Merge pull request #161 from enoch3712/139-change-name-documentloader…

…documentai

name convention changed with warning

extract_thinker/__init__.py

@@ -22,7 +22,10 @@
 from .document_loader.document_loader_doc2txt import DocumentLoaderDoc2txt
 from .document_loader.document_loader_aws_textract import DocumentLoaderAWSTextract
 from .document_loader.document_loader_llm_image import DocumentLoaderLLMImage
-from .document_loader.document_loader_google_document_ai import DocumentLoaderDocumentAI
+from .document_loader.document_loader_google_document_ai import (
+    DocumentLoaderGoogleDocumentAI,
+    DocumentLoaderDocumentAI,
+)
 
 __all__ = [
     'Extractor',
@@ -39,6 +42,7 @@
     'DocumentLoaderTxt',
     'DocumentLoaderDoc2txt',
     'DocumentLoaderAWSTextract',
+    'DocumentLoaderGoogleDocumentAI',
     'DocumentLoaderDocumentAI',
     'Classification',
     'ClassificationResponse',

extract_thinker/document_loader/document_loader_google_document_ai.py

@@ -10,6 +10,7 @@
 from cachetools.keys import hashkey
 from cachetools import cachedmethod
 from operator import attrgetter
+import warnings
 
 class Config:
     def __init__(
@@ -21,7 +22,7 @@ def __init__(
         self.page_range = page_range
 
 
-class DocumentLoaderDocumentAI(CachedDocumentLoader):
+class DocumentLoaderGoogleDocumentAI(CachedDocumentLoader):
     """Loader for documents using Google Document AI."""
 
     SUPPORTED_FORMATS = [
@@ -179,3 +180,14 @@ def _get_page_key_value_pairs(page: documentai.Document.Page) -> List[Dict[str,
             for kv_pair in page.key_value_pairs
         ]
 
+
+# Create an alias with deprecation warning
+class DocumentLoaderDocumentAI(DocumentLoaderGoogleDocumentAI):
+    def __init__(self, *args, **kwargs):
+        warnings.warn(
+            "DocumentLoaderDocumentAI is deprecated and will be removed in 0.1.0"
+            "Use DocumentLoaderGoogleDocumentAI instead.",
+            DeprecationWarning,
+            stacklevel=2
+        )
+        super().__init__(*args, **kwargs)

tests/test_document_loader_google_document_ai.py

@@ -1,21 +1,40 @@
 import os
 import pytest
+import warnings
 from dotenv import load_dotenv
-from extract_thinker.document_loader.document_loader_google_document_ai import DocumentLoaderDocumentAI
+from extract_thinker.document_loader.document_loader_google_document_ai import (
+    DocumentLoaderGoogleDocumentAI,
+    DocumentLoaderDocumentAI
+)
 from tests.test_document_loader_base import BaseDocumentLoaderTest
 
 load_dotenv()
 
 class TestDocumentLoaderGoogleDocumentAI(BaseDocumentLoaderTest):
     @pytest.fixture
     def loader(self):
-        return DocumentLoaderDocumentAI(
+        return DocumentLoaderGoogleDocumentAI(
             project_id=os.getenv("DOCUMENTAI_PROJECT_ID"),
             location=os.getenv("DOCUMENTAI_LOCATION"),
             processor_id=os.getenv("DOCUMENTAI_PROCESSOR_ID"),
             credentials=os.getenv("DOCUMENTAI_GOOGLE_CREDENTIALS")
         )
 
+    def test_deprecation_warning(self):
+        """Test that using old class name raises deprecation warning"""
+        with pytest.warns(DeprecationWarning) as record:
+            DocumentLoaderDocumentAI(
+                project_id=os.getenv("DOCUMENTAI_PROJECT_ID"),
+                location=os.getenv("DOCUMENTAI_LOCATION"),
+                processor_id=os.getenv("DOCUMENTAI_PROCESSOR_ID"),
+                credentials=os.getenv("DOCUMENTAI_GOOGLE_CREDENTIALS")
+            )
+
+        # Verify the warning message
+        assert len(record) == 1
+        assert "DocumentLoaderDocumentAI is deprecated" in str(record[0].message)
+        assert "Use DocumentLoaderGoogleDocumentAI instead" in str(record[0].message)
+
     @pytest.fixture
     def test_file_path(self):
         current_dir = os.path.dirname(os.path.abspath(__file__))