Introduce a centralized search service with OpenAI-powered intent parsing and caching

2026-03-12 20:14:06 +01:00
parent f651309d6a
commit 6e887b315f
16 changed files with 3164 additions and 5 deletions
@@ -0,0 +1,192 @@
+<?php
+
+app_require('interfaces/system_search_intent_parser_i.php');
+app_require('classes/system_search_cache.php');
+app_require('classes/system_search_openai_intent_parser.php');
+
+use classes\system_search_cache;
+use classes\system_search_openai_intent_parser;
+
+if (!class_exists('SystemSearchTestRedisAdapter')) {
+    class SystemSearchTestRedisAdapter
+    {
+        private array $store = [];
+
+        public function reset(): void
+        {
+            $this->store = [];
+        }
+
+        public function get(string $key): mixed
+        {
+            return $this->store[$key] ?? null;
+        }
+
+        public function set(string $key, string $value): void
+        {
+            $this->store[$key] = $value;
+        }
+
+        public function setEx(string $key, string $value, int $ttl): void
+        {
+            $this->store[$key] = $value;
+        }
+
+        public function delete(string $key): void
+        {
+            unset($this->store[$key]);
+        }
+
+        public function expire(string $key, int $ttl): void
+        {
+            // TTL is not simulated in unit tests.
+        }
+
+        public function set_array(string $key, array $value): void
+        {
+            $this->store[$key] = json_encode($value, JSON_UNESCAPED_UNICODE);
+        }
+
+        public function get_array(string $key): ?array
+        {
+            $value = $this->store[$key] ?? null;
+            if (!is_string($value)) {
+                return null;
+            }
+            $decoded = json_decode($value, true);
+            return is_array($decoded) ? $decoded : null;
+        }
+
+        public function clear_keys(string $pattern): void
+        {
+            $regex = '/^' . str_replace('\*', '.*', preg_quote($pattern, '/')) . '$/';
+            foreach (array_keys($this->store) as $key) {
+                if (preg_match($regex, $key)) {
+                    unset($this->store[$key]);
+                }
+            }
+        }
+    }
+}
+
+beforeEach(function (): void {
+    system_search_cache::setAdapterForTests(new SystemSearchTestRedisAdapter());
+});
+
+afterEach(function (): void {
+    system_search_cache::setAdapterForTests(null);
+});
+
+it('redacts obvious sensitive fragments before sending query to intent parser', function (): void {
+    $query = 'Contact alice@example.com at +45 12 34 56 78, cvr 12345678';
+    $redacted = system_search_openai_intent_parser::redactSensitiveQuery($query);
+
+    expect($redacted)->toContain('[email]');
+    expect($redacted)->toContain('[phone]');
+    expect($redacted)->toContain('[cvr]');
+    expect($redacted)->not->toContain('alice@example.com');
+    expect($redacted)->not->toContain('12345678');
+});
+
+it('builds payload with redacted query and parses strict JSON output', function (): void {
+    $capturedPrompt = '';
+    $parser = new system_search_openai_intent_parser(
+        function (array $payload, string $apiKey) use (&$capturedPrompt): array {
+            $capturedPrompt = (string)($payload['input'][0]['content'][0]['text'] ?? '');
+            return [
+                'output' => [
+                    [
+                        'content' => [
+                            [
+                                'text' => json_encode([
+                                    'success' => true,
+                                    'normalized_query' => 'acme unpaid invoices',
+                                    'aliases' => ['acme', 'invoice overdue'],
+                                    'entity_hints' => ['customers', 'invoices'],
+                                    'confidence' => 0.93,
+                                    'association_hint' => true,
+                                ], JSON_UNESCAPED_UNICODE),
+                            ],
+                        ],
+                    ],
+                ],
+            ];
+        },
+        true,
+        'test-key'
+    );
+
+    $result = $parser->parse(
+        'find unpaid invoices for alice@example.com',
+        ['customers', 'invoices'],
+        ['customers' => ['account'], 'invoices' => ['billing']]
+    );
+
+    expect($capturedPrompt)->toContain('[email]');
+    expect($capturedPrompt)->not->toContain('alice@example.com');
+    expect($result['success'])->toBeTrue();
+    expect($result['source'])->toBe('openai');
+    expect($result['confidence'])->toBe(0.93);
+    expect($result['entity_hints'])->toBe(['customers', 'invoices']);
+});
+
+it('falls back safely when OpenAI is disabled', function (): void {
+    $parser = new system_search_openai_intent_parser(null, false, null);
+    $result = $parser->parse('find acme invoices', ['customers', 'invoices']);
+
+    expect($result['success'])->toBeFalse();
+    expect($result['source'])->toBe('none');
+    expect($result['fallback_reason'])->toBe('openai_disabled');
+});
+
+it('handles malformed OpenAI response payloads without throwing', function (): void {
+    $parser = new system_search_openai_intent_parser(
+        function (array $payload, string $apiKey): array {
+            return ['output' => []];
+        },
+        true,
+        'test-key'
+    );
+
+    $result = $parser->parse('find acme invoices', ['invoices']);
+
+    expect($result['success'])->toBeFalse();
+    expect($result['source'])->toBe('openai');
+    expect((string)$result['fallback_reason'])->toContain('Invalid response format');
+});
+
+it('uses parser cache for identical query and allowed type combinations', function (): void {
+    $calls = 0;
+    $parser = new system_search_openai_intent_parser(
+        function (array $payload, string $apiKey) use (&$calls): array {
+            $calls++;
+            return [
+                'output' => [
+                    [
+                        'content' => [
+                            [
+                                'text' => json_encode([
+                                    'success' => true,
+                                    'normalized_query' => 'acme invoices',
+                                    'aliases' => ['acme'],
+                                    'entity_hints' => ['invoices'],
+                                    'confidence' => 0.8,
+                                    'association_hint' => false,
+                                ], JSON_UNESCAPED_UNICODE),
+                            ],
+                        ],
+                    ],
+                ],
+            ];
+        },
+        true,
+        'test-key'
+    );
+
+    $first = $parser->parse('acme invoices', ['invoices']);
+    $second = $parser->parse('acme invoices', ['invoices']);
+
+    expect($first['source'])->toBe('openai');
+    expect($second['source'])->toBe('cache');
+    expect($calls)->toBe(1);
+});