- Set socket non-blocking during SSL handshake
- Use select with 5-second timeout for SSL_connect
- Fallback to plain TCP if SSL fails for 'web' services

- Add SSL support to web_proxy.c for connecting to wsssht daemon
- Use HTTPS for 'https' services, HTTP for 'http' services
- Try HTTPS first with HTTP fallback for 'web' services
- Accept invalid/expired certificates as requested

- Added support for proxying requests to CLIENT_NAME.dominio
- Implemented service type detection (web, http, https) with priority ordering
- Added tunnel establishment and management for client services
- Implemented HTTPS support with invalid certificate acceptance
- Added tunnel reuse and timeout management (5 minutes)
- Added SSL connection functions for HTTPS tunneling
- Fixed build issues and updated dependencies

This commit adds a new web proxy server feature to wssshd that:
- Adds --web-proxy [PORT] command line option (default port: 9090)
- Listens on 127.0.0.1 only for security
- Proxies HTTP requests to registered clients based on hostname
- Supports multiple concurrent connections

Changes:
- config.h: Add web_proxy_enabled and web_proxy_port fields
- config.c: Add --web-proxy option parsing and config file support
- web_proxy.h: New header with function declarations
- web_proxy.c: New implementation with TCP server and hostname-based routing
- main.c: Add web_proxy_start/stop_server calls
- wssshd.1: Update man page with new option
- README.md: Add documentation for web proxy feature

- Updated README.md with comprehensive config sections, examples, and detailed web interface description
- Added new CHANGELOG entry for v1.7.1 with RDP fixes
- Updated TODO.md with completed RDP improvements
- Updated man pages with missing --enc options and corrected versions
- Added detailed web interface documentation covering dashboard, SSH terminal, VNC, and RDP functionality

- Created test_rdp_client.c based on test_vnc_client.c
- Implements RDP protocol negotiation including Connection Request/Confirm
- Sends MCS Connect Initial and Client Info PDU with username
- Supports both wsssht --pipe tunneling and direct TCP connection with -p PORT
- Tests RDP connection and reports results in output

- Increased RDP HTML buffer size from 32768 to 65536 bytes to prevent truncation of the main script
- Removed debug alert statements from RDP page
- RDP page should now fully load and execute JavaScript

The RDP connection issue was caused by the HTML being truncated due to insufficient buffer size, preventing the main script from being served to the browser.

- Added keyboard classes (Keyboard, EmulatedKeyboard, UnicodeKeyboard) from example.html
- Added mouse event handling functions
- Integrated WebAssembly module loading and RDP client setup
- Added event listeners for canvas focus, keyboard, and mouse events
- Fixed RDP connection flow to match example.html structure
- Added proper input handling and keyboard layout support
- Implemented canvas event management for proper user interaction

This should resolve the RDP connection issue where it was stopping at negotiation phase due to missing user input handling.

- Wrapped the initial script execution in try-catch
- Added alert for critical errors to show if script fails
- This will help identify if JavaScript syntax errors are preventing execution
- If alert appears, it means script is running but has errors
- If no alert appears, script is completely blocked

- Added checks to verify if DOM elements exist before attaching listeners
- Added detailed logging for each event listener attachment
- Log format shows which elements are found and which listeners are attached
- This will reveal if buttons exist in DOM and if listeners are properly attached

- Added alert('RDP page script is loading!') at the very beginning
- This will show if the script is being executed or blocked
- If no alert appears, the issue is script loading or CSP
- If alert appears but no console logs, there's a JavaScript error

- Added console.log statements throughout RDP page JavaScript
- Log WebAssembly module loading progress
- Log event listener attachment
- Log button click events (connect, disconnect, window controls)
- Log function calls (showRdpConfig, startRdpConnection, connect)
- This will help identify where the execution stops when buttons are clicked

- Added complete keyboard and mouse event handling from example.html
- Implemented Keyboard, EmulatedKeyboard, and UnicodeKeyboard classes
- Added mouse event handlers for movement, clicks, and wheel
- Added proper WebAssembly module loading with async/await
- Added missing constants (KeyAcquire, KeyRelease, SyncFlags)
- Fixed event listener attachments for canvas and input focus
- RDP page should now respond to user interactions

- Removed Content Security Policy header for RDP pages
- CSP was blocking WebAssembly eval() and other required functionality
- RDP pages now load without CSP restrictions
- WebAssembly and RDP client can now execute properly

- Simplified Content Security Policy to use wildcard https://* for external resources
- Reduced CSP length to prevent parsing issues
- Maintained 'unsafe-eval' for WebAssembly functionality
- CSP now allows all HTTPS resources while maintaining security

- Added fonts.gstatic.com to font-src for Google Fonts
- Added fonts.googleapis.com to script-src and style-src for Google Fonts API
- Expanded img-src to allow all HTTPS images
- CSP now properly allows all required resources for RDP functionality

- Added full CSP header allowing WebAssembly, external scripts/styles, WebSocket connections
- Includes 'unsafe-eval' and 'unsafe-inline' for RDP functionality
- Allows CDN resources from jsdelivr and cloudflare
- Permits WebSocket connections for RDP protocol

- Added https://cdn.jsdelivr.net and https://cdnjs.cloudflare.com to script-src, style-src, and font-src
- Required for Bootstrap and FontAwesome loading
- Maintains security while allowing necessary external resources

- Added Content-Security-Policy header with 'unsafe-eval' for script-src
- Required for WebAssembly module loading in RDP client
- Only applied to RDP HTML responses

- Made connect() function async to properly load WallixModule
- Moved WebAssembly initialization into connect function
- Set module constants after loading
- Ensured module is ready before creating RDP client

- Added WallixModule initialization with logging
- Integrated keyboard classes (Keyboard, EmulatedKeyboard, UnicodeKeyboard) from example
- Added mouse event handling functions
- Implemented event listeners for canvas and focus
- Updated connect function to properly set up RDP client with input handling
- Added updateInput and keyboard layout management
- Cleaned up event listeners on disconnect

- WallixModule is not a Promise, Module is the global WebAssembly object
- Check if newRdpGraphics2D is defined before using it
- Use Module.RdpClient directly like in rdpwasm/mstsc.js
- Rebuild project to embed updated HTML template

- WallixModule is the Promise returned by the IIFE, not an object with .ready property
- Use WallixModule.then((module) => { ... }) to access the loaded module
- Rebuild project to embed updated HTML template

- Replace undefined GraphicsDevice with newRdpGraphics2D
- Replace undefined RdpClient with WallixModule.RdpClient
- Add proper WebAssembly module loading with WallixModule.ready promise
- Rebuild project to embed updated HTML template

The client was failing WebSocket connection because the server was sending
Sec-WebSocket-Protocol: binary header in the handshake response, but the
client did not request any protocol. According to WebSocket RFC 6455,
servers should not send protocol headers if not requested by the client.

Changed the handshake response for RDP to not include the protocol header,
matching the terminal WebSocket behavior.

- Changed the path check in get_embedded_asset from /rdp/ to /rdpwasm/
- Rebuilt the project to include the updated asset serving logic

- Renamed templates/rdp/ to templates/rdpwasm/
- Updated embed_assets.sh to embed from rdpwasm and map to /rdpwasm/ paths
- Updated rdp.html to load scripts from /rdpwasm/
- Regenerated Makefile and rebuilt successfully

- Add patterns for wssshd2 build artifacts (*.o, *.a, *.so, wssshd, Makefile, etc.)
- Consolidate wssshtools executable patterns to wss*/test_*
- Ensure all build artifacts are properly ignored to prevent accidental commits

- Add Bootstrap modal for VNC password input
- Implement credentialsrequired event handler to show password prompt
- Handle password submission and resume authentication
- Allow Enter key to submit password
- Disconnect if user cancels without providing password

- Remove WebSocket-level chunking of VNC data (breaks streaming protocol)
- Send VNC data as complete WebSocket frames instead of chunks
- Enhance SSL error handling with better retry logic and exponential backoff
- Make VNC handler resilient to temporary SSL buffer issues
- Add comprehensive debug logging for WebSocket frame transmission
- Add client-side debug logging for VNC rectangle processing

Root cause: WebSocket chunking interrupted VNC TightDecoder which expects
complete compressed data blocks. Client accumulated partial data until
connection reset with ECONNRESET.

- Remove wsProtocols: ['binary'] setting that was causing reserved bits error
- WebSocket extension negotiation was failing with 'reserved3 = 1'
- Use default WebSocket protocol negotiation instead