static bool
read_0_9(SerdReader* reader, Ref str, bool at_least_one)
{
unsigned count = 0;
for (uint8_t c; is_digit((c = peek_byte(reader))); ++count) {
push_byte(reader, str, eat_byte_safe(reader, c));
}
if (at_least_one && count == 0) {
r_err(reader, SERD_ERR_BAD_SYNTAX, "expected digit\n");
}
return count;
}
// Read ECHAR escape, initial \ is already eaten by caller
static inline bool
read_ECHAR(SerdReader* reader, Ref dest, SerdNodeFlags* flags)
{
const uint8_t c = peek_byte(reader);
switch (c) {
case 't':
eat_byte_safe(reader, 't');
push_byte(reader, dest, '\t');
return true;
case 'b':
eat_byte_safe(reader, 'b');
push_byte(reader, dest, '\b');
return true;
case 'n':
*flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'n');
push_byte(reader, dest, '\n');
return true;
case 'r':
*flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'r');
push_byte(reader, dest, '\r');
return true;
case 'f':
eat_byte_safe(reader, 'f');
push_byte(reader, dest, '\f');
return true;
case '\\': case '"': case '\'':
push_byte(reader, dest, eat_byte_safe(reader, c));
return true;
default:
return false;
}
}
static bool
read_blank(SerdReader* reader, ReadContext ctx, bool subject, Ref* dest, bool* ate_dot)
{
const SerdStatementFlags old_flags = *ctx.flags;
bool empty;
switch (peek_byte(reader)) {
case '_':
return (*dest = read_BLANK_NODE_LABEL(reader, ate_dot));
case '[':
eat_byte_safe(reader, '[');
if ((empty = peek_delim(reader, ']'))) {
*ctx.flags |= (subject) ? SERD_EMPTY_S : SERD_EMPTY_O;
} else {
*ctx.flags |= (subject) ? SERD_ANON_S_BEGIN : SERD_ANON_O_BEGIN;
if (peek_delim(reader, '=')) {
if (!(*dest = read_blankName(reader)) ||
!eat_delim(reader, ';')) {
return false;
}
}
}
if (!*dest) {
*dest = blank_id(reader);
}
if (ctx.subject) {
TRY_RET(emit_statement(reader, ctx, *dest, 0, 0));
}
ctx.subject = *dest;
if (!empty) {
*ctx.flags &= ~(SERD_LIST_CONT);
if (!subject) {
*ctx.flags |= SERD_ANON_CONT;
}
bool ate_dot_in_list = false;
read_predicateObjectList(reader, ctx, &ate_dot_in_list);
if (ate_dot_in_list) {
return r_err(reader, SERD_ERR_BAD_SYNTAX, "`.' inside blank\n");
}
read_ws_star(reader);
if (reader->end_sink) {
reader->end_sink(reader->handle, deref(reader, *dest));
}
*ctx.flags = old_flags;
}
return (eat_byte_check(reader, ']') == ']');
case '(':
return read_collection(reader, ctx, dest);
default: return false; // never reached
}
}
static bool
read_PERCENT(SerdReader* reader, Ref dest)
{
push_byte(reader, dest, eat_byte_safe(reader, '%'));
const uint8_t h1 = read_HEX(reader);
const uint8_t h2 = read_HEX(reader);
if (h1 && h2) {
push_byte(reader, dest, h1);
push_byte(reader, dest, h2);
return true;
}
return false;
}
static SerdStatus
read_PN_LOCAL(SerdReader* reader, Ref dest, bool* ate_dot)
{
uint8_t c = peek_byte(reader);
SerdStatus st;
switch (c) {
case '0': case '1': case '2': case '3': case '4': case '5':
case '6': case '7': case '8': case '9': case ':': case '_':
push_byte(reader, dest, eat_byte_safe(reader, c));
break;
default:
if ((st = read_PLX(reader, dest)) > SERD_FAILURE) {
return st;
} else if (st != SERD_SUCCESS && !read_PN_CHARS_BASE(reader, dest)) {
return SERD_FAILURE;
}
}
while ((c = peek_byte(reader))) { // Middle: (PN_CHARS | '.' | ';')*
if (c == '.' || c == ':') {
push_byte(reader, dest, eat_byte_safe(reader, c));
} else if ((st = read_PLX(reader, dest)) > SERD_FAILURE) {
return st;
} else if (st != SERD_SUCCESS && !read_PN_CHARS(reader, dest)) {
break;
}
}
SerdNode* const n = deref(reader, dest);
if (n->buf[n->n_bytes - 1] == '.') {
// Ate trailing dot, pop it from stack/node and inform caller
--n->n_bytes;
serd_stack_pop(&reader->stack, 1);
*ate_dot = true;
}
return SERD_SUCCESS;
}
static inline SerdStatus
bad_char(SerdReader* reader, Ref dest, const char* fmt, uint8_t c)
{
r_err(reader, SERD_ERR_BAD_SYNTAX, fmt, c);
push_replacement(reader, dest);
// Skip bytes until the next start byte
for (uint8_t b = peek_byte(reader); (b & 0x80);) {
eat_byte_safe(reader, b);
b = peek_byte(reader);
}
return SERD_SUCCESS;
}
static Ref
read_blankName(SerdReader* reader)
{
eat_byte_safe(reader, '=');
if (eat_byte_check(reader, '=') != '=') {
return r_err(reader, SERD_ERR_BAD_SYNTAX, "expected `='\n");
}
Ref subject = 0;
bool ate_dot = false;
read_ws_star(reader);
read_iri(reader, &subject, &ate_dot);
return subject;
}
// STRING_LITERAL_LONG_QUOTE and STRING_LITERAL_LONG_SINGLE_QUOTE
// Initial triple quotes are already eaten by caller
static Ref
read_STRING_LITERAL_LONG(SerdReader* reader, SerdNodeFlags* flags, uint8_t q)
{
Ref ref = push_node(reader, SERD_LITERAL, "", 0);
while (true) {
const uint8_t c = peek_byte(reader);
uint32_t code;
switch (c) {
case '\\':
eat_byte_safe(reader, c);
if (!read_ECHAR(reader, ref, flags) &&
!read_UCHAR(reader, ref, &code)) {
r_err(reader, SERD_ERR_BAD_SYNTAX,
"invalid escape `\\%c'\n", peek_byte(reader));
return pop_node(reader, ref);
}
break;
default:
if (c == q) {
eat_byte_safe(reader, q);
const uint8_t q2 = eat_byte_safe(reader, peek_byte(reader));
const uint8_t q3 = peek_byte(reader);
if (q2 == q && q3 == q) { // End of string
eat_byte_safe(reader, q3);
return ref;
} else {
*flags |= SERD_HAS_QUOTE;
push_byte(reader, ref, c);
read_character(reader, ref, flags, q2);
}
} else {
read_character(reader, ref, flags, eat_byte_safe(reader, c));
}
}
}
return ref;
}
static SerdStatus
read_PLX(SerdReader* reader, Ref dest)
{
uint8_t c = peek_byte(reader);
switch (c) {
case '%':
if (!read_PERCENT(reader, dest)) {
return SERD_ERR_BAD_SYNTAX;
}
return SERD_SUCCESS;
case '\\':
eat_byte_safe(reader, c);
if (is_alpha(c = peek_byte(reader))) {
// Escapes like \u \n etc. are not supported
return SERD_ERR_BAD_SYNTAX;
} else {
// Allow escaping of pretty much any other character
push_byte(reader, dest, eat_byte_safe(reader, c));
return SERD_SUCCESS;
}
default:
return SERD_FAILURE;
}
}
// [24] ws ::= #x9 | #xA | #xD | #x20 | comment
static inline bool
read_ws(SerdReader* reader)
{
const uint8_t c = peek_byte(reader);
switch (c) {
case 0x9: case 0xA: case 0xD: case 0x20:
eat_byte_safe(reader, c);
return true;
case '#':
read_comment(reader);
return true;
default:
return false;
}
}
// Read the remainder of a PN_PREFIX after some initial characters
static SerdStatus
read_PN_PREFIX_tail(SerdReader* reader, Ref dest)
{
uint8_t c;
while ((c = peek_byte(reader))) { // Middle: (PN_CHARS | '.')*
if (c == '.') {
push_byte(reader, dest, eat_byte_safe(reader, c));
} else if (!read_PN_CHARS(reader, dest)) {
break;
}
}
const SerdNode* const n = deref(reader, dest);
if (n->buf[n->n_bytes - 1] == '.' && !read_PN_CHARS(reader, dest)) {
r_err(reader, SERD_ERR_BAD_SYNTAX, "prefix ends with `.'\n");
return SERD_ERR_BAD_SYNTAX;
}
return SERD_SUCCESS;
}
static bool
read_number(SerdReader* reader, Ref* dest, Ref* datatype, bool* ate_dot)
{
#define XSD_DECIMAL NS_XSD "decimal"
#define XSD_DOUBLE NS_XSD "double"
#define XSD_INTEGER NS_XSD "integer"
Ref ref = push_node(reader, SERD_LITERAL, "", 0);
uint8_t c = peek_byte(reader);
bool has_decimal = false;
if (c == '-' || c == '+') {
push_byte(reader, ref, eat_byte_safe(reader, c));
}
if ((c = peek_byte(reader)) == '.') {
has_decimal = true;
// decimal case 2 (e.g. '.0' or `-.0' or `+.0')
push_byte(reader, ref, eat_byte_safe(reader, c));
TRY_THROW(read_0_9(reader, ref, true));
} else {
// all other cases ::= ( '-' | '+' ) [0-9]+ ( . )? ( [0-9]+ )? ...
TRY_THROW(is_digit(c));
read_0_9(reader, ref, true);
if ((c = peek_byte(reader)) == '.') {
has_decimal = true;
// Annoyingly, dot can be end of statement, so tentatively eat
eat_byte_safe(reader, c);
c = peek_byte(reader);
if (!is_digit(c) && c != 'e' && c != 'E') {
*dest = ref;
*ate_dot = true; // Force caller to deal with stupid grammar
return true; // Next byte is not a number character, done
}
push_byte(reader, ref, '.');
read_0_9(reader, ref, false);
}
}
c = peek_byte(reader);
if (c == 'e' || c == 'E') {
// double
push_byte(reader, ref, eat_byte_safe(reader, c));
switch ((c = peek_byte(reader))) {
case '+': case '-':
push_byte(reader, ref, eat_byte_safe(reader, c));
default: break;
}
TRY_THROW(read_0_9(reader, ref, true));
*datatype = push_node(reader, SERD_URI,
XSD_DOUBLE, sizeof(XSD_DOUBLE) - 1);
} else if (has_decimal) {
*datatype = push_node(reader, SERD_URI,
XSD_DECIMAL, sizeof(XSD_DECIMAL) - 1);
} else {
*datatype = push_node(reader, SERD_URI,
XSD_INTEGER, sizeof(XSD_INTEGER) - 1);
}
*dest = ref;
return true;
except:
pop_node(reader, *datatype);
pop_node(reader, ref);
return false;
}
// Read UCHAR escape, initial \ is already eaten by caller
static inline bool
read_UCHAR(SerdReader* reader, Ref dest, uint32_t* char_code)
{
const uint8_t b = peek_byte(reader);
unsigned length = 0;
switch (b) {
case 'U':
length = 8;
break;
case 'u':
length = 4;
break;
default:
return false;
}
eat_byte_safe(reader, b);
uint8_t buf[9] = { 0, 0, 0, 0, 0, 0, 0, 0, 0 };
for (unsigned i = 0; i < length; ++i) {
if (!(buf[i] = read_HEX(reader))) {
return false;
}
}
uint32_t code;
sscanf((const char*)buf, "%X", &code);
unsigned size = 0;
if (code < 0x00000080) {
size = 1;
} else if (code < 0x00000800) {
size = 2;
} else if (code < 0x00010000) {
size = 3;
} else if (code < 0x00110000) {
size = 4;
} else {
r_err(reader, SERD_ERR_BAD_SYNTAX,
"unicode character 0x%X out of range\n", code);
push_replacement(reader, dest);
*char_code = 0xFFFD;
return true;
}
// Build output in buf
// (Note # of bytes = # of leading 1 bits in first byte)
uint32_t c = code;
switch (size) {
case 4:
buf[3] = 0x80 | (uint8_t)(c & 0x3F);
c >>= 6;
c |= (16 << 12); // set bit 4
case 3:
buf[2] = 0x80 | (uint8_t)(c & 0x3F);
c >>= 6;
c |= (32 << 6); // set bit 5
case 2:
buf[1] = 0x80 | (uint8_t)(c & 0x3F);
c >>= 6;
c |= 0xC0; // set bits 6 and 7
case 1:
buf[0] = (uint8_t)c;
}
for (unsigned i = 0; i < size; ++i) {
push_byte(reader, dest, buf[i]);
}
*char_code = code;
return true;
}