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

Advance immediately upon evaluating a character

This commit is contained in:
John Keiser 2020-08-03 13:26:56 -07:00
parent 099bb1afef
commit e6762f9b48
3 changed files with 41 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/generic/stage2/logger.h
View File

@ -7,10 +7,10 @@ namespace logger {
static constexpr const char * DASHES = "----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------";
static constexpr const bool LOG_ENABLED = false;
static constexpr const int LOG_EVENT_LEN = 30;
static constexpr const int LOG_BUFFER_LEN = 20;
static constexpr const int LOG_DETAIL_LEN = 50;
static constexpr const int LOG_INDEX_LEN = 10;
static constexpr const int LOG_EVENT_LEN = 20;
static constexpr const int LOG_BUFFER_LEN = 10;
static constexpr const int LOG_SMALL_BUFFER_LEN = 10;
static constexpr const int LOG_INDEX_LEN = 5;
static int log_depth; // Not threadsafe. Log only.
@ -28,8 +28,8 @@ namespace logger {
if (LOG_ENABLED) {
log_depth = 0;
printf("\n");
printf("| %-*s | %-*s | %*s | %*s | %*s | %-*s | %-*s | %-*s |\n", LOG_EVENT_LEN, "Event", LOG_BUFFER_LEN, "Buffer", 4, "Curr", 4, "Next", 5, "Next#", 5, "Tape#", LOG_DETAIL_LEN, "Detail", LOG_INDEX_LEN, "index");
printf("|%.*s|%.*s|%.*s|%.*s|%.*s|%.*s|%.*s|%.*s|\n", LOG_EVENT_LEN+2, DASHES, LOG_BUFFER_LEN+2, DASHES, 4+2, DASHES, 4+2, DASHES, 5+2, DASHES, 5+2, DASHES, LOG_DETAIL_LEN+2, DASHES, LOG_INDEX_LEN+2, DASHES);
printf("| %-*s | %-*s | %-*s | %-*s | %-*s | Detail |\n", LOG_EVENT_LEN, "Event", LOG_BUFFER_LEN, "Buffer", LOG_SMALL_BUFFER_LEN, "Next", 5, "Next#", 5, "Tape#");
printf("|%.*s|%.*s|%.*s|%.*s|%.*s|--------|\n", LOG_EVENT_LEN+2, DASHES, LOG_BUFFER_LEN+2, DASHES, LOG_SMALL_BUFFER_LEN+2, DASHES, 5+2, DASHES, 5+2, DASHES);
}
}
@ -44,22 +44,35 @@ namespace logger {
static really_inline void log_line(S &structurals, const char *title_prefix, const char *title, const char *detail) {
if (LOG_ENABLED) {
printf("| %*s%s%-*s ", log_depth*2, "", title_prefix, LOG_EVENT_LEN - log_depth*2 - int(strlen(title_prefix)), title);
auto current_index = structurals.at_beginning() ? nullptr : structurals.next_structural-1;
auto next_index = structurals.next_structural;
auto current = current_index ? &structurals.buf[*current_index] : (const uint8_t*)" ";
auto next = &structurals.buf[*next_index];
{
// Print the next N characters in the buffer.
printf("| ");
// Otherwise, print the characters starting from the buffer position.
// Print spaces for unprintable or newline characters.
for (int i=0;i<LOG_BUFFER_LEN;i++) {
printf("%c", printable_char(structurals.current()[i]));
printf("%c", printable_char(current[i]));
}
printf(" ");
// Print the next N characters in the buffer.
printf("| ");
// Otherwise, print the characters starting from the buffer position.
// Print spaces for unprintable or newline characters.
for (int i=0;i<LOG_SMALL_BUFFER_LEN;i++) {
printf("%c", printable_char(next[i]));
}
printf(" ");
}
printf("| %c ", printable_char(structurals.current_char()));
printf("| %c ", printable_char(structurals.peek_next_char()));
printf("| %5u ", structurals.parser.structural_indexes[*(structurals.current_structural+1)]);
printf("| %5u ", structurals.next_tape_index());
printf("| %-*s ", LOG_DETAIL_LEN, detail);
printf("| %*u ", LOG_INDEX_LEN, *structurals.current_structural);
if (current_index) {
printf("| %*u ", LOG_INDEX_LEN, *current_index);
} else {
printf("| %-*s ", LOG_INDEX_LEN, "");
}
printf("| %*u ", LOG_INDEX_LEN, structurals.next_tape_index());
printf("| %-s ", detail);
printf("|\n");
}
}

25
src/generic/stage2/structural_iterator.h
View File

@ -5,44 +5,45 @@ namespace stage2 {
class structural_iterator {
public:
const uint8_t* const buf;
uint32_t *current_structural;
uint32_t *next_structural;
dom_parser_implementation &parser;
// Start a structural
really_inline structural_iterator(dom_parser_implementation &_parser, size_t start_structural_index)
: buf{_parser.buf},
current_structural{&_parser.structural_indexes[start_structural_index]},
next_structural{&_parser.structural_indexes[start_structural_index]},
parser{_parser} {
}
// Get the buffer position of the current structural character
really_inline const uint8_t* current() {
return &buf[*current_structural];
return &buf[*(next_structural-1)];
}
// Get the current structural character
really_inline char current_char() {
return buf[*current_structural];
return buf[*(next_structural-1)];
}
// Get the next structural character without advancing
really_inline char peek_next_char() {
return buf[*(current_structural+1)];
return buf[*next_structural];
}
really_inline const uint8_t* peek() {
return &buf[*next_structural];
}
really_inline const uint8_t* advance() {
current_structural++;
return &buf[*current_structural];
return &buf[*(next_structural++)];
}
really_inline char advance_char() {
current_structural++;
return buf[*current_structural];
return buf[*(next_structural++)];
}
really_inline size_t remaining_len() {
return parser.len - *current_structural;
return parser.len - *(next_structural-1);
}
really_inline bool at_end() {
return current_structural == &parser.structural_indexes[parser.n_structural_indexes];
return next_structural == &parser.structural_indexes[parser.n_structural_indexes];
}
really_inline bool at_beginning() {
return current_structural == parser.structural_indexes.get();
return next_structural == parser.structural_indexes.get();
}
};

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

@ -238,7 +238,7 @@ struct structural_parser : structural_iterator {
WARN_UNUSED really_inline error_code finish() {
end_document();
parser.next_structural_index = uint32_t(current_structural + 1 - &parser.structural_indexes[0]);
parser.next_structural_index = uint32_t(next_structural - &parser.structural_indexes[0]);
if (depth != 0) {
log_error("Unclosed objects or arrays!");
@ -279,7 +279,7 @@ WARN_UNUSED static really_inline error_code parse_structurals(dom_parser_impleme
// Read first value
//
{
switch (parser.current_char()) {
switch (parser.advance_char()) {
case '{': {
if (parser.empty_object()) { goto document_end; }
SIMDJSON_TRY( parser.start_object() );