C RevEngAI Toolkit

creait is a library for writing C applications that interact with the RevEngAI API. creait is currently under development.

Installation

# first install libtoml (https://github.com/brightprogrammer/tomlc99)

# Clone this repo and cd into it
git clone [email protected]:RevEngAI/creait.git && cd creait

# Add the install path of libtoml
# export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/lib

# Configre the build using ninja. Remove -G Ninja if you prefer usign GNU Makefiles (make required)
cmake -B Build -G Ninja

# Build and install creait.
ninja -C Build && sudo ninja -C Build install

You can just copy paste this directly in your terminal and it will do everyting for you, given the following dependencies are installed already.

Dependencies

Before building, user/developer must have libcurl (development package), git, cmake, make, ninja and pkg-config installed on host system. The package names differ from OS to OS.

When building and installing cJSON from source on Mac OS based systems, add CMAKE_INSTALL_NAME_DIR=<install_prefix>/lib Where <install_prefix> is the prefix path you chose to install the library. This can be something like /usr or /usr/local or if you're installing it some other place then that directory path. For example, my command looks like this : cmake -B build -G Ninja -D CMAKE_BUILD_TYPE=Debug -D CMAKE_INSTALL_NAME_DIR=/usr/local/lib

On Windows

• You need Visual Studio.
• Git can be manually installed by downloading the installation setup.
• Python must be installed and then use pip install meson to install meson.
• pkg-config can be installed by running choco install pkgconfiglite after installing chocolatey package manager for windows.

Usage

All struct type names start with Reai and all function name starts with reai_. The naming convention for function is in snake_case and is in the following format : reai_<object-name>_<command>(...). Examples are reai_analysis_info_vec_create() or reai_request(), where Reai is the object name.

There are three main objects you need to interact with.

• First one is opaque Reai object. This is the main object that connects you to RevEng.AI servers.
• Second is ReaiRequest which you must use to build your request.
• Last is ReaiResponse where you get the response from RevEng.AI servers.

Configurations

To connect with RevEng.AI servers, you first need to load a config file. The config file must usually be present in your home directory and must have name ~/.reai-rz.toml. When using a plugin, this file can be auto-generated using one of the commands. A very basic config is

apikey = "libr3"                   # Replace this with your own API key
host = "https://api.reveng.ai/v1"  # API version and base endpoint
model = "binnet-0.3-x86"           # Set the latest AI model here.
db_dir_path = "/home/<user>/.reai" # This path may change depending on your OS
log_dir_path = "/tmp"              # This path may change depending on your OS

To load the config, you must create a ReaiConfig object that parses this toml file and stores the data in it for you to use easily.

#include <Reai/Config.h>

int main() {
    ReaiConfig *cfg = reai_config_load (NULL);
    RETURN_VALUE_IF(!cfg, EXIT_FAILURE, "Failed to load configuration.");

    return EXIT_SUCCESS;
}

Note that NULL is passed to reai_config_load. This means the config will be loaded from it's default expected path. Alternatively, you can load the config from your given path as well.

Making Contact

Next step is to connect with RevEng.AI servers. This is done using a single function call

#include <Reai/Api/Api.h>

int main() {
    // load config

    // These values can be passed without loading config as well if
    // it is desired to be hardcoded ihe the program itself.
    Reai *reai = reai_create (cfg->host, cfg->apikey, cfg->model);
    RETURN_VALUE_IF(!reai, EXIT_FAILURE, "Failed to connect to RevEng.AI servers.");

    return EXIT_SUCCESS;
}

If you already have host and apikey present in your codebase then you don't need to load the configuration as well.

Getting Responses

Before you make any request, you must create a response structure where you'll get responses from the server. This is done by initializing one.

#include <Reai/Api/Api.h>

int main() {
    // other code

    ReaiResponse response;
    reai_response_init (&response);

}

Making Requests

There are different types of requests and you can go through them in RevEng.AI API docs, or you can go through them in Include/Reai/Api/Request.h header. Name of each request type is very closely related to the endpoint it'll communicate with, and to create a request, you need to fill valid data in correspondigly name structure inside ReaiRequest object.

For example, if you want to search a binary you'll use the request type REAI_REQUEST_TYPE_SEARCH and fill data in ReaiRequest::search.

#include <Reai/Api/Api.h>

int main() {
    // other code

    ReaiRequest request = {0};

    // Build request
    request.type                   = REAI_REQUEST_TYPE_SEARCH;
    request.search.collection_name = "trojan";
    reai_request (reai, &request, &response);

    // Verify that you got correct response
    RETURN_VALUE_IF (
        response.type != REAI_RESPONSE_TYPE_SEARCH,
        EXIT_FAILURE,
        "Failed to perform search : %s.\n",
        response.raw.data
    );

    // Go over all query results and print values
    if (response.search.success) {
        REAI_VEC_FOREACH (response.search.query_results, result, {
            PRINT_ERR ("%s\n", result->binary_name);
        });
    } else {
        PRINT_ERR ("Search failed.\n");
    }
}

PRINT_ERR is just a macro to help debugging. It prepends the name of the function which issued the message and hence helps finding errors quickly.

Extra Notes

Respones are reset everytime a new request is made. This means anything inside the response you want to keep, you must clone. For vectors there are clone_create methods already implemented, and whatever you clone is your responsibility and you must destroy those explicitly.

Same goes for any other structure. If you called init then after use you must issue a deinit on the same object. If you called create then you must explicitly issue a destroy.