
(*pos)++;
|
|
__eor = format[*pos];
|
|
bracket_set_range( __sor, __eor, !__neg );
|
|
}
|
|
}
|
|
// we find the char ']' so it means we reach the end of this set
|
|
else if (format[*pos]==']') return 0;
|
|
// if we reach the '\0' we have a syntax problem
|
|
else if (format[*pos]=='\0') return -1;
|
|
// we are considering one particular char and prepare for a potential range if we find the char '-' later on
|
|
else {
|
|
__sor = format[*pos];
|
|
bracket_set_range(__sor, __sor, !__neg);
|
|
(*pos)++;
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
|
|
int __scanf(
|
|
struct __scanf_input * __restrict__ in,
|
|
char const * __restrict__ format,
|
|
va_list *args)
|
|
{
|
|
|
|
bool skip = false;
|
|
int MOD = sizeof(int);
|
|
|
|
in->bytes_read = 0; // we haven't started to read char from the input stream
|
|
|
|
int validrets = 0; // to be incremented each time we successfully read and store an input as per the format
|
|
int err = 0; // err control on __strto_xx( ) functions
|
|
int user_length = 0; // length defined by user with a %xx modifier
|
|
int pos = 0; // current pos in the format string
|
|
|
|
__scanf_start( in );
|
|
|
|
// TODO: No __scanf_end() in any of the "return validrets"!!
|
|
|
|
while( format[pos] != 0 ) {
|
|
user_length = 0;
|
|
MOD = sizeof(int);
|
|
skip = false;
|
|
|
|
if( format[pos] == ' ' ) {
|
|
__purge_space(in);
|
|
}
|
|
// we will have to manage a given format
|
|
else if( format[pos] == '%' ) {
|
|
|
|
int readmaxlength = INT_MAX;
|
|
char size_letter = 0;
|
|
|
|
// main loop
|
|
loopagain:
|
|
|
|
pos++;
|
|
|
|
switch(format[pos]) {
|
|
// we need to decrypt the corresponding scanf set of character
|
|
case '[': {
|
|
err = bracket_set_parse( format, &pos );
|
|
if (err!=0) return validrets;
|
|
int currentlength = 0;
|
|
// we need to assign the read char to the corresponding pointer
|
|
char *c = skip ? NULL : va_arg(*args, char *);
|
|
for(int u=0; u<readmaxlength; u++) {
|
|
int temp = __scanf_peek(in);
|
|
if(bracket_set_test(temp)) {
|
|
__scanf_in(in);
|
|
if(c) *c++ = temp;
|
|
currentlength++;
|
|
}
|
|
else if(temp==EOF && !currentlength && !validrets)
|
|
return EOF;
|
|
else break;
|
|
}
|
|
if(!currentlength)
|
|
return validrets;
|
|
*c = '\0';
|
|
validrets += !skip;
|
|
break;
|
|
}
|
|
|
|
// return the number of char read so far (cannot be skipped %*n is not valid)
|
|
case 'n':
|
|
*va_arg(*args, int *) = in->bytes_read;
|
|
break;
|
|
|
|
// we are expecting the char '%' to be in the input stream, if not err and return
|
|
case '%': {
|
|
if (__scanf_peek(in) != '%') return validrets;
|
|
else __scanf_in( in );
|
|
break;
|
|
}
|
|
|
|
// the next read, even if valid, will not be stored
|
|
case '*':
|
|
skip = true;
|
|
goto loopagain;
|
|
|
|
case 'h':
|
|
MOD = (size_letter == 'h') ? sizeof(char) : sizeof(short);
|
|
size_letter = 'h';
|
|
goto loopagain;
|
|
case 'l':
|
|
MOD = (size_letter == 'l') ? sizeof(long long) : sizeof(long);
|
|
/* FP conversions will adjust to sizeof(double) later */
|
|
size_letter = 'l';
|
|
goto loopagain;
|
|
case 'L':
|
|
MOD = sizeof(long double);
|
|
size_letter = 'L';
|
|
goto loopagain;
|
|
case 'j':
|
|
MOD = sizeof(intmax_t);
|
|
goto loopagain;
|
|
case 'z':
|
|
MOD = sizeof(size_t);
|
|
goto loopagain;
|
|
case 't':
|
|
MOD = sizeof(ptrdiff_t);
|
|
goto loopagain;
|
|
|
|
case '0' ... '9': {
|
|
user_length = user_length * 10 + (int) ( format[pos] - '0' );
|
|
readmaxlength = user_length;
|
|
goto loopagain;
|
|
break;
|
|
}
|
|
|
|
case 'd':
|
|
case 'i':
|
|
case 'o':
|
|
case 'u':
|
|
case 'x':
|
|
case 'X': {
|
|
int f = format[pos];
|
|
int base = (f == 'd' || f == 'u') ? 10 :
|
|
(f == 'o') ? 8:
|
|
(f == 'x' || f == 'X') ? 16 : 0;
|
|
bool use_unsigned = (f == 'o' || f == 'x' || f == 'X');
|
|
|
|
long long int temp;
|
|
err = __strto_int(in, base, NULL, &temp, use_unsigned,
|
|
readmaxlength);
|
|
if (err == EOF && validrets == 0) return EOF;
|
|
if (err != 0) return validrets;
|
|
if (skip) __scanf_store_i( temp, 0, args );
|
|
else __scanf_store_i( temp, MOD, args );
|
|
validrets++;
|
|
break;
|
|
}
|
|
|
|
case 'a':
|
|
case 'A':
|
|
case 'e':
|
|
case 'E':
|
|
case 'f':
|
|
case 'F':
|
|
case 'g':
|
|
case 'G': {
|
|
/* Adjust interpretation of no size / 'l' size */
|
|
if(size_letter == 0)
|
|
MOD = sizeof(float);
|
|
if(size_letter == 'l')
|
|
MOD = sizeof(double);
|
|
|
|
// read a double from the current input stream
|
|
// and store in the corresponding arg as a char by reference
|
|
long double temp;
|
|
err = __strto_fp( in, NULL, NULL, &temp,
|
|
readmaxlength);
|
|
if (err == EOF && validrets == 0) return EOF;
|
|
if (err != 0) return validrets;
|
|
if (skip) __scanf_store_d( temp, 0, args );
|
|
else __scanf_store_d( temp, MOD, args );
|
|
validrets++;
|
|
break;
|
|
}
|
|
|
|
case 'p': {
|
|
long int temp;
|
|
if (!skip) {
|
|
void *p = (void *) va_arg( *args, void** ); // get the adress of the target pointer (void**)
|
|
err = __strto_int( in, 0, p, NULL, true,
|
|
readmaxlength);
|
|
}
|
|
else err = __strto_int( in, 0, &temp, NULL, true,
|
|
readmaxlength);
|
|
if (err == 0) validrets++;
|
|
else return validrets;
|
|
skip = false;
|
|
break;
|
|
}
|
|
|
|
case 'c': {
|
|
if(readmaxlength == INT_MAX)
|
|
readmaxlength = 1;
|
|
char *c = skip ? NULL : va_arg(*args, char *);
|
|
|
|
for(int u = 0; u < readmaxlength; u++) {
|
|
int temp = __scanf_in(in);
|
|
if(temp==EOF) return EOF;
|
|
else if(c) *c++ = temp;
|
|
}
|
|
validrets += !skip;
|
|
break;
|
|
}
|
|
|
|
case 's': {
|
|
char temp;
|
|
int curstrlength = 0;
|
|
__purge_space(in);
|
|
|
|
char *c = skip ? NULL : va_arg(*args, char *);
|
|
for(int u = 0; u < readmaxlength; u++) {
|
|
temp = __scanf_peek(in);
|
|
if(temp==EOF && curstrlength==0) return validrets;
|
|
if(isspace(temp) || ((temp==EOF && curstrlength!=0))) {
|
|
if(c) {
|
|
*c = 0;
|
|
validrets++;
|
|
}
|
|
break;
|
|
}
|
|
else {
|
|
int temp = __scanf_in( in );
|
|
if(c)
|
|
*c++ = temp;
|
|
curstrlength++;
|
|
}
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
// we are looking for a specific character in the input stream
|
|
else {
|
|
// if the next char of the stream is corresponding, we validate the read and go to the following char
|
|
if(format[pos] == __scanf_peek( in )) {
|
|
__scanf_in( in );
|
|
pos++;
|
|
}
|
|
else return validrets; // else we return the number of valid read
|
|
}
|
|
pos++;
|
|
}
|
|
|
|
__scanf_end( in );
|
|
return validrets;
|
|
}
|