Spaces:
Runtime error
Runtime error
Commit
·
5e73cde
1
Parent(s):
c7e22a1
Add application file
Browse files- summarizer_app.py +66 -0
summarizer_app.py
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from langchain import OpenAI, PromptTemplate, LLMChain
|
| 2 |
+
from langchain.text_splitter import CharacterTextSplitter
|
| 3 |
+
from langchain.chains.mapreduce import MapReduceChain
|
| 4 |
+
from langchain.prompts import PromptTemplate
|
| 5 |
+
from langchain.docstore.document import Document
|
| 6 |
+
from langchain.chains.summarize import load_summarize_chain
|
| 7 |
+
import json
|
| 8 |
+
import gradio as gr
|
| 9 |
+
|
| 10 |
+
# Configure votre clé API
|
| 11 |
+
openai.api_key = os.environ['OpenaiKey']
|
| 12 |
+
|
| 13 |
+
#définition du LLM
|
| 14 |
+
llm = OpenAI(temperature=0)
|
| 15 |
+
|
| 16 |
+
#chargement des paramètres
|
| 17 |
+
with open("parametres.json", "r") as p:
|
| 18 |
+
params = json.load(p)
|
| 19 |
+
taille_max = params["taille_max"]
|
| 20 |
+
chunks_max = taille_max//4000+1
|
| 21 |
+
|
| 22 |
+
#résumé d'un texte
|
| 23 |
+
|
| 24 |
+
def summarize_text(text_to_summarize, llm):
|
| 25 |
+
|
| 26 |
+
#préparation du texte
|
| 27 |
+
text_splitter = CharacterTextSplitter()
|
| 28 |
+
texts = text_splitter.split_text(text_to_summarize)
|
| 29 |
+
print(len(texts))
|
| 30 |
+
docs = [Document(page_content=t) for t in texts[:chunks_max]]
|
| 31 |
+
print(len(docs))
|
| 32 |
+
|
| 33 |
+
#résumé
|
| 34 |
+
chain = load_summarize_chain(llm, chain_type="map_reduce")
|
| 35 |
+
chain.run(docs)
|
| 36 |
+
|
| 37 |
+
chain = load_summarize_chain(llm, chain_type="map_reduce", return_intermediate_steps=True)
|
| 38 |
+
steps = chain({"input_documents": docs}, return_only_outputs=True)
|
| 39 |
+
|
| 40 |
+
print(len(steps['intermediate_steps']))
|
| 41 |
+
print(steps['intermediate_steps'])
|
| 42 |
+
|
| 43 |
+
return steps['output_text']
|
| 44 |
+
|
| 45 |
+
# Lecture et résumé d'un fichier texte
|
| 46 |
+
|
| 47 |
+
def summarize_uploaded_file(file):
|
| 48 |
+
if not file.name.endswith('.txt'):
|
| 49 |
+
return ("Le fichier doit être un fichier texte (.txt)")
|
| 50 |
+
with open(file.name, "r", encoding = "latin-1") as f:
|
| 51 |
+
text = f.read()
|
| 52 |
+
summary = summarize_text(text, llm)
|
| 53 |
+
return summary
|
| 54 |
+
|
| 55 |
+
# Création de l'interface Gradio
|
| 56 |
+
|
| 57 |
+
iface = gr.Interface(
|
| 58 |
+
fn=summarize_uploaded_file,
|
| 59 |
+
inputs="file",
|
| 60 |
+
outputs=gr.outputs.Textbox(label="Résumé"),
|
| 61 |
+
title="Text File Summarizer",
|
| 62 |
+
description=f"Résume un fichier texte de longueur jusqu'à {taille_max} tokens",
|
| 63 |
+
allow_flagging = "never")
|
| 64 |
+
|
| 65 |
+
# Lancer l'interface
|
| 66 |
+
iface.launch()
|