Update app.py

app.py

@@ -5,8 +5,9 @@ import json
 import re
 import os
 import time
+import tempfile
 
-from main import extract_key_phrases, score_sentences, summarize_text  # Optional, keep if used elsewhere
+from main import extract_key_phrases, score_sentences, summarize_text  # Optional
 
 st.set_page_config(page_title="PDF Tools", layout="wide")
 
@@ -239,22 +240,28 @@ UNSTRACT_API_KEY = os.getenv("UNSTRACT_API_KEY")  # Set this in your environment
 def extract_text_from_pdf_unstract(pdf_file):
     headers = {"unstract-key": UNSTRACT_API_KEY}
     pdf_bytes = pdf_file.read()
-    filename = getattr(pdf_file, "name", "uploaded.pdf")
-    files = {
-        "file": (filename, io.BytesIO(pdf_bytes), "application/pdf"),
-    }
-    whisper_url = f"{UNSTRACT_BASE}/whisper"
-    with st.spinner("Uploading and processing PDF with Unstract..."):
-        # For debugging, uncomment next line:
-        # st.write("DEBUG UPLOAD:", files)
-        r = requests.post(whisper_url, files=files, headers=headers)
-        if r.status_code != 202:
-            st.error(f"Unstract: Error uploading PDF: {r.status_code} - {r.text}")
-            return None
-        whisper_hash = r.json().get("whisper_hash")
-        if not whisper_hash:
-            st.error("Unstract: No whisper_hash received.")
-            return None
+    original_name = getattr(pdf_file, "name", "uploaded.pdf")
+    if not original_name.lower().endswith(".pdf"):
+        original_name = "uploaded.pdf"
+
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
+        tmp.write(pdf_bytes)
+        tmp.flush()
+        tmp.seek(0)
+        with open(tmp.name, "rb") as f:
+            files = {
+                "file": (original_name, f, "application/pdf"),
+            }
+            whisper_url = f"{UNSTRACT_BASE}/whisper"
+            with st.spinner("Uploading and processing PDF with Unstract..."):
+                r = requests.post(whisper_url, files=files, headers=headers)
+                if r.status_code != 202:
+                    st.error(f"Unstract: Error uploading PDF: {r.status_code} - {r.text}")
+                    return None
+                whisper_hash = r.json().get("whisper_hash")
+                if not whisper_hash:
+                    st.error("Unstract: No whisper_hash received.")
+                    return None
 
     # Step 2: Poll /whisper-status until processed
     status_url = f"{UNSTRACT_BASE}/whisper-status?whisper_hash={whisper_hash}"