Page MenuHomePhabricator
Differential D18492 Diff 44442 src/search/PhutilSearchQueryCompiler.php

Changeset View

Standalone View

View Options

src/search/PhutilSearchQueryCompiler.php

<?php <?php
final class PhutilSearchQueryCompiler final class PhutilSearchQueryCompiler
extends Phobject { extends Phobject {
private $operators = '+ -><()~*:""&|'; private $operators = '+ -><()~*:""&|';
private $query; private $query;
private $stemmer; private $stemmer;
private $enableFunctions = false;
const OPERATOR_NOT = 'not'; const OPERATOR_NOT = 'not';
const OPERATOR_AND = 'and'; const OPERATOR_AND = 'and';
const OPERATOR_SUBSTRING = 'sub';
const OPERATOR_EXACT = 'exact';
public function setOperators($operators) { public function setOperators($operators) {
$this->operators = $operators; $this->operators = $operators;
return $this; return $this;
} }
public function getOperators() { public function getOperators() {
return $this->operators; return $this->operators;
} }
public function setStemmer(PhutilSearchStemmer $stemmer) { public function setStemmer(PhutilSearchStemmer $stemmer) {
$this->stemmer = $stemmer; $this->stemmer = $stemmer;
return $this; return $this;
} }
public function getStemmer() { public function getStemmer() {
return $this->stemmer; return $this->stemmer;
} }
public function setEnableFunctions($enable_functions) {
$this->enableFunctions = $enable_functions;
return $this;
}
public function getEnableFunctions() {
return $this->enableFunctions;
}
public function compileQuery(array $tokens) { public function compileQuery(array $tokens) {
assert_instances_of($tokens, 'PhutilSearchQueryToken'); assert_instances_of($tokens, 'PhutilSearchQueryToken');
$result = array(); $result = array();
foreach ($tokens as $token) { foreach ($tokens as $token) {
$result[] = $this->renderToken($token); $result[] = $this->renderToken($token);
} }
▲ Show 20 Lines • Show All 58 Lines • ▼ Show 20 Lines if ($query_bytes > $maximum_bytes) {
'Query is too long (%s bytes, maximum is %s bytes).', 'Query is too long (%s bytes, maximum is %s bytes).',
new PhutilNumber($query_bytes), new PhutilNumber($query_bytes),
new PhutilNumber($maximum_bytes))); new PhutilNumber($maximum_bytes)));
} }
$query = phutil_utf8v($query); $query = phutil_utf8v($query);
$length = count($query); $length = count($query);
$enable_functions = $this->getEnableFunctions();
$mode = 'scan'; $mode = 'scan';
$current_operator = array(); $current_operator = array();
$current_token = array(); $current_token = array();
$current_function = null;
$is_quoted = false; $is_quoted = false;
$tokens = array(); $tokens = array();
if ($enable_functions) {
$operator_characters = '[~=+-]';
} else {
$operator_characters = '[+-]';
}
for ($ii = 0; $ii < $length; $ii++) { for ($ii = 0; $ii < $length; $ii++) {
$character = $query[$ii]; $character = $query[$ii];
if ($mode == 'scan') { if ($mode == 'scan') {
if (preg_match('/^\s\z/u', $character)) { if (preg_match('/^\s\z/u', $character)) {
continue; continue;
} }
$mode = 'function';
}
if ($mode == 'function') {
$mode = 'operator'; $mode = 'operator';
if ($enable_functions) {
$found = false;
for ($jj = $ii; $jj < $length; $jj++) {
if (preg_match('/^[a-zA-Z]\z/u', $query[$jj])) {
continue;
}
if ($query[$jj] == ':') {
$found = $jj;
}
break;
}
if ($found !== false) {
$function = array_slice($query, $ii, ($jj - $ii));
$current_function = implode('', $function);
if (!strlen($current_function)) {
$current_function = null;
}
$ii = $jj;
continue;
}
}
} }
if ($mode == 'operator') { if ($mode == 'operator') {
if (preg_match('/^\s\z/u', $character)) { if (preg_match('/^\s\z/u', $character)) {
continue; continue;
} }
if (preg_match('/^[+-]\z/', $character)) { if (preg_match('/^'.$operator_characters.'\z/', $character)) {
$current_operator[] = $character; $current_operator[] = $character;
continue; continue;
} }
$mode = 'quote'; $mode = 'quote';
} }
if ($mode == 'quote') { if ($mode == 'quote') {
Show All 24 Lines for ($ii = 0; $ii < $length; $ii++) {
if (preg_match('/^"\z/', $character)) { if (preg_match('/^"\z/', $character)) {
$capture = true; $capture = true;
$mode = 'token'; $mode = 'token';
$is_quoted = true; $is_quoted = true;
} }
} }
if ($capture) { if ($capture) {
$tokens[] = array( $token = array(
'operator' => $current_operator, 'operator' => $current_operator,
'quoted' => $was_quoted, 'quoted' => $was_quoted,
'value' => $current_token, 'value' => $current_token,
); );
if ($enable_functions) {
$token['function'] = $current_function;
}
$tokens[] = $token;
$current_operator = array(); $current_operator = array();
$current_token = array(); $current_token = array();
$current_function = null;
continue; continue;
} else { } else {
$current_token[] = $character; $current_token[] = $character;
} }
} }
} }
if ($is_quoted) { if ($is_quoted) {
throw new PhutilSearchQueryCompilerSyntaxException( throw new PhutilSearchQueryCompilerSyntaxException(
pht( pht(
'Query contains unmatched double quotes.')); 'Query contains unmatched double quotes.'));
} }
if ($mode == 'operator') { if ($mode == 'operator') {
throw new PhutilSearchQueryCompilerSyntaxException( throw new PhutilSearchQueryCompilerSyntaxException(
pht( pht(
'Query contains operator ("%s") with no search term.', 'Query contains operator ("%s") with no search term.',
implode('', $current_operator))); implode('', $current_operator)));
} }
$tokens[] = array( $token = array(
'operator' => $current_operator, 'operator' => $current_operator,
'quoted' => false, 'quoted' => false,
'value' => $current_token, 'value' => $current_token,
); );
if ($enable_functions) {
$token['function'] = $current_function;
}
$tokens[] = $token;
$results = array(); $results = array();
foreach ($tokens as $token) { foreach ($tokens as $token) {
$value = implode('', $token['value']); $value = implode('', $token['value']);
$operator_string = implode('', $token['operator']); $operator_string = implode('', $token['operator']);
if (!strlen($value)) { if (!strlen($value)) {
continue; continue;
} }
switch ($operator_string) { switch ($operator_string) {
case '-': case '-':
$operator = self::OPERATOR_NOT; $operator = self::OPERATOR_NOT;
break; break;
case '~':
$operator = self::OPERATOR_SUBSTRING;
break;
case '=':
$operator = self::OPERATOR_EXACT;
break;
case '': case '':
case '+': case '+':
$operator = self::OPERATOR_AND; $operator = self::OPERATOR_AND;
break; break;
default: default:
throw new PhutilSearchQueryCompilerSyntaxException( throw new PhutilSearchQueryCompilerSyntaxException(
pht( pht(
'Query has an invalid sequence of operators ("%s").', 'Query has an invalid sequence of operators ("%s").',
$operator_string)); $operator_string));
} }
$results[] = array( $result = array(
'operator' => $operator, 'operator' => $operator,
'quoted' => $token['quoted'], 'quoted' => $token['quoted'],
'value' => $value, 'value' => $value,
); );
if ($enable_functions) {
$result['function'] = $token['function'];
}
$results[] = $result;
} }
return $results; return $results;
} }
private function renderToken( private function renderToken(
PhutilSearchQueryToken $token, PhutilSearchQueryToken $token,
PhutilSearchStemmer $stemmer = null) { PhutilSearchStemmer $stemmer = null) {
▲ Show 20 Lines • Show All 49 Lines • Show Last 20 Lines
Phabricator · Built by Phacility