File size: 4,040 Bytes
196c8fb
 
 
 
 
46f2531
196c8fb
b330d7a
196c8fb
 
 
 
f588f27
196c8fb
 
b330d7a
 
 
 
 
 
 
 
196c8fb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aeb7c17
196c8fb
cd8a160
196c8fb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fd3203e
 
196c8fb
440ebd9
196c8fb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
94cf8b4
196c8fb
 
 
bf94a11
196c8fb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from PyPDF2 import PdfReader,PdfWriter
import gradio as gr
from langchain.embeddings import CohereEmbeddings
from langchain.prompts import PromptTemplate
from langchain import OpenAI
from langchain_cohere import ChatCohere
from langchain.text_splitter import RecursiveCharacterTextSplitter
import os
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

import spacy
spacy.cli.download("en_core_web_md")
nlp = spacy.load('en_core_web_md')


from dotenv import load_dotenv

load_dotenv()


OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
COHERE_API_KEY = os.getenv('COHERE_API_KEY')
text_splitter = RecursiveCharacterTextSplitter(chunk_size = 200, chunk_overlap = 0)
embedding = CohereEmbeddings(model='embed-multilingual-v3.0',cohere_api_key=COHERE_API_KEY)



def recieve_pdf(filename):
    reader = PdfReader(filename)
    writer = PdfWriter()

    for page in reader.pages:
        writer.add_page(page)
    

    with open('processed_file.pdf','wb') as f:
        writer.write(f)

    read = PdfReader('processed_file.pdf')
    extracted_file =[page.extract_text(0) for page in read.pages]       
    extracted_text = ''.join(extracted_file)

    global file
    file = extracted_text

    # summary_prompt_formated = summary_prompt.format(document = extracted_text)

    return 'Document succesfully uploaded'


def chatbot(query,history):
    similarity_array =[]
    embeded_query = embedding.embed_documents([query])
    
    doc = nlp(file)
    sentences_1 = [str(sentence) for sentence in doc.sents]
    embedded_text = embedding.embed_documents(sentences_1)
        


    similarity_score = cosine_similarity(embeded_query,embedded_text)
    similarity_array.append(similarity_score) 

    

    most_similar_index = np.argmax(similarity_array)
    most_similar_documents = sentences_1[most_similar_index]



    splitter_text = text_splitter.split_text(file)
    recursive_embedded_text = embedding.embed_documents(splitter_text)

    most_similar_embed = embedding.embed_documents([most_similar_documents])
    final_similarity_score = cosine_similarity(most_similar_embed,recursive_embedded_text)

    final_similarity_index = np.argmax(final_similarity_score)
    final_document = splitter_text[final_similarity_index]

    prompt_formated = prompt.format(context = final_document, query = query)
    response = llm.invoke(prompt_formated).content


    history.append((query, response))

    
    return '', history

summary_template = """ You an article summarizer and have been provided with this file

{document}

provide a one line summary of the content of the provides file.

"""

summary_prompt = PromptTemplate(input_variables= ['document'], template=summary_template)
template = """ You are a knowledgeable chatbot that gently answers questions.

You know the following context information.

{context}

Answer to the following question from a user. Use only information from the previous context. Do not invent or assume stuff.


Question: {query}

Answer:"""

prompt = PromptTemplate(input_variables= ['context', 'query'], template= template)

llm =ChatCohere(cohere_api_key=os.getenv('COHERE_API_KEY'))

with gr.Blocks(theme='finlaymacklon/smooth_slate') as demo:
    signal = gr.Markdown('''# Welcome to Chat with Docs
                              I am an AI that recieves a **PDF** and can answer questions on the content of the document.''')
    inp = gr.File()
    out = gr.Textbox(label= 'Summary')
    inp.upload(fn= recieve_pdf,inputs= inp,outputs=out,show_progress=True)
    signal_1 = gr.Markdown('Use the Textbox below to chat. **Ask** questions regarding the pdf you uploaded')
    chat = gr.Chatbot()
    msg = gr.Textbox(info='input your chat')
        
    with gr.Row():
        submit = gr.Button('Send')
        clear = gr.ClearButton([msg,chat])

    msg.submit(chatbot, [msg, chat], [msg ,chat])
    submit.click(chatbot, [msg, chat], [msg ,chat])

    feedback = gr.Markdown('# [Please use this to provide feedback](https://forms.gle/oNZKx4nL7DmmJ64g8)')



demo.launch()