jasder
/
simdjson
1
0
Fork 0
Code Issues Pull Requests Releases 2 Wiki Activity

Stop caching current structural, keep current index around instead of 

next
This commit is contained in:
John Keiser 2020-06-06 13:40:34 -07:00
parent 5f00b37e21
commit d178e089a6
3 changed files with 30 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/generic/stage2/logger.h
View File

@ -62,9 +62,9 @@ namespace logger {
}
printf("| %c ", printable_char(structurals.at_beginning() ? ' ' : structurals.current_char()));
printf("| %c ", printable_char(structurals.peek_char()));
printf("| %5u ", structurals.parser.structural_indexes[*structurals.next_structural]);
printf("| %5u ", structurals.parser.structural_indexes[*(structurals.current_structural+1)]);
printf("| %-*s ", LOG_DETAIL_LEN, detail);
printf("| %*u ", LOG_INDEX_LEN, *(structurals.next_structural-1));
printf("| %*u ", LOG_INDEX_LEN, *structurals.current_structural);
printf("|\n");
}
}

47
src/generic/stage2/structural_iterator.h
View File

@ -3,34 +3,33 @@ namespace stage2 {
class structural_iterator {
public:
const uint8_t* const buf;
uint32_t *next_structural;
uint8_t c{0}; // used to track the (structural) character we are looking at
uint32_t *current_structural;
dom_parser_implementation &parser;
really_inline structural_iterator(dom_parser_implementation &_parser, size_t next_structural_index)
// Start a structural
really_inline structural_iterator(dom_parser_implementation &_parser, size_t start_structural_index)
: buf{_parser.buf},
next_structural{&_parser.structural_indexes[next_structural_index]},
current_structural{&_parser.structural_indexes[start_structural_index]},
parser{_parser} {
}
really_inline char advance_char() {
c = buf[*next_structural];
next_structural++;
return c;
}
really_inline char current_char() {
return c;
}
really_inline char peek_char() {
return buf[*next_structural];
}
// Get the buffer position of the current structural character
really_inline const uint8_t* current() {
return &buf[current_structural_index()];
return &buf[*current_structural];
}
// Get the current structural character
really_inline char current_char() {
return buf[*current_structural];
}
// Get the next structural character without advancing
really_inline char peek_char() {
return buf[*(current_structural+1)];
}
really_inline char advance_char() {
current_structural++;
return buf[*current_structural];
}
really_inline size_t remaining_len() {
return parser.len - current_structural_index();
}
really_inline uint32_t current_structural_index() {
return *(next_structural-1);
return parser.len - *current_structural;
}
template<typename F>
really_inline bool with_space_terminated_copy(const F& f) {
@ -53,18 +52,18 @@ public:
}
memcpy(copy, buf, parser.len);
memset(copy + parser.len, ' ', SIMDJSON_PADDING);
bool result = f(reinterpret_cast<const uint8_t*>(copy), current_structural_index());
bool result = f(reinterpret_cast<const uint8_t*>(copy), *current_structural);
free(copy);
return result;
}
really_inline bool past_end(uint32_t n_structural_indexes) {
return next_structural > &parser.structural_indexes[n_structural_indexes];
return current_structural >= &parser.structural_indexes[n_structural_indexes];
}
really_inline bool at_end(uint32_t n_structural_indexes) {
return next_structural == &parser.structural_indexes[n_structural_indexes];
return current_structural == &parser.structural_indexes[n_structural_indexes];
}
really_inline bool at_beginning() {
return next_structural == &parser.structural_indexes[0];
return current_structural == parser.structural_indexes.get();
}
};

11
src/generic/stage2/structural_parser.h
View File

@ -72,11 +72,12 @@ struct number_writer {
struct structural_parser : structural_iterator {
/** Next write location in the string buf for stage 2 parsing */
uint8_t *current_string_buf_loc{};
/** Current depth (nested objects and arrays) */
uint32_t depth;
// For non-streaming, to pass an explicit 0 as next_structural, which enables optimizations
really_inline structural_parser(dom_parser_implementation &_parser, uint32_t _next_structural)
: structural_iterator(_parser, _next_structural),
really_inline structural_parser(dom_parser_implementation &_parser, uint32_t start_structural_index)
: structural_iterator(_parser, start_structural_index),
depth{0} {
}
@ -232,7 +233,7 @@ struct structural_parser : structural_iterator {
WARN_UNUSED really_inline error_code finish() {
end_document();
parser.next_structural_index = uint32_t(next_structural - &parser.structural_indexes[0]);
parser.next_structural_index = uint32_t(current_structural + 1 - &parser.structural_indexes[0]);
if (depth != 0) {
log_error("Unclosed objects or arrays!");
@ -283,6 +284,7 @@ struct structural_parser : structural_iterator {
}
really_inline void init() {
log_start();
current_string_buf_loc = parser.doc->string_buf.get();
parser.current_loc = 0;
parser.error = UNINITIALIZED;
@ -294,10 +296,7 @@ struct structural_parser : structural_iterator {
return parser.error = EMPTY;
}
log_start();
init();
// Advance to the first character as soon as possible
advance_char();
// Push the root scope (there is always at least one scope)
if (start_document(finish_state)) {
return parser.error = DEPTH_ERROR;