Path.write_bytes

response = requests.get(raw_url, timeout=30)

                    if response.status_code == 200:
                        filename = f"{repo['name']}_{Path(pdf_path).name}"
                        dest = direct_dir / filename
                        dest.write_bytes(response.content)
                        logger.info(f"   ✓ Downloaded: {filename} ({len(response.content) / 1024:.1f} KB)")
                        self.pdfs_extracted += 1
                    else:
                        logger.warning(f"   ✗ Failed to download {pdf_path} (HTTP {response.status_code})")

                except Exception as e:
                    logger.warning(f"   ✗ Error downloading {pdf_path}: {e}")
                    continue

            time.sleep(1)

    def run(self, max_repos: int = 50, include_direct: bool = True):
        """Run complete download pipeline"""
        start_time = time.time()

        logger.info(f"\n{'='*80}")
        logger.info("RESUME DATASET BULK DOWNLOADER")
        logger.info(f"{'='*80}\n")

temp_path = Path(temp_dir)

                    # Save uploaded files
                    for uploaded_file in uploaded_files:
                        file_path = temp_path / uploaded_file.name
                        file_path.write_bytes(uploaded_file.getvalue())

                    # Scan temp directory
                    findings = scan_path(str(temp_path))
                    st.session_state.scan_results = findings
                    st.rerun()

    with tab2:
        st.header("Scan GitHub Repositories")

        # GitHub authentication (required)
        with st.expander("🔐 GitHub Authentication (Required)", expanded=True):
            st.markdown("""
            **Required:** Authenticate with your GitHub Personal Access Token to scan public repositories.

            **How to create a PAT:**
            1. Go to [GitHub Settings > Tokens](https://github.com/settings/tokens)
            2. Click "Generate new token (classic)"
            3. Give it a name (e.g., "Secret Scanner")
            4. Select scope: **`public_repo`** (read-only access to public repositories)