Skip to content

Do not reallocate whole input string when matching next token #119

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Jun 11, 2024
Merged

Do not reallocate whole input string when matching next token #119

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
4c9a805
Refactor Tokenizer::createNextToken() to accept string and offset
mvorisek Jun 11, 2024
15887a8
Change "^" to "\G" in regexes to match offset start
mvorisek Jun 11, 2024
b3b1c95
Refactor Tokenizer::getNextQuotedString() to accept string and offset
mvorisek Jun 11, 2024
459f48f
Refactor all "strtoupper($stringSlow)" code
mvorisek May 31, 2024
af658af
Refactor all remaining "$stringSlow" code
mvorisek Jun 11, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
83 changes: 46 additions & 37 deletions src/Tokenizer.php
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -800,17 +800,17 @@ public function tokenize(string $string): Cursor
{
$tokens = [];

$token = null;
$upper = strtoupper($string);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

given that this optimizations in this method are begin kept, do you have a benchmark to understand the impact of all the other optimizations you are suggesting?

Copy link
Contributor Author

@mvorisek mvorisek Jun 4, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Previously the uppercased matching was already done... and done very poorly as the string was upper cased again and again. The overall speedup of this PR is more than -90% runtime (10x faster) - see https://github.com/doctrine/sql-formatter/actions/runs/9366932237/job/25785513399?pr=119#step:7:25 now and before. This is discussed in #119 (comment) and the PR description.

$offset = 0;
$token = null;

// Keep processing the string until it is empty
while ($string !== '') {
while ($offset < strlen($string)) {
// Get the next token and the token type
$token = $this->createNextToken($string, $token);
$token = $this->createNextToken($string, $upper, $offset, $token);
$offset += strlen($token->value());

$tokens[] = $token;

// Advance the string
$string = substr($string, strlen($token->value()));
}

return new Cursor($tokens);
Expand All @@ -822,30 +822,31 @@ public function tokenize(string $string): Cursor
* are all their own tokens.
*
* @param string $string The SQL string
* @param string $upper The SQL string in upper case
* @param Token|null $previous The result of the previous createNextToken() call
*
* @return Token An associative array containing the type and value of the token.
*/
private function createNextToken(string $string, Token|null $previous = null): Token
private function createNextToken(string $string, string $upper, int $offset, Token|null $previous = null): Token
{
$matches = [];
// Whitespace
if (preg_match('/^\s+/', $string, $matches)) {
if (preg_match('/\G\s+/', $string, $matches, 0, $offset)) {
return new Token(Token::TOKEN_TYPE_WHITESPACE, $matches[0]);
}

// Comment
if (
$string[0] === '#' ||
(isset($string[1]) && ($string[0] === '-' && $string[1] === '-') ||
(isset($string[1]) && $string[0] === '/' && $string[1] === '*'))
$string[$offset] === '#' ||
(isset($string[$offset + 1]) && ($string[$offset] === '-' && $string[$offset + 1] === '-') ||
(isset($string[$offset + 1]) && $string[$offset] === '/' && $string[$offset + 1] === '*'))
) {
// Comment until end of line
if ($string[0] === '-' || $string[0] === '#') {
$last = strpos($string, "\n");
if ($string[$offset] === '-' || $string[$offset] === '#') {
$last = strpos($string, "\n", $offset);
$type = Token::TOKEN_TYPE_COMMENT;
} else { // Comment until closing comment tag
$pos = strpos($string, '*/', 2);
$pos = strpos($string, '*/', $offset + 2);
$last = $pos !== false
? $pos + 2
: false;
Expand All @@ -856,30 +857,30 @@ private function createNextToken(string $string, Token|null $previous = null): T
$last = strlen($string);
}

return new Token($type, substr($string, 0, $last));
return new Token($type, substr($string, $offset, $last - $offset));
}

// Quoted String
if ($string[0] === '"' || $string[0] === '\'' || $string[0] === '`' || $string[0] === '[') {
if ($string[$offset] === '"' || $string[$offset] === '\'' || $string[$offset] === '`' || $string[$offset] === '[') {
return new Token(
($string[0] === '`' || $string[0] === '['
($string[$offset] === '`' || $string[$offset] === '['
? Token::TOKEN_TYPE_BACKTICK_QUOTE
: Token::TOKEN_TYPE_QUOTE),
$this->getNextQuotedString($string),
$this->getNextQuotedString($string, $offset),
);
}

// User-defined Variable
if (($string[0] === '@' || $string[0] === ':') && isset($string[1])) {
if (($string[$offset] === '@' || $string[$offset] === ':') && isset($string[$offset + 1])) {
$value = null;
$type = Token::TOKEN_TYPE_VARIABLE;

// If the variable name is quoted
if ($string[1] === '"' || $string[1] === '\'' || $string[1] === '`') {
$value = $string[0] . $this->getNextQuotedString(substr($string, 1));
if ($string[$offset + 1] === '"' || $string[$offset + 1] === '\'' || $string[$offset + 1] === '`') {
$value = $string[$offset] . $this->getNextQuotedString($string, $offset + 1);
} else {
// Non-quoted variable name
preg_match('/^(' . $string[0] . '[\w.$]+)/', $string, $matches);
preg_match('/\G(' . $string[$offset] . '[\w.$]+)/', $string, $matches, 0, $offset);
if ($matches) {
$value = $matches[1];
}
Expand All @@ -893,79 +894,85 @@ private function createNextToken(string $string, Token|null $previous = null): T
// Number (decimal, binary, or hex)
if (
preg_match(
'/^(\d+(\.\d+)?|0x[\da-fA-F]+|0b[01]+)($|\s|"\'`|' . $this->regexBoundaries . ')/',
'/\G(\d+(\.\d+)?|0x[\da-fA-F]+|0b[01]+)($|\s|"\'`|' . $this->regexBoundaries . ')/',
$string,
$matches,
0,
$offset,
)
) {
return new Token(Token::TOKEN_TYPE_NUMBER, $matches[1]);
}

// Boundary Character (punctuation and symbols)
if (preg_match('/^(' . $this->regexBoundaries . ')/', $string, $matches)) {
if (preg_match('/\G(' . $this->regexBoundaries . ')/', $string, $matches, 0, $offset)) {
return new Token(Token::TOKEN_TYPE_BOUNDARY, $matches[1]);
}

// A reserved word cannot be preceded by a '.'
// this makes it so in "mytable.from", "from" is not considered a reserved word
if ($previous === null || $previous->value() !== '.') {
$upper = strtoupper($string);
// Top Level Reserved Word
if (
preg_match(
'/^(' . $this->regexReservedToplevel . ')($|\s|' . $this->regexBoundaries . ')/',
'/\G(' . $this->regexReservedToplevel . ')($|\s|' . $this->regexBoundaries . ')/',
$upper,
$matches,
0,
$offset,
)
) {
return new Token(
Token::TOKEN_TYPE_RESERVED_TOPLEVEL,
substr($string, 0, strlen($matches[1])),
substr($string, $offset, strlen($matches[1])),
);
}

// Newline Reserved Word
if (
preg_match(
'/^(' . $this->regexReservedNewline . ')($|\s|' . $this->regexBoundaries . ')/',
'/\G(' . $this->regexReservedNewline . ')($|\s|' . $this->regexBoundaries . ')/',
$upper,
$matches,
0,
$offset,
)
) {
return new Token(
Token::TOKEN_TYPE_RESERVED_NEWLINE,
substr($string, 0, strlen($matches[1])),
substr($string, $offset, strlen($matches[1])),
);
}

// Other Reserved Word
if (
preg_match(
'/^(' . $this->regexReserved . ')($|\s|' . $this->regexBoundaries . ')/',
'/\G(' . $this->regexReserved . ')($|\s|' . $this->regexBoundaries . ')/',
$upper,
$matches,
0,
$offset,
)
) {
return new Token(
Token::TOKEN_TYPE_RESERVED,
substr($string, 0, strlen($matches[1])),
substr($string, $offset, strlen($matches[1])),
);
}
}

// A function must be succeeded by '('
// this makes it so "count(" is considered a function, but "count" alone is not
$upper = strtoupper($string);
// function
if (preg_match('/^(' . $this->regexFunction . '[(]|\s|[)])/', $upper, $matches)) {
if (preg_match('/\G(' . $this->regexFunction . '[(]|\s|[)])/', $upper, $matches, 0, $offset)) {
return new Token(
Token::TOKEN_TYPE_RESERVED,
substr($string, 0, strlen($matches[1]) - 1),
substr($string, $offset, strlen($matches[1]) - 1),
);
}

// Non reserved word
preg_match('/^(.*?)($|\s|["\'`]|' . $this->regexBoundaries . ')/', $string, $matches);
preg_match('/\G(.*?)($|\s|["\'`]|' . $this->regexBoundaries . ')/', $string, $matches, 0, $offset);

return new Token(Token::TOKEN_TYPE_WORD, $matches[1]);
}
Expand All @@ -985,7 +992,7 @@ private function quoteRegex(array $strings): array
);
}

private function getNextQuotedString(string $string): string
private function getNextQuotedString(string $string, int $offset): string
{
$ret = '';

Expand All @@ -997,7 +1004,7 @@ private function getNextQuotedString(string $string): string
if (
preg_match(
<<<'EOD'
~^(?>(?sx)
~\G(?>(?sx)
(?:`[^`]*(?:$|`))+
|(?:\[[^\]]*($|\]))(?:\][^\]]*(?:$|\]))*
|(?:"[^"\\]*(?:\\.[^"\\]*)*(?:"|$))+
Expand All @@ -1006,6 +1013,8 @@ private function getNextQuotedString(string $string): string
EOD,
$string,
$matches,
0,
$offset,
)
) {
$ret = $matches[0];
Expand Down