marek/MtoRagSystem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

p43A

Browse Source
This commit is contained in:
team 1
2026-05-05 12:12:51 +02:00
parent da374edcf4
commit 2c041a88c0
12 changed files with 429 additions and 282 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
src/Agent/AgentRunner.php
View File

@@ -1238,7 +1238,7 @@ final readonly class AgentRunner
* 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 Testomat model reference and the first explicit °dH value
* 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.
*
@@ -1261,12 +1261,12 @@ final readonly class AgentRunner
$anchors = [];
$model = $this->extractFirstTestomatModelAnchor($answer);
$model = $this->extractFirstProductModelAnchor($answer);
if ($model !== '') {
$anchors[] = $model;
}
$hardnessValue = $this->extractFirstHardnessValueAnchor($answer);
$hardnessValue = $this->extractFirstMeasurementValueAnchor($answer);
if ($hardnessValue !== '') {
$anchors[] = $hardnessValue;
}
@@ -1325,9 +1325,9 @@ final readonly class AgentRunner
return array_reverse($turns);
}
private function extractFirstTestomatModelAnchor(string $text): string
private function extractFirstProductModelAnchor(string $text): string
{
if (preg_match($this->agentRunnerConfig->getFollowUpReferenceAnchorTestomatModelPattern(), $text, $matches) !== 1) {
if (preg_match($this->agentRunnerConfig->getFollowUpReferenceAnchorProductModelPattern(), $text, $matches) !== 1) {
return '';
}
@@ -1337,9 +1337,9 @@ final readonly class AgentRunner
return trim(str_replace('®', '', $value));
}
private function extractFirstHardnessValueAnchor(string $text): string
private function extractFirstMeasurementValueAnchor(string $text): string
{
if (preg_match($this->agentRunnerConfig->getFollowUpReferenceAnchorHardnessValuePattern(), $text, $matches) !== 1) {
if (preg_match($this->agentRunnerConfig->getFollowUpReferenceAnchorMeasurementValuePattern(), $text, $matches) !== 1) {
return '';
}
@@ -1500,7 +1500,7 @@ final readonly class AgentRunner
return true;
}
if ($this->extractFirstTestomatModelAnchor($prompt) !== '') {
if ($this->extractFirstProductModelAnchor($prompt) !== '') {
return false;
}
@@ -1564,7 +1564,7 @@ final readonly class AgentRunner
private function hasStandaloneConcreteShopSubject(string $prompt): bool
{
if ($this->extractFirstTestomatModelAnchor($prompt) !== '') {
if ($this->extractFirstProductModelAnchor($prompt) !== '') {
return true;
}
@@ -1622,7 +1622,7 @@ final readonly class AgentRunner
return $prompt;
}
if ($this->extractFirstTestomatModelAnchor($prompt) === '') {
if ($this->extractFirstProductModelAnchor($prompt) === '') {
return $optimizedShopQuery;
}
@@ -2249,7 +2249,7 @@ final readonly class AgentRunner
continue;
}
$model = $this->extractFirstTestomatModelAnchor($turn);
$model = $this->extractFirstProductModelAnchor($turn);
if ($model !== '') {
$query = str_replace(
@@ -2334,7 +2334,7 @@ final readonly class AgentRunner
}
}
$modelAnchor = $this->extractFirstTestomatModelAnchor($turn);
$modelAnchor = $this->extractFirstProductModelAnchor($turn);
if ($modelAnchor !== '' && !$this->isMetaOnlyShopQuery($modelAnchor)) {
return mb_strtolower($modelAnchor, 'UTF-8');

24
src/Config/AgentRunnerConfig.php
View File

@@ -125,14 +125,34 @@ final class AgentRunnerConfig
return $this->getRequiredString('follow_up_context.history_question_strip_pattern');
}
public function getFollowUpReferenceAnchorProductModelPattern(): string
{
$value = $this->optionalValue('follow_up_context.reference_anchor.product_model_pattern');
if (is_string($value) && trim($value) !== '') {
return $value;
}
return $this->getRequiredString('follow_up_context.reference_anchor.testomat_model_pattern');
}
public function getFollowUpReferenceAnchorMeasurementValuePattern(): string
{
$value = $this->optionalValue('follow_up_context.reference_anchor.measurement_value_pattern');
if (is_string($value) && trim($value) !== '') {
return $value;
}
return $this->getRequiredString('follow_up_context.reference_anchor.hardness_value_pattern');
}
public function getFollowUpReferenceAnchorTestomatModelPattern(): string
{
return $this->getRequiredString('follow_up_context.reference_anchor.testomat_model_pattern');
return $this->getFollowUpReferenceAnchorProductModelPattern();
}
public function getFollowUpReferenceAnchorHardnessValuePattern(): string
{
return $this->getRequiredString('follow_up_context.reference_anchor.hardness_value_pattern');
return $this->getFollowUpReferenceAnchorMeasurementValuePattern();
}

41
src/Config/CommerceQueryParserConfig.php
View File

@@ -13,6 +13,7 @@ final class CommerceQueryParserConfig
*/
public function __construct(
private readonly array $config = [],
private readonly ?DomainVocabularyConfig $vocabulary = null,
) {
}
@@ -268,7 +269,10 @@ final class CommerceQueryParserConfig
/** @return string[] */
public function getSemanticShopSearchTokens(): array
{
return $this->stringList('semantic_shop_search_tokens');
return $this->configuredStringListOrVocabularyView(
'semantic_shop_search_tokens',
'vocabulary_views.semantic_shop_search_tokens'
);
}
public function buildExactTokenRemovalPattern(string $token): string
@@ -319,6 +323,27 @@ final class CommerceQueryParserConfig
return $out;
}
/** @return string[] */
private function configuredStringListOrVocabularyView(string $configPath, string $viewPathConfigPath): array
{
if ($this->hasPath($configPath)) {
return $this->stringList($configPath);
}
if ($this->vocabulary === null) {
throw $this->missing($configPath);
}
$viewPath = $this->string($viewPathConfigPath);
$terms = $this->vocabulary->view($viewPath, []);
if ($terms === []) {
throw $this->invalid($viewPathConfigPath, sprintf('references empty vocabulary view "%s"', $viewPath));
}
return $terms;
}
/** @return array<string, string> */
private function stringMap(string $path): array
{
@@ -372,6 +397,20 @@ final class CommerceQueryParserConfig
return $value;
}
private function hasPath(string $path): bool
{
$current = $this->config;
foreach (explode('.', $path) as $segment) {
if (!is_array($current) || !array_key_exists($segment, $current)) {
return false;
}
$current = $current[$segment];
}
return true;
}
private function value(string $path): mixed
{
$current = $this->config;

114
src/Config/NdjsonHybridRetrieverConfig.php
View File

@@ -13,6 +13,7 @@ final class NdjsonHybridRetrieverConfig
*/
public function __construct(
private array $config = [],
private ?DomainVocabularyConfig $vocabulary = null,
) {
}
@@ -216,55 +217,82 @@ final class NdjsonHybridRetrieverConfig
/** @return string[] */
public function genericProductTokens(): array
{
return $this->requiredStringList('generic_product_tokens');
return $this->configuredStringListOrVocabularyView(
'generic_product_tokens',
'vocabulary_views.generic_product_tokens'
);
}
/** @return string[] */
public function importantShortModelTokens(): array
{
return $this->requiredStringList('important_short_model_tokens');
return $this->configuredStringListOrVocabularyView(
'important_short_model_tokens',
'vocabulary_views.important_short_model_tokens'
);
}
/** @return string[] */
public function familyDescriptorTokens(): array
{
return $this->requiredStringList('family_descriptor_tokens');
return $this->configuredStringListOrVocabularyView(
'family_descriptor_tokens',
'vocabulary_views.family_descriptor_tokens'
);
}
/** @return string[] */
public function looksLikeReagentTokens(): array
{
return $this->requiredStringList('looks_like_reagent_tokens');
return $this->configuredStringListOrVocabularyView(
'looks_like_reagent_tokens',
'vocabulary_views.looks_like_reagent_tokens'
);
}
/** @return string[] */
public function looksLikeSafetyDocs(): array
{
return $this->requiredStringList('looks_like_safety_docs');
return $this->configuredStringListOrVocabularyView(
'looks_like_safety_docs',
'vocabulary_views.looks_like_safety_docs'
);
}
/** @return string[] */
public function looksLikeReagentWords(): array
{
return $this->requiredStringList('looks_like_reagent_words');
return $this->configuredStringListOrVocabularyView(
'looks_like_reagent_words',
'vocabulary_views.looks_like_reagent_words'
);
}
/** @return string[] */
public function looksLikeDocumentWords(): array
{
return $this->requiredStringList('looks_like_document_words');
return $this->configuredStringListOrVocabularyView(
'looks_like_document_words',
'vocabulary_views.looks_like_document_words'
);
}
/** @return string[] */
public function looksLikeSafetyWords(): array
{
return $this->requiredStringList('looks_like_safety_words');
return $this->configuredStringListOrVocabularyView(
'looks_like_safety_words',
'vocabulary_views.looks_like_safety_words'
);
}
/** @return string[] */
public function looksLikeDeviceWords(): array
{
return $this->requiredStringList('looks_like_device_words');
return $this->configuredStringListOrVocabularyView(
'looks_like_device_words',
'vocabulary_views.looks_like_device_words'
);
}
/**
@@ -471,6 +499,74 @@ final class NdjsonHybridRetrieverConfig
return $out;
}
/** @return string[] */
private function configuredStringListOrVocabularyView(string $configPath, string $viewPathConfigPath): array
{
if ($this->hasKey($configPath)) {
return $this->requiredStringList($configPath);
}
if ($this->vocabulary === null) {
throw $this->missing($configPath);
}
$viewPath = $this->requiredPathString($viewPathConfigPath);
$terms = $this->vocabulary->view($viewPath, []);
if ($terms === []) {
throw $this->invalid($viewPathConfigPath, sprintf('references empty vocabulary view "%s"', $viewPath));
}
return $terms;
}
private function requiredPathString(string $key): string
{
$value = $this->requiredPathValue($key);
if (!is_scalar($value)) {
throw $this->invalid($key, 'must be a non-empty string');
}
$value = trim((string) $value);
if ($value === '') {
throw $this->invalid($key, 'must be a non-empty string');
}
return $value;
}
private function requiredPathValue(string $key): mixed
{
$current = $this->config;
foreach (explode('.', $key) as $segment) {
if (!is_array($current) || !array_key_exists($segment, $current)) {
throw $this->missing($key);
}
$current = $current[$segment];
}
return $current;
}
private function hasKey(string $key): bool
{
$current = $this->config;
foreach (explode('.', $key) as $segment) {
if (!is_array($current) || !array_key_exists($segment, $current)) {
return false;
}
$current = $current[$segment];
}
return true;
}
private function requiredValue(string $key): mixed
{
if (!array_key_exists($key, $this->config)) {

55
src/Config/PromptBuilderConfig.php
View File

@@ -11,6 +11,7 @@ final class PromptBuilderConfig
*/
public function __construct(
private readonly array $config = [],
private readonly ?DomainVocabularyConfig $vocabulary = null,
) {
}
@@ -159,6 +160,35 @@ final class PromptBuilderConfig
return $out;
}
/**
* @return string[]
*/
private function getConfiguredStringListOrVocabularyView(string $configPath, string $viewPathConfigPath): array
{
if ($this->hasPath($configPath)) {
return $this->getRequiredStringList($configPath);
}
if ($this->vocabulary === null) {
throw new \InvalidArgumentException(sprintf(
'RetrieX prompt config path "%s" is missing and no vocabulary resolver is available.',
$configPath
));
}
$viewPath = $this->getRequiredString($viewPathConfigPath);
$terms = $this->vocabulary->view($viewPath, []);
if ($terms === []) {
throw new \InvalidArgumentException(sprintf(
'RetrieX prompt vocabulary view "%s" resolved to an empty list.',
$viewPath
));
}
return $terms;
}
/**
* @return string[]
*/
@@ -193,6 +223,21 @@ final class PromptBuilderConfig
private function hasPath(string $path): bool
{
$current = $this->config;
foreach (explode('.', $path) as $segment) {
if (!is_array($current) || !array_key_exists($segment, $current)) {
return false;
}
$current = $current[$segment];
}
return true;
}
private function getOptionalValue(string $path): mixed
{
$current = $this->config;
@@ -573,7 +618,10 @@ final class PromptBuilderConfig
*/
public function getTechnicalProductKeywords(): array
{
return $this->getRequiredStringList('technical_product_keywords');
return $this->getConfiguredStringListOrVocabularyView(
'technical_product_keywords',
'vocabulary_views.technical_product_keywords'
);
}
/**
@@ -581,7 +629,10 @@ final class PromptBuilderConfig
*/
public function getAccessoryRequestKeywords(): array
{
return $this->getRequiredStringList('accessory_request_keywords');
return $this->getConfiguredStringListOrVocabularyView(
'accessory_request_keywords',
'vocabulary_views.accessory_request_keywords'
);
}
public function getMeasurementEvidenceSectionLabel(): string