Skip to content

Commit d731a7d

Browse files
jkeiserjkeiser
committed
Privatize structural_parser
1 parent 059468b commit d731a7d

1 file changed

Lines changed: 8 additions & 141 deletions

File tree

‎src/generic/stage2/structural_parser.h‎

Lines changed: 8 additions & 141 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,7 @@
44
// "simdjson/stage2.h" (this simplifies amalgation)
55

66
namespace stage2 {
7+
namespace { // Make everything here private
78

89
#ifdef SIMDJSON_USE_COMPUTED_GOTO
910
#define INIT_ADDRESSES() { &&array_begin, &&array_continue, &&error, &&finish, &&object_begin, &&object_continue }
@@ -81,10 +82,6 @@ struct structural_parser {
8182
parser{_parser},
8283
depth{0} {
8384
}
84-
// For streaming: pick up after the previous document
85-
really_inline structural_parser(dom_parser_implementation &_parser)
86-
: structural_parser(_parser, _parser.next_structural_index) {
87-
}
8885

8986
WARN_UNUSED really_inline bool start_scope(ret_address_t continue_state) {
9087
parser.containing_scope[depth].tape_index = parser.current_loc;
@@ -272,24 +269,6 @@ struct structural_parser {
272269
}
273270
}
274271

275-
// override to add streaming
276-
WARN_UNUSED really_inline error_code finish() {
277-
if ( structurals.past_end(parser.n_structural_indexes) ) {
278-
log_error("IMPOSSIBLE: past the end of the JSON!");
279-
return parser.error = TAPE_ERROR;
280-
}
281-
end_document();
282-
if (depth != 0) {
283-
log_error("Unclosed objects or arrays!");
284-
return parser.error = TAPE_ERROR;
285-
}
286-
if (parser.containing_scope[depth].tape_index != 0) {
287-
log_error("IMPOSSIBLE: root scope tape index did not start at 0!");
288-
return parser.error = TAPE_ERROR;
289-
}
290-
return SUCCESS;
291-
}
292-
293272
template<bool STREAMING>
294273
WARN_UNUSED really_inline error_code finish() {
295274
// Check if we're at (or past) the end
@@ -426,11 +405,12 @@ struct structural_parser {
426405
#undef FAIL_IF
427406
#define FAIL_IF(EXPR) { if (EXPR) { goto error; } }
428407

408+
template<bool STREAMING>
429409
WARN_UNUSED static error_code parse_structurals(dom_parser_implementation &dom_parser, dom::document &doc) noexcept {
430410
dom_parser.doc = &doc;
431411
static constexpr stage2::unified_machine_addresses addresses = INIT_ADDRESSES();
432-
stage2::structural_parser parser(dom_parser, 0);
433-
error_code result = parser.start<false>(dom_parser.len, addresses.finish);
412+
stage2::structural_parser parser(dom_parser, STREAMING ? dom_parser.next_structural_index : 0);
413+
error_code result = parser.start<STREAMING>(dom_parser.len, addresses.finish);
434414
if (result) { return result; }
435415

436416
//
@@ -545,140 +525,27 @@ WARN_UNUSED static error_code parse_structurals(dom_parser_implementation &dom_p
545525
}
546526

547527
finish:
548-
return parser.finish<false>();
528+
return parser.finish<STREAMING>();
549529

550530
error:
551531
return parser.error();
552532
}
553533

534+
} // namespace {}
554535
} // namespace stage2
555536

556537
/************
557538
* The JSON is parsed to a tape, see the accompanying tape.md file
558539
* for documentation.
559540
***********/
560541
WARN_UNUSED error_code dom_parser_implementation::stage2(dom::document &_doc) noexcept {
561-
return stage2::parse_structurals(*this, _doc);
542+
return stage2::parse_structurals<false>(*this, _doc);
562543
}
563544

564545
/************
565546
* The JSON is parsed to a tape, see the accompanying tape.md file
566547
* for documentation.
567548
***********/
568549
WARN_UNUSED error_code dom_parser_implementation::stage2_next(dom::document &_doc) noexcept {
569-
this->doc = &_doc;
570-
static constexpr stage2::unified_machine_addresses addresses = INIT_ADDRESSES();
571-
stage2::structural_parser parser(*this, next_structural_index);
572-
error_code result = parser.start<true>(len, addresses.finish);
573-
if (result) { return result; }
574-
//
575-
// Read first value
576-
//
577-
switch (parser.structurals.current_char()) {
578-
case '{':
579-
FAIL_IF( parser.start_object(addresses.finish) );
580-
goto object_begin;
581-
case '[':
582-
FAIL_IF( parser.start_array(addresses.finish) );
583-
goto array_begin;
584-
case '"':
585-
FAIL_IF( parser.parse_string() );
586-
goto finish;
587-
case 't': case 'f': case 'n':
588-
FAIL_IF( parser.parse_single_atom() );
589-
goto finish;
590-
case '0': case '1': case '2': case '3': case '4':
591-
case '5': case '6': case '7': case '8': case '9':
592-
FAIL_IF(
593-
parser.structurals.with_space_terminated_copy([&](const uint8_t *copy, size_t idx) {
594-
return parser.parse_number(&copy[idx], false);
595-
})
596-
);
597-
goto finish;
598-
case '-':
599-
FAIL_IF(
600-
parser.structurals.with_space_terminated_copy([&](const uint8_t *copy, size_t idx) {
601-
return parser.parse_number(&copy[idx], true);
602-
})
603-
);
604-
goto finish;
605-
default:
606-
parser.log_error("Document starts with a non-value character");
607-
goto error;
608-
}
609-
610-
//
611-
// Object parser parsers
612-
//
613-
object_begin:
614-
switch (parser.advance_char()) {
615-
case '"': {
616-
FAIL_IF( parser.parse_string(true) );
617-
goto object_key_parser;
618-
}
619-
case '}':
620-
parser.end_object();
621-
goto scope_end;
622-
default:
623-
parser.log_error("Object does not start with a key");
624-
goto error;
625-
}
626-
627-
object_key_parser:
628-
if (parser.advance_char() != ':' ) { parser.log_error("Missing colon after key in object"); goto error; }
629-
parser.increment_count();
630-
parser.advance_char();
631-
GOTO( parser.parse_value(addresses, addresses.object_continue) );
632-
633-
object_continue:
634-
switch (parser.advance_char()) {
635-
case ',':
636-
if (parser.advance_char() != '"' ) { parser.log_error("Key string missing at beginning of field in object"); goto error; }
637-
FAIL_IF( parser.parse_string(true) );
638-
goto object_key_parser;
639-
case '}':
640-
parser.end_object();
641-
goto scope_end;
642-
default:
643-
parser.log_error("No comma between object fields");
644-
goto error;
645-
}
646-
647-
scope_end:
648-
CONTINUE( parser.parser.ret_address[parser.depth] );
649-
650-
//
651-
// Array parser parsers
652-
//
653-
array_begin:
654-
if (parser.advance_char() == ']') {
655-
parser.end_array();
656-
goto scope_end;
657-
}
658-
parser.increment_count();
659-
660-
main_array_switch:
661-
/* we call update char on all paths in, so we can peek at parser.c on the
662-
* on paths that can accept a close square brace (post-, and at start) */
663-
GOTO( parser.parse_value(addresses, addresses.array_continue) );
664-
665-
array_continue:
666-
switch (parser.advance_char()) {
667-
case ',':
668-
parser.increment_count();
669-
parser.advance_char();
670-
goto main_array_switch;
671-
case ']':
672-
parser.end_array();
673-
goto scope_end;
674-
default:
675-
parser.log_error("Missing comma between array values");
676-
goto error;
677-
}
678-
679-
finish:
680-
return parser.finish<true>();
681-
682-
error:
683-
return parser.error();
550+
return stage2::parse_structurals<true>(*this, _doc);
684551
}

0 commit comments

Comments
 (0)