add history to shop search

src/Commerce/CommerceQueryParser.php

@@ -21,29 +21,61 @@ final readonly class CommerceQueryParser
     {
     }
 
-    public function parse(string $originalPrompt, string $intent): CommerceSearchQuery
+    public function parse(
+        string $originalPrompt,
+        string $intent,
+        string $historyContext = ''
+    ): CommerceSearchQuery
     {
-        $normalized = $this->normalize($originalPrompt);
-        [$priceMin, $priceMax] = $this->extractPriceRange($normalized);
-        $sizes = $this->extractSizes($normalized);
-        $brand = $this->extractBrand($normalized);
-        $properties = [];
+        $normalizedPrompt = $this->normalize($originalPrompt);
+
+        [$priceMin, $priceMax] = $this->extractPriceRange($normalizedPrompt);
+        $sizes = $this->extractSizes($normalizedPrompt);
+        $brand = $this->extractBrand($normalizedPrompt);
 
         $searchText = $this->buildSearchText(
-            $normalized,
+            $normalizedPrompt,
             $sizes,
             $brand,
             $priceMin,
             $priceMax
         );
 
+        if ($historyContext !== '' && $this->shouldUseHistoryContext($normalizedPrompt)) {
+            $latestHistoryQuestion = $this->extractLatestQuestionFromHistory($historyContext);
+
+            if ($latestHistoryQuestion !== '') {
+                $normalizedHistoryPrompt = $this->normalize($latestHistoryQuestion);
+
+                [$historyPriceMin, $historyPriceMax] = $this->extractPriceRange($normalizedHistoryPrompt);
+                $historySizes = $this->extractSizes($normalizedHistoryPrompt);
+                $historyBrand = $this->extractBrand($normalizedHistoryPrompt);
+
+                $historySearchText = $this->buildSearchText(
+                    $normalizedHistoryPrompt,
+                    $historySizes,
+                    $historyBrand,
+                    $historyPriceMin,
+                    $historyPriceMax
+                );
+
+                $searchText = $this->mergeSearchTexts($historySearchText, $searchText);
+
+                if (($brand === null || $brand === '') && $historyBrand !== null && $historyBrand !== '') {
+                    $brand = $historyBrand;
+                }
+            }
+        }
+
+        $finalSearchText = $searchText !== '' ? $searchText : $normalizedPrompt;
+
         return new CommerceSearchQuery(
             originalPrompt: $originalPrompt,
-            normalizedPrompt: $normalized,
-            searchText: $searchText !== '' ? $searchText : $normalized,
+            normalizedPrompt: $normalizedPrompt,
+            searchText: $finalSearchText,
             brand: $brand,
             sizes: $sizes,
-            properties: $properties,
+            properties: [],
             priceMin: $priceMin,
             priceMax: $priceMax,
             intent: $intent,
@@ -138,9 +170,7 @@ final readonly class CommerceQueryParser
     {
         $text = ' ' . $prompt . ' ';
 
-        $phrasesToRemove = $this->config->getPhrasesToRemove();
-
-        foreach ($phrasesToRemove as $phrase) {
+        foreach ($this->config->getPhrasesToRemove() as $phrase) {
             $text = str_replace($phrase, ' ', $text);
         }
 
@@ -155,16 +185,83 @@ final readonly class CommerceQueryParser
         if ($priceMin !== null || $priceMax !== null) {
             $text = preg_replace('/\bzwischen\s+\d+(?:[.,]\d+)?\s+und\s+\d+(?:[.,]\d+)?\s*euro\b/u', ' ', $text) ?? $text;
             $text = preg_replace('/\b(?:unter|bis|max(?:imal)?|ab|mindestens|min)\s+\d+(?:[.,]\d+)?\s*euro\b/u', ' ', $text) ?? $text;
-            $text = preg_replace('/\b'.$this->intentConfig->getPricePattern().'\b/u', ' ', $text) ?? $text;
+            $text = preg_replace('/\b' . $this->intentConfig->getPricePattern() . '\b/u', ' ', $text) ?? $text;
         }
 
         $text = preg_replace('/\s+/u', ' ', $text) ?? $text;
         $text = trim($text, " \t\n\r\0\x0B-.,");
-        $tokens = array_filter(explode(' ', $text), static fn(string $token): bool => mb_strlen($token) > 1);
+
+        $tokens = array_filter(
+            explode(' ', $text),
+            static fn(string $token): bool => mb_strlen($token) > 1
+        );
+
+        $tokens = $this->filterSearchTokens($tokens);
 
         return trim(implode(' ', $tokens));
     }
 
+    private function shouldUseHistoryContext(string $prompt): bool
+    {
+        return preg_match(
+                '/\b(' . $this->config->getHistoryContextPattern() . ')\b/u',
+                $prompt
+            ) === 1;
+    }
+
+    private function extractLatestQuestionFromHistory(string $historyContext): string
+    {
+        if (preg_match_all('/^Question:\s*(.+)$/m', $historyContext, $matches) !== 1 && preg_match_all('/^Question:\s*(.+)$/m', $historyContext, $matches) === false) {
+            return '';
+        }
+
+        $questions = $matches[1] ?? [];
+        if ($questions === []) {
+            return '';
+        }
+
+        $lastQuestion = end($questions);
+
+        return is_string($lastQuestion) ? trim($lastQuestion) : '';
+    }
+
+    private function mergeSearchTexts(string $historySearchText, string $currentSearchText): string
+    {
+        $tokens = [];
+
+        foreach ([$historySearchText, $currentSearchText] as $text) {
+            if ($text === '') {
+                continue;
+            }
+
+            foreach (explode(' ', $text) as $token) {
+                $token = trim($token);
+
+                if ($token === '' || mb_strlen($token) <= 1) {
+                    continue;
+                }
+
+                $tokens[$token] = $token;
+            }
+        }
+
+        return implode(' ', array_values($tokens));
+    }
+
+    /**
+     * @param string[] $tokens
+     * @return string[]
+     */
+    private function filterSearchTokens(array $tokens): array
+    {
+        $stopWords = $this->config->getFilterSearchTokensPattern();
+
+        return array_values(array_filter(
+            $tokens,
+            static fn(string $token): bool => !in_array($token, $stopWords, true)
+        ));
+    }
+
     private function toFloat(string $value): ?float
     {
         $value = str_replace(',', '.', trim($value));