github.com/decred/dcrlnd@v0.7.6/docs/grpc/python.md

# How to write a Python gRPC client for the Lightning Network Daemon

This section enumerates what you need to do to write a client that communicates
with `lnd` in Python.

## Setup and Installation

Lnd uses the gRPC protocol for communication with clients like lncli. gRPC is
based on protocol buffers and as such, you will need to compile the lnd proto
file in Python before you can use it to communicate with lnd.

1. Create a virtual environment for your project
    ```
    $ virtualenv lnd
    ```
2. Activate the virtual environment
    ```
    $ source lnd/bin/activate
    ```
3. Install dependencies (googleapis-common-protos is required due to the use of
  google/api/annotations.proto)
    ```
    (lnd)$ pip install grpcio grpcio-tools googleapis-common-protos
    ```
4. Clone the google api's repository (required due to the use of
  google/api/annotations.proto)
    ```
    (lnd)$ git clone https://github.com/googleapis/googleapis.git
    ```
5. Copy the lnd lightning.proto file (you'll find this at
  [lnrpc/lightning.proto](https://github.com/lightningnetwork/lnd/blob/master/lnrpc/lightning.proto))
  or just download it
    ```shell
    lnd ⛰  curl -o lightning.proto -s https://raw.githubusercontent.com/lightningnetwork/lnd/master/lnrpc/lightning.proto
    ```
6. Compile the proto file
    ```shell
    lnd ⛰  python -m grpc_tools.protoc --proto_path=googleapis:. --python_out=. --grpc_python_out=. lightning.proto
    ```

After following these steps, two files `lightning_pb2.py` and
`lightning_pb2_grpc.py` will be generated. These files will be imported in your
project anytime you use Python gRPC.

### Generating RPC modules for subservers

If you want to use any of the subservers' functionality, you also need to
generate the python modules for them.

For example, if you want to generate the RPC modules for the `Router` subserver
(located/defined in `routerrpc/router.proto`), you need to run the following two
extra steps (after completing all 6 step described above) to get the
`router_pb2.py` and `router_pb2_grpc.py`:

```
(lnd)$ curl -o router.proto -s https://raw.githubusercontent.com/lightningnetwork/lnd/master/lnrpc/routerrpc/router.proto
(lnd)$ python -m grpc_tools.protoc --proto_path=googleapis:. --python_out=. --grpc_python_out=. router.proto
```

### Imports and Client

Every time you use Python gRPC, you will have to import the generated rpc modules
and set up a channel and stub to your connect to your `lnd` node.

Note that when an IP address is used to connect to the node (e.g. 192.168.1.21 instead of localhost) you need to add `--tlsextraip=192.168.1.21` to your `lnd` configuration and re-generate the certificate (delete tls.cert and tls.key and restart lnd).

```python
import lightning_pb2 as ln
import lightning_pb2_grpc as lnrpc
import grpc
import os

# Due to updated ECDSA generated tls.cert we need to let gprc know that
# we need to use that cipher suite otherwise there will be a handhsake
# error when we communicate with the lnd rpc server.
os.environ["GRPC_SSL_CIPHER_SUITES"] = 'HIGH+ECDSA'

# Lnd cert is at ~/.lnd/tls.cert on Linux and
# ~/Library/Application Support/Lnd/tls.cert on Mac
cert = open(os.path.expanduser('~/.lnd/tls.cert'), 'rb').read()
creds = grpc.ssl_channel_credentials(cert)
channel = grpc.secure_channel('localhost:10009', creds)
stub = lnrpc.LightningStub(channel)
```

## Examples

Let's walk through some examples of Python gRPC clients. These examples assume
that you have at least two `lnd` nodes running, the RPC location of one of which
is at the default `localhost:10009`, with an open channel between the two nodes.

### Simple RPC

```python
# Retrieve and display the wallet balance
response = stub.WalletBalance(ln.WalletBalanceRequest())
print(response.total_balance)
```

### Response-streaming RPC

```python
request = ln.InvoiceSubscription()
for invoice in stub.SubscribeInvoices(request):
    print(invoice)
```

Now, create an invoice for your node at `localhost:10009`and send a payment to
it from another node.
```bash
$ lncli addinvoice --amt=100
{
	"r_hash": <R_HASH>,
	"pay_req": <PAY_REQ>
}
$ lncli sendpayment --pay_req=<PAY_REQ>
```

Your Python console should now display the details of the recently satisfied
invoice.

### Bidirectional-streaming RPC

```python
from time import sleep
import codecs

def request_generator(dest, amt):
      # Initialization code here
      counter = 0
      print("Starting up")
      while True:
          request = ln.SendRequest(
              dest=dest,
              amt=amt,
          )
          yield request
          # Alter parameters here
          counter += 1
          sleep(2)

# Outputs from lncli are hex-encoded
dest_hex = <RECEIVER_ID_PUBKEY>
dest_bytes = codecs.decode(dest_hex, 'hex')

request_iterable = request_generator(dest=dest_bytes, amt=100)

for payment in stub.SendPayment(request_iterable):
    print(payment)
```
This example will send a payment of 100 atoms every 2 seconds.

### Using Macaroons

To authenticate using macaroons you need to include the macaroon in the metadata of the request.

```python
import codecs

# Lnd admin macaroon is at ~/.lnd/data/chain/bitcoin/simnet/admin.macaroon on Linux and
# ~/Library/Application Support/Lnd/data/chain/bitcoin/simnet/admin.macaroon on Mac
with open(os.path.expanduser('~/.lnd/data/chain/bitcoin/simnet/admin.macaroon'), 'rb') as f:
    macaroon_bytes = f.read()
    macaroon = codecs.encode(macaroon_bytes, 'hex')
```

The simplest approach to use the macaroon is to include the metadata in each request as shown below.

```python
stub.GetInfo(ln.GetInfoRequest(), metadata=[('macaroon', macaroon)])
```

However, this can get tiresome to do for each request, so to avoid explicitly including the macaroon we can update the credentials to include it automatically.

```python
def metadata_callback(context, callback):
    # for more info see grpc docs
    callback([('macaroon', macaroon)], None)


# build ssl credentials using the cert the same as before
cert_creds = grpc.ssl_channel_credentials(cert)

# now build meta data credentials
auth_creds = grpc.metadata_call_credentials(metadata_callback)

# combine the cert credentials and the macaroon auth credentials
# such that every call is properly encrypted and authenticated
combined_creds = grpc.composite_channel_credentials(cert_creds, auth_creds)

# finally pass in the combined credentials when creating a channel
channel = grpc.secure_channel('localhost:10009', combined_creds)
stub = lnrpc.LightningStub(channel)

# now every call will be made with the macaroon already included
stub.GetInfo(ln.GetInfoRequest())
```


## Conclusion

With the above, you should have all the `lnd` related `gRPC` dependencies
installed locally into your virtual environment. In order to get up to speed
with `protofbuf` usage from Python, see [this official `protobuf` tutorial for
Python](https://developers.google.com/protocol-buffers/docs/pythontutorial).
Additionally, [this official gRPC
resource](http://www.grpc.io/docs/tutorials/basic/python.html) provides more
details around how to drive `gRPC` from Python.

## API documentation

There is an [online API documentation](https://api.lightning.community?python)
available that shows all currently existing RPC methods, including code snippets
on how to use them.

## Special Scenarios

Due to a conflict between lnd's `UpdateChannelPolicy` gRPC endpoint and the python reserved word list, the follow syntax is required in order to use `PolicyUpdateRequest` with the `global` variable.
Here is an example of a working format that allows for use of a reserved word `global` in this scenario.

```
args = {'global': True, 'base_fee_msat': 1000, 'fee_rate': 0.000001, 'time_lock_delta': 40}
stub.UpdateChannelPolicy(ln.PolicyUpdateRequest(**args))
```