BxBaseStudioAgentsVectorStoreData.php 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117
  1. <?php defined('BX_DOL') or die('hack attempt');
  2. /**
  3. * Copyright (c) UNA, Inc - https://una.io
  4. * MIT License - https://opensource.org/licenses/MIT
  5. *
  6. * @defgroup UnaView UNA Studio Representation classes
  7. * @ingroup UnaStudio
  8. * @{
  9. */
  10. class BxBaseStudioAgentsVectorStoreData extends BxDolStudioAgentsVectorStoreData
  11. {
  12. public function __construct ($aOptions, $oTemplate = false)
  13. {
  14. parent::__construct ($aOptions, $oTemplate);
  15. $this->addMarkers($this->_aBrowseParams);
  16. }
  17. protected function _delete ($mixedId)
  18. {
  19. // TODO: delete files from vector store
  20. $r = $this->getVectorStoreDataById($mixedId);
  21. $oVectorStore = BxDolAI::getVectorStoreInstance($r['vector_store_id']);
  22. if ($oVectorStore) {
  23. $oVectorStore->deleteBySource($r['vector_store_id'], $r['id']);
  24. }
  25. $mixedResult = parent::_delete($mixedId);
  26. return $mixedResult;
  27. }
  28. protected function _getCellSize ($mixedValue, $sKey, $aField, $aRow)
  29. {
  30. $mixedValue = bx_format_bytes($mixedValue);
  31. return parent::_getCellDefault ($mixedValue, $sKey, $aField, $aRow);
  32. }
  33. protected function _getCellStatus ($mixedValue, $sKey, $aField, $aRow)
  34. {
  35. switch ($mixedValue) {
  36. case 'pending':
  37. $mixedValue = '<span title="Pending">⏳</span>';
  38. break;
  39. case 'processing':
  40. $mixedValue = '<span title="Processing">⚙️</span>';
  41. break;
  42. case 'ready':
  43. $mixedValue = '<span title="Ready">✅</span>';
  44. break;
  45. case 'error':
  46. $mixedValue = '<span title="Error">❌</span>';
  47. break;
  48. }
  49. return parent::_getCellDefault ($mixedValue, $sKey, $aField, $aRow);
  50. }
  51. static public function processPendingData()
  52. {
  53. $iLimit = 5; // TODO: add to settings
  54. for ($i = 0; $i < $iLimit; $i++) {
  55. $a = BxDolAiQuery::getVectorStorePendingData(1);
  56. if (!$a)
  57. break;
  58. foreach ($a as $r) {
  59. BxDolAiQuery::updateVectorStoreDataStatus($r['id'], 'processing');
  60. $aVectorStore = BxDolAiQuery::getVectorStoreObject($r['vector_store_id']);
  61. $oVectorStore = BxDolAI::getVectorStoreInstance($aVectorStore['id']);
  62. $oEmbedder = BxDolAI::getModelInstance($aVectorStore['embedding_provider_id']);
  63. $sError = '';
  64. if ($oEmbedder && $oVectorStore) {
  65. try {
  66. $aSettings = $r['settings'] ? json_decode($r['settings'], true) : [];
  67. $aMetadata = $r['metadata'] ? json_decode($r['metadata'], true) : [];
  68. $documents = NeuronAI\RAG\DataLoader\StringDataLoader::for($r['content'])
  69. ->withSplitter(
  70. new NeuronAI\RAG\Splitter\DelimiterTextSplitter(
  71. maxLength: isset($aSettings['chunk_size']) ? $aSettings['chunk_size'] : 512,
  72. separator: isset($aSettings['delimiter']) ? $aSettings['delimiter'] : '.',
  73. wordOverlap: isset($aSettings['overlap']) ? $aSettings['overlap'] : 2
  74. )
  75. )
  76. ->getDocuments();
  77. foreach ($documents as $document) {
  78. $document->sourceType = $r['vector_store_id'];
  79. $document->sourceName = $r['id'];
  80. if (!empty($aMetadata))
  81. $document->metadata = $aMetadata;
  82. }
  83. $oVectorStore->addDocuments(
  84. $oEmbedder->embedDocuments($documents)
  85. );
  86. }
  87. catch (Exception $e) {
  88. $sError = "Error process vector store data ({$r['id']}): " . $e->getMessage();
  89. }
  90. if (!$sError)
  91. BxDolAiQuery::updateVectorStoreDataStatus($r['id'], 'ready');
  92. }
  93. else {
  94. $sError = "Can't process vector store data with id {$r['id']}, missing embedder({$aVectorStore['embedding_provider_id']}) or vector store({$aVectorStore['id']}) instance";
  95. }
  96. if ($sError) {
  97. bx_log('sys_agents', $sError, BX_LOG_ERR);
  98. BxDolAiQuery::updateVectorStoreDataStatus($r['id'], 'error');
  99. }
  100. }
  101. }
  102. }
  103. }
  104. /** @} */