Spaces:
Sleeping
Sleeping
File size: 5,282 Bytes
284c80a 382488b 284c80a 720c8eb 284c80a 720c8eb 284c80a 720c8eb 284c80a 4ebb24d 284c80a 720c8eb f520f6e 720c8eb 284c80a bc29c51 c4ee0af b6f4209 e279307 284c80a 39cf970 284c80a 39cf970 284c80a 720c8eb c102673 720c8eb 71df0ea 470d207 c8ad4b9 71df0ea 720c8eb 3d86e91 284c80a 95bc4d4 |
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 |
import datetime
from urllib.request import Request, urlopen
from pypdf import PdfReader
from io import StringIO
import io
import pandas as pd
import os
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
from transformers import pipeline
from openai import OpenAI
from groq import Groq
import time
import json
from openai import OpenAI
openai_key = "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"
client = OpenAI(api_key = openai_key)
desc = pd.read_excel('Descriptor.xlsx',header = None)
desc_list = desc.iloc[:,0].to_list()
def callAzure(prompt,text):
url = "https://Meta-Llama-3-70B-Instruct-fkqip-serverless.eastus2.inference.ai.azure.com"
api_key = "o5yaLhTIvg0s5zuYVInBpyneEZO8oonY"
client = OpenAI(base_url=url, api_key=api_key)
msg = "{} {}".format(prompt, text)
response = client.chat.completions.create(
messages=[
{
"role": "user",
"content": msg,
}
],
model="azureai",
max_tokens = 1000
)
return response.choices[0].message.content
def filter(input_json):
sym = pd.read_excel('symbol.xlsx',header = None)
sym_list = sym.iloc[:,0].to_list()
if input_json['FileURL']==None or input_json['FileURL'].lower()=='null':
return [0,"File_URL"]
if input_json['symbol']== 'null' or input_json['symbol'] not in sym_list:
return [0,"symbol"]
if input_json['TypeofAnnouncement'] not in ['General_Announcements','Outcome','General']:
return [0,"Annoucement"]
if input_json['Descriptor'] not in desc_list:
return [0,"Desc"]
url = 'https://www.bseindia.com/xml-data/corpfiling/AttachLive/'+ input_json['FileURL'].split('Pname=')[-1]
req = Request(url, headers={'User-Agent': 'Mozilla/5.0'})
html = urlopen(req)
cont = html.read()
reader = PdfReader(io.BytesIO(cont))
content = ''
for i in range(len(reader.pages)):
content+= reader.pages[i].extract_text()
document = content
return [1, document]
def summary(input_json):
prompt = pd.read_excel('DescriptorPrompt.xlsx')
promptShort = prompt.iloc[:,1].to_list()
promptLong = prompt.iloc[:,2].to_list()
output = {}
filtering_results = filter(input_json)
if filtering_results[0] == 0:
#return 0
return filtering_results[1]
id = desc_list.index(input_json['Descriptor'])
long_text = filtering_results[1]
long_text = long_text.lstrip()
long_text = long_text.rstrip()
long_text = long_text[:6000]
url = 'https://www.bseindia.com/xml-data/corpfiling/AttachLive/'+ input_json['FileURL'].split('Pname=')[-1]
output["Link to BSE website"] = url
output["Date of time of receiving data from BSE"] = input_json["newsdate"] + "Z"
output["Stock Ticker"] = input_json['symbol']
answer = callAzure("You are an financial expert. Wherever possible, mention the name of the company " + promptShort[id] + " Do not exceed over 400 characters", long_text)
try:
idx = answer.index("\n")
except:
idx = -2
output['Short Summary'] = answer[idx+2:]
answer = callAzure("Make sure the following summary of a news article is not more than 80 words. Rewrite it and make it below 80 words ", output['Short Summary'])
try:
idx = answer.index("\n")
except:
idx = -2
output['Short Summary'] = answer[idx+2:]
prompt = "Provide the main topic of the news article strictly as a tag, using only one or two words, with only the first word capitalized and the rest in lowercase. No additional text or explanation."
answer = callAzure(prompt, output['Short Summary'])
output['Tag'] = answer
prompt = "Generate a precise headline for the news article that includes the name of the company. Be very careful about correctly representing any financial figures mentioned in lakhs and crores. Provide only the headline, with no additional text or explanation."
answer = callAzure(prompt, output['Short Summary'])
output['Headline'] = answer
utc_now = datetime.datetime.utcnow()
ist_now = utc_now.astimezone(datetime.timezone(datetime.timedelta(hours=5, minutes=30)))
Date = ist_now.strftime("%Y-%m-%d")
time = ist_now.strftime("%X")
output['Date and time of data delivery from Skylark'] = Date+"T"+time+"Z"
prompt = "Answer in one word the sentiment of this News out of Positive, Negative or Neutral {}"
output['Sentiment'] = callAzure(prompt, output['Short Summary'])
completion = client.chat.completions.create(
model="gpt-4o",
messages=[
{"role": "system", "content": "You are a financial expert. Help the client with summarizing the financial newsletter. Write the summary in max 500 words. Do not truncate."},
{"role": "user", "content": "{} {}".format(promptLong[id], long_text)}
],
temperature=0,
max_tokens=4000,
)
output['Long summary'] = completion.choices[0].message.content
# response = client.images.generate(
# model="dall-e-3",
# prompt=headline.text,
# size="1024x1024",
# quality="standard",
# n=1
# )
# output["Link to Infographic (data visualization only)] = response.data[0].url
return output |