Files
Investidor-V2/app/Services/SearchTerms.php
T

113 lines
4.4 KiB
PHP

<?php
namespace App\Services;
use Illuminate\Support\Str;
/** Text matching shared by the database and the built-in page catalogue. */
final class SearchTerms
{
/** @return list<string> */
public function tokens(string $query): array
{
$normalized = mb_strtolower(Str::ascii($query));
$tokens = array_values(array_filter(array_unique(array_map(
fn ($word) => trim($word, ".,;:!?\"'()[]{}"),
preg_split('/\s+/u', trim($normalized), -1, PREG_SPLIT_NO_EMPTY) ?: []
)), fn ($word) => $word !== ''));
if (count($tokens) > 1) {
$meaningful = array_values(array_diff($tokens, ['a', 'as', 'o', 'os', 'de', 'do', 'da', 'dos', 'das', 'em', 'no', 'na', 'nos', 'nas', 'para', 'por', 'e', 'como', 'qual', 'quais', 'um', 'uma']));
return $meaningful === [] ? $tokens : $meaningful;
}
return $tokens;
}
/** Known Portuguese spelling and number variants; do not infer semantic equivalence. */
public function variants(string $token): array
{
foreach ([['sector', 'sectores', 'setor', 'setores'], ['projecto', 'projectos', 'projeto', 'projetos'],
['investir', 'investimento', 'investimentos'], ['investidor', 'investidores'], ['oportunidade', 'oportunidades'], ['documento', 'documentos'],
['empresa', 'empresas'], ['parceria', 'parcerias'], ['zona', 'zonas'], ['formulario', 'formularios'],
['provincia', 'provincias'], ['corredor', 'corredores'], ['cluster', 'clusters'], ['imposto', 'impostos'], ['licenca', 'licencas']] as $group) {
if (in_array($token, $group, true)) {
return $group;
}
}
return [$token];
}
public function pattern(string $token): string
{
return '%'.str_replace(['!', '%', '_'], ['!!', '!%', '!_'], $token).'%';
}
/** Columns are trusted application constants, never user input. */
public function orFields($query, array $columns, string $token): void
{
foreach ($columns as $column) {
foreach ($this->variants($token) as $variant) {
$query->orWhereRaw($this->normalizedColumn($column)." LIKE ? ESCAPE '!'", [$this->pattern($variant)]);
}
}
}
public function constrain($query, array $columns, array $tokens): void
{
foreach ($tokens as $token) {
$query->where(fn ($match) => $this->orFields($match, $columns, $token));
}
}
public function matches(string $text, array $tokens): bool
{
$text = mb_strtolower(Str::ascii($text));
foreach ($tokens as $token) {
if (! array_filter($this->variants($token), fn ($variant) => str_contains($text, $variant))) {
return false;
}
}
return true;
}
private function normalizedColumn(string $column): string
{
return "translate(lower(coalesce({$column}::text, '')), 'áàâãäéèêëíìîïóòôõöúùûüçñ', 'aaaaaeeeeiiiiooooouuuucn')";
}
public function score(string $title, string $phrase, array $tokens): int
{
$title = mb_strtolower(Str::ascii($title));
$phrase = mb_strtolower(Str::ascii(trim($phrase)));
$score = $title === $phrase ? 1000 : (str_contains($title, $phrase) ? 400 : 0);
foreach ($tokens as $token) {
if ($this->matches($title, [$token])) {
$score += 20;
}
}
return $score;
}
/** The same title score used for in-memory pages, with bound search terms. */
public function orderByRelevance($query, string $titleColumn, string $phrase, array $tokens, string $tieColumn): void
{
$column = $this->normalizedColumn($titleColumn);
$phrase = mb_strtolower(Str::ascii(trim($phrase)));
$sql = "CASE WHEN {$column} = ? THEN 1000 WHEN {$column} LIKE ? ESCAPE '!' THEN 400 ELSE 0 END";
$bindings = [$phrase, $this->pattern($phrase)];
foreach ($tokens as $token) {
$conditions = [];
foreach ($this->variants($token) as $variant) {
$conditions[] = "{$column} LIKE ? ESCAPE '!'";
$bindings[] = $this->pattern($variant);
}
$sql .= ' + CASE WHEN ('.implode(' OR ', $conditions).') THEN 20 ELSE 0 END';
}
$query->orderByRaw('('.$sql.') DESC', $bindings)->orderBy($titleColumn)->orderBy($tieColumn);
}
}