20 KiB
NATSBridge.jl Tutorial
A comprehensive tutorial for learning how to use NATSBridge.jl for bi-directional communication between Julia and JavaScript services using NATS.
Table of Contents
- What is NATSBridge.jl?
- Key Concepts
- Installation
- Basic Usage
- Payload Types
- Transport Strategies
- Advanced Features
- Complete Examples
What is NATSBridge.jl?
NATSBridge.jl is a Julia module that provides a high-level API for sending and receiving data across network boundaries using NATS as the message bus. It implements the Claim-Check pattern for handling large payloads efficiently.
Core Features
- Bi-directional communication: Julia ↔ JavaScript
- Smart transport selection: Automatic direct vs link transport based on payload size
- Multi-payload support: Send multiple payloads of different types in a single message
- Claim-check pattern: Upload large files to HTTP server, send only URLs via NATS
- Type-aware serialization: Different serialization strategies for different data types
Key Concepts
1. msgEnvelope_v1 (Message Envelope)
The msgEnvelope_v1 structure provides a comprehensive message format for bidirectional communication:
struct msgEnvelope_v1
correlationId::String # Unique identifier to track messages
msgId::String # This message id
timestamp::String # Message published timestamp
sendTo::String # Topic/subject the sender sends to
msgPurpose::String # Purpose (ACK | NACK | updateStatus | shutdown | chat)
senderName::String # Sender name (e.g., "agent-wine-web-frontend")
senderId::String # Sender id (uuid4)
receiverName::String # Message receiver name (e.g., "agent-backend")
receiverId::String # Message receiver id (uuid4 or nothing for broadcast)
replyTo::String # Topic to reply to
replyToMsgId::String # Message id this message is replying to
brokerURL::String # NATS server address
metadata::Dict{String, Any}
payloads::AbstractArray{msgPayload_v1} # Multiple payloads stored here
end
2. msgPayload_v1 (Payload Structure)
The msgPayload_v1 structure provides flexible payload handling:
struct msgPayload_v1
id::String # Id of this payload (e.g., "uuid4")
dataname::String # Name of this payload (e.g., "login_image")
type::String # "text | dictionary | table | image | audio | video | binary"
transport::String # "direct | link"
encoding::String # "none | json | base64 | arrow-ipc"
size::Integer # Data size in bytes
data::Any # Payload data in case of direct transport or a URL in case of link
metadata::Dict{String, Any} # Dict("checksum" => "sha256_hash", ...)
end
3. Standard API Format
The system uses a standardized list-of-tuples format for all payload operations:
# Input format for smartsend (always a list of tuples with type info)
[(dataname1, data1, type1), (dataname2, data2, type2), ...]
# Output format for smartreceive (always returns a list of tuples)
[(dataname1, data1, type1), (dataname2, data2, type2), ...]
Important: Even when sending a single payload, you must wrap it in a list.
Installation
using Pkg
Pkg.add("NATS")
Pkg.add("JSON")
Pkg.add("Arrow")
Pkg.add("HTTP")
Pkg.add("UUIDs")
Pkg.add("Dates")
Pkg.add("Base64")
Pkg.add("PrettyPrinting")
Pkg.add("DataFrames")
Then include the NATSBridge module:
include("NATSBridge.jl")
using .NATSBridge
Basic Usage
Sending Data (smartsend)
using NATSBridge
# Send a simple dictionary
data = Dict("key" => "value")
env = NATSBridge.smartsend("my.subject", [("dataname1", data, "dictionary")])
Receiving Data (smartreceive)
using NATSBridge
# Subscribe to a NATS subject
NATS.subscribe("my.subject") do msg
# Process the message
result = NATSBridge.smartreceive(
msg,
max_retries = 5,
base_delay = 100,
max_delay = 5000
)
# result is a list of (dataname, data, type) tuples
for (dataname, data, type) in result
println("Received $dataname of type $type")
println("Data: $data")
end
end
Payload Types
NATSBridge.jl supports the following payload types:
| Type | Description | Serialization |
|---|---|---|
text |
Plain text | UTF-8 encoding |
dictionary |
JSON-serializable data (Dict, NamedTuple) | JSON |
table |
Tabular data (DataFrame, array of structs) | Apache Arrow IPC |
image |
Image data (Bitmap, PNG/JPG bytes) | Binary |
audio |
Audio data (WAV, MP3 bytes) | Binary |
video |
Video data (MP4, AVI bytes) | Binary |
binary |
Generic binary data | Binary |
Transport Strategies
NATSBridge.jl automatically selects the appropriate transport strategy based on payload size:
Direct Transport (< 1MB)
Small payloads are encoded as Base64 and sent directly over NATS.
# Small data (< 1MB) - uses direct transport
small_data = rand(1000) # ~8KB
env = NATSBridge.smartsend("small", [("data", small_data, "table")])
Link Transport (≥ 1MB)
Large payloads are uploaded to an HTTP file server, and only the URL is sent via NATS.
# Large data (≥ 1MB) - uses link transport
large_data = rand(10_000_000) # ~80MB
env = NATSBridge.smartsend("large", [("data", large_data, "table")])
Complete Examples
Example 1: Text Message
Sender:
using NATSBridge
using UUIDs
const SUBJECT = "/NATSBridge_text_test"
const NATS_URL = "nats://localhost:4222"
const FILESERVER_URL = "http://localhost:8080"
function plik_upload_handler(fileserver_url::String, dataname::String, data::Vector{UInt8})::Dict{String, Any}
url_getUploadID = "$fileserver_url/upload"
headers = ["Content-Type" => "application/json"]
body = """{ "OneShot" : true }"""
httpResponse = HTTP.request("POST", url_getUploadID, headers, body; body_is_form=false)
responseJson = JSON.parse(String(httpResponse.body))
uploadid = responseJson["id"]
uploadtoken = responseJson["uploadToken"]
file_multipart = HTTP.Multipart(dataname, IOBuffer(data), "application/octet-stream")
url_upload = "$fileserver_url/file/$uploadid"
headers = ["X-UploadToken" => uploadtoken]
form = HTTP.Form(Dict("file" => file_multipart))
httpResponse = HTTP.post(url_upload, headers, form)
responseJson = JSON.parse(String(httpResponse.body))
fileid = responseJson["id"]
url = "$fileserver_url/file/$uploadid/$fileid/$dataname"
return Dict("status" => httpResponse.status, "uploadid" => uploadid, "fileid" => fileid, "url" => url)
end
function test_text_send()
small_text = "Hello, this is a small text message."
large_text = join(["Line $i: " for i in 1:50000], "")
data1 = ("small_text", small_text, "text")
data2 = ("large_text", large_text, "text")
env = NATSBridge.smartsend(
SUBJECT,
[data1, data2],
nats_url = NATS_URL,
fileserver_url = FILESERVER_URL,
fileserverUploadHandler = plik_upload_handler,
size_threshold = 1_000_000,
correlation_id = string(uuid4()),
msg_purpose = "chat",
sender_name = "text_sender"
)
end
Receiver:
using NATSBridge
const SUBJECT = "/NATSBridge_text_test"
const NATS_URL = "nats://localhost:4222"
function test_text_receive()
conn = NATS.connect(NATS_URL)
NATS.subscribe(conn, SUBJECT) do msg
result = NATSBridge.smartreceive(
msg,
max_retries = 5,
base_delay = 100,
max_delay = 5000
)
for (dataname, data, data_type) in result
if data_type == "text"
println("Received text: $data")
write("./received_$dataname.txt", data)
end
end
end
sleep(120)
NATS.drain(conn)
end
Example 2: Dictionary (JSON) Message
Sender:
using NATSBridge
using UUIDs
const SUBJECT = "/NATSBridge_dict_test"
const NATS_URL = "nats://localhost:4222"
const FILESERVER_URL = "http://localhost:8080"
function plik_upload_handler(fileserver_url::String, dataname::String, data::Vector{UInt8})::Dict{String, Any}
url_getUploadID = "$fileserver_url/upload"
headers = ["Content-Type" => "application/json"]
body = """{ "OneShot" : true }"""
httpResponse = HTTP.request("POST", url_getUploadID, headers, body; body_is_form=false)
responseJson = JSON.parse(String(httpResponse.body))
uploadid = responseJson["id"]
uploadtoken = responseJson["uploadToken"]
file_multipart = HTTP.Multipart(dataname, IOBuffer(data), "application/octet-stream")
url_upload = "$fileserver_url/file/$uploadid"
headers = ["X-UploadToken" => uploadtoken]
form = HTTP.Form(Dict("file" => file_multipart))
httpResponse = HTTP.post(url_upload, headers, form)
responseJson = JSON.parse(String(httpResponse.body))
fileid = responseJson["id"]
url = "$fileserver_url/file/$uploadid/$fileid/$dataname"
return Dict("status" => httpResponse.status, "uploadid" => uploadid, "fileid" => fileid, "url" => url)
end
function test_dict_send()
small_dict = Dict("name" => "Alice", "age" => 30)
large_dict = Dict("ids" => collect(1:50000), "names" => ["User_$i" for i in 1:50000])
data1 = ("small_dict", small_dict, "dictionary")
data2 = ("large_dict", large_dict, "dictionary")
env = NATSBridge.smartsend(
SUBJECT,
[data1, data2],
nats_url = NATS_URL,
fileserver_url = FILESERVER_URL,
fileserverUploadHandler = plik_upload_handler,
size_threshold = 1_000_000,
correlation_id = string(uuid4()),
msg_purpose = "chat"
)
end
Receiver:
using NATSBridge
const SUBJECT = "/NATSBridge_dict_test"
const NATS_URL = "nats://localhost:4222"
function test_dict_receive()
conn = NATS.connect(NATS_URL)
NATS.subscribe(conn, SUBJECT) do msg
result = NATSBridge.smartreceive(
msg,
max_retries = 5,
base_delay = 100,
max_delay = 5000
)
for (dataname, data, data_type) in result
if data_type == "dictionary"
println("Received dictionary: $data")
write("./received_$dataname.json", JSON.json(data, 2))
end
end
end
sleep(120)
NATS.drain(conn)
end
Example 3: DataFrame (Table) Message
Sender:
using NATSBridge
using DataFrames
using UUIDs
const SUBJECT = "/NATSBridge_table_test"
const NATS_URL = "nats://localhost:4222"
const FILESERVER_URL = "http://localhost:8080"
function plik_upload_handler(fileserver_url::String, dataname::String, data::Vector{UInt8})::Dict{String, Any}
url_getUploadID = "$fileserver_url/upload"
headers = ["Content-Type" => "application/json"]
body = """{ "OneShot" : true }"""
httpResponse = HTTP.request("POST", url_getUploadID, headers, body; body_is_form=false)
responseJson = JSON.parse(String(httpResponse.body))
uploadid = responseJson["id"]
uploadtoken = responseJson["uploadToken"]
file_multipart = HTTP.Multipart(dataname, IOBuffer(data), "application/octet-stream")
url_upload = "$fileserver_url/file/$uploadid"
headers = ["X-UploadToken" => uploadtoken]
form = HTTP.Form(Dict("file" => file_multipart))
httpResponse = HTTP.post(url_upload, headers, form)
responseJson = JSON.parse(String(httpResponse.body))
fileid = responseJson["id"]
url = "$fileserver_url/file/$uploadid/$fileid/$dataname"
return Dict("status" => httpResponse.status, "uploadid" => uploadid, "fileid" => fileid, "url" => url)
end
function test_table_send()
small_df = DataFrame(id = 1:10, name = ["Alice", "Bob", "Charlie"], score = [95, 88, 92])
large_df = DataFrame(id = 1:50000, name = ["User_$i" for i in 1:50000], score = rand(1:100, 50000))
data1 = ("small_table", small_df, "table")
data2 = ("large_table", large_df, "table")
env = NATSBridge.smartsend(
SUBJECT,
[data1, data2],
nats_url = NATS_URL,
fileserver_url = FILESERVER_URL,
fileserverUploadHandler = plik_upload_handler,
size_threshold = 1_000_000,
correlation_id = string(uuid4()),
msg_purpose = "chat"
)
end
Receiver:
using NATSBridge
using DataFrames
const SUBJECT = "/NATSBridge_table_test"
const NATS_URL = "nats://localhost:4222"
function test_table_receive()
conn = NATS.connect(NATS_URL)
NATS.subscribe(conn, SUBJECT) do msg
result = NATSBridge.smartreceive(
msg,
max_retries = 5,
base_delay = 100,
max_delay = 5000
)
for (dataname, data, data_type) in result
if data_type == "table"
data = DataFrame(data)
println("Received DataFrame with $(size(data, 1)) rows")
display(data[1:min(5, size(data, 1)), :])
end
end
end
sleep(120)
NATS.drain(conn)
end
Example 4: Mixed Content (Chat with Text, Image, Audio)
Sender:
using NATSBridge
using DataFrames
using UUIDs
const SUBJECT = "/NATSBridge_mix_test"
const NATS_URL = "nats://localhost:4222"
const FILESERVER_URL = "http://localhost:8080"
function plik_upload_handler(fileserver_url::String, dataname::String, data::Vector{UInt8})::Dict{String, Any}
url_getUploadID = "$fileserver_url/upload"
headers = ["Content-Type" => "application/json"]
body = """{ "OneShot" : true }"""
httpResponse = HTTP.request("POST", url_getUploadID, headers, body; body_is_form=false)
responseJson = JSON.parse(String(httpResponse.body))
uploadid = responseJson["id"]
uploadtoken = responseJson["uploadToken"]
file_multipart = HTTP.Multipart(dataname, IOBuffer(data), "application/octet-stream")
url_upload = "$fileserver_url/file/$uploadid"
headers = ["X-UploadToken" => uploadtoken]
form = HTTP.Form(Dict("file" => file_multipart))
httpResponse = HTTP.post(url_upload, headers, form)
responseJson = JSON.parse(String(httpResponse.body))
fileid = responseJson["id"]
url = "$fileserver_url/file/$uploadid/$fileid/$dataname"
return Dict("status" => httpResponse.status, "uploadid" => uploadid, "fileid" => fileid, "url" => url)
end
function test_mix_send()
# Text data
text_data = "Hello! This is a test chat message. 🎉"
# Dictionary data
dict_data = Dict("type" => "chat", "sender" => "serviceA")
# Small table data
table_data_small = DataFrame(id = 1:10, name = ["msg_$i" for i in 1:10])
# Large table data (link transport)
table_data_large = DataFrame(id = 1:150_000, name = ["msg_$i" for i in 1:150_000])
# Small image data (direct transport)
image_data = UInt8[rand(1:255) for _ in 1:100]
# Large image data (link transport)
large_image_data = UInt8[rand(1:255) for _ in 1:1_500_000]
# Small audio data (direct transport)
audio_data = UInt8[rand(1:255) for _ in 1:100]
# Large audio data (link transport)
large_audio_data = UInt8[rand(1:255) for _ in 1:1_500_000]
# Small video data (direct transport)
video_data = UInt8[rand(1:255) for _ in 1:150]
# Large video data (link transport)
large_video_data = UInt8[rand(1:255) for _ in 1:1_500_000]
# Small binary data (direct transport)
binary_data = UInt8[rand(1:255) for _ in 1:200]
# Large binary data (link transport)
large_binary_data = UInt8[rand(1:255) for _ in 1:1_500_000]
# Create payloads list - mixed content
payloads = [
# Small data (direct transport)
("chat_text", text_data, "text"),
("chat_json", dict_data, "dictionary"),
("chat_table_small", table_data_small, "table"),
# Large data (link transport)
("chat_table_large", table_data_large, "table"),
("user_image_large", large_image_data, "image"),
("audio_clip_large", large_audio_data, "audio"),
("video_clip_large", large_video_data, "video"),
("binary_file_large", large_binary_data, "binary")
]
env = NATSBridge.smartsend(
SUBJECT,
payloads,
nats_url = NATS_URL,
fileserver_url = FILESERVER_URL,
fileserverUploadHandler = plik_upload_handler,
size_threshold = 1_000_000,
correlation_id = string(uuid4()),
msg_purpose = "chat",
sender_name = "mix_sender"
)
end
Receiver:
using NATSBridge
using DataFrames
const SUBJECT = "/NATSBridge_mix_test"
const NATS_URL = "nats://localhost:4222"
function test_mix_receive()
conn = NATS.connect(NATS_URL)
NATS.subscribe(conn, SUBJECT) do msg
result = NATSBridge.smartreceive(
msg,
max_retries = 5,
base_delay = 100,
max_delay = 5000
)
println("Received $(length(result)) payloads")
for (dataname, data, data_type) in result
println("\n=== Payload: $dataname (type: $data_type) ===")
if data_type == "text"
println(" Type: String")
println(" Length: $(length(data)) characters")
elseif data_type == "dictionary"
println(" Type: JSON Object")
println(" Keys: $(keys(data))")
elseif data_type == "table"
data = DataFrame(data)
println(" Type: DataFrame")
println(" Dimensions: $(size(data, 1)) rows x $(size(data, 2)) columns")
elseif data_type == "image"
println(" Type: Vector{UInt8}")
println(" Size: $(length(data)) bytes")
write("./received_$dataname.bin", data)
elseif data_type == "audio"
println(" Type: Vector{UInt8}")
println(" Size: $(length(data)) bytes")
write("./received_$dataname.bin", data)
elseif data_type == "video"
println(" Type: Vector{UInt8}")
println(" Size: $(length(data)) bytes")
write("./received_$dataname.bin", data)
elseif data_type == "binary"
println(" Type: Vector{UInt8}")
println(" Size: $(length(data)) bytes")
write("./received_$dataname.bin", data)
end
end
end
sleep(120)
NATS.drain(conn)
end
Best Practices
- Always wrap payloads in a list - Even for single payloads:
[("dataname", data, "type")] - Use appropriate transport - Let NATSBridge handle size-based routing (default 1MB threshold)
- Customize size threshold - Use
size_thresholdparameter to adjust the direct/link split - Provide fileserver handler - Implement
fileserverUploadHandlerfor link transport - Include correlation IDs - Track messages across distributed systems
- Handle errors - Implement proper error handling for network failures
- Close connections - Ensure NATS connections are properly closed using
NATS.drain()
Conclusion
NATSBridge.jl provides a powerful abstraction for bi-directional communication between Julia and JavaScript services. By understanding the key concepts and following the best practices, you can build robust, scalable applications that leverage the full power of NATS messaging.
For more information, see:
docs/architecture.md- Detailed architecture documentationdocs/implementation.md- Implementation details