p43C

src/Agent/ReferenceAnchorExtractor.php

@@ -0,0 +1,129 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Agent;
+
+use App\Config\AgentRunnerConfig;
+
+/**
+ * Extracts generic reference anchors used to resolve follow-up questions.
+ *
+ * The extractor is deliberately domain-neutral: product model anchors and
+ * measurement value anchors are configured through AgentRunnerConfig. Current
+ * water-analysis patterns remain supported by configuration, but the calling
+ * code no longer needs Testomat- or hardness-specific concepts.
+ */
+final readonly class ReferenceAnchorExtractor
+{
+    public function __construct(
+        private AgentRunnerConfig $config,
+    ) {
+    }
+
+    /**
+     * Extract stable reference anchors from the latest assistant answer.
+     *
+     * These anchors are only used to resolve follow-up references such as
+     * "der Wert" or "welcher Indikator". They are not factual evidence for
+     * the final answer. To avoid propagating wrong earlier answers, only the
+     * first explicit product-model reference and the first explicit measurement
+     * value are kept. Indicator names, reagent codes, prices, URLs and product
+     * numbers are intentionally ignored here.
+     *
+     * @return string[]
+     */
+    public function extractLatestAssistantReferenceAnchors(string $history): array
+    {
+        $turn = $this->extractLatestHistoryTurn($history);
+
+        if ($turn === '') {
+            return [];
+        }
+
+        $answer = preg_replace($this->config->getFollowUpHistoryQuestionStripPattern(), '', $turn, 1) ?? '';
+        $answer = trim($answer);
+
+        if ($answer === '') {
+            return [];
+        }
+
+        $anchors = [];
+
+        $model = $this->extractFirstProductModelAnchor($answer);
+        if ($model !== '') {
+            $anchors[] = $model;
+        }
+
+        $measurementValue = $this->extractFirstMeasurementValueAnchor($answer);
+        if ($measurementValue !== '') {
+            $anchors[] = $measurementValue;
+        }
+
+        return array_values(array_unique($anchors));
+    }
+
+    public function extractFirstProductModelAnchor(string $text): string
+    {
+        if (preg_match($this->config->getFollowUpReferenceAnchorProductModelPattern(), $text, $matches) !== 1) {
+            return '';
+        }
+
+        $value = $this->sanitizeAnchor((string) ($matches[0] ?? ''));
+        $value = preg_replace('/\s+/u', ' ', $value) ?? $value;
+
+        return trim(str_replace('®', '', $value));
+    }
+
+    public function extractFirstMeasurementValueAnchor(string $text): string
+    {
+        if (preg_match($this->config->getFollowUpReferenceAnchorMeasurementValuePattern(), $text, $matches) !== 1) {
+            return '';
+        }
+
+        $value = preg_replace('/\s+/u', ' ', (string) ($matches[0] ?? '')) ?? '';
+
+        return trim($value);
+    }
+
+    private function extractLatestHistoryTurn(string $history): string
+    {
+        $history = trim($history);
+
+        if ($history === '') {
+            return '';
+        }
+
+        $parts = preg_split($this->config->getFollowUpHistoryTurnSplitPattern(), $history);
+
+        if ($parts === false || $parts === []) {
+            return '';
+        }
+
+        $turns = array_values(array_filter(
+            array_map(static fn(string $part): string => trim($part), $parts),
+            static fn(string $part): bool => $part !== ''
+        ));
+
+        if ($turns === []) {
+            return '';
+        }
+
+        return (string) end($turns);
+    }
+
+    private function sanitizeAnchor(string $value): string
+    {
+        $value = trim((string) preg_replace('/\s+/u', ' ', $value));
+
+        if ($value === '') {
+            return '';
+        }
+
+        if (mb_strlen($value, 'UTF-8') <= 500) {
+            return $value;
+        }
+
+        return rtrim(mb_substr($value, 0, 497, 'UTF-8')) . '...';
+    }
+}