Add new loaders to constants.py

This commit is contained in:
Gabriel Luiz Freitas Almeida 2023-11-23 11:48:14 -03:00
commit 7a521fa6cb

View file

@ -1,3 +1,5 @@
from typing import Any, Dict, List
OPENAI_MODELS = [
"text-davinci-003",
"text-davinci-002",
@ -59,3 +61,117 @@ DIRECT_TYPES = [
"code",
"NestedDict",
]
LOADERS_INFO: List[Dict[str, Any]] = [
{
"loader": "AirbyteJSONLoader",
"name": "Airbyte JSON (.jsonl)",
"import": "langchain.document_loaders.AirbyteJSONLoader",
"defaultFor": ["jsonl"],
"allowdTypes": ["jsonl"],
},
{
"loader": "JSONLoader",
"name": "JSON (.json)",
"import": "langchain.document_loaders.JSONLoader",
"defaultFor": ["json"],
"allowdTypes": ["json"],
},
{
"loader": "BSHTMLLoader",
"name": "BeautifulSoup4 HTML (.html, .htm)",
"import": "langchain.document_loaders.BSHTMLLoader",
"allowdTypes": ["html", "htm"],
},
{
"loader": "CSVLoader",
"name": "CSV (.csv)",
"import": "langchain.document_loaders.CSVLoader",
"defaultFor": ["csv"],
"allowdTypes": ["csv"],
},
{
"loader": "CoNLLULoader",
"name": "CoNLL-U (.conllu)",
"import": "langchain.document_loaders.CoNLLULoader",
"defaultFor": ["conllu"],
"allowdTypes": ["conllu"],
},
{
"loader": "EverNoteLoader",
"name": "EverNote (.enex)",
"import": "langchain.document_loaders.EverNoteLoader",
"defaultFor": ["enex"],
"allowdTypes": ["enex"],
},
{
"loader": "FacebookChatLoader",
"name": "Facebook Chat (.json)",
"import": "langchain.document_loaders.FacebookChatLoader",
"allowdTypes": ["json"],
},
{
"loader": "OutlookMessageLoader",
"name": "Outlook Message (.msg)",
"import": "langchain.document_loaders.OutlookMessageLoader",
"defaultFor": ["msg"],
"allowdTypes": ["msg"],
},
{
"loader": "PyPDFLoader",
"name": "PyPDF (.pdf)",
"import": "langchain.document_loaders.PyPDFLoader",
"defaultFor": ["pdf"],
"allowdTypes": ["pdf"],
},
{
"loader": "STRLoader",
"name": "Subtitle (.str)",
"import": "langchain.document_loaders.STRLoader",
"defaultFor": ["str"],
"allowdTypes": ["str"],
},
{
"loader": "TextLoader",
"name": "Text (.txt)",
"import": "langchain.document_loaders.TextLoader",
"defaultFor": ["txt"],
"allowdTypes": ["txt"],
},
{
"loader": "UnstructuredEmailLoader",
"name": "Unstructured Email (.eml)",
"import": "langchain.document_loaders.UnstructuredEmailLoader",
"defaultFor": ["eml"],
"allowdTypes": ["eml"],
},
{
"loader": "UnstructuredHTMLLoader",
"name": "Unstructured HTML (.html, .htm)",
"import": "langchain.document_loaders.UnstructuredHTMLLoader",
"defaultFor": ["html", "htm"],
"allowdTypes": ["html", "htm"],
},
{
"loader": "UnstructuredMarkdownLoader",
"name": "Unstructured Markdown (.md)",
"import": "langchain.document_loaders.UnstructuredMarkdownLoader",
"defaultFor": ["md"],
"allowdTypes": ["md"],
},
{
"loader": "UnstructuredPowerPointLoader",
"name": "Unstructured PowerPoint (.pptx)",
"import": "langchain.document_loaders.UnstructuredPowerPointLoader",
"defaultFor": ["pptx"],
"allowdTypes": ["pptx"],
},
{
"loader": "UnstructuredWordLoader",
"name": "Unstructured Word (.docx)",
"import": "langchain.document_loaders.UnstructuredWordLoader",
"defaultFor": ["docx"],
"allowdTypes": ["docx"],
},
]