Update app.py
Browse files
app.py
CHANGED
|
@@ -13,8 +13,26 @@ api_key = "sk-proj-HYw82rX0O01CJrobRFj13r3BFxN3H1qOfswMSK1NGtGpfNyylWzMV_lQXw8Ee
|
|
| 13 |
|
| 14 |
client = OpenAI(api_key=api_key)
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
# ------------------------------------------------------------------
|
| 17 |
-
# 2 —
|
| 18 |
# ------------------------------------------------------------------
|
| 19 |
|
| 20 |
extraction_schema: Dict[str, Any] = {
|
|
@@ -61,50 +79,64 @@ extraction_schema: Dict[str, Any] = {
|
|
| 61 |
},
|
| 62 |
}
|
| 63 |
|
| 64 |
-
# Emballage dans la structure tool
|
| 65 |
TOOLS = [{"type": "function", "function": extraction_schema}]
|
| 66 |
|
| 67 |
SYSTEM_PROMPT = (
|
| 68 |
"Tu es un assistant administratif. À partir d’un texte mal rédigé, "
|
| 69 |
"tu dois extraire les informations personnelles sous forme de JSON "
|
| 70 |
-
"conforme au schéma fourni. Ne réponds JAMAIS hors JSON."
|
|
|
|
| 71 |
)
|
| 72 |
|
| 73 |
# ------------------------------------------------------------------
|
| 74 |
-
# 3 —
|
| 75 |
# ------------------------------------------------------------------
|
| 76 |
|
| 77 |
def extraire_infos(texte: str) -> str:
|
| 78 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
|
| 80 |
-
|
| 81 |
-
# Première (et unique) requête : on force le choix de l'outil
|
| 82 |
response = client.chat.completions.create(
|
| 83 |
-
model="gpt-4o-mini", # ou
|
| 84 |
temperature=0,
|
| 85 |
-
messages=
|
| 86 |
-
{"role": "system", "content": SYSTEM_PROMPT},
|
| 87 |
-
{"role": "user", "content": texte},
|
| 88 |
-
],
|
| 89 |
tools=TOOLS,
|
| 90 |
-
tool_choice={
|
|
|
|
|
|
|
|
|
|
| 91 |
)
|
| 92 |
|
| 93 |
choice = response.choices[0]
|
| 94 |
-
if choice.finish_reason
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
|
| 103 |
-
|
| 104 |
-
|
|
|
|
|
|
|
| 105 |
|
| 106 |
# ------------------------------------------------------------------
|
| 107 |
-
# 4 —
|
| 108 |
# ------------------------------------------------------------------
|
| 109 |
|
| 110 |
texte_exemple = (
|
|
@@ -117,7 +149,10 @@ texte_exemple = (
|
|
| 117 |
)
|
| 118 |
|
| 119 |
with gr.Blocks(title="Extracteur intelligent de données CERFA") as demo:
|
| 120 |
-
gr.Markdown(
|
|
|
|
|
|
|
|
|
|
| 121 |
|
| 122 |
input_box = gr.Textbox(
|
| 123 |
lines=14,
|
|
@@ -131,8 +166,9 @@ with gr.Blocks(title="Extracteur intelligent de données CERFA") as demo:
|
|
| 131 |
extract_btn.click(extraire_infos, input_box, output_box)
|
| 132 |
|
| 133 |
# ------------------------------------------------------------------
|
| 134 |
-
# 5 —
|
| 135 |
# ------------------------------------------------------------------
|
| 136 |
|
| 137 |
if __name__ == "__main__":
|
| 138 |
demo.launch()
|
|
|
|
|
|
| 13 |
|
| 14 |
client = OpenAI(api_key=api_key)
|
| 15 |
|
| 16 |
+
:
|
| 17 |
+
export OPENAI_API_KEY=sk-...
|
| 18 |
+
pip install openai>=1.0.0 gradio
|
| 19 |
+
"""
|
| 20 |
+
|
| 21 |
+
import os
|
| 22 |
+
import json
|
| 23 |
+
from typing import Dict, Any
|
| 24 |
+
|
| 25 |
+
import gradio as gr
|
| 26 |
+
from openai import OpenAI
|
| 27 |
+
|
| 28 |
+
# ------------------------------------------------------------------
|
| 29 |
+
# 1 — Initialisation du client OpenAI
|
| 30 |
+
# ------------------------------------------------------------------
|
| 31 |
+
|
| 32 |
+
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
|
| 33 |
+
|
| 34 |
# ------------------------------------------------------------------
|
| 35 |
+
# 2 — Définition du schéma JSON (outil)
|
| 36 |
# ------------------------------------------------------------------
|
| 37 |
|
| 38 |
extraction_schema: Dict[str, Any] = {
|
|
|
|
| 79 |
},
|
| 80 |
}
|
| 81 |
|
|
|
|
| 82 |
TOOLS = [{"type": "function", "function": extraction_schema}]
|
| 83 |
|
| 84 |
SYSTEM_PROMPT = (
|
| 85 |
"Tu es un assistant administratif. À partir d’un texte mal rédigé, "
|
| 86 |
"tu dois extraire les informations personnelles sous forme de JSON "
|
| 87 |
+
"conforme au schéma fourni. Ne réponds JAMAIS hors JSON. "
|
| 88 |
+
"Utilise IMPÉRATIVEMENT l'outil extract_user_info."
|
| 89 |
)
|
| 90 |
|
| 91 |
# ------------------------------------------------------------------
|
| 92 |
+
# 3 — Fonction d'extraction avec relance forcée
|
| 93 |
# ------------------------------------------------------------------
|
| 94 |
|
| 95 |
def extraire_infos(texte: str) -> str:
|
| 96 |
+
"""Renvoie le JSON extrait ou une erreur si l’outil n’est jamais appelé."""
|
| 97 |
+
|
| 98 |
+
messages = [
|
| 99 |
+
{"role": "system", "content": SYSTEM_PROMPT},
|
| 100 |
+
{"role": "user", "content": texte},
|
| 101 |
+
]
|
| 102 |
|
| 103 |
+
for attempt in range(2): # on tente deux fois maximum
|
|
|
|
| 104 |
response = client.chat.completions.create(
|
| 105 |
+
model="gpt-4o-mini", # ou gpt-3.5-turbo-0125
|
| 106 |
temperature=0,
|
| 107 |
+
messages=messages,
|
|
|
|
|
|
|
|
|
|
| 108 |
tools=TOOLS,
|
| 109 |
+
tool_choice={
|
| 110 |
+
"type": "function",
|
| 111 |
+
"function": {"name": "extract_user_info"},
|
| 112 |
+
},
|
| 113 |
)
|
| 114 |
|
| 115 |
choice = response.choices[0]
|
| 116 |
+
if choice.finish_reason == "tool_calls":
|
| 117 |
+
args_str = choice.message.tool_calls[0].function.arguments
|
| 118 |
+
parsed = json.loads(args_str)
|
| 119 |
+
return json.dumps(parsed, indent=2, ensure_ascii=False)
|
| 120 |
+
|
| 121 |
+
# Pas d'appel d'outil : on renforce l'instruction et on réessaie
|
| 122 |
+
messages.insert(
|
| 123 |
+
0,
|
| 124 |
+
{
|
| 125 |
+
"role": "system",
|
| 126 |
+
"content": (
|
| 127 |
+
"⚠️ Utilise OBLIGATOIREMENT l'outil extract_user_info et "
|
| 128 |
+
"ne renvoie jamais de texte libre."
|
| 129 |
+
),
|
| 130 |
+
},
|
| 131 |
+
)
|
| 132 |
|
| 133 |
+
return (
|
| 134 |
+
"❌ Erreur : le modèle n'a pas renvoyé d'appel d'outil après deux "
|
| 135 |
+
"tentatives."
|
| 136 |
+
)
|
| 137 |
|
| 138 |
# ------------------------------------------------------------------
|
| 139 |
+
# 4 — Interface Gradio
|
| 140 |
# ------------------------------------------------------------------
|
| 141 |
|
| 142 |
texte_exemple = (
|
|
|
|
| 149 |
)
|
| 150 |
|
| 151 |
with gr.Blocks(title="Extracteur intelligent de données CERFA") as demo:
|
| 152 |
+
gr.Markdown(
|
| 153 |
+
"Copie-colle un texte mal écrit contenant des infos personnelles.\n"
|
| 154 |
+
"L'IA le transforme en JSON prêt pour l'administration."
|
| 155 |
+
)
|
| 156 |
|
| 157 |
input_box = gr.Textbox(
|
| 158 |
lines=14,
|
|
|
|
| 166 |
extract_btn.click(extraire_infos, input_box, output_box)
|
| 167 |
|
| 168 |
# ------------------------------------------------------------------
|
| 169 |
+
# 5 — Lancement
|
| 170 |
# ------------------------------------------------------------------
|
| 171 |
|
| 172 |
if __name__ == "__main__":
|
| 173 |
demo.launch()
|
| 174 |
+
|