publish_DSD_via_WebService: Publish a Data Stream via a Web Service

In streamConnect: Connecting Stream Mining Components Using Sockets and Web Services

Publish a Data Stream via a Web Service

Description

Uses the package plumber to publish a data stream as a web service.

Usage

publish_DSD_via_WebService(
  dsd,
  port,
  task_file = NULL,
  serializer = "csv",
  serve = TRUE,
  background = TRUE,
  debug = FALSE
)

Arguments

dsd	A character string that creates a DSD.
port	port used to serve the DSD.
task_file	name of the plumber task script file.
serializer	method used to serialize the data. By default csv (comma separated values) is used. Other methods are json and rds (see plumber::serializer_csv).
serve	if TRUE, then a task file is written and a server started, otherwise, only a plumber task file is written.
background	logical; start a background process?
debug	if TRUE, then the service is started locally and a web client is started to explore the interface.

Details

The function writes a plumber task script file and starts the web server to serve the content of the stream using the endpoints

• ⁠http://localhost:port/get_points?n=100⁠ and

• ⁠http://localhost:port/info⁠.

APIs generated using plumber can be easily deployed. See: Hosting. By setting a task_file and serve = FALSE a plumber task script file is generated that can be deployment.

A convenient reader for stream data over web services is available as DSD_ReadWebService.

Value

a processx::process object created with callr::r_bg() which runs the plumber server in the background. The process can be stopped with rp$kill() or by killing the process using the operating system with the appropriate PID. rp$get_result() can be used to check for errors in the server process (e.g., when it terminates unexpectedly).

See Also

Other WebService: DSC_WebService(), DSD_ReadWebService(), publish_DSC_via_WebService()

Other dsd: DSD_ReadSocket(), DSD_ReadWebService(), publish_DSD_via_Socket()

Examples

# find a free port
port <- httpuv::randomPort()
port

# create a background DSD process sending data to the port
rp1 <- publish_DSD_via_WebService("DSD_Gaussians(k = 3, d = 3)", port = port)
rp1

# connect to the port and read manually. See DSD_ReadWebService for
# a more convenient way to connect to the WebService in R.
library("httr")

# we use RETRY to give the server time to spin up
resp <- RETRY("GET", paste0("http://localhost:", port, "/info"))
d <- content(resp, show_col_types = FALSE)
d

# example: Get 100 points and plot them
resp <- GET(paste0("http://localhost:", port, "/get_points?n=100"))
d <- content(resp, show_col_types = FALSE)
head(d)

dsd <- DSD_Memory(d)
dsd
plot(dsd, n = -1)

# end the DSD process. Note: that closing the connection above
# may already kill the process.
rp1$kill()
rp1

# Publish using json

rp2 <- publish_DSD_via_WebService("DSD_Gaussians(k = 3, d = 3)", 
           port = port, serializer = "json")
rp2

# connect to the port and read
# we use RETRY to give the server time to spin up
resp <- RETRY("GET", paste0("http://localhost:", port, "/info"))
content(resp, as = "text")

resp <- GET(paste0("http://localhost:", port, "/get_points?n=5"))
content(resp, as = "text")

# cleanup
rp2$kill()
rp2

# Debug the interface (run the service and start a web interface)
if (interactive())
  publish_DSD_via_WebService("DSD_Gaussians(k = 3, d = 3)", port = port, 
         debug = TRUE)

streamConnect documentation built on June 22, 2024, 9:55 a.m.