File size: 16,417 Bytes
f0798cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3c68453
f0798cc
 
 
 
 
 
3c68453
 
 
 
 
 
7ebdef9
 
f0798cc
 
 
 
 
 
 
 
 
 
 
 
7ebdef9
48537c4
5b8c5fb
3c68453
 
f0798cc
3c68453
f0798cc
3c68453
 
 
 
 
f0798cc
3c68453
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7ebdef9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f0798cc
3c68453
f0798cc
3c68453
 
f0798cc
3c68453
 
f0798cc
3c68453
7ebdef9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3c68453
 
 
 
f0798cc
3c68453
f0798cc
3c68453
 
 
 
 
 
 
 
 
f0798cc
3c68453
f0798cc
3c68453
 
 
 
 
 
f0798cc
3c68453
 
 
 
f0798cc
3c68453
f0798cc
3c68453
f0798cc
3c68453
f0798cc
3c68453
 
 
f0798cc
3c68453
 
f0798cc
3c68453
 
f0798cc
3c68453
 
 
 
f0798cc
3c68453
f0798cc
3c68453
 
 
 
 
 
f0798cc
3c68453
f0798cc
3c68453
f0798cc
3c68453
 
f0798cc
3c68453
f0798cc
3c68453
f0798cc
3c68453
f0798cc
 
 
 
3c68453
f0798cc
 
 
 
 
 
 
 
 
 
3c68453
f0798cc
 
 
 
 
 
 
 
3c68453
f0798cc
3c68453
f0798cc
3c68453
 
f0798cc
3c68453
 
 
 
 
 
f0798cc
3c68453
f0798cc
3c68453
f0798cc
3c68453
 
 
 
 
48537c4
3c68453
 
 
f0798cc
 
 
 
 
 
 
 
 
 
 
 
 
3c68453
f0798cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3c68453
f0798cc
3c68453
f0798cc
 
 
 
3c68453
 
 
 
f0798cc
3c68453
f0798cc
6b8138e
 
 
 
f0798cc
 
 
 
 
e515aa4
f0798cc
 
3c68453
6b8138e
 
 
 
 
 
 
 
 
 
 
f0798cc
 
 
 
 
a3de190
f0798cc
a3de190
3c68453
 
a3de190
3c68453
 
fac3d77
3c68453
 
a3de190
3c68453
 
 
f0798cc
3c68453
fac3d77
f0798cc
 
fac3d77
f0798cc
 
3c68453
 
 
f0798cc
 
 
 
 
 
 
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
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
import openai
import os
import pdfplumber
from langchain.chains.mapreduce import MapReduceChain
from langchain.text_splitter import CharacterTextSplitter
from langchain.chains.summarize import load_summarize_chain
from langchain.chat_models import ChatOpenAI
from langchain.document_loaders import UnstructuredFileLoader
from langchain.prompts import PromptTemplate
import logging
import json
from typing import List
import mimetypes
import validators
import requests
import tempfile
from langchain.chains import create_extraction_chain
from GoogleNews import GoogleNews
import pandas as pd
import requests
import gradio as gr
import re
from langchain.document_loaders import WebBaseLoader
from langchain.chains.combine_documents.stuff import StuffDocumentsChain
from transformers import pipeline
import plotly.express as px
from langchain.document_loaders.csv_loader import CSVLoader
from langchain.chains.llm import LLMChain
import yfinance as yf
import pandas as pd
import nltk
from nltk.tokenize import sent_tokenize
from openai import OpenAI


class KeyValueExtractor:

    def __init__(self):

        """
        Initialize the ContractSummarizer object.

        Parameters:
            pdf_file_path (str): The path to the input PDF file.
        """
        self.model = "facebook/bart-large-mnli"
        self.client = OpenAI()
        
        
    def get_url(self,keyword):
      return f"https://finance.yahoo.com/quote/{keyword}?p={keyword}"

    def get_each_link_summary(self,url):

      loader = WebBaseLoader(url)
      docs = loader.load()
      text_splitter = CharacterTextSplitter.from_tiktoken_encoder(
      chunk_size=3000, chunk_overlap=200
        )

      # Split the documents into chunks
      split_docs = text_splitter.split_documents(docs)

      # Prepare the prompt template for summarization
      prompt_template = """The give text is Finance Stock Details for one company i want to get values for
      Previous Close : [value]
      Open : [value]
      Bid : [value]
      Ask : [value]
      Day's Range : [value]
      52 Week Range : [value]
      Volume : [value]
      Avg. Volume : [value]
      Market Cap : [value]
      Beta (5Y Monthly) : [value]
      PE Ratio (TTM) : [value]
      EPS (TTM) : [value]
      Earnings Date : [value]
      Forward Dividend & Yield : [value]
      Ex-Dividend Date : [value]
      1y Target Est : [value]
      these details form that and Write a abractive summary about those details:
      Given Text: {text}
      CONCISE SUMMARY:"""
      prompt = PromptTemplate.from_template(prompt_template)

      # Prepare the template for refining the summary with additional context
      refine_template = (
          "Your job is to produce a final summary\n"
          "We have provided an existing summary up to a certain point: {existing_answer}\n"
          "We have the opportunity to refine the existing summary"
          "(only if needed) with some more context below.\n"
          "------------\n"
          "{text}\n"
          "------------\n"
          "Given the new context, refine the original summary"
          "If the context isn't useful, return the original summary."
      )
      refine_prompt = PromptTemplate.from_template(refine_template)

      # Load the summarization chain using the ChatOpenAI language model
      chain = load_summarize_chain(
          llm = ChatOpenAI(temperature=0),
          chain_type="refine",
          question_prompt=prompt,
          refine_prompt=refine_prompt,
          return_intermediate_steps=True,
          input_key="input_documents",
          output_key="output_text",
      )

      # Generate the refined summary using the loaded summarization chain
      result = chain({"input_documents": split_docs}, return_only_outputs=True)
      print(result["output_text"])

      return result["output_text"]

    def one_day_summary(self,content) -> None:
        conversation = [
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": f"i want detailed Summary from given finance details. i want information like what happen today comparing last day good or bad Bullish or Bearish like these details i want summary. content in backticks.```{content}```."}
        ]
        
        # Call OpenAI GPT-3.5-turbo
        chat_completion = self.client.chat.completions.create(
            model = "gpt-3.5-turbo",
            messages = conversation,
            max_tokens=1000,
            temperature=0
        )
        
        response = chat_completion.choices[0].message.content
        return response   

        # # Use OpenAI's Completion API to analyze the text and extract key-value pairs
        # response = openai.Completion.create(
        #   engine="text-davinci-003",  # You can choose a different engine as well
        #   temperature = 0,
        #   prompt=f"i want detailed Summary from given finance details. i want information like what happen today comparing last day good or bad Bullish or Bearish like these details i want summary. content in backticks.```{content}```.",
        #   max_tokens=1000 # You can adjust the length of the response
        # )
        
        # # Extract and return the chatbot's reply
        # result = response['choices'][0]['text'].strip()
        # print(result)
        # return result

    def extract_key_value_pair(self,content) -> None:

        """
        Extract key-value pairs from the refined summary.

        Prints the extracted key-value pairs.
        """

        try:
            conversation = [
                {"role": "system", "content": "You are a helpful assistant."},
                {"role": "user", "content": f"Get maximum count meaningfull key value pairs. content in backticks.```{content}```."}
            ]
            
            # Call OpenAI GPT-3.5-turbo
            chat_completion = self.client.chat.completions.create(
                model = "gpt-3.5-turbo",
                messages = conversation,
                max_tokens=1000,
                temperature=0
            )
            response = chat_completion.choices[0].message.content
            return response            
          # # Use OpenAI's Completion API to analyze the text and extract key-value pairs
          # response = openai.Completion.create(
          #     engine="text-davinci-003",  # You can choose a different engine as well
          #     temperature = 0,
          #     prompt=f"Get maximum count meaningfull key value pairs. content in backticks.```{content}```.",
          #     max_tokens=1000 # You can adjust the length of the response
          # )

          # # Extract and return the chatbot's reply
          # result = response['choices'][0]['text'].strip()
          # return result
        except Exception as e:
            # If an error occurs during the key-value extraction process, log the error
            logging.error(f"Error while extracting key-value pairs: {e}")
            print("Error:", e)

    def analyze_sentiment_for_graph(self, text):

        pipe = pipeline("zero-shot-classification", model=self.model)
        label=["Positive", "Negative", "Neutral"]
        result = pipe(text, label)
        sentiment_scores = {
            result['labels'][0]: result['scores'][0],
            result['labels'][1]: result['scores'][1],
            result['labels'][2]: result['scores'][2]
        }
        return sentiment_scores

    def display_graph(self,text):

        sentiment_scores = self.analyze_sentiment_for_graph(text)
        labels = sentiment_scores.keys()
        scores = sentiment_scores.values()
        fig = px.bar(x=scores, y=labels, orientation='h', color=labels, color_discrete_map={"Negative": "red", "Positive": "green", "Neutral": "gray"})
        fig.update_traces(texttemplate='%{x:.2f}%', textposition='outside')
        fig.update_layout(title="Sentiment Analysis",width=800)

        formatted_pairs = []
        for key, value in sentiment_scores.items():
            formatted_value = round(value, 2)  # Round the value to two decimal places
            formatted_pairs.append(f"{key} : {formatted_value}")

        result_string = '\t'.join(formatted_pairs)

        return fig

    def get_finance_data(self,symbol):

        # Define the stock symbol and date range
        start_date = '2022-08-19'
        end_date = '2023-08-19'

        # Fetch historical OHLC data using yfinance
        data = yf.download(symbol, start=start_date, end=end_date)

        # Select only the OHLC columns
        ohlc_data = data[['Open', 'High', 'Low', 'Close']]

        csv_path = "ohlc_data.csv"
        # Save the OHLC data to a CSV file
        ohlc_data.to_csv(csv_path)
        return csv_path

    def csv_to_dataframe(self,csv_path):

      # Replace 'your_file.csv' with the actual path to your CSV file
      csv_file_path = csv_path
      # Read the CSV file into a DataFrame
      df = pd.read_csv(csv_file_path)
      # Now you can work with the 'df' DataFrame
      return df  # Display the first few rows of the DataFrame

    def save_dataframe_in_text_file(self,df):

        output_file_path = 'output.txt'

        # Convert the DataFrame to a text file
        df.to_csv(output_file_path, sep='\t', index=False)

        return output_file_path

    def csv_loader(self,output_file_path):

        loader = UnstructuredFileLoader(output_file_path, strategy="fast")
        docs = loader.load()

        return docs

    def document_text_spilliter(self,docs):

        """
        Split documents into chunks for efficient processing.

        Returns:
            List[str]: List of split document chunks.
        """

        # Initialize the text splitter with specified chunk size and overlap
        text_splitter = CharacterTextSplitter.from_tiktoken_encoder(
            chunk_size=1000, chunk_overlap=200
        )

        # Split the documents into chunks
        split_docs = text_splitter.split_documents(docs)

        # Return the list of split document chunks
        return split_docs

    def change_bullet_points(self,text):

        nltk.download('punkt')  # Download the sentence tokenizer data (only need to run this once)

        # Example passage
        passage = text

        # Tokenize the passage into sentences
        sentences = sent_tokenize(passage)
        bullet_string = ""
        # Print the extracted sentences
        for sentence in sentences:
            bullet_string+="* "+sentence+"\n"

        return bullet_string

    def one_year_summary(self,keyword):

        csv_path = self.get_finance_data(keyword)
        df = self.csv_to_dataframe(csv_path)
        output_file_path = self.save_dataframe_in_text_file(df)
        docs = self.csv_loader(output_file_path)
        split_docs = self.document_text_spilliter(docs)
        
        prompt_template = """Analyze the Financial Details and Write a abractive quick short summary how the company perform up and down,Bullish/Bearish of the following:
                {text}
                CONCISE SUMMARY:"""
        prompt = PromptTemplate.from_template(prompt_template)

        # Prepare the template for refining the summary with additional context
        refine_template = (
            "Your job is to produce a final summary\n"
            "We have provided an existing summary up to a certain point: {existing_answer}\n"
            "We have the opportunity to refine the existing summary"
            "(only if needed) with some more context below.\n"
            "------------\n"
            "{text}\n"
            "------------\n"
            "Given the new context, refine the original summary"
            "If the context isn't useful, return the original summary."
            "10 line summary is enough"
        )
        refine_prompt = PromptTemplate.from_template(refine_template)

        # Load the summarization chain using the ChatOpenAI language model
        chain = load_summarize_chain(
            llm = ChatOpenAI(temperature=0),
            chain_type="refine",
            question_prompt=prompt,
            refine_prompt=refine_prompt,
            return_intermediate_steps=True,
            input_key="input_documents",
            output_key="output_text",
        )

        # Generate the refined summary using the loaded summarization chain
        result = chain({"input_documents": split_docs}, return_only_outputs=True)
        one_year_perfomance_summary = self.change_bullet_points(result["output_text"])
        # Return the refined summary
        return one_year_perfomance_summary

    def main(self,keyword):


      clean_url = self.get_url(keyword)
      link_summary  =  self.get_each_link_summary(clean_url)
      clean_summary = self.one_day_summary(link_summary)
      key_value = self.extract_key_value_pair(clean_summary)

      return clean_summary, key_value

    def company_names(self,input_text):
        words = input_text.split("-")
        return words[1]

    def gradio_interface(self):

        with gr.Blocks(css="style.css",theme= 'karthikeyan-adople/hudsonhayes-gray') as app:
              gr.HTML("""<center class="darkblue" style='background-color:rgb(0,1,36); text-align:center;padding:25px;'><center><h1 class ="center">
                <img src="file=logo.png" height="110px" width="280px"></h1></center>
                  <br><h1 style="color:#fff"> Finance Tool for Investors </h1></center>""")
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150, ):
                  input_news = gr.Textbox(label="Company Name")
              with gr.Accordion("List_of_Companies", open = False):
                  with gr.Row(elem_id="col-container"):
                    with gr.Column(scale=1.0, min_width=150 ):
                          gr.Examples(
                            [["Apple Inc. - AAPL"], ["Microsoft Corporation - MSFT"],["Amazon.com Inc. - AMZN"],["Facebook Inc. - FB"],["Tesla Inc. - TSLA"]],
                            [input_news],
                            input_news,
                            fn=self.company_names,
                            cache_examples=True,
                        )

              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150):
                  analyse = gr.Button("Analyse")
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=0.50, min_width=150):
                  result_summary = gr.Textbox(label="Summary For Last Day Perfomance", lines = 12)
                with gr.Column(scale=0.50, min_width=150):
                  key_value_pair_result = gr.Textbox(label="Discussed Topics", lines = 12)
              with gr.Row(elem_id="col-container"):
                  with gr.Column(scale=1.0, min_width=0):
                      plot_for_day =gr.Plot(label="Sentiment for Last Day", size=(500, 600))
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150):
                  analyse_sentiment = gr.Button("Analyse Sentiment For Last Day")
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150, ):
                  one_year_summary = gr.Textbox(label="Summary For One Year Perfomance",lines = 12)
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150):
                  one_year = gr.Button("Analyse One Year Summary")
              with gr.Row(elem_id="col-container"):
                  with gr.Column(scale=1.0, min_width=0):
                      plot_for_year =gr.Plot(label="Sentiment for One Year", size=(500, 600))
              with gr.Row(elem_id="col-container"):
                with gr.Column(scale=1.0, min_width=150):
                  analyse_sentiment_for_year = gr.Button("Analyse Sentiment For One Year")

              analyse.click(self.main, input_news, [result_summary,key_value_pair_result])
              analyse_sentiment.click(self.display_graph,result_summary,[plot_for_day])
              one_year.click(self.one_year_summary,input_news,one_year_summary)
              analyse_sentiment_for_year.click(self.display_graph,one_year_summary,[plot_for_year])

        app.launch(debug=True)

if __name__ == "__main__":

  text_process = KeyValueExtractor()
  text_process.gradio_interface()