char *
PJS_ConvertUC(
pTHX_
SV *sv,
STRLEN *len
) {
dSP;
char *ret;
STRLEN elen;
SvPV_force(sv, elen);
if(SvUTF8(sv) && !sv_utf8_downgrade(sv, 1)) {
SV *svtmp;
ENTER; SAVETMPS;
PUSHMARK(SP);
XPUSHs(sv_2mortal(newSVpv(PJS_STR_ENCODING, 0)));
XPUSHs(sv_mortalcopy(sv));
PUTBACK;
call_pv("Encode::encode", G_SCALAR);
SPAGAIN;
svtmp = newSVsv(POPs);
SAVEMORTALIZESV(svtmp);
ret = SvPV(svtmp, elen);
PUTBACK;
FREETMPS; LEAVE;
*len = -(elen / 2);
} else {
ret = SvPV(sv, elen);
*len = elen;
}
return ret;
}
char *
Perl_sv_pvbyten(pTHX_ SV *sv, STRLEN *lp)
{
PERL_ARGS_ASSERT_SV_PVBYTEN;
sv_utf8_downgrade(sv, FALSE);
return sv_pvn(sv,lp);
}
char *
Perl_sv_pvbyte(pTHX_ SV *sv)
{
PERL_ARGS_ASSERT_SV_PVBYTE;
sv_utf8_downgrade(sv, FALSE);
return sv_pv(sv);
}
/* This is the main routine to deserialize a structure.
* It rolls up all the other "top level" routines into one
*/
SV *
srl_decode_into(pTHX_ srl_decoder_t *dec, SV *src, SV* into, UV start_offset)
{
assert(dec != NULL);
if (SvUTF8(src))
sv_utf8_downgrade(src, 0);
srl_begin_decoding(aTHX_ dec, src, start_offset);
srl_read_header(aTHX_ dec);
if (SRL_DEC_HAVE_OPTION(dec, SRL_F_DECODER_DECOMPRESS_SNAPPY)) {
/* uncompress */
uint32_t dest_len;
SV *buf_sv;
unsigned char *buf;
unsigned char *old_pos;
const ptrdiff_t sereal_header_len = dec->pos - dec->buf_start;
const STRLEN compressed_packet_len =
( dec->proto_version_and_flags & SRL_PROTOCOL_ENCODING_MASK ) == SRL_PROTOCOL_ENCODING_SNAPPY_INCREMENTAL
? (STRLEN)srl_read_varint_uv_length(aTHX_ dec, " while reading compressed packet size")
: (STRLEN)(dec->buf_end - dec->pos);
int decompress_ok;
int header_len;
/* all decl's above here, or we break C89 compilers */
dec->bytes_consumed= compressed_packet_len + (dec->pos - dec->buf_start);
header_len = csnappy_get_uncompressed_length(
(char *)dec->pos,
compressed_packet_len,
&dest_len
);
if (header_len == CSNAPPY_E_HEADER_BAD)
SRL_ERROR("Invalid Snappy header in Snappy-compressed Sereal packet");
/* Let perl clean this up. Yes, it's not the most efficient thing
* ever, but it's just one mortal per full decompression, so not
* a bottle-neck. */
buf_sv = sv_2mortal( newSV(sereal_header_len + dest_len + 1 ));
buf = (unsigned char *)SvPVX(buf_sv);
/* FIXME probably unnecessary to copy the Sereal header! */
Copy(dec->buf_start, buf, sereal_header_len, unsigned char);
old_pos = dec->pos;
dec->buf_start = buf;
dec->pos = buf + sereal_header_len;
dec->buf_end = dec->pos + dest_len;
dec->buf_len = dest_len + sereal_header_len;
decompress_ok = csnappy_decompress_noheader((char *)(old_pos + header_len),
compressed_packet_len - header_len,
(char *)dec->pos,
&dest_len);
if (expect_false( decompress_ok != 0 ))
{
SRL_ERRORf1("Snappy decompression of Sereal packet payload failed with error %i!", decompress_ok);
}
}
if (expect_true(!into)) {
into= sv_2mortal(newSV_type(SVt_NULL));
}
srl_read_single_value(aTHX_ dec, into);
/* assert(dec->pos == dec->buf_end); For now we disable this */
if (expect_false(SRL_DEC_HAVE_OPTION(dec, SRL_F_DECODER_NEEDS_FINALIZE))) {
srl_finalize_structure(aTHX_ dec);
}
/* If we aren't reading from a decompressed buffer we have to remember the number
* of bytes used for the user to query. */
if (dec->bytes_consumed == 0)
dec->bytes_consumed = dec->pos - dec->buf_start;
if (SRL_DEC_HAVE_OPTION(dec, SRL_F_DECODER_DESTRUCTIVE_INCREMENTAL)) {
STRLEN len;
char *pv= SvPV(src,len);
/* check the length here? do something different if the string is now exhausted? */
sv_chop(src, pv + dec->bytes_consumed);
}
srl_clear_decoder(aTHX_ dec);
return into;
}
static char *
my_sv_2pvbyte(pTHX_ register SV *sv, STRLEN *lp)
{
sv_utf8_downgrade(sv,0);
return SvPV(sv,*lp);
}
/*
* This is the main Load function.
* It takes a yaml stream and turns it into 0 or more Perl objects.
*/
void
Load(SV *yaml_sv)
{
dXSARGS;
perl_yaml_loader_t loader;
SV *node;
const unsigned char *yaml_str;
STRLEN yaml_len;
yaml_str = (const unsigned char *)SvPV_const(yaml_sv, yaml_len);
if (DO_UTF8(yaml_sv)) {
yaml_sv = sv_mortalcopy(yaml_sv);
if (!sv_utf8_downgrade(yaml_sv, TRUE))
croak("%s", "Wide character in YAML::XS::Load()");
yaml_str = (const unsigned char *)SvPV_const(yaml_sv, yaml_len);
}
sp = mark;
if (0 && (items || ax)) {} /* XXX Quiet the -Wall warnings for now. */
yaml_parser_initialize(&loader.parser);
loader.document = 0;
yaml_parser_set_input_string(
&loader.parser,
yaml_str,
yaml_len
);
/* Get the first event. Must be a STREAM_START */
if (!yaml_parser_parse(&loader.parser, &loader.event))
goto load_error;
if (loader.event.type != YAML_STREAM_START_EVENT)
croak("%sExpected STREAM_START_EVENT; Got: %d != %d",
ERRMSG,
loader.event.type,
YAML_STREAM_START_EVENT
);
loader.anchors = newHV();
sv_2mortal((SV *)loader.anchors);
/* Keep calling load_node until end of stream */
while (1) {
loader.document++;
/* We are through with the previous event - delete it! */
yaml_event_delete(&loader.event);
if (!yaml_parser_parse(&loader.parser, &loader.event))
goto load_error;
if (loader.event.type == YAML_STREAM_END_EVENT)
break;
node = load_node(&loader);
/* We are through with the previous event - delete it! */
yaml_event_delete(&loader.event);
hv_clear(loader.anchors);
if (! node) break;
XPUSHs(sv_2mortal(node));
if (!yaml_parser_parse(&loader.parser, &loader.event))
goto load_error;
if (loader.event.type != YAML_DOCUMENT_END_EVENT)
croak("%sExpected DOCUMENT_END_EVENT", ERRMSG);
}
/* Make sure the last event is a STREAM_END */
if (loader.event.type != YAML_STREAM_END_EVENT)
croak("%sExpected STREAM_END_EVENT; Got: %d != %d",
ERRMSG,
loader.event.type,
YAML_STREAM_END_EVENT
);
yaml_parser_delete(&loader.parser);
PUTBACK;
return;
load_error:
croak("%s", loader_error_msg(&loader, NULL));
}
EXTERN void
parse(pTHX_
PSTATE* p_state,
SV* chunk,
SV* self)
{
char *s, *beg, *end;
U32 utf8 = 0;
STRLEN len;
if (!chunk) {
/* eof */
char empty[1];
if (p_state->buf && SvOK(p_state->buf)) {
/* flush it */
s = SvPV(p_state->buf, len);
end = s + len;
utf8 = SvUTF8(p_state->buf);
assert(len);
while (s < end) {
if (p_state->literal_mode) {
if (strEQ(p_state->literal_mode, "plaintext") && !p_state->closing_plaintext)
break;
p_state->pending_end_tag = p_state->literal_mode;
p_state->literal_mode = 0;
s = parse_buf(aTHX_ p_state, s, end, utf8, self);
continue;
}
if (!p_state->strict_comment && !p_state->no_dash_dash_comment_end && *s == '<') {
p_state->no_dash_dash_comment_end = 1;
s = parse_buf(aTHX_ p_state, s, end, utf8, self);
continue;
}
if (!p_state->strict_comment && *s == '<') {
/* some kind of unterminated markup. Report rest as as comment */
token_pos_t token;
token.beg = s + 1;
token.end = end;
report_event(p_state, E_COMMENT, s, end, utf8, &token, 1, self);
s = end;
}
break;
}
if (s < end) {
/* report rest as text */
report_event(p_state, E_TEXT, s, end, utf8, 0, 0, self);
}
SvREFCNT_dec(p_state->buf);
p_state->buf = 0;
}
if (p_state->pend_text && SvOK(p_state->pend_text))
flush_pending_text(p_state, self);
if (p_state->ignoring_element) {
/* document not balanced */
SvREFCNT_dec(p_state->ignoring_element);
p_state->ignoring_element = 0;
}
report_event(p_state, E_END_DOCUMENT, empty, empty, 0, 0, 0, self);
/* reset state */
p_state->offset = 0;
if (p_state->line)
p_state->line = 1;
p_state->column = 0;
p_state->literal_mode = 0;
p_state->is_cdata = 0;
return;
}
#ifdef UNICODE_HTML_PARSER
if (p_state->utf8_mode)
sv_utf8_downgrade(chunk, 0);
#endif
if (p_state->buf && SvOK(p_state->buf)) {
sv_catsv(p_state->buf, chunk);
beg = SvPV(p_state->buf, len);
utf8 = SvUTF8(p_state->buf);
}
else {
beg = SvPV(chunk, len);
utf8 = SvUTF8(chunk);
if (p_state->offset == 0) {
report_event(p_state, E_START_DOCUMENT, beg, beg, 0, 0, 0, self);
/* Print warnings if we find unexpected Unicode BOM forms */
#ifdef UNICODE_HTML_PARSER
if (DOWARN &&
p_state->argspec_entity_decode &&
!p_state->utf8_mode && (
(!utf8 && len >= 3 && strnEQ(beg, "\xEF\xBB\xBF", 3)) ||
(utf8 && len >= 6 && strnEQ(beg, "\xC3\xAF\xC2\xBB\xC2\xBF", 6)) ||
(!utf8 && probably_utf8_chunk(aTHX_ beg, len))
)
)
{
warn("Parsing of undecoded UTF-8 will give garbage when decoding entities");
}
if (DOWARN && utf8 && len >= 2 && strnEQ(beg, "\xFF\xFE", 2)) {
warn("Parsing string decoded with wrong endianess");
}
#endif
if (DOWARN) {
if (!utf8 && len >= 4 &&
(strnEQ(beg, "\x00\x00\xFE\xFF", 4) ||
strnEQ(beg, "\xFE\xFF\x00\x00", 4))
)
{
warn("Parsing of undecoded UTF-32");
}
else if (!utf8 && len >= 2 &&
(strnEQ(beg, "\xFE\xFF", 2) || strnEQ(beg, "\xFF\xFE", 2))
)
{
warn("Parsing of undecoded UTF-16");
}
}
}
}
if (!len)
return; /* nothing to do */
end = beg + len;
s = parse_buf(aTHX_ p_state, beg, end, utf8, self);
if (s == end || p_state->eof) {
if (p_state->buf) {
SvOK_off(p_state->buf);
}
}
else {
/* need to keep rest in buffer */
if (p_state->buf) {
/* chop off some chars at the beginning */
if (SvOK(p_state->buf)) {
sv_chop(p_state->buf, s);
}
else {
sv_setpvn(p_state->buf, s, end - s);
if (utf8)
SvUTF8_on(p_state->buf);
else
SvUTF8_off(p_state->buf);
}
}
else {
p_state->buf = newSVpv(s, end - s);
if (utf8)
SvUTF8_on(p_state->buf);
}
}
return;
}
