zentraR

zentraR is an R client for the ZENTRA Cloud v5 API. It gets you from "I have an API key" to "I have a tidy, up-to-date data frame of my sensor readings" in a few lines — handling authentication, pagination, rate limits, and reshaping so you can get straight to analysis.

Features

• Device discovery — zc_list_devices() lists every device your key can access.
• Tidy readings — zc_get_readings() returns one row per measurement, ready for dplyr / ggplot2, with a zc_pivot_wider() helper for a spreadsheet layout.
• Automatic pagination & rate-limit handling — large pulls just work; the client paces itself and retries on throttling.
• Incremental sync with pluggable storage — zc_sync() fetches only what's new and appends it to your store of choice: native R files (zc_store_rds()), plain CSV (zc_store_csv()), or nothing at all (return-only, for loading into your own database).
• Schedule-friendly — run zc_sync() on file open, daily, or weekly. See the Scheduling automatic syncs vignette.

Installation

Install the released version from CRAN:

install.packages("zentraR")

Or install the development version from the METER Group public packages group on GitLab:

# install.packages("remotes")
remotes::install_gitlab("meter-group-inc/pubpackages/zentraR")

Authentication

Get your API key from ZENTRA Cloud: your name (top right) → Integrations → Show Token (https://app.zentracloud.io/profile/integrations). Then either set it for the session, or save it to your .Renviron so it's always available:

library(zentraR)

zc_set_key("your-api-key")                 # this session only
zc_set_key("your-api-key", install = TRUE) # persist across sessions

Quick start

library(zentraR)

# 1. What devices can I see?
devices <- zc_list_devices(expand = "max_min_timestamp")
devices

# 2. Pull the last week of readings for one device (tidy long format).
readings <- zc_get_readings("z6-00930", start = Sys.Date() - 7)
readings

# 3. Reshape to one column per measurement.
zc_pivot_wider(readings)

# 4. Add human-readable quality flags.
zc_label_errors(readings)

Keeping a local record up to date

zc_sync() remembers what you already have and fetches only newer readings, so you can run it on a routine. Pick where the data lives:

# Persist as CSV (accessible to non-R tools):
store <- zc_store_csv("data/zentra")

# First run backfills history; later runs fetch only what's new:
zc_sync("z6-00930", store = store, start = Sys.Date() - 30)
zc_sync("z6-00930", store = store)   # incremental

# Sync every device your key can access:
zc_sync(store = store)

# Read your accumulated data back:
zc_store_read(store)

Prefer native R objects for an RStudio project? Use zc_store_rds("data/zentra"). Piping into your own database? Pass store = NULL and zc_sync() simply returns the new readings.

Rate limits

The v5 API is limited per key: roughly a 5-request burst, then about one request per minute. zentraR respects this automatically (pacing and retrying), but very large historical backfills will take time. Because zc_sync() is incremental and resumable, routine top-ups stay well within the limit.

Learn more

Built-in vignettes:

vignette("getting-started", package = "zentraR")
vignette("working-with-data", package = "zentraR")
vignette("scheduling", package = "zentraR")

Online (kept current with the API) on ZENTRA Cloud:

• Getting Started with zentraR
• Working with Your Data in R
• Scheduling Automatic Syncs

Development

The test suite runs against mocked responses, so R CMD check needs no network and no API key.

One file is different: tests/testthat/test-live-contract.R calls the real v5 API and asserts the response shape the parsers depend on, so that an upstream change is caught by CI rather than by users. It is skipped unless you opt in:

export ZENTRACLOUD_API_KEY=your-key
export ZENTRAR_TEST_DEVICE_SN=z6-00930
export ZENTRAR_LIVE_TESTS=true
Rscript -e 'devtools::test(filter = "live-contract")'

In CI these run as the scheduled api-contract job; see .gitlab-ci.yml. That job also sets ZENTRAR_REQUIRE_LIVE=true, which turns a missing key into a failure rather than a skip, so a credential problem can't masquerade as a pass.

License

MIT © METER Group, Inc.