Upload 2 files
Browse files- api.py +81 -0
- requirements.txt +4 -0
api.py
ADDED
@@ -0,0 +1,81 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
!pip install transformers accelerate torch PyPDF2 gradio --quiet
|
2 |
+
|
3 |
+
import gradio as gr
|
4 |
+
from transformers import pipeline
|
5 |
+
import PyPDF2
|
6 |
+
|
7 |
+
# ===== مدلها =====
|
8 |
+
# مدل خلاصهسازی فارسی (عمومی)
|
9 |
+
persian_summarizer = pipeline(
|
10 |
+
"summarization",
|
11 |
+
model="m3hrdadfi/bert2bert-fa-summarization",
|
12 |
+
tokenizer="m3hrdadfi/bert2bert-fa-summarization"
|
13 |
+
)
|
14 |
+
|
15 |
+
# مدل خلاصهسازی انگلیسی (عمومی و سبک)
|
16 |
+
english_summarizer = pipeline(
|
17 |
+
"summarization",
|
18 |
+
model="facebook/bart-large-cnn"
|
19 |
+
)
|
20 |
+
|
21 |
+
# ===== توابع =====
|
22 |
+
def extract_text_from_pdf(pdf_file):
|
23 |
+
"""استخراج متن از فایل PDF"""
|
24 |
+
pdf_reader = PyPDF2.PdfReader(pdf_file)
|
25 |
+
text = ""
|
26 |
+
for page in pdf_reader.pages:
|
27 |
+
page_text = page.extract_text()
|
28 |
+
if page_text:
|
29 |
+
text += page_text + "\n"
|
30 |
+
return text
|
31 |
+
|
32 |
+
def summarize_document(language, text, pdf):
|
33 |
+
"""خلاصهسازی متن یا PDF بر اساس زبان انتخابی"""
|
34 |
+
if pdf is not None:
|
35 |
+
text = extract_text_from_pdf(pdf)
|
36 |
+
|
37 |
+
if not text or len(text.strip()) < 50:
|
38 |
+
return "❗ لطفاً یک متن یا فایل PDF معتبر و حداقل ۵۰ کاراکتر وارد کنید."
|
39 |
+
|
40 |
+
# انتخاب مدل بر اساس زبان
|
41 |
+
if language == "Persian":
|
42 |
+
summarizer = persian_summarizer
|
43 |
+
max_chunk = 500
|
44 |
+
else:
|
45 |
+
summarizer = english_summarizer
|
46 |
+
max_chunk = 1024
|
47 |
+
|
48 |
+
# شکستن متن به بخشهای کوچکتر
|
49 |
+
text = text.strip().replace("\n", " ")
|
50 |
+
chunks = [text[i:i+max_chunk] for i in range(0, len(text), max_chunk)]
|
51 |
+
|
52 |
+
summary = ""
|
53 |
+
for chunk in chunks:
|
54 |
+
result = summarizer(
|
55 |
+
chunk,
|
56 |
+
max_length=150,
|
57 |
+
min_length=40,
|
58 |
+
do_sample=False
|
59 |
+
)[0]['summary_text']
|
60 |
+
summary += result + "\n"
|
61 |
+
|
62 |
+
return summary.strip()
|
63 |
+
|
64 |
+
# ===== رابط Gradio =====
|
65 |
+
with gr.Blocks() as demo:
|
66 |
+
gr.Markdown("## 📄 SummarizeX – AI Text & PDF Summarizer 🌐")
|
67 |
+
gr.Markdown("خلاصهساز هوش مصنوعی برای متن و PDF – پشتیبانی از فارسی و انگلیسی")
|
68 |
+
|
69 |
+
with gr.Row():
|
70 |
+
language = gr.Radio(["Persian", "English"], label="زبان خلاصهسازی", value="Persian")
|
71 |
+
|
72 |
+
with gr.Row():
|
73 |
+
pdf_input = gr.File(label="آپلود PDF", file_types=[".pdf"])
|
74 |
+
text_input = gr.Textbox(label="یا متن را وارد کنید", lines=8, placeholder="متن را اینجا وارد کنید...")
|
75 |
+
|
76 |
+
summarize_button = gr.Button("خلاصه کن / Summarize")
|
77 |
+
output_box = gr.Textbox(label="خلاصه نهایی", lines=8)
|
78 |
+
|
79 |
+
summarize_button.click(summarize_document, inputs=[language, text_input, pdf_input], outputs=output_box)
|
80 |
+
|
81 |
+
demo.launch(share=True)
|
requirements.txt
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
gradio
|
2 |
+
transformers
|
3 |
+
pypdf2
|
4 |
+
torch
|