* Added Display for FragmentIdentifier
* AckDelayQueue as wrapper for DelayQueue
* First pass on ack controller
* Controller very initially incorporated into the client
* Cleanup
* Using socks5 build.rs
* Extra (temporary) log statements
* Decreased logging levels
* Adding the beginnings of a socks5 crate
* Removing unused import
* Adding built.rs
* Figured out test failure, stuck a note in detailing under what conditions it fails.
* Added lib
* wip on the way to compile
* First compile with much of the client in place
* Comment reflow to 80 lines
* Changed Socks5 client help message
* Latest changes to from develop applied to socks5 client
* Minor cleanup on unused code
* Adding snafu dependency
* Adding socks library
* Getting socks into the module structure
* Tokio conversion for socks code nearly completed.
* Starting traffic controllers again
* Bitcoin SP starting to breathe with Socks5 proxy. Responses not yet being sent.
* Adding in some hugely verbose print action so we can see things happening
* WIP refactor of socks code.
* Renamed structs to be more rubyish
* Refactored the run command a bit.
* handle_client doesn't need to be public
* Starting to split the handle method up into smaller, refactorable chunks
* Renamed a test
* Finished initial refactor
* Minor cleanup
* Made a few notes for my future self
* Being a bit more explicit in authtentication test
* Ensuring that user/password authentication attempts fail if that auth mode is off
* Documentation
* Refactord types into a types module
* Sending the request ID across and reading the response when it comes back.
* Added the request_id to the response header
* Adding exception handling to websocket send
* Semi-working...
* Removing non-functional examples.
* Minor output clarification
* Adding a Socks5 service provider
* Websocket connection is now being made.
* Added some simple and ungraceful websocket connection error handling
* Renamed socks5_proxy back to proxy
* ibid
* Nicer websocket start method
* Receiving messages via websocket
* Socks requests work in the simple case, SSL requests don't (yet).
* Minor cleanup, renaming variables and moving private functions around
* Comments on try_read_request
* Moved some code around
* Removed commented code and printlns
* Comments sp request
* Commented response data read
* Changing Request to Connection
* ibid
* Added a controller and split connection / request parsing
* Built out error handling on requests a bit
* Initial router action
* Request deserialization tests back in action
* Request constructors
* Constructor for controller
* Renamed message_router to controller
* Starting to build out the responses
* Returning proxied connection data
* Moving towards new Socks5 request crate
* Sending Socks5 multi-part requests through mixnet
* Removed the detritus of exploratory coding.
* Breaking the socks client read loop when empty bytes are read
* Documenting the message format for serialized socks requests
* Returning a response from the socks proxy
* Removing unused import
* Removing more detritus
* Restarting loop if no response is received
* The off-by-one change that fixed it all
* Removing unused response.rs module
* Removed unused import
* Comment cleanup
* More detritus
* Cleaning...
* Docs for socks5-requests
* Using the simple-socks5-requests crate Response in the socks5 client
* Removing unused error types
* Split request/response into their own files and wrote more tests
* Removing temporary README notes
* Renamed all instances of request_id to connection_id
* Docs on Connection struct
* Caving in to connecting inside the constructor for the moment
* Fixing up comments on socks5 service provider start
* Simplified errors in the Socks5 requests crate
* Flattened service provider module hierarchy a bit.
* Removed println
* Comment to explain return on timeout
* Logging controller connect errors
* Renamed websocket reads and writes to make them a bit more understandable
* Renamed TodoError to ConnectionError
* Logging errors instead of panicking on connection read/write failures
* Fixed error handling in controller
* Removing dead comments
* Cargo fmt applied
* Removing print statements
* Removed more comments, prints, etc
Co-authored-by: jstuczyn <jedrzej.stuczynski@gmail.com>
* Changed identity keypair to use ed25519
* Encryption key is now x25519 based + compatibiltiy with sphinx
* Pathing and import fixes
* Moved all asymmetric keys to sub-module in crypto
* Extracted aes to separate module
* kdf module in crypto
* Ability to perform diffie hellman on encryption keys
* ecdsa on identity keys
* Extremely rough and incomplete registration handshake
* Authentication primitives
* Creating new random authenticationIV
* Wrapper type for the derived shared key
* Removed AuthToken in favour of using SharedKey for authentication
* Gateway identity keys
* Registration handshake without error mapping
* Gateway address in client config
* Added extra key for gateway presence
* Updated pemstore to work on borrows instead
* Gateway client trying to perform the handshake
* Gateway changes to allow for handshake and shared key
* Debug trait on sharedkey
* native client using updated gateway client
* Slightly updated gateway API
* Minor cleanup
* Fixed pemstore to correctly save multiple keypairs
* Gateway actually deriving shared key during handshake
* Gateway sending correct mid-handshake message
* Missing quotation mark in client config template
* Fixed template for correct shared key serialization
* Fixed gateway authentication
* Fixed tests
* Using correct gateway key when converting to sphinx node
* "get_all_clients" takes them from gateways as opposed to providers now
* cargo fmt
* Renamed pemstore methods
* Unused import
* Encryption of forward requests between client and gateway
* Updated sphinx dependency to use public revision
* Sending 'error' on handshake processing error
* Removed some dead code
* Dead code I forgot to remove before
* Extracted AckAes128Key into a struct
* Slight pemstore revamp allowing for symmetric key store
* ibid.
* PemStorableKey for SharedKey
* Introduced single location responsible for key management for client
* WIP
* Sphinx version update
* Stop using NodeAddressBytes for two distinct and confusing purposes
* Abstracting away SocketAddr from sphinx forwarding
* Passing the bool for reply surbs
* Attack plan for replies + encryption
* Comment + removed variable binding
* ReplySURB usage
* Topology import in nymsphinx
* Sphinx version update
* Changed 'Recipient' to contain client's encryption key
* Message preparation taking shape!
* reply surb also containing the encryption key
* Very initial message receiver
* Sphinx version update
* A possibly working way of receiving surbs
* Fixed incorrect field name in client config template
* camel casing all request arguments
* Renamed and moved `MessageMode` to more appropriate file
* Restored reconstruction tests
* Removed dead code from chunking
* Made rust examples compilable
* reply SURB key storage
* Replies as an InputMessage
* Forgotten commented code
* No retransmission processing for cover or replies
* Received reply processing
* Renamed client pathfinder to something more appropriate
* Made HasherOutputSize public
* Added key store path to config
* Reply surb attaching key digest when used
* Changes due to previous renaming
* Removed comment
* Fixed insert_encryption_key
* Assigning initial value of key store path
* Computing key digest with correct algorithm
* Initial and presumably temporary request serialization
* hacky way of introducing 'FragmentIdentifier' for replies
* Moved responsibility of reply encryption, padding, etc, to message preparer
* Optional recipient in try_get_valid_topology_ref
* Handling new reply surbs with acks and padding
* Updated go and python examples to include replies in text and binary cases
* Updated rust examples + binaryserverresponse
* Helpers in rust examples
* And updated JS example
* Moved shared key generation function to crypto crate
* Cover traffic encryption!
* hmac computation in crypto
* Updated aes imports due to new dependencies
* hkdf made more generic
* crypto cleanup + algorithms in params
* Clippy cleanup pass
* Generating encryption+mac shared keys between client and gateway
* MACs attached to forward requests to gateway
* Gateway messages encrypted and mac'd
* Lowered logging level
* compiler warning cleanup
* Some minor cleanup
* Generic stream cipher
* Generic shared key derivation + algorithm definitions
* Project-wide AES clean-up
* Comment fix
* Removed commented imports
* Updated comments
* Fixed topology test fixture
* VersionFilterable for HashMap
* Removed NymTopology trait in favour of concrete type
* Removed providers from NymTopology
* Made gateway conversion use reference, similarly to mixes
* Using more concrete types in topology rather than b58 strings
* Allowing gateways to have DNS-resolvable mix listener address
* Error propagation for gateway key conversion
* Reducing ACK packet size to exactly what we need
* Made fragmentidentifier into an array
* Padding all sent messages so they'd be split into constant length packets
* Ability to send sphinx packets of different sizes + more efficient decoding
* Closing connection on connection corruption
* Missing semicolons
* Missing license notices
* Default for packetsize
* Split nymsphinx
* Replaced Mutex with RwLock for TopologyAccessor + impl Deref
* Sphinx update + import cleanup
* Moved packet_sizes file
* Updated NymTopology API
* sphinx version bump
* Missing license notice
* nymsphinx-params crate
* Changes due to ibid.
* Chunking rework to allow variable size fragments
* Initial ack crate
* Version bump to new dev build
* Cargo lock changes
* random_route_to_gateway by node address
* exposing getting read permit
* Very initial draft on ack control
* Correctly dereferencing out of topology read permit
* All pending changes + compilation todo!s
* Restricted scope of deref on TopologyAccessorInner
* Type path alias for generate_key
* Derived traits for MessageChunker
* Ack control starting to take shape!
* Awaiting callbacks
* Most of work done on acks. Now to wire it all together
* Import cleanup
* rng generalization
* Connected real traffic together; only acks from gateway left
* Removed redundant things from nymsphinx::utils
* nymsphinx-cover crate
* Ack-related fields in client config
* Decreased packet store log level
* Restored forward sphinx request
* Slight adjustements to surb acks
* Changed TopologyReadPermit from type alias into a struct
* Changes due to ibid.
* Sphinx version upgrade
* Gateway being able to understand and handle acks
* Special Cover FragmentIdentifier + removal of dead code
* Initial packet router for gateway client
* Kill client if it fails to send to gateway too many times
* Cover messages with acks
* Moved out gateway client errors
* Ignoring cover traffic acks
* Changes in ack control
* Another sphinx version upgrade
* websocket handler delegating message chunking
* Using config defined ack wait additions
* Other minor changes I should have been more dilligent with splitting
* Import path fix
* sphinx_receiver => mixnet_receiver
* Missing renamed variable instance
* Updated aes-ctr to 0.4.0
* Removed concept of 'unfragmented' single fragment
* Replay fragments detection
* Long method split
* typo
* Cleaner client init
* Fixed race condition
* Fixed similar issue for retransmission
* Cargo fmt
* Minor clenaup
Jędrzej Stuczyński