Skip to content

Commit e744c2b

Browse files
rahim-kanjirahim-kanji
committed
Optimize transaction command parsing to avoid unnecessary tokenization
Previously, the parser always tokenized the full command, even when we only needed to check whether it was a transaction command. Now, it first extracts the first word to determine relevance and performs full tokenization only when necessary.
1 parent 7c665b9 commit e744c2b

File tree

6 files changed

+149
-49
lines changed

6 files changed

+149
-49
lines changed

‎include/PgSQL_ExplicitTxnStateMgr.h‎

Lines changed: 14 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -56,6 +56,9 @@ struct TxnCmd {
5656
*/
5757
class PgSQL_TxnCmdParser {
5858
public:
59+
PgSQL_TxnCmdParser() noexcept { tokens.reserve(16); }
60+
~PgSQL_TxnCmdParser() noexcept = default;
61+
5962
TxnCmd parse(std::string_view input, bool in_transaction_mode) noexcept;
6063

6164
private:
@@ -67,14 +70,20 @@ class PgSQL_TxnCmdParser {
6770
TxnCmd parse_start(size_t& pos) noexcept;
6871

6972
// Helpers
70-
static std::string to_lower(std::string_view s) noexcept {
71-
std::string s_copy(s);
72-
std::transform(s_copy.begin(), s_copy.end(), s_copy.begin(), ::tolower);
73-
return s_copy;
73+
inline static bool iequals(std::string_view a, std::string_view b) noexcept {
74+
if (a.size() != b.size()) return false;
75+
for (size_t i = 0; i < a.size(); ++i) {
76+
char ca = a[i];
77+
char cb = b[i];
78+
if (ca >= 'A' && ca <= 'Z') ca += 32;
79+
if (cb >= 'A' && cb <= 'Z') cb += 32;
80+
if (ca != cb) return false;
81+
}
82+
return true;
7483
}
7584

7685
inline static bool contains(std::vector<std::string_view>&& list, std::string_view value) noexcept {
77-
for (const auto& item : list) if (item == value) return true;
86+
for (const auto& item : list) if (iequals(item, value)) return true;
7887
return false;
7988
}
8089
};

‎lib/PgSQL_ExplicitTxnStateMgr.cpp‎

Lines changed: 117 additions & 26 deletions
Original file line numberDiff line numberDiff line change
@@ -327,23 +327,105 @@ bool PgSQL_ExplicitTxnStateMgr::handle_transaction(std::string_view input) {
327327
return true;
328328
}
329329

330-
331330
TxnCmd PgSQL_TxnCmdParser::parse(std::string_view input, bool in_transaction_mode) noexcept {
332-
tokens.clear();
333331
TxnCmd cmd;
334-
bool in_quote = false;
332+
333+
if (input.empty()) return cmd;
334+
335+
// Extract first word without full tokenization
335336
size_t start = 0;
337+
size_t end = 0;
338+
339+
while (start < input.size() && fast_isspace(input[start])) {
340+
start++;
341+
}
342+
343+
if (start >= input.size()) return cmd;
344+
345+
// Find end of first word
346+
end = start;
347+
bool in_quote = false;
336348
char quote_char = 0;
337349

338-
// Tokenize with quote handling
339-
for (size_t i = 0; i <= input.size(); ++i) {
350+
while (end < input.size()) {
351+
char c = input[end];
352+
353+
if (!in_quote && (c == '"' || c == '\'')) {
354+
// If we hit a quote at the start, this isn't a transaction command
355+
return cmd;
356+
}
357+
358+
if (fast_isspace(c) || c == ';') {
359+
break;
360+
}
361+
362+
end++;
363+
}
364+
365+
std::string_view first_word = input.substr(start, end - start);
366+
367+
// Check if this is a transaction command we care about
368+
TxnCmd::Type cmd_type = TxnCmd::UNKNOWN;
369+
370+
if (in_transaction_mode) {
371+
if (iequals(first_word, "begin")) {
372+
cmd.type = TxnCmd::BEGIN;
373+
return cmd;
374+
}
375+
376+
if (iequals(first_word, "start")) {
377+
cmd_type = TxnCmd::BEGIN;
378+
} else if (iequals(first_word, "savepoint")) {
379+
cmd_type = TxnCmd::SAVEPOINT;
380+
} else if (iequals(first_word, "release")) {
381+
cmd_type = TxnCmd::RELEASE;
382+
} else if (iequals(first_word, "rollback")) {
383+
cmd_type = TxnCmd::ROLLBACK;
384+
}
385+
} else {
386+
387+
if (iequals(first_word, "commit") || iequals(first_word, "end")) {
388+
cmd.type = TxnCmd::COMMIT;
389+
return cmd;
390+
}
391+
392+
if (iequals(first_word, "abort")) {
393+
cmd.type = TxnCmd::ROLLBACK;
394+
return cmd;
395+
}
396+
397+
if (iequals(first_word, "rollback")) {
398+
cmd_type = TxnCmd::ROLLBACK;
399+
}
400+
}
401+
402+
// If not a transaction command, return early
403+
if (cmd_type == TxnCmd::UNKNOWN) {
404+
return cmd;
405+
}
406+
407+
// Continue tokenization from where we left off
408+
tokens.clear();
409+
410+
// Continue tokenizing the rest of the input
411+
in_quote = false;
412+
quote_char = 0;
413+
start = end; // Continue from after the first word
414+
415+
while (start < input.size() && fast_isspace(input[start])) {
416+
start++;
417+
}
418+
419+
// Tokenize the remaining input
420+
for (size_t i = start; i <= input.size(); ++i) {
340421
const bool at_end = i == input.size();
341422
const char c = at_end ? 0 : input[i];
342423

343424
if (in_quote) {
344425
if (c == quote_char || at_end) {
345426
tokens.emplace_back(input.substr(start + 1, i - start - 1));
346427
in_quote = false;
428+
start = i + 1;
347429
}
348430
continue;
349431
}
@@ -353,41 +435,50 @@ TxnCmd PgSQL_TxnCmdParser::parse(std::string_view input, bool in_transaction_mod
353435
quote_char = c;
354436
start = i;
355437
}
356-
else if (isspace(c) || c == ';' || at_end) {
438+
else if (fast_isspace(c) || c == ';' || at_end) {
357439
if (start < i) tokens.emplace_back(input.substr(start, i - start));
358440
start = i + 1;
359441
}
360442
}
361443

362-
if (tokens.empty()) return cmd;
363-
364444
size_t pos = 0;
365-
const std::string first = to_lower(tokens[pos++]);
366-
367-
if (in_transaction_mode == true) {
368-
if (first == "begin") cmd.type = TxnCmd::BEGIN;
369-
else if (first == "start") cmd = parse_start(pos);
370-
else if (first == "savepoint") cmd = parse_savepoint(pos);
371-
else if (first == "release") cmd = parse_release(pos);
372-
else if (first == "rollback") cmd = parse_rollback(pos);
445+
446+
if (in_transaction_mode) {
447+
448+
switch (cmd_type) {
449+
case TxnCmd::BEGIN:
450+
cmd = parse_start(pos);
451+
break;
452+
case TxnCmd::SAVEPOINT:
453+
cmd = parse_savepoint(pos);
454+
break;
455+
case TxnCmd::RELEASE:
456+
cmd = parse_release(pos);
457+
break;
458+
case TxnCmd::ROLLBACK:
459+
cmd = parse_rollback(pos);
460+
break;
461+
default:
462+
break;
463+
}
373464
} else {
374-
if (first == "commit" || first == "end") cmd.type = TxnCmd::COMMIT;
375-
else if (first == "abort") cmd.type = TxnCmd::ROLLBACK;
376-
else if (first == "rollback") cmd = parse_rollback(pos);
465+
if (cmd_type == TxnCmd::ROLLBACK)
466+
cmd = parse_rollback(pos);
377467
}
468+
378469
return cmd;
379470
}
380471

381472
TxnCmd PgSQL_TxnCmdParser::parse_rollback(size_t& pos) noexcept {
382473
TxnCmd cmd{ TxnCmd::ROLLBACK };
383-
while (pos < tokens.size() && contains({ "work", "transaction" }, to_lower(tokens[pos]))) pos++;
474+
while (pos < tokens.size() && contains({ "work", "transaction" }, tokens[pos])) pos++;
384475

385-
if (pos < tokens.size() && to_lower(tokens[pos]) == "to") {
476+
if (pos < tokens.size() && iequals(tokens[pos], "to")) {
386477
cmd.type = TxnCmd::ROLLBACK_TO;
387-
if (++pos < tokens.size() && to_lower(tokens[pos]) == "savepoint") pos++;
478+
if (++pos < tokens.size() && iequals(tokens[pos], "savepoint")) pos++;
388479
if (pos < tokens.size()) cmd.savepoint = tokens[pos++];
389-
} else if (pos < tokens.size() && to_lower(tokens[pos]) == "and") {
390-
if (++pos < tokens.size() && to_lower(tokens[pos]) == "chain") {
480+
} else if (pos < tokens.size() && iequals(tokens[pos], "and")) {
481+
if (++pos < tokens.size() && iequals(tokens[pos], "chain")) {
391482
cmd.type = TxnCmd::ROLLBACK_AND_CHAIN;
392483
pos++;
393484
}
@@ -403,14 +494,14 @@ TxnCmd PgSQL_TxnCmdParser::parse_savepoint(size_t& pos) noexcept {
403494

404495
TxnCmd PgSQL_TxnCmdParser::parse_release(size_t& pos) noexcept {
405496
TxnCmd cmd{ TxnCmd::RELEASE };
406-
if (pos < tokens.size() && to_lower(tokens[pos]) == "savepoint") pos++;
497+
if (pos < tokens.size() && iequals(tokens[pos], "savepoint")) pos++;
407498
if (pos < tokens.size()) cmd.savepoint = tokens[pos++];
408499
return cmd;
409500
}
410501

411502
TxnCmd PgSQL_TxnCmdParser::parse_start(size_t& pos) noexcept {
412503
TxnCmd cmd{ TxnCmd::UNKNOWN };
413-
if (pos < tokens.size() && to_lower(tokens[pos]) == "transaction") {
504+
if (pos < tokens.size() && iequals(tokens[pos], "transaction")) {
414505
cmd.type = TxnCmd::BEGIN;
415506
pos++;
416507
}

‎lib/PgSQL_Protocol.cpp‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -741,7 +741,7 @@ std::vector<std::pair<std::string, std::string>> PgSQL_Protocol::parse_options(c
741741

742742
while (pos < input.size()) {
743743
// Skip leading spaces
744-
while (pos < input.size() && std::isspace(input[pos])) {
744+
while (pos < input.size() && fast_isspace(input[pos])) {
745745
++pos;
746746
}
747747

@@ -751,7 +751,7 @@ std::vector<std::pair<std::string, std::string>> PgSQL_Protocol::parse_options(c
751751
pos += 2; // Skip "-c", "--"
752752
}
753753

754-
while (pos < input.size() && std::isspace(input[pos])) {
754+
while (pos < input.size() && fast_isspace(input[pos])) {
755755
++pos;
756756
}
757757

@@ -772,7 +772,7 @@ std::vector<std::pair<std::string, std::string>> PgSQL_Protocol::parse_options(c
772772
bool last_was_escape = false;
773773
while (pos < input.size()) {
774774
char c = input[pos];
775-
if (std::isspace(c) && !last_was_escape) {
775+
if (fast_isspace(c) && !last_was_escape) {
776776
break;
777777
}
778778
if (c == '\\' && !last_was_escape) {

‎lib/PgSQL_Session.cpp‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -6557,7 +6557,7 @@ std::vector<std::string> PgSQL_DateStyle_Util::split_datestyle(std::string_view
65576557
int* lastNonSpace = (currentToken == 1) ? &lastNonSpace1 : &lastNonSpace2;
65586558

65596559
// Cache is-space check.
6560-
bool is_space = std::isspace(static_cast<unsigned char>(c));
6560+
bool is_space = fast_isspace(static_cast<unsigned char>(c));
65616561
// Skip leading whitespace for a new token.
65626562
if (currentStr->empty() && is_space) {
65636563
continue;

‎lib/PgSQL_Variables_Validator.cpp‎

Lines changed: 13 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -183,7 +183,7 @@ bool pgsql_variable_validate_maintenance_work_mem(const char* value, const param
183183
if (transformed_value) *transformed_value = nullptr;
184184

185185
// Skip leading whitespace
186-
while (isspace((unsigned char)*p)) p++;
186+
while (fast_isspace((unsigned char)*p)) p++;
187187

188188
// Parse numeric part
189189
num = strtoll(p, &endptr, 10);
@@ -196,11 +196,11 @@ bool pgsql_variable_validate_maintenance_work_mem(const char* value, const param
196196
p = endptr;
197197

198198
// Skip whitespace after number
199-
while (isspace((unsigned char)*p)) p++;
199+
while (fast_isspace((unsigned char)*p)) p++;
200200

201201
// Parse unit
202202
if (*p != '\0') {
203-
char tmp_unit = tolower(*p);
203+
char tmp_unit = ::tolower(*p);
204204
switch (tmp_unit) {
205205
case 'k':
206206
case 'm':
@@ -210,15 +210,15 @@ bool pgsql_variable_validate_maintenance_work_mem(const char* value, const param
210210
unit = toupper(*p++);
211211
has_unit = true;
212212
// Check optional 'b'/'B'
213-
if (tolower(*p) == 'b') p++;
213+
if (::tolower(*p) == 'b') p++;
214214
break;
215215
default:
216216
return false;
217217
}
218218
}
219219

220220
// Skip trailing whitespace
221-
while (isspace((unsigned char)*p)) p++;
221+
while (fast_isspace((unsigned char)*p)) p++;
222222

223223
// Validate entire string consumed
224224
if (*p != '\0') return false;
@@ -241,7 +241,7 @@ bool pgsql_variable_validate_maintenance_work_mem_v2(const char* value, const pa
241241
const char* input = value;
242242

243243
/* Trim leading whitespace */
244-
while (isspace((unsigned char)*input)) input++;
244+
while (fast_isspace((unsigned char)*input)) input++;
245245

246246
/* Parse numeric part */
247247
uint64_t number;
@@ -256,7 +256,7 @@ bool pgsql_variable_validate_maintenance_work_mem_v2(const char* value, const pa
256256
//num_len = endptr - input;
257257

258258
// Skip whitespace after number
259-
while (isspace((unsigned char)*endptr)) endptr++;
259+
while (fast_isspace((unsigned char)*endptr)) endptr++;
260260

261261
/* Parse unit part */
262262
const char* unit_ptr = endptr;
@@ -273,7 +273,7 @@ bool pgsql_variable_validate_maintenance_work_mem_v2(const char* value, const pa
273273
/* Convert unit to lowercase for validation */
274274
char u[3] = { 0 };
275275
for (int i = 0; i < 2 && unit_ptr[i]; i++)
276-
u[i] = tolower((unsigned char)unit_ptr[i]);
276+
u[i] = ::tolower((unsigned char)unit_ptr[i]);
277277

278278
/* Validate unit and set multiplier */
279279
if (unit_len == 1 && u[0] == 'b') {
@@ -332,7 +332,7 @@ bool pgsql_variable_validate_maintenance_work_mem_v3(const char* value, const pa
332332
(void)session;
333333

334334
// Trim leading whitespace
335-
while (isspace((unsigned char)*value)) value++;
335+
while (fast_isspace((unsigned char)*value)) value++;
336336

337337
char* endptr;
338338
const char* num_start = value;
@@ -371,7 +371,7 @@ bool pgsql_variable_validate_maintenance_work_mem_v3(const char* value, const pa
371371
// Convert unit to lowercase for validation
372372
char u[3] = { 0 };
373373
for (int i = 0; i < 2 && unit_ptr[i]; i++)
374-
u[i] = tolower((unsigned char)unit_ptr[i]);
374+
u[i] = ::tolower((unsigned char)unit_ptr[i]);
375375

376376
// Validate units and set multipliers
377377
if (unit_len == 1 && u[0] == 'b') {
@@ -471,7 +471,7 @@ bool pgsql_variable_validate_search_path(const char* value, const params_t* para
471471

472472
while (*token && result) {
473473
/* skip leading whitespace */
474-
while (*token && isspace((unsigned char)*token)) token++;
474+
while (*token && fast_isspace((unsigned char)*token)) token++;
475475
if (*token == '\0') break;
476476

477477
const char* part_start = token;
@@ -508,7 +508,7 @@ bool pgsql_variable_validate_search_path(const char* value, const params_t* para
508508
}
509509
} else {
510510
// unquoted identifier or $user
511-
while (*token && *token != ',' && !isspace(*token)) token++;
511+
while (*token && *token != ',' && !fast_isspace(*token)) token++;
512512
part_len = (size_t)(token - part_start);
513513
if (part_len == 0 || part_len > 63) {
514514
result = false;
@@ -543,7 +543,7 @@ bool pgsql_variable_validate_search_path(const char* value, const params_t* para
543543
normalized[norm_pos] = '\0';
544544

545545
// skip whitespace after part
546-
while (*token && isspace(*token)) token++;
546+
while (*token && fast_isspace(*token)) token++;
547547

548548
// expect comma or end
549549
if (*token == ',') {

‎lib/ProxySQL_Config.cpp‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -92,7 +92,7 @@ int ProxySQL_Config::Read_Global_Variables_from_configfile(const char *prefix) {
9292
char *query=(char *)malloc(strlen(q)+strlen(prefix)+strlen(n)+strlen(value_string.c_str()));
9393
sprintf(query,q, prefix, n, value_string.c_str());
9494
//fprintf(stderr, "%s\n", query);
95-
admindb->execute(query);
95+
admindb->execute(query);
9696
free(query);
9797
}
9898
admindb->execute("PRAGMA foreign_keys = ON");

0 commit comments

Comments
 (0)