Spaces:
Runtime error
Runtime error
DrishtiSharma
commited on
Commit
•
cff2190
1
Parent(s):
d3d9576
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import dotenv
|
3 |
+
import gradio as gr
|
4 |
+
import lancedb
|
5 |
+
import logging
|
6 |
+
from langchain.embeddings.cohere import CohereEmbeddings
|
7 |
+
from langchain.llms import Cohere
|
8 |
+
from langchain.prompts import PromptTemplate
|
9 |
+
from langchain.chains import RetrievalQA
|
10 |
+
from langchain.vectorstores import LanceDB
|
11 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
12 |
+
# Assume these loaders are implemented based on your specific requirements
|
13 |
+
from custom_document_loaders import TextLoader, PyPDFLoader, DocxLoader, ImageLoader
|
14 |
+
import argostranslate.package
|
15 |
+
import argostranslate.translate
|
16 |
+
import shutil
|
17 |
+
|
18 |
+
# Configuration and Logging
|
19 |
+
dotenv.load_dotenv(".env")
|
20 |
+
DB_PATH = "/tmp/lancedb"
|
21 |
+
COHERE_MODEL_NAME = "multilingual-22-12"
|
22 |
+
LANGUAGE_ISO_CODES = {"English": "en", "Hindi": "hi", "Turkish": "tr", "French": "fr"}
|
23 |
+
|
24 |
+
logging.basicConfig(level=logging.INFO)
|
25 |
+
logger = logging.getLogger(__name__)
|
26 |
+
|
27 |
+
# Initialize argostranslate
|
28 |
+
argostranslate.package.update_package_index()
|
29 |
+
|
30 |
+
def initialize_documents_and_embeddings(input_file_path):
|
31 |
+
logger.info(f"Processing file: {input_file_path}")
|
32 |
+
file_extension = os.path.splitext(input_file_path)[1].lower()
|
33 |
+
loader = None
|
34 |
+
|
35 |
+
if file_extension in [".txt"]:
|
36 |
+
loader = TextLoader(input_file_path)
|
37 |
+
elif file_extension in [".pdf"]:
|
38 |
+
loader = PyPDFLoader(input_file_path)
|
39 |
+
elif file_extension in [".doc", ".docx"]:
|
40 |
+
loader = DocxLoader(input_file_path)
|
41 |
+
elif file_extension in [".jpg", ".jpeg", ".png"]:
|
42 |
+
loader = ImageLoader(input_file_path)
|
43 |
+
else:
|
44 |
+
raise ValueError("Unsupported file type. Supported files are .txt, .pdf, .docx, and image files.")
|
45 |
+
|
46 |
+
documents = loader.load()
|
47 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size=300, chunk_overlap=50)
|
48 |
+
texts = text_splitter.split_documents(documents)
|
49 |
+
embeddings = CohereEmbeddings(model=COHERE_MODEL_NAME)
|
50 |
+
return texts, embeddings
|
51 |
+
|
52 |
+
def initialize_database(texts, embeddings):
|
53 |
+
if os.path.exists(DB_PATH):
|
54 |
+
shutil.rmtree(DB_PATH) # Ensure a fresh start
|
55 |
+
db = lancedb.connect(DB_PATH)
|
56 |
+
table = db.create_table("multiling-rag", mode="overwrite")
|
57 |
+
return LanceDB.from_documents(texts, embeddings, connection=table)
|
58 |
+
|
59 |
+
def translate_text(text, from_code, to_code):
|
60 |
+
installed_languages = argostranslate.translate.get_installed_languages()
|
61 |
+
from_lang = next((lang for lang in installed_languages if lang.code == from_code), None)
|
62 |
+
to_lang = next((lang for lang in installed_languages if lang.code == to_code), None)
|
63 |
+
if not from_lang or not to_lang:
|
64 |
+
logger.error("Translation languages not installed.")
|
65 |
+
return "Translation error"
|
66 |
+
translation = from_lang.get_translation(to_lang)
|
67 |
+
return translation.translate(text)
|
68 |
+
|
69 |
+
def answer_question(question, input_language, output_language, db):
|
70 |
+
try:
|
71 |
+
input_lang_code = LANGUAGE_ISO_CODES[input_language]
|
72 |
+
output_lang_code = LANGUAGE_ISO_CODES[output_language]
|
73 |
+
question_in_english = translate_text(question, input_lang_code, "en") if input_language != "English" else question
|
74 |
+
# Simplified retrieval and response logic for demonstration
|
75 |
+
response = "This is a simulated response based on the question."
|
76 |
+
result_in_target_language = translate_text(response, "en", output_lang_code) if output_language != "English" else response
|
77 |
+
return result_in_target_language
|
78 |
+
except Exception as e:
|
79 |
+
logger.error(f"Error in answer_question: {str(e)}")
|
80 |
+
return "An error occurred while processing your question."
|
81 |
+
|
82 |
+
def document_analysis_and_feedback(document_path, feedback):
|
83 |
+
# Placeholder for document analysis logic
|
84 |
+
response = "Document analysis and feedback processing is not fully implemented."
|
85 |
+
return response
|
86 |
+
|
87 |
+
def setup_gradio_interface(db):
|
88 |
+
with gr.Blocks() as demo:
|
89 |
+
gr.Markdown("# Multilingual Health and Wellness Chatbot")
|
90 |
+
with gr.Tab("Ask a Question"):
|
91 |
+
with gr.Row():
|
92 |
+
input_language = gr.Dropdown(list(LANGUAGE_ISO_CODES.keys()), label="Input Language")
|
93 |
+
output_language = gr.Dropdown(list(LANGUAGE_ISO_CODES.keys()), label="Output Language")
|
94 |
+
question = gr.Textbox(label="Your question")
|
95 |
+
answer = gr.Textbox(label="Answer")
|
96 |
+
question.submit(lambda q, i, o: answer_question(q, i, o, db), inputs=[question, input_language, output_language], outputs=answer)
|
97 |
+
|
98 |
+
with gr.Tab("Upload Document"):
|
99 |
+
with gr.Row():
|
100 |
+
document = gr.File(label="Upload your health document")
|
101 |
+
feedback_box = gr.Textbox(label="Feedback (optional)")
|
102 |
+
upload_response = gr.Textbox(label="Analysis Result")
|
103 |
+
document.submit(document_analysis_and_feedback, inputs=[document, feedback_box], outputs=upload_response)
|
104 |
+
|
105 |
+
return demo
|
106 |
+
|
107 |
+
def main():
|
108 |
+
INPUT_FILE_PATH = "sample-text.txt" # Placeholder file path
|
109 |
+
texts, embeddings = initialize_documents_and_embeddings(INPUT_FILE_PATH)
|
110 |
+
db = initialize_database(texts, embeddings)
|
111 |
+
demo = setup_gradio_interface(db)
|
112 |
+
demo.launch()
|
113 |
+
|
114 |
+
if __name__ == "__main__":
|
115 |
+
main()
|