Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,63 +2,85 @@ import os
|
|
| 2 |
import cv2
|
| 3 |
import pandas as pd
|
| 4 |
from paddleocr import PaddleOCR
|
|
|
|
| 5 |
|
| 6 |
-
# Initialisation OCR (
|
| 7 |
-
ocr = PaddleOCR(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
def extract_second_column(image_path):
|
| 10 |
"""
|
| 11 |
-
Extrait le texte de la 2e colonne
|
| 12 |
"""
|
| 13 |
result = ocr.ocr(image_path, cls=True)
|
| 14 |
|
| 15 |
-
|
|
|
|
| 16 |
|
| 17 |
-
|
| 18 |
-
text = line[1][0]
|
| 19 |
-
column_2_text.append(text)
|
| 20 |
|
| 21 |
-
|
|
|
|
| 22 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
# dossier persistant Hugging face
|
| 27 |
-
os.makedirs("/data", exist_ok=True)
|
| 28 |
-
if not os.path.exists(images_dir):
|
| 29 |
-
raise FileNotFoundError(
|
| 30 |
-
f"Le dossier '{images_dir}' est introuvable. "
|
| 31 |
-
"Vérifiez qu'il est bien copié dans le conteneur Docker."
|
| 32 |
-
)
|
| 33 |
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
|
| 37 |
-
|
|
|
|
| 38 |
|
| 39 |
-
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
-
|
| 42 |
|
| 43 |
for filename in sorted(os.listdir(images_dir)):
|
| 44 |
if filename.lower().endswith((".jpg", ".jpeg", ".png")):
|
| 45 |
image_path = os.path.join(images_dir, filename)
|
| 46 |
-
|
| 47 |
|
| 48 |
-
for
|
| 49 |
all_results.append({
|
| 50 |
"image": filename,
|
| 51 |
-
"colonne_2":
|
| 52 |
})
|
| 53 |
|
| 54 |
df = pd.DataFrame(all_results)
|
| 55 |
-
|
| 56 |
-
output_path="resultats_colonne_2.csv"
|
| 57 |
df.to_csv(output_path, index=False)
|
| 58 |
-
print(f"✅ Extraction terminée:{output_path}")
|
| 59 |
-
print("Fichiers presents :",os.listdir("."))
|
| 60 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
-
if __name__ == "__main__":
|
| 63 |
-
main()
|
| 64 |
|
|
|
|
| 2 |
import cv2
|
| 3 |
import pandas as pd
|
| 4 |
from paddleocr import PaddleOCR
|
| 5 |
+
import gradio as gr
|
| 6 |
|
| 7 |
+
# Initialisation OCR (une seule fois)
|
| 8 |
+
ocr = PaddleOCR(
|
| 9 |
+
use_angle_cls=True,
|
| 10 |
+
lang="fr",
|
| 11 |
+
show_log=False
|
| 12 |
+
)
|
| 13 |
|
| 14 |
def extract_second_column(image_path):
|
| 15 |
"""
|
| 16 |
+
Extrait le texte de la 2e colonne du tableau (approche par position X)
|
| 17 |
"""
|
| 18 |
result = ocr.ocr(image_path, cls=True)
|
| 19 |
|
| 20 |
+
if not result or not result[0]:
|
| 21 |
+
return []
|
| 22 |
|
| 23 |
+
boxes = result[0]
|
|
|
|
|
|
|
| 24 |
|
| 25 |
+
# Trier par position horizontale (x)
|
| 26 |
+
boxes_sorted_x = sorted(boxes, key=lambda b: min(p[0] for p in b[0]))
|
| 27 |
|
| 28 |
+
# Regrouper en colonnes (heuristique)
|
| 29 |
+
columns = {}
|
| 30 |
+
for box in boxes_sorted_x:
|
| 31 |
+
x_coords = [p[0] for p in box[0]]
|
| 32 |
+
x_center = sum(x_coords) / len(x_coords)
|
| 33 |
+
columns.setdefault(int(x_center // 100), []).append(box)
|
| 34 |
|
| 35 |
+
# Trier les colonnes
|
| 36 |
+
sorted_cols = sorted(columns.items(), key=lambda x: x[0])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
+
# Vérifier qu'il y a au moins 2 colonnes
|
| 39 |
+
if len(sorted_cols) < 2:
|
| 40 |
+
return []
|
| 41 |
+
|
| 42 |
+
second_col = sorted_cols[1][1]
|
| 43 |
|
| 44 |
+
# Trier verticalement
|
| 45 |
+
second_col_sorted = sorted(
|
| 46 |
+
second_col,
|
| 47 |
+
key=lambda b: min(p[1] for p in b[0])
|
| 48 |
+
)
|
| 49 |
|
| 50 |
+
texts = [b[1][0] for b in second_col_sorted]
|
| 51 |
+
return texts
|
| 52 |
|
| 53 |
+
def main():
|
| 54 |
+
images_dir = "images"
|
| 55 |
+
output_dir = "/data"
|
| 56 |
+
os.makedirs(output_dir, exist_ok=True)
|
| 57 |
|
| 58 |
+
all_results = []
|
| 59 |
|
| 60 |
for filename in sorted(os.listdir(images_dir)):
|
| 61 |
if filename.lower().endswith((".jpg", ".jpeg", ".png")):
|
| 62 |
image_path = os.path.join(images_dir, filename)
|
| 63 |
+
col2_values = extract_second_column(image_path)
|
| 64 |
|
| 65 |
+
for val in col2_values:
|
| 66 |
all_results.append({
|
| 67 |
"image": filename,
|
| 68 |
+
"colonne_2": val
|
| 69 |
})
|
| 70 |
|
| 71 |
df = pd.DataFrame(all_results)
|
| 72 |
+
output_path = os.path.join(output_dir, "resultats_colonne_2.csv")
|
|
|
|
| 73 |
df.to_csv(output_path, index=False)
|
|
|
|
|
|
|
| 74 |
|
| 75 |
+
return output_path
|
| 76 |
+
|
| 77 |
+
# Interface Gradio
|
| 78 |
+
gr.Interface(
|
| 79 |
+
fn=main,
|
| 80 |
+
inputs=[],
|
| 81 |
+
outputs=gr.File(label="Télécharger le fichier CSV"),
|
| 82 |
+
title="Extraction OCR – Colonne 2 des tableaux",
|
| 83 |
+
description="Cliquez sur le bouton pour lancer l'OCR et télécharger le CSV."
|
| 84 |
+
).launch()
|
| 85 |
|
|
|
|
|
|
|
| 86 |
|