カテゴリー: Python

import os, logging, sys
from pathlib import Path
from PyQt6.QtWidgets import QLabel,QWidget,QApplication,QTextEdit,QLineEdit,QPushButton
from PyQt6.QtCore import Qt
from llama_index import download_loader,LLMPredictor, GPTSimpleVectorIndex, ServiceContext
from langchain import OpenAI
import datetime

class LlamaIndex(QWidget):
    def __init__(self):
        super().__init__()
        self.setWindowTitle("LlamaIndex")
        self.setGeometry(100,100,480,480)
        self.setStyleSheet('background-color: #483D8B')
        self.setAcceptDrops(True)
        
        file = QLabel('FILE',self)
        file.setGeometry(10,15,34,16)
        file.setStyleSheet('color: #FFFFFF; font-size: 14pt;')
        file.setAlignment(Qt.AlignmentFlag.AlignCenter)
        
        self.input= QLineEdit('',self)
        self.input.setGeometry(55,10,355,25)
        self.input.setAcceptDrops(True)
        
        idxBtn = QPushButton('IDX作成',self)
        idxBtn.setGeometry(420,10,50,25)
        idxBtn.setStyleSheet('color: #FFFFFF; font-size: 12pt;')
        idxBtn.released.connect(self.makeIDX)
        
        idx = QLabel('IDX',self)
        idx.setGeometry(10,45,26,16)
        idx.setStyleSheet('color: #FFFFFF; font-size: 14pt;')
        idx.setAlignment(Qt.AlignmentFlag.AlignCenter)
        
        self.input2= QLineEdit('',self)
        self.input2.setGeometry(55,40,355,25)
        self.input2.setAcceptDrops(True)
        
        self.idxBtn2 = QPushButton('IDX読込',self)
        self.idxBtn2.setGeometry(420,40,50,25)
        self.idxBtn2.setStyleSheet('color: #FFFFFF; font-size: 12pt;')
        self.idxBtn2.released.connect(self.loadIDX)
        
        send = QPushButton('送信',self)
        send.setGeometry(420,70,50,25)
        send.setStyleSheet('color: #FFFFFF; font-size: 14pt;')
        send.released.connect(self.sendQuestion)
        
        clear = QPushButton('CL',self)
        clear.setGeometry(420,100,50,25)
        clear.setStyleSheet('color: #FFFFFF; font-size: 14pt;')
        clear.released.connect(self.clear)
            
        self.questionInput = QTextEdit('',self)
        self.questionInput.setGeometry(10,70,400,95)
        self.questionInput.setAcceptDrops(False)
        
        self.output = QTextEdit('',self)
        self.output.setGeometry(10,170,460,305)
        self.output.setAcceptDrops(False)
        
        # APIキーを環境変数から取得
        apiKey = os.getenv("CHATGPT_API_KEY")
        os.environ["OPENAI_API_KEY"] = apiKey

        # ログレベルの設定(DEBUG)
        logging.basicConfig(stream=sys.stdout, level=logging.DEBUG, force=True)
        logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

    def makeIDX(self):
        # インデックスの作成および保存
        llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-embedding-ada-002"))
        service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)
        
        data_path = self.input.text()
        data_path2 = data_path.replace("'", "") # 拡張子判定用
        data_path3 = Path(data_path2) # loader用

        if data_path2.endswith('.csv'):
            SimpleCSVReader = download_loader("SimpleCSVReader")
            loader = SimpleCSVReader()
        elif data_path2.endswith('.pdf'):
            PDFReader = download_loader("PDFReader")
            loader = PDFReader()
        else:
            print('ファイルがcsv,pdfではありません')
            sys.exit()

        nodes = loader.load_data(file=data_path3)

        index = GPTSimpleVectorIndex.from_documents(nodes, service_context=service_context)
        
        now = datetime.datetime.now()
        formatted_time = now.strftime('%y%m%d_%H%M%S')

        index_file = "/AI/LlamaIndex/index/" + formatted_time + "_index.json"
        index.save_to_disk(index_file)
        
        self.input2.setText(index_file)
        self.idxBtn2.click()

    def loadIDX(self):
        # インデックスの読込
        llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-davinci-003", max_tokens=3500))
        service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)
        
        index_path = self.input2.text()

        self.index = GPTSimpleVectorIndex.load_from_disk(save_path= index_path, service_context=service_context)
        self.output.setText("IDX読込完了")
        
    def sendQuestion(self):
        question = self.questionInput.toPlainText()
        response = self.index.query(question)
        
        print(response)
        self.output.setText(str(response))

    def clear(self):
        self.input.clear()
        self.input2.clear()
        self.questionInput.clear()
        self.output.clear()

app = QApplication(sys.argv)
window = LlamaIndex()
window.show()
app.exec()

import os, logging, sys
from pathlib import Path
# llama_index改変版
from llama_index_fork import download_loader,LLMPredictor, GPTSimpleVectorIndex, ServiceContext
from langchain import OpenAI
from mylib import MySimpleCSVReader
import pdb

# APIキーを環境変数から取得
apiKey = os.getenv("CHATGPT_API_KEY")
os.environ["OPENAI_API_KEY"] = apiKey

# ログレベルの設定(DEBUG)
logging.basicConfig(stream=sys.stdout, level=logging.DEBUG, force=True)
logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

# インデックスの作成および保存
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-embedding-ada-002"))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

print("ファイルパスを入力して下さい")
file_path = input()
file_path2 = file_path.replace("'", "") # 拡張子判定用
file_path3 = Path(file_path2) # loader用

if file_path2.endswith('.csv'):
	# SimpleCSVReader = download_loader("SimpleCSVReader")
	loader = MySimpleCSVReader.MySimpleCSVReader()
elif file_path2.endswith('.pdf'):
	PDFReader = download_loader("PDFReader")
	loader = PDFReader()
else:
	print('ファイルがcsv,pdfではありません')
	sys.exit()

# documentsの型は、List[Document]
documents = loader.load_data(file=file_path3)

print("documentsの内容")
num = 1
for doc in documents:
    print(str(num) + ": " + str(doc) + "\n")
    num = num + 1

# ブレークポイント設定
pdb.set_trace()

index = GPTSimpleVectorIndex.from_documents(documents, service_context=service_context)
print("\nindex: " + str(index) + "\n")

index.save_to_disk('index.json')

# インデックスの読込
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-davinci-003", max_tokens=3500))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

index = GPTSimpleVectorIndex.load_from_disk(save_path="index.json", service_context=service_context)

# 質問(Ctrl+cで終了)
while True :
  print("質問を入力して下さい")
  question = input()
  
  print(index.query(question))
  

# Python3.10の仮想環境envを作成
virtualenv -p python3.10 env

# ライブラリをインストール
./env/bin/pip install ライブラリ名

# 仮想環境envの有効化
source env/bin/activate

# 仮想環境envの無効化
deactivate

import os, logging, sys
from pathlib import Path
from llama_index import download_loader,LLMPredictor, GPTSimpleVectorIndex, ServiceContext
from langchain import OpenAI

# APIキーを環境変数から取得
apiKey = os.getenv("CHATGPT_API_KEY")
os.environ["OPENAI_API_KEY"] = apiKey

# ログレベルの設定(DEBUG)
logging.basicConfig(stream=sys.stdout, level=logging.DEBUG, force=True)
logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

# インデックスの作成および保存
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-embedding-ada-002"))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

print("ファイルパスを入力して下さい")
file_path = input()
file_path2 = file_path.replace("'", "") # 拡張子判定用
file_path3 = Path(file_path2) # loader用

if file_path2.endswith('.csv'):
	SimpleCSVReader = download_loader("SimpleCSVReader")
	loader = SimpleCSVReader()
elif file_path2.endswith('.pdf'):
	PDFReader = download_loader("PDFReader")
	loader = PDFReader()
else:
	print('ファイルがcsv,pdfではありません')
	sys.exit()

nodes = loader.load_data(file=file_path3)

index = GPTSimpleVectorIndex.from_documents(nodes, service_context=service_context)
index.save_to_disk('index.json')

# インデックスの読込
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-davinci-003", max_tokens=3500))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

index = GPTSimpleVectorIndex.load_from_disk(save_path="index.json", service_context=service_context)

# 質問(Ctrl+cで終了)
while True :
  print("質問を入力して下さい")
  question = input()
  
  print(index.query(question))
  

import os, logging, sys
from pathlib import Path
from llama_index import download_loader,LLMPredictor, GPTSimpleVectorIndex, ServiceContext
from langchain import OpenAI

# APIキーを環境変数から取得
apiKey = os.getenv("CHATGPT_API_KEY")
os.environ["OPENAI_API_KEY"] = apiKey

# ログレベルの設定(DEBUG)
logging.basicConfig(stream=sys.stdout, level=logging.DEBUG, force=True)
logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

# インデックスの作成および保存
SimpleCSVReader = download_loader("SimpleCSVReader")

llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-embedding-ada-002"))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

csv_path = Path('fine_tuning_data.csv')

loader = SimpleCSVReader()
documents = loader.load_data(file=csv_path)

index = GPTSimpleVectorIndex.from_documents(documents, service_context=service_context)
index.save_to_disk('fine_tuning_data.json')

# インデックスの読込
llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name="text-davinci-003", max_tokens=3500))
service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)

index = GPTSimpleVectorIndex.load_from_disk(save_path="fine_tuning_data.json", service_context=service_context)

# 質問(Ctrl+cで終了)
while True :
  print("質問を入力して下さい")
  question = input()
  
  print(index.query(question))