Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,9 +6,9 @@ from huggingface_hub import InferenceClient
|
|
| 6 |
from langdetect import detect
|
| 7 |
import os
|
| 8 |
|
| 9 |
-
#
|
| 10 |
-
|
| 11 |
-
client = InferenceClient("google/gemma-3-4b-it")
|
| 12 |
|
| 13 |
def get_language(text):
|
| 14 |
try:
|
|
@@ -17,62 +17,76 @@ def get_language(text):
|
|
| 17 |
return "ar"
|
| 18 |
|
| 19 |
def search_islamweb(query):
|
| 20 |
-
|
| 21 |
full_query = f"site:islamweb.net {query}"
|
| 22 |
try:
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
return links
|
| 25 |
-
except:
|
|
|
|
| 26 |
return []
|
| 27 |
|
| 28 |
def scrape_content(url):
|
| 29 |
try:
|
| 30 |
-
|
|
|
|
| 31 |
response = requests.get(url, headers=headers, timeout=10)
|
| 32 |
response.encoding = 'utf-8'
|
| 33 |
soup = BeautifulSoup(response.text, 'html.parser')
|
| 34 |
|
| 35 |
-
#
|
| 36 |
-
main_content = soup.find('div', {'class': 'item'})
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
|
|
|
|
|
|
|
|
|
| 42 |
return ""
|
| 43 |
|
| 44 |
def islamic_ai_search(question):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
lang = get_language(question)
|
| 46 |
|
| 47 |
# 1. البحث عن الروابط
|
| 48 |
links = search_islamweb(question)
|
| 49 |
if not links:
|
| 50 |
-
|
|
|
|
| 51 |
|
| 52 |
-
# 2. استخراج السياق
|
| 53 |
context = ""
|
| 54 |
source_link = ""
|
| 55 |
for link in links:
|
| 56 |
content = scrape_content(link)
|
| 57 |
-
if len(content) >
|
| 58 |
context = content
|
| 59 |
source_link = link
|
| 60 |
break
|
| 61 |
|
| 62 |
if not context:
|
| 63 |
-
|
|
|
|
| 64 |
|
| 65 |
-
# 3.
|
| 66 |
system_message = (
|
| 67 |
"You are an expert Islamic researcher. Answer the user's question ONLY using the provided context from IslamWeb. "
|
| 68 |
"Strict rules: \n"
|
| 69 |
"1. If the answer is not in the context, say you don't know.\n"
|
| 70 |
-
"2.
|
| 71 |
-
"3. Do not add outside information.\n"
|
| 72 |
f"Context: {context}"
|
| 73 |
)
|
| 74 |
|
| 75 |
-
# 4. طلب الإجابة
|
| 76 |
response = ""
|
| 77 |
try:
|
| 78 |
for message in client.chat_completion(
|
|
@@ -84,24 +98,24 @@ def islamic_ai_search(question):
|
|
| 84 |
stream=True,
|
| 85 |
):
|
| 86 |
token = message.choices[0].delta.content
|
| 87 |
-
|
| 88 |
-
|
|
|
|
| 89 |
except Exception as e:
|
| 90 |
-
|
|
|
|
| 91 |
|
| 92 |
-
#
|
| 93 |
-
with gr.Blocks(
|
| 94 |
-
gr.Markdown("# 🕋 الباحث الإسلامي الذكي
|
| 95 |
-
gr.Markdown("اسأل أي سؤال ديني، وسيقوم الذكاء الاصطناعي بالبحث في إسلام ويب وتلخيص الإجابة لك.")
|
| 96 |
|
| 97 |
with gr.Row():
|
| 98 |
-
input_text = gr.Textbox(label="سؤالك / Sorunuz"
|
| 99 |
|
| 100 |
submit_btn = gr.Button("بحث وتحليل", variant="primary")
|
| 101 |
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
output_source = gr.Textbox(label="المصدر الأصلي (إسلام ويب)")
|
| 105 |
|
| 106 |
submit_btn.click(
|
| 107 |
fn=islamic_ai_search,
|
|
|
|
| 6 |
from langdetect import detect
|
| 7 |
import os
|
| 8 |
|
| 9 |
+
# محاولة جلب التوكن من الإعدادات
|
| 10 |
+
hf_token = os.getenv("HF_TOKEN")
|
| 11 |
+
client = InferenceClient("google/gemma-3-4b-it", token=hf_token)
|
| 12 |
|
| 13 |
def get_language(text):
|
| 14 |
try:
|
|
|
|
| 17 |
return "ar"
|
| 18 |
|
| 19 |
def search_islamweb(query):
|
| 20 |
+
print(f"جاري البحث عن: {query}") # تظهر في سجلات الـ Logs
|
| 21 |
full_query = f"site:islamweb.net {query}"
|
| 22 |
try:
|
| 23 |
+
# تقليل عدد النتائج لسرعة الاستجابة وتجنب الحظر
|
| 24 |
+
links = []
|
| 25 |
+
for j in search(full_query, num=3, stop=3, pause=2):
|
| 26 |
+
links.append(j)
|
| 27 |
+
print(f"الروابط المستخرجة: {links}")
|
| 28 |
return links
|
| 29 |
+
except Exception as e:
|
| 30 |
+
print(f"فشل البحث في جوجل: {e}")
|
| 31 |
return []
|
| 32 |
|
| 33 |
def scrape_content(url):
|
| 34 |
try:
|
| 35 |
+
print(f"جاري استخراج محتوى: {url}")
|
| 36 |
+
headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'}
|
| 37 |
response = requests.get(url, headers=headers, timeout=10)
|
| 38 |
response.encoding = 'utf-8'
|
| 39 |
soup = BeautifulSoup(response.text, 'html.parser')
|
| 40 |
|
| 41 |
+
# بنية إسلام ويب قد تختلف، نحاول أكثر من مكان
|
| 42 |
+
main_content = soup.find('div', {'class': 'item'}) or soup.find('div', {'id': 'fullitem'}) or soup.find('article')
|
| 43 |
+
|
| 44 |
+
if main_content:
|
| 45 |
+
text = main_content.get_text(strip=True)[:4000]
|
| 46 |
+
print("تم استخراج النص بنجاح.")
|
| 47 |
+
return text
|
| 48 |
+
return ""
|
| 49 |
+
except Exception as e:
|
| 50 |
+
print(f"فشل الاستخراج: {e}")
|
| 51 |
return ""
|
| 52 |
|
| 53 |
def islamic_ai_search(question):
|
| 54 |
+
if not question.strip():
|
| 55 |
+
yield "الرجاء كتابة سؤال أولاً.", ""
|
| 56 |
+
return
|
| 57 |
+
|
| 58 |
lang = get_language(question)
|
| 59 |
|
| 60 |
# 1. البحث عن الروابط
|
| 61 |
links = search_islamweb(question)
|
| 62 |
if not links:
|
| 63 |
+
yield "عذراً، جوجل حظر طلب البحث حالياً. جرب مرة أخرى بعد قليل أو تأكد من إعدادات البحث.", ""
|
| 64 |
+
return
|
| 65 |
|
| 66 |
+
# 2. استخراج السياق
|
| 67 |
context = ""
|
| 68 |
source_link = ""
|
| 69 |
for link in links:
|
| 70 |
content = scrape_content(link)
|
| 71 |
+
if len(content) > 150:
|
| 72 |
context = content
|
| 73 |
source_link = link
|
| 74 |
break
|
| 75 |
|
| 76 |
if not context:
|
| 77 |
+
yield "لم أتمكن من قراءة محتوى الفتوى من الروابط الموجودة.", ""
|
| 78 |
+
return
|
| 79 |
|
| 80 |
+
# 3. نظام الرسائل للنموذج
|
| 81 |
system_message = (
|
| 82 |
"You are an expert Islamic researcher. Answer the user's question ONLY using the provided context from IslamWeb. "
|
| 83 |
"Strict rules: \n"
|
| 84 |
"1. If the answer is not in the context, say you don't know.\n"
|
| 85 |
+
"2. Response language MUST be the same as input question language.\n"
|
|
|
|
| 86 |
f"Context: {context}"
|
| 87 |
)
|
| 88 |
|
| 89 |
+
# 4. طلب الإجابة (Streaming)
|
| 90 |
response = ""
|
| 91 |
try:
|
| 92 |
for message in client.chat_completion(
|
|
|
|
| 98 |
stream=True,
|
| 99 |
):
|
| 100 |
token = message.choices[0].delta.content
|
| 101 |
+
if token:
|
| 102 |
+
response += token
|
| 103 |
+
yield response, source_link
|
| 104 |
except Exception as e:
|
| 105 |
+
print(f"خطأ الـ AI: {e}")
|
| 106 |
+
yield f"حدث خطأ أثناء توليد الإجابة. تأكد من إضافة HF_TOKEN في الـ Secrets. الخطأ: {e}", source_link
|
| 107 |
|
| 108 |
+
# واجهة Gradio
|
| 109 |
+
with gr.Blocks() as demo:
|
| 110 |
+
gr.Markdown("# 🕋 الباحث الإسلامي الذكي")
|
|
|
|
| 111 |
|
| 112 |
with gr.Row():
|
| 113 |
+
input_text = gr.Textbox(label="سؤالك / Sorunuz")
|
| 114 |
|
| 115 |
submit_btn = gr.Button("بحث وتحليل", variant="primary")
|
| 116 |
|
| 117 |
+
output_answer = gr.Markdown(label="الإجابة")
|
| 118 |
+
output_source = gr.Textbox(label="المصدر الأصلي")
|
|
|
|
| 119 |
|
| 120 |
submit_btn.click(
|
| 121 |
fn=islamic_ai_search,
|