|
3 | 3 | from .llm import LLM |
4 | 4 | from .document_loader.document_loader import DocumentLoader |
5 | 5 | from .document_loader.cached_document_loader import CachedDocumentLoader |
6 | | -from .document_loader.document_loader_tesseract import DocumentLoaderTesseract |
| 6 | +from .document_loader.document_loader_tesseract import DocumentLoaderTesseract, TesseractConfig |
7 | 7 | from .document_loader.document_loader_spreadsheet import DocumentLoaderSpreadSheet |
8 | | -from .document_loader.document_loader_azure_document_intelligence import DocumentLoaderAzureForm |
| 8 | +from .document_loader.document_loader_azure_document_intelligence import DocumentLoaderAzureForm, AzureConfig |
9 | 9 | from .document_loader.document_loader_pypdf import DocumentLoaderPyPdf, PyPDFConfig |
10 | | -from .document_loader.document_loader_pdfplumber import DocumentLoaderPdfPlumber |
11 | | -from .document_loader.document_loader_beautiful_soup import DocumentLoaderBeautifulSoup |
12 | | -from .document_loader.document_loader_markitdown import DocumentLoaderMarkItDown |
13 | | -from .document_loader.document_loader_docling import DocumentLoaderDocling |
| 10 | +from .document_loader.document_loader_pdfplumber import DocumentLoaderPdfPlumber, PDFPlumberConfig |
| 11 | +from .document_loader.document_loader_beautiful_soup import DocumentLoaderBeautifulSoup, BeautifulSoupConfig |
| 12 | +from .document_loader.document_loader_markitdown import DocumentLoaderMarkItDown, MarkItDownConfig |
| 13 | +from .document_loader.document_loader_docling import DocumentLoaderDocling, DoclingConfig |
14 | 14 | from .models.classification import Classification |
15 | 15 | from .models.classification_response import ClassificationResponse |
16 | 16 | from .process import Process |
|
21 | 21 | from .models.splitting_strategy import SplittingStrategy |
22 | 22 | from .models.completion_strategy import CompletionStrategy |
23 | 23 | from .batch_job import BatchJob |
24 | | -from .document_loader.document_loader_txt import DocumentLoaderTxt |
25 | | -from .document_loader.document_loader_doc2txt import DocumentLoaderDoc2txt |
26 | | -from .document_loader.document_loader_aws_textract import DocumentLoaderAWSTextract |
27 | | -from .document_loader.document_loader_llm_image import DocumentLoaderLLMImage |
28 | | -from .document_loader.document_loader_data import DocumentLoaderData |
| 24 | +from .document_loader.document_loader_txt import DocumentLoaderTxt, TxtConfig |
| 25 | +from .document_loader.document_loader_doc2txt import DocumentLoaderDoc2txt, Doc2txtConfig |
| 26 | +from .document_loader.document_loader_aws_textract import DocumentLoaderAWSTextract, TextractConfig |
| 27 | +from .document_loader.document_loader_llm_image import DocumentLoaderLLMImage, LLMImageConfig |
| 28 | +from .document_loader.document_loader_data import DocumentLoaderData, DataLoaderConfig |
29 | 29 | from .document_loader.document_loader_google_document_ai import ( |
30 | 30 | DocumentLoaderGoogleDocumentAI, |
31 | 31 | DocumentLoaderDocumentAI, |
| 32 | + GoogleDocAIConfig |
32 | 33 | ) |
33 | 34 | from .warning import filter_pydantic_v2_warnings |
34 | 35 | from .document_loader.document_loader_mistral_ocr import DocumentLoaderMistralOCR, MistralOCRConfig |
|
40 | 41 | 'DocumentLoader', |
41 | 42 | 'CachedDocumentLoader', |
42 | 43 | 'DocumentLoaderTesseract', |
| 44 | + 'TesseractConfig', |
43 | 45 | 'DocumentLoaderSpreadSheet', |
44 | 46 | 'DocumentLoaderAzureForm', |
| 47 | + 'AzureConfig', |
45 | 48 | 'DocumentLoaderPyPdf', |
46 | 49 | 'PyPDFConfig', |
47 | 50 | 'DocumentLoaderPdfPlumber', |
| 51 | + 'PDFPlumberConfig', |
48 | 52 | 'DocumentLoaderBeautifulSoup', |
| 53 | + 'BeautifulSoupConfig', |
49 | 54 | 'DocumentLoaderLLMImage', |
| 55 | + 'LLMImageConfig', |
50 | 56 | 'DocumentLoaderTxt', |
| 57 | + 'TxtConfig', |
51 | 58 | 'DocumentLoaderDoc2txt', |
| 59 | + 'Doc2txtConfig', |
52 | 60 | 'DocumentLoaderAWSTextract', |
| 61 | + 'TextractConfig', |
53 | 62 | 'DocumentLoaderGoogleDocumentAI', |
54 | 63 | 'DocumentLoaderDocumentAI', |
55 | | - 'DocumentLoaderMarkItDown', |
56 | | - 'DocumentLoaderData', |
| 64 | + 'GoogleDocAIConfig', |
| 65 | + 'DocumentLoaderMarkItDown', |
| 66 | + 'MarkItDownConfig', |
| 67 | + 'DocumentLoaderData', |
| 68 | + 'DataLoaderConfig', |
57 | 69 | 'Classification', |
58 | 70 | 'CompletionStrategy', |
59 | 71 | 'DocumentLoaderDocling', |
| 72 | + 'DoclingConfig', |
60 | 73 | 'ClassificationResponse', |
61 | 74 | 'Process', |
62 | 75 | 'ClassificationStrategy', |
|
65 | 78 | 'TextSplitter', |
66 | 79 | 'Contract', |
67 | 80 | 'SplittingStrategy', |
68 | | - 'BatchJob', |
| 81 | + 'BatchJob', |
69 | 82 | 'DocumentLoaderMistralOCR', |
70 | 83 | 'MistralOCRConfig', |
71 | 84 | ] |
0 commit comments