Spaces:
Runtime error
Runtime error
Upload file_cleaning_ui.py
Browse files- file_cleaning_ui.py +142 -0
file_cleaning_ui.py
ADDED
|
@@ -0,0 +1,142 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python
|
| 2 |
+
|
| 3 |
+
import os
|
| 4 |
+
import re
|
| 5 |
+
import tempfile
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
|
| 8 |
+
import pdfplumber
|
| 9 |
+
import docx
|
| 10 |
+
import gradio as gr
|
| 11 |
+
|
| 12 |
+
def clean_text_for_rag(text: str) -> str:
|
| 13 |
+
"""Normalise et nettoie le texte pour un usage RAG."""
|
| 14 |
+
# Normalisation des caractères typographiques
|
| 15 |
+
text = re.sub(
|
| 16 |
+
r"[’‘“”«»–—\u00A0\u202F…œŒæÆ©®™§°±×÷]",
|
| 17 |
+
lambda m: {
|
| 18 |
+
"’": "'", "‘": "'", "“": '"', "”": '"',
|
| 19 |
+
"«": '"', "»": '"', "–": "-", "—": "-",
|
| 20 |
+
"…": "...", "œ": "oe", "Œ": "OE",
|
| 21 |
+
"æ": "ae", "Æ": "AE", "©": "(c)", "®": "(R)",
|
| 22 |
+
"™": "TM", "§": "§", "°": "°", "±": "+/-",
|
| 23 |
+
"×": "x", "÷": "/"
|
| 24 |
+
}.get(m.group(0), m.group(0)),
|
| 25 |
+
text,
|
| 26 |
+
)
|
| 27 |
+
# Conserver uniquement les caractères suivants
|
| 28 |
+
text = re.sub(r'[^a-zA-ZÀ-ÿæ-œ0-9\s\.\,\:\;\!\?\-\_\'\"\\\(\)]', '', text)
|
| 29 |
+
# Réduire les espaces multiples
|
| 30 |
+
return re.sub(r'\s+', ' ', text).strip()
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
def extract_and_clean_pdf(pdf_path: str) -> str:
|
| 34 |
+
"""Ouvre le PDF, récupère le texte et le nettoie."""
|
| 35 |
+
print(f"[+] Extraction du PDF : {pdf_path}")
|
| 36 |
+
all_pages = []
|
| 37 |
+
with pdfplumber.open(pdf_path) as pdf:
|
| 38 |
+
for page in pdf.pages:
|
| 39 |
+
txt = page.extract_text()
|
| 40 |
+
if txt:
|
| 41 |
+
all_pages.append(txt)
|
| 42 |
+
return clean_text_for_rag(" ".join(all_pages))
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
def extract_and_clean_docx(docx_path: str) -> str:
|
| 46 |
+
"""Lit un fichier DOCX et le nettoie."""
|
| 47 |
+
print(f"[+] Extraction du DOCX : {docx_path}")
|
| 48 |
+
doc = docx.Document(docx_path)
|
| 49 |
+
paragraphs = []
|
| 50 |
+
for para in doc.paragraphs:
|
| 51 |
+
text = para.text.strip()
|
| 52 |
+
if text:
|
| 53 |
+
paragraphs.append(text)
|
| 54 |
+
return clean_text_for_rag(" ".join(paragraphs))
|
| 55 |
+
|
| 56 |
+
def extract_and_clean_txt(txt_path: str) -> str:
|
| 57 |
+
"""Lit un fichier texte (txt, md, …) et le nettoie."""
|
| 58 |
+
print(f"[+] Lecture du fichier texte : {txt_path}")
|
| 59 |
+
with open(txt_path, "r", encoding="utf-8") as f:
|
| 60 |
+
lines = f.readlines()
|
| 61 |
+
cleaned = [
|
| 62 |
+
clean_text_for_rag(line.strip())
|
| 63 |
+
for line in lines
|
| 64 |
+
if line.strip()
|
| 65 |
+
]
|
| 66 |
+
return "\n".join(cleaned)
|
| 67 |
+
|
| 68 |
+
def process_file(input_file: gr.File, output_name: str) -> str:
|
| 69 |
+
"""
|
| 70 |
+
- Detecte le type (PDF ou texte)
|
| 71 |
+
- Effectue l'extraction + nettoyage
|
| 72 |
+
- Crée un fichier temporaire **avec le nom choisi** (output_name)
|
| 73 |
+
- Retourne le chemin du fichier temporaire (Gradio le propose en téléchargement)
|
| 74 |
+
"""
|
| 75 |
+
input_path = input_file.name
|
| 76 |
+
_, ext = os.path.splitext(input_path.lower())
|
| 77 |
+
|
| 78 |
+
if ext == ".pdf":
|
| 79 |
+
cleaned_text = extract_and_clean_pdf(input_path)
|
| 80 |
+
elif ext == ".docx":
|
| 81 |
+
cleaned_text = extract_and_clean_docx(input_path)
|
| 82 |
+
else:
|
| 83 |
+
cleaned_text = extract_and_clean_txt(input_path)
|
| 84 |
+
|
| 85 |
+
output_name = output_name.strip()
|
| 86 |
+
if not output_name.lower().endswith(".md"):
|
| 87 |
+
output_name = f"{output_name}.md"
|
| 88 |
+
|
| 89 |
+
temp_dir = tempfile.mkdtemp()
|
| 90 |
+
out_path = os.path.join(temp_dir, output_name)
|
| 91 |
+
|
| 92 |
+
with open(out_path, "w", encoding="utf-8") as f:
|
| 93 |
+
f.write(cleaned_text)
|
| 94 |
+
|
| 95 |
+
return out_path
|
| 96 |
+
|
| 97 |
+
with gr.Blocks(title="Nettoyage de texte pour RAG") as demo:
|
| 98 |
+
gr.Markdown("# 📄 Nettoyage d'un fichier pour utilisation RAG")
|
| 99 |
+
gr.Markdown(
|
| 100 |
+
"Déposez simplement votre fichier : nous nous chargeons d’extraire son contenu textuel, de le nettoyer "
|
| 101 |
+
"puis de vous le restituer en format markdown **sous le nom que vous choisissez.**"
|
| 102 |
+
)
|
| 103 |
+
|
| 104 |
+
with gr.Row():
|
| 105 |
+
with gr.Column(scale=1):
|
| 106 |
+
input_file = gr.File(
|
| 107 |
+
label="Déposez votre fichier ici",
|
| 108 |
+
file_types=["pdf", "txt", "md", "docx"],
|
| 109 |
+
)
|
| 110 |
+
output_name = gr.Textbox(
|
| 111 |
+
value="output.md",
|
| 112 |
+
label="Nom du fichier de sortie (en .md)",
|
| 113 |
+
placeholder="exemple.md",
|
| 114 |
+
interactive=True,
|
| 115 |
+
)
|
| 116 |
+
submit_btn = gr.Button("Traiter le fichier", variant="primary")
|
| 117 |
+
with gr.Column(scale=1):
|
| 118 |
+
output_file = gr.File(
|
| 119 |
+
label="Fichier nettoyé (.md)",
|
| 120 |
+
file_types=["md"],
|
| 121 |
+
)
|
| 122 |
+
|
| 123 |
+
submit_btn.click(
|
| 124 |
+
fn=process_file,
|
| 125 |
+
inputs=[input_file, output_name],
|
| 126 |
+
outputs=output_file,
|
| 127 |
+
)
|
| 128 |
+
|
| 129 |
+
gr.Markdown(
|
| 130 |
+
"""
|
| 131 |
+
---
|
| 132 |
+
**Prétraitements effectués :**
|
| 133 |
+
- Suppression des symboles non imprimables et des caractères parasites
|
| 134 |
+
- Conservation des lettres (y compris accentuées), chiffres, espaces et ponctuation simple
|
| 135 |
+
- Normalisation des espaces pour un texte harmonieux
|
| 136 |
+
- Export automatique au format **`.md`**
|
| 137 |
+
|
| 138 |
+
"""
|
| 139 |
+
)
|
| 140 |
+
|
| 141 |
+
if __name__ == "__main__":
|
| 142 |
+
demo.launch(share=True)
|