Spaces:
Build error
Build error
app.py
CHANGED
|
@@ -14,7 +14,6 @@ st.set_page_config(page_title="📑 Contract Analyzer", layout="wide")
|
|
| 14 |
# ========== FUNCTIONS ==========
|
| 15 |
|
| 16 |
# Tải mô hình Hugging Face từ Hub
|
| 17 |
-
@st.cache_resource
|
| 18 |
def load_inference_client():
|
| 19 |
return InferenceClient(repo_id="HuggingFaceH4/zephyr-7b-beta") # Mô hình Zephyr
|
| 20 |
|
|
@@ -23,19 +22,23 @@ inference_client = load_inference_client()
|
|
| 23 |
def extract_text_from_pdf(uploaded_file):
|
| 24 |
try:
|
| 25 |
with pdfplumber.open(uploaded_file) as pdf:
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
return ""
|
| 34 |
|
| 35 |
def load_text(file):
|
| 36 |
if not file:
|
| 37 |
return ""
|
| 38 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
ext = file.name.split('.')[-1].lower()
|
| 40 |
if ext == 'txt':
|
| 41 |
return StringIO(file.getvalue().decode("utf-8")).read()
|
|
@@ -79,7 +82,11 @@ def query_zephyr_model(text1, text2, question):
|
|
| 79 |
prompt = f"Compare the following two contracts and answer the question:\nText 1: {text1}\nText 2: {text2}\nQuestion: {question}"
|
| 80 |
try:
|
| 81 |
result = inference_client(inputs=prompt)
|
| 82 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
except Exception as e:
|
| 84 |
st.error(f"Error querying the model: {e}")
|
| 85 |
return None
|
|
@@ -129,16 +136,15 @@ def main():
|
|
| 129 |
user_question = st.text_input("Enter your question about the contracts:")
|
| 130 |
|
| 131 |
if user_question and st.button("Analyze Question"):
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
with st.spinner("Analyzing..."):
|
| 137 |
-
try:
|
| 138 |
-
pred = query_zephyr_model(text1, text2, user_question)
|
| 139 |
st.success(pred)
|
| 140 |
-
|
| 141 |
-
st.error(
|
|
|
|
|
|
|
| 142 |
|
| 143 |
if __name__ == "__main__":
|
| 144 |
main()
|
|
|
|
| 14 |
# ========== FUNCTIONS ==========
|
| 15 |
|
| 16 |
# Tải mô hình Hugging Face từ Hub
|
|
|
|
| 17 |
def load_inference_client():
|
| 18 |
return InferenceClient(repo_id="HuggingFaceH4/zephyr-7b-beta") # Mô hình Zephyr
|
| 19 |
|
|
|
|
| 22 |
def extract_text_from_pdf(uploaded_file):
|
| 23 |
try:
|
| 24 |
with pdfplumber.open(uploaded_file) as pdf:
|
| 25 |
+
text = "\n".join(page.extract_text() or "" for page in pdf.pages)
|
| 26 |
+
if not text.strip():
|
| 27 |
+
raise ValueError("No extractable text found in the PDF")
|
| 28 |
+
return text
|
| 29 |
+
except Exception as e:
|
| 30 |
+
st.error(f"Error reading PDF: {e}")
|
| 31 |
+
return ""
|
|
|
|
| 32 |
|
| 33 |
def load_text(file):
|
| 34 |
if not file:
|
| 35 |
return ""
|
| 36 |
try:
|
| 37 |
+
# Check file size (e.g., limit to 10MB)
|
| 38 |
+
if file.size > 10 * 1024 * 1024: # 10MB
|
| 39 |
+
st.warning("File is too large. Please upload a smaller file.")
|
| 40 |
+
return ""
|
| 41 |
+
|
| 42 |
ext = file.name.split('.')[-1].lower()
|
| 43 |
if ext == 'txt':
|
| 44 |
return StringIO(file.getvalue().decode("utf-8")).read()
|
|
|
|
| 82 |
prompt = f"Compare the following two contracts and answer the question:\nText 1: {text1}\nText 2: {text2}\nQuestion: {question}"
|
| 83 |
try:
|
| 84 |
result = inference_client(inputs=prompt)
|
| 85 |
+
if 'generated_text' in result:
|
| 86 |
+
return result['generated_text']
|
| 87 |
+
else:
|
| 88 |
+
st.error("No generated text found in the response.")
|
| 89 |
+
return None
|
| 90 |
except Exception as e:
|
| 91 |
st.error(f"Error querying the model: {e}")
|
| 92 |
return None
|
|
|
|
| 136 |
user_question = st.text_input("Enter your question about the contracts:")
|
| 137 |
|
| 138 |
if user_question and st.button("Analyze Question"):
|
| 139 |
+
with st.spinner("Analyzing..."):
|
| 140 |
+
try:
|
| 141 |
+
pred = query_zephyr_model(text1, text2, user_question)
|
| 142 |
+
if pred:
|
|
|
|
|
|
|
|
|
|
| 143 |
st.success(pred)
|
| 144 |
+
else:
|
| 145 |
+
st.error("Failed to get a valid answer from the model.")
|
| 146 |
+
except Exception as e:
|
| 147 |
+
st.error(f"Failed on Document: {e}")
|
| 148 |
|
| 149 |
if __name__ == "__main__":
|
| 150 |
main()
|