Spaces:
Sleeping
Sleeping
Upload app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,197 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import re
|
| 3 |
+
import tempfile
|
| 4 |
+
import streamlit as st
|
| 5 |
+
from PIL import Image
|
| 6 |
+
from pdf2image import convert_from_path, pdfinfo_from_path
|
| 7 |
+
import pytesseract
|
| 8 |
+
|
| 9 |
+
# ----------------------------------------------------------------------
|
| 10 |
+
# تنظیمات اولیه
|
| 11 |
+
# ----------------------------------------------------------------------
|
| 12 |
+
ZWNJ = '\u200c'
|
| 13 |
+
|
| 14 |
+
# ----------------------------------------------------------------------
|
| 15 |
+
# تابع نرمالسازی متن فارسی
|
| 16 |
+
# ----------------------------------------------------------------------
|
| 17 |
+
def normalize_text(text: str) -> str:
|
| 18 |
+
if not text:
|
| 19 |
+
return ""
|
| 20 |
+
|
| 21 |
+
arabic_to_persian = {
|
| 22 |
+
'ك': 'ک', 'ي': 'ی', 'ئ': 'ی', 'أ': 'ا', 'ة': 'ه',
|
| 23 |
+
'ؤ': 'و', 'إ': 'ا', 'ٰ': '', 'ٔ': '', 'ّ': ''
|
| 24 |
+
}
|
| 25 |
+
for ar, pr in arabic_to_persian.items():
|
| 26 |
+
text = text.replace(ar, pr)
|
| 27 |
+
|
| 28 |
+
corrections = {
|
| 29 |
+
'پسرده': 'پرده', 'اینن': 'این', 'خلسوت': 'خلوت',
|
| 30 |
+
'نضورد': 'نخورد', 'سبصد': 'سیصد', 'صایون': 'صابون',
|
| 31 |
+
'وشد': 'شد', 'میکنند': 'میکنند', 'میشود': 'میشود',
|
| 32 |
+
'میکرد': 'میکرد', 'میکنم': 'میکنم', 'میکنی': 'میکنی',
|
| 33 |
+
'میکسرد': 'میکرد', 'میکسند': 'میکنند', 'اِقتصاد': 'اقتصاد',
|
| 34 |
+
'اِجتماع': 'اجتماع', 'اِنسان': 'انسان', 'اِمکان': 'امکان'
|
| 35 |
+
}
|
| 36 |
+
for wrong, correct in corrections.items():
|
| 37 |
+
text = text.replace(wrong, correct)
|
| 38 |
+
|
| 39 |
+
text = re.sub(r'\b[0-9a-zA-Z\-]+\b', '', text)
|
| 40 |
+
text = re.sub(r'[^\w\s\u200c\u200d\u200e\u200f\u0600-\u06FF]', ' ', text)
|
| 41 |
+
text = re.sub(r'\s+', ' ', text).strip()
|
| 42 |
+
text = re.sub(r'(می|نمی)\s+(باشد|کند|شود|روم|کنم|کنی|کنید|کنیم|رسد|گردد|دهد)',
|
| 43 |
+
lambda m: m.group(1) + ZWNJ + m.group(2), text)
|
| 44 |
+
text = re.sub(r'(ها)\s+', r'\1' + ZWNJ + ' ', text)
|
| 45 |
+
text = re.sub(r'\s+([؟؟،،.:!;])', r'\1', text)
|
| 46 |
+
return text
|
| 47 |
+
|
| 48 |
+
# ----------------------------------------------------------------------
|
| 49 |
+
# حذف نویز و اعداد وسط متن
|
| 50 |
+
# ----------------------------------------------------------------------
|
| 51 |
+
def remove_noise(text: str) -> str:
|
| 52 |
+
text = re.sub(r'\b[0-9۰-۹]+\b', '', text)
|
| 53 |
+
text = re.sub(r'\s+', ' ', text).strip()
|
| 54 |
+
return text
|
| 55 |
+
|
| 56 |
+
# ----------------------------------------------------------------------
|
| 57 |
+
# فرمت شبیه کتاب
|
| 58 |
+
# ----------------------------------------------------------------------
|
| 59 |
+
def format_as_book(text: str) -> str:
|
| 60 |
+
text = re.sub(r'\n+', '\n', text)
|
| 61 |
+
text = re.sub(r'(?<![.؟!])\n\s*', ' ', text)
|
| 62 |
+
text = re.sub(r'\s+', ' ', text).strip()
|
| 63 |
+
text = re.sub(r'([.؟!])\s+', r'\1\n', text)
|
| 64 |
+
return text
|
| 65 |
+
|
| 66 |
+
# ----------------------------------------------------------------------
|
| 67 |
+
# OCR از تصویر
|
| 68 |
+
# ----------------------------------------------------------------------
|
| 69 |
+
def ocr_from_image(image: Image.Image) -> str:
|
| 70 |
+
image = image.convert('L')
|
| 71 |
+
config = r'--oem 3 --psm 6 -c preserve_interword_spaces=1'
|
| 72 |
+
raw = pytesseract.image_to_string(image, lang='fas+eng', config=config)
|
| 73 |
+
text = normalize_text(raw)
|
| 74 |
+
text = remove_noise(text)
|
| 75 |
+
text = format_as_book(text)
|
| 76 |
+
return text
|
| 77 |
+
|
| 78 |
+
# ----------------------------------------------------------------------
|
| 79 |
+
# OCR از PDF (با نمایش پیشرفت)
|
| 80 |
+
# ----------------------------------------------------------------------
|
| 81 |
+
def ocr_from_pdf(pdf_path: str, start_page: int, end_page: int) -> str:
|
| 82 |
+
try:
|
| 83 |
+
info = pdfinfo_from_path(pdf_path)
|
| 84 |
+
total_pages = int(info["Pages"])
|
| 85 |
+
end_page = min(end_page, total_pages)
|
| 86 |
+
start_page = max(1, start_page)
|
| 87 |
+
|
| 88 |
+
if start_page > total_pages:
|
| 89 |
+
return "❌ شماره صفحه شروع از تعداد کل صفحات بیشتر است."
|
| 90 |
+
|
| 91 |
+
images = convert_from_path(pdf_path, dpi=300, first_page=start_page, last_page=end_page)
|
| 92 |
+
all_text = f"📊 استخراج صفحات {start_page} تا {end_page} از {total_pages} صفحه:\n\n"
|
| 93 |
+
|
| 94 |
+
# 🟢 نوار پیشرفت و متن وضعیت
|
| 95 |
+
progress_bar = st.progress(0)
|
| 96 |
+
status_text = st.empty()
|
| 97 |
+
|
| 98 |
+
total_to_process = len(images)
|
| 99 |
+
for i, img in enumerate(images):
|
| 100 |
+
page_num = start_page + i
|
| 101 |
+
status_text.text(f"در حال پردازش صفحه {page_num} از {end_page} ...")
|
| 102 |
+
|
| 103 |
+
# OCR صفحه
|
| 104 |
+
text = ocr_from_image(img)
|
| 105 |
+
all_text += f"--- صفحه {page_num} ---\n{text}\n\n"
|
| 106 |
+
|
| 107 |
+
# بروزرسانی درصد
|
| 108 |
+
percent = int(((i + 1) / total_to_process) * 100)
|
| 109 |
+
progress_bar.progress(percent)
|
| 110 |
+
|
| 111 |
+
status_text.text("✅ پردازش کامل شد.")
|
| 112 |
+
return all_text.strip()
|
| 113 |
+
|
| 114 |
+
except Exception as e:
|
| 115 |
+
return f"❌ خطا در پردازش PDF: {str(e)}"
|
| 116 |
+
|
| 117 |
+
# ----------------------------------------------------------------------
|
| 118 |
+
# رابط کاربری Streamlit
|
| 119 |
+
# ----------------------------------------------------------------------
|
| 120 |
+
def main():
|
| 121 |
+
st.set_page_config(page_title="OCR فارسی - کتابی", layout="wide")
|
| 122 |
+
st.markdown(
|
| 123 |
+
"""
|
| 124 |
+
<style>
|
| 125 |
+
body {direction: rtl; text-align: right; font-family: "Vazir", "Tahoma", sans-serif;}
|
| 126 |
+
textarea {
|
| 127 |
+
direction: rtl !important;
|
| 128 |
+
text-align: justify !important;
|
| 129 |
+
font-family: "Vazir", "Tahoma", sans-serif !important;
|
| 130 |
+
line-height: 1.8 !important;
|
| 131 |
+
white-space: pre-wrap !important;
|
| 132 |
+
}
|
| 133 |
+
</style>
|
| 134 |
+
""", unsafe_allow_html=True
|
| 135 |
+
)
|
| 136 |
+
|
| 137 |
+
st.title(" OCR فارسی - خروجی شبیه کتاب")
|
| 138 |
+
st.markdown("استخراج متن از **PDF** یا **تصویر** بدون اعداد و نویز وسط متن")
|
| 139 |
+
|
| 140 |
+
with st.sidebar:
|
| 141 |
+
st.header("⚙️ تنظیمات استخراج")
|
| 142 |
+
uploaded_file = st.file_uploader("📁 فایل خود را آپلود کنید (PDF، JPG، PNG)", type=["pdf","jpg","jpeg","png"])
|
| 143 |
+
start_page, end_page = 1, 1
|
| 144 |
+
tmp_path = None
|
| 145 |
+
|
| 146 |
+
if uploaded_file is not None:
|
| 147 |
+
file_ext = uploaded_file.name.lower().split('.')[-1]
|
| 148 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=f".{file_ext}") as tmp_file:
|
| 149 |
+
tmp_file.write(uploaded_file.getvalue())
|
| 150 |
+
tmp_path = tmp_file.name
|
| 151 |
+
|
| 152 |
+
if file_ext == "pdf":
|
| 153 |
+
try:
|
| 154 |
+
info = pdfinfo_from_path(tmp_path)
|
| 155 |
+
total_pages = int(info["Pages"])
|
| 156 |
+
st.success(f"📄 فایل PDF با **{total_pages} صفحه** شناسایی شد.")
|
| 157 |
+
extract_all = st.checkbox("✅ استخراج همه صفحات", value=True)
|
| 158 |
+
if not extract_all:
|
| 159 |
+
start_page = st.number_input("صفحه شروع", 1, total_pages, 1)
|
| 160 |
+
end_page = st.number_input("صفحه پایان", 1, total_pages, total_pages)
|
| 161 |
+
else:
|
| 162 |
+
start_page, end_page = 1, total_pages
|
| 163 |
+
except Exception as e:
|
| 164 |
+
st.error(f"❌ خطا در خواندن PDF: {e}")
|
| 165 |
+
else:
|
| 166 |
+
st.info("📷 فایل تصویری — تمام محتوا پردازش خواهد شد.")
|
| 167 |
+
|
| 168 |
+
# Main
|
| 169 |
+
if uploaded_file is not None and tmp_path:
|
| 170 |
+
file_ext = uploaded_file.name.lower().split('.')[-1]
|
| 171 |
+
|
| 172 |
+
if file_ext == "pdf":
|
| 173 |
+
if st.button("🚀 استخراج متن از PDF", use_container_width=True):
|
| 174 |
+
with st.spinner("در حال پردازش..."):
|
| 175 |
+
result = ocr_from_pdf(tmp_path, start_page, end_page)
|
| 176 |
+
st.markdown("### 📝 متن استخراجشده ")
|
| 177 |
+
st.text_area("📘 خروجی OCR", result, height=600)
|
| 178 |
+
st.download_button("📥 دانلود متن", result, file_name="extracted_text.txt")
|
| 179 |
+
|
| 180 |
+
else:
|
| 181 |
+
st.image(uploaded_file, caption="تصویر آپلود شده", use_column_width=True)
|
| 182 |
+
if st.button("🚀 استخراج متن از تصویر", use_container_width=True):
|
| 183 |
+
with st.spinner("در حال پردازش..."):
|
| 184 |
+
image = Image.open(tmp_path)
|
| 185 |
+
result = ocr_from_image(image)
|
| 186 |
+
st.markdown("### 📝 متن استخراجشده")
|
| 187 |
+
st.text_area("📘 خروجی OCR", result, height=600)
|
| 188 |
+
st.download_button("📥 دانلود متن", result, file_name="extracted_text.txt")
|
| 189 |
+
|
| 190 |
+
if os.path.exists(tmp_path):
|
| 191 |
+
os.unlink(tmp_path)
|
| 192 |
+
|
| 193 |
+
else:
|
| 194 |
+
st.info("📁 لطفاً یک فایل آپلود کنید.")
|
| 195 |
+
|
| 196 |
+
if __name__ == "__main__":
|
| 197 |
+
main()
|