For a general introduction to Hydra Ecosystem, see hydraecosystem.org.
hydra-python-agent is a smart Hydra client implemented in Python which works with hydrus. Reference implementation is Heracles.ts. Smart clients are generic automated clients that establish resilient connected data networks leveraging knowledge graphs.
Our Hydra Agent has different interfaces that you can try:
- A Web-based GUI - which shows how Hydra APIs are connected and how the Smart Agent is generic and can automatically build requests.
- A Python package - so you can use to communicate with a Hydra API in your code/software.
- Natural-language-like command line tool - this is still a GET only implementation
The Agent is designed to:
- Provide a seamless Client that can be used to communicate with Hydra APIs
- Cache metadata from the Hydra server it connects to, to allow querying on the client-side
- Maintain a syncrhonization mechanism which makes sure cached resources are consistent
- The graph can be queried using OpenCypher.
The final goal is to create a Client that can connected to multiple hydrus servers and operate between them while caching information in a graph-based database(Redis). This should enable the client to have an "aggregated view" of the connected network (the network of all the servers it connects to) and make complex sematic queries to it.
NOTE: You'll need to use python3. Using venv(virtual environment) is recommended.
Clone and setup a virtual environment:
git clone https://github.com/HTTP-APIs/hydra-python-agent.git
cd hydra-python-agent
python3 -m venv venv
source venv/bin/activate
Install dependencies and setup Agent:
pip3 install --upgrade pip
pip3 install -r requirements.txt
python3 setup.py install
Setup Redis which is used as caching layer(if permission denied use sudo):
./redis_setup.sh
Setup hydrus
Since this is an API Client, we need an appropriate Hydra Server to query to. To setup a localhost follow the instructions at https://github.com/HTTP-APIs/hydrus#demo. You might want to run hydrus serve --no-auth to skip setting up headers.
After installing the Agent and running Redis, as per instructions above, you can do something like:
from hydra_agent.agent import Agent
agent = Agent("http://localhost:8080/serverapi/") # <- hydrus Server URL
agent.get("http://localhost:8080/serverapi/DroneCollection/")
The agent supports GET, PUT, POST or DELETE:
- GET - used to READ resources or collections
- PUT - used to CREATE new resources in the Server
- POST - used to UPDATE resources in the Server
- DELETE - used to DELETE resources in the Server
To GET a existing resource you should:
agent.get("http://localhost:8080/serverapi/<CollectionType>/<Resource-ID>")
agent.get("http://localhost:8080/serverapi/<CollectionType>/")
To PUT a new resource you should:
new_resource = {"@type": "Drone", "name": "Drone 1", "model": "Model S", ...}
agent.put("http://localhost:8080/serverapi/<CollectionType>/", new_resource)
To UPDATE a resource you should:
existing_resource["name"] = "Updated Name"
agent.post("http://localhost:8080/serverapi/<CollectionType>/<Resource-ID>", existing_resource)
To DELETE a resource you should:
agent.delete("http://localhost:8080/serverapi/<CollectionType>/<Resource-ID>")
More than that, Agent extends Session from https://2.python-requests.org/en/master/api/#request-sessions, so all methods like auth, cookies, headers and so on can also be used.
If you've followed the installation instructions you can run:
python hydra_agent/querying_mechanism.py
Another alternative to run the CLT is using docker componse. To run both Redis server and the client(stop any Redis instance before), you can run the command:
docker-compose run client
To query you should provide a hydrus URL first:
url>>> http://localhost:8080/serverapi/
Obs.: If failing to connect to localhost running the Agent via Docker, head to issue #104.
- Natural Language querying format
Run help inside the CLT to get the querying format.
>>>help # it will provide the querying format
You can query the server with the following format:
Get all endpoints:- show endpoints Get all class_endpoints:- show classEndpoints Get all collection_endpoints:- show collectionEndpoints Get all members of collection_endpoint:- show < collection_endpoint > members Get all properties of objects:- show objects< endpoint_type > properties Get all properties of any member:- **show object< id_of_member > properties ** Get all classes properties:- show class< class_endpoint > properties Get data with compare properties:- show < key > < value > and/or < key1 > < value1 > Get data by using both opeartions(and,or) you should use brackets like:- show model xyz and (name Drone1 or name Drone2) or, show < key > < value > and (< key > < value > or < key > < value >)
For more detail take a look at wiki file
- For the Web GUI there's a list of future enhacements here.
- For now the Natural language CLT it is a proof-of-concept, only
GETfunctionality