File size: 5,623 Bytes
ef19859 42279cd ef19859 42279cd 6036494 ef19859 54c3758 ef19859 9f72f91 ef19859 6036494 ef19859 6036494 ef19859 6036494 ef19859 6036494 ef19859 9f72f91 ef19859 5935bc1 ef19859 7323674 6036494 5935bc1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 |
from langchain_community.document_loaders import TextLoader
from langchain_community.docstore.document import Document
from langchain.text_splitter import CharacterTextSplitter, RecursiveCharacterTextSplitter
from langchain_community.vectorstores import Chroma
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.retrievers import BM25Retriever
from langchain_community.llms import OpenAI
from langchain_openai import ChatOpenAI
from langchain.chains import RetrievalQA
from langchain.schema import AIMessage, HumanMessage
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
import os
def split_with_source(text, source):
splitter = CharacterTextSplitter(
separator = "\n",
chunk_size = 512,
chunk_overlap = 128,
add_start_index = True,
)
documents = splitter.create_documents([text])
# print(documents)
for doc in documents:
doc.metadata["source"] = source
# print(doc.metadata)
return documents
def count_files_in_folder(folder_path):
# Kiểm tra xem đường dẫn thư mục có tồn tại không
if not os.path.isdir(folder_path):
print("Đường dẫn không hợp lệ.")
return None
# Sử dụng os.listdir() để lấy danh sách các tập tin và thư mục trong thư mục
files = os.listdir(folder_path)
# Đếm số lượng tập tin trong danh sách
file_count = len(files)
return file_count
def get_document_from_raw_text():
documents = [Document(page_content="", metadata={'source': 0})]
files = os.listdir(os.path.join(os.getcwd(), "raw_data"))
# print(files)
for i in files:
file_path = i
with open(os.path.join(os.path.join(os.getcwd(), "raw_data"),file_path), 'r', encoding="utf-8") as file:
# Xử lý bằng text_spliter
# Tiền xử lý văn bản
content = file.read().replace('\n\n', "\n")
# content = ''.join(content.split('.'))
new_doc = content
texts = split_with_source(new_doc, i)
documents = documents + texts
##Xử lý mỗi khi xuống dòng
# for line in file:
# # Loại bỏ khoảng trắng thừa và ký tự xuống dòng ở đầu và cuối mỗi dòng
# line = line.strip()
# documents.append(Document(page_content=line, metadata={"source": i}))
# print(documents)
return documents
def load_the_embedding_retrieve(is_ready = False, k = 3, model= 'sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'):
embeddings = HuggingFaceEmbeddings(model_name=model)
if is_ready:
retriever = Chroma(persist_directory=os.path.join(os.getcwd(), "Data"), embedding_function=embeddings).as_retriever(
search_kwargs={"k": k}
)
else:
documents = get_document_from_raw_text()
print(type(documents))
retriever = Chroma.from_documents(documents, embeddings).as_retriever(
search_kwargs={"k": k}
)
return retriever
def load_the_bm25_retrieve(k = 3):
documents = get_document_from_raw_text()
bm25_retriever = BM25Retriever.from_documents(documents)
bm25_retriever.k = k
return bm25_retriever
def get_qachain(llm_name = "gpt-3.5-turbo-0125", chain_type = "stuff", retriever = None, return_source_documents = True):
llm = ChatOpenAI(temperature=0,
model_name=llm_name)
return RetrievalQA.from_chain_type(llm=llm,
chain_type=chain_type,
retriever=retriever,
return_source_documents=return_source_documents)
def summarize_messages(demo_ephemeral_chat_history, llm):
stored_messages = demo_ephemeral_chat_history.messages
if len(stored_messages) == 0:
return False
summarization_prompt = ChatPromptTemplate.from_messages(
[
MessagesPlaceholder(variable_name="chat_history"),
(
"user", os.environ['SUMARY_MESSAGE_PROMPT'],
),
]
)
summarization_chain = summarization_prompt | llm
summary_message = summarization_chain.invoke({"chat_history": stored_messages})
demo_ephemeral_chat_history.clear()
demo_ephemeral_chat_history.add_message(summary_message)
return demo_ephemeral_chat_history
def get_question_from_summarize(summary, question, llm):
new_qa_prompt = ChatPromptTemplate.from_messages([
("system", os.environ['NEW_QUESTION_PROMPT']),
("human",
'''
Sumary: {summary}
Question: {question}
Output:
'''
)
]
)
new_qa_chain = new_qa_prompt | llm
return new_qa_chain.invoke({'summary': summary, 'question': question}).content
def get_final_answer(question, context, chat_history, prompt, llm):
qa_prompt = ChatPromptTemplate.from_messages(
[
MessagesPlaceholder("chat_history"),
("system", prompt),
("human", '''
Context: {context}
Question: {question}
Output:
'''),
]
)
answer_chain = qa_prompt | llm
answer = answer_chain.invoke({'question': question, 'context': context, 'chat_history': chat_history})
return answer.content
def process_llm_response(llm_response):
print(llm_response['result'])
print('\n\nSources:')
for source in llm_response["source_documents"]:
print(source.metadata['source'])
|