Spaces:
Sleeping
Sleeping
Upload 8 files
Browse files- app/main.py +17 -11
app/main.py
CHANGED
|
@@ -2,7 +2,6 @@ from fastapi import FastAPI, File, UploadFile, Form, HTTPException
|
|
| 2 |
from fastapi.middleware.cors import CORSMiddleware
|
| 3 |
from fastapi.responses import HTMLResponse
|
| 4 |
from fastapi.staticfiles import StaticFiles
|
| 5 |
-
import pytesseract
|
| 6 |
from PIL import Image
|
| 7 |
import PyPDF2
|
| 8 |
from PyPDF2 import PdfFileReader
|
|
@@ -13,6 +12,9 @@ import os
|
|
| 13 |
import easyocr
|
| 14 |
import numpy as np
|
| 15 |
|
|
|
|
|
|
|
|
|
|
| 16 |
app = FastAPI()
|
| 17 |
|
| 18 |
# Servir les fichiers statiques (CSS, JS, etc.)
|
|
@@ -37,21 +39,25 @@ def extract_text_from_pdf(file):
|
|
| 37 |
# Passer l'objet fichier à PyPDF2
|
| 38 |
reader = PyPDF2.PdfReader(pdf_file)
|
| 39 |
text = ""
|
| 40 |
-
for
|
| 41 |
-
text +=
|
| 42 |
return text
|
| 43 |
|
| 44 |
|
| 45 |
|
| 46 |
def extract_text_from_pptx(file):
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
#
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
|
| 56 |
|
| 57 |
|
|
|
|
| 2 |
from fastapi.middleware.cors import CORSMiddleware
|
| 3 |
from fastapi.responses import HTMLResponse
|
| 4 |
from fastapi.staticfiles import StaticFiles
|
|
|
|
| 5 |
from PIL import Image
|
| 6 |
import PyPDF2
|
| 7 |
from PyPDF2 import PdfFileReader
|
|
|
|
| 12 |
import easyocr
|
| 13 |
import numpy as np
|
| 14 |
|
| 15 |
+
from pptx import Presentation
|
| 16 |
+
from io import BytesIO
|
| 17 |
+
|
| 18 |
app = FastAPI()
|
| 19 |
|
| 20 |
# Servir les fichiers statiques (CSS, JS, etc.)
|
|
|
|
| 39 |
# Passer l'objet fichier à PyPDF2
|
| 40 |
reader = PyPDF2.PdfReader(pdf_file)
|
| 41 |
text = ""
|
| 42 |
+
for page in reader.pages:
|
| 43 |
+
text += page.extract_text()
|
| 44 |
return text
|
| 45 |
|
| 46 |
|
| 47 |
|
| 48 |
def extract_text_from_pptx(file):
|
| 49 |
+
# Lire le contenu du fichier
|
| 50 |
+
file_content = file.read()
|
| 51 |
+
# Utiliser BytesIO pour créer un objet fichier en mémoire
|
| 52 |
+
pptx_file = BytesIO(file_content)
|
| 53 |
+
# Passer l'objet fichier à python-pptx
|
| 54 |
+
prs = Presentation(pptx_file)
|
| 55 |
+
text = ""
|
| 56 |
+
for slide in prs.slides:
|
| 57 |
+
for shape in slide.shapes:
|
| 58 |
+
if hasattr(shape, "text"):
|
| 59 |
+
text += shape.text + "\n"
|
| 60 |
+
return text
|
| 61 |
|
| 62 |
|
| 63 |
|