fixes

2026-04-04 18:19:34 +03:00
parent 5fb17f717d
commit 7c09846bd6
22 changed files with 1229 additions and 94 deletions
@@ -46,9 +46,16 @@ class AdzunaTransformer implements JobTransformer

    private function sanitizeDescription(string $description): string
    {
-        $clean = strip_tags($description);
+        // Adzuna returns truncated descriptions with ellipsis (...)
+        // Keep the description as-is but strip excessive HTML while preserving basic formatting
+        $clean = strip_tags($description, '<p><br><ul><ol><li><strong><b><em><i><h1><h2><h3><h4>');
        $clean = html_entity_decode($clean, ENT_QUOTES | ENT_HTML5, 'UTF-8');
-        $clean = preg_replace('/\s+/', ' ', $clean);
+
+        // Convert multiple spaces to single space but preserve newlines
+        $clean = preg_replace('/[^\S\n]+/', ' ', $clean);
+
+        // Note: Adzuna API only provides truncated descriptions in search results.
+        // Full descriptions would require individual job detail API calls.
        return trim($clean);
    }

@@ -70,9 +70,32 @@ class RemotiveTransformer implements JobTransformer
            return null;
        }

-        // Try to extract numbers from salary string
-        if (preg_match_all('/[\d,]+/', $salary, $matches)) {
-            $numbers = array_map(fn($n) => (float) str_replace(',', '', $n), $matches[0]);
+        // Normalize the salary string
+        $salary = strtolower($salary);
+
+        // Check if it's hourly rate - if so, convert to annual (assuming 40hr/week * 52 weeks = 2080 hours/year)
+        $isHourly = str_contains($salary, '/hour') || str_contains($salary, 'per hour') || str_contains($salary, '/hr');
+
+        // Try to extract numbers with optional k/K suffix
+        // Pattern: number followed by optional comma-separated digits, then optional 'k'
+        if (preg_match_all('/\$?\s*([\d,]+)\s*([kK])?/', $salary, $matches, PREG_SET_ORDER)) {
+            $numbers = [];
+
+            foreach ($matches as $match) {
+                $number = (float) str_replace(',', '', $match[1]);
+
+                // If there's a 'k' suffix, multiply by 1000
+                if (!empty($match[2]) && strtolower($match[2]) === 'k') {
+                    $number *= 1000;
+                }
+
+                // If it's hourly, convert to annual
+                if ($isHourly) {
+                    $number *= 2080; // 40 hours/week * 52 weeks
+                }
+
+                $numbers[] = $number;
+            }

            if (count($numbers) >= 2) {
                return $type === 'min' ? min($numbers) : max($numbers);