inegiR — submission candidate

inegiR retrieves INEGI indicator data through the authenticated JSON API v2.0. This contribution prepares version 3.1.0, a submission candidate repairing Eduardo Flores’s inegiR, pinned to 64d33221507cce2f86151a96fffc3283526a49cf. Eduardo Flores’s authorship and the upstream CC0 license are retained. This artifact has not been published to CRAN or accepted upstream. See baseline and scope, changes, and verification. Inicio en español.

Install this repair

Install the community contribution from the personal GitHub fork:

install.packages("remotes")
remotes::install_github("jadrk040507/inegiR")

This fork is a community contribution and does not imply upstream acceptance. For a local build, install the supplied source package from this workspace (with its dependencies already installed):

install.packages("dist/inegiR_3.1.0.tar.gz", repos = NULL, type = "source")
library(inegiR)

Installing from CRAN or the upstream GitHub repository does not install these local changes. Indicator access uses the API; it does not scrape HTML pages. Legacy service entry points remain available. DENUE grid/statistics coordinate guards now accept integer and classed numeric inputs; their live responses remain unverified.

Token and quick start

Register for an indicator API token through INEGI’s API page. For one R session, call inegi_set_token("your-token"); it sets INEGI_TOKEN invisibly and writes no file. To persist it yourself, add this line to your personal .Renviron, then restart R:

INEGI_TOKEN=your-token

Keep that file private and outside version control. inegi_get_token() returns the configured token; calling it at the console displays the returned value. The helpers never print credentials as a side effect. A missing or blank token raises inegi_input_error.

The following calls require your token and network access. IDs are examples from legacy documentation; check their current definition and bank with INEGI.

x <- inegi_data(c("216064", "216097"),
  start_date = "2020-01-01", end_date = "2025-12-31",
  language = "es", geography = "00", database = "BIE")
latest <- inegi_latest("216064")
meta <- inegi_metadata(unique(x$series_id))
freq <- inegi_catalog("CL_FREQ")

# Explicit title join that preserves observation order and avoids suffixes.
# Use the same database and language for data and metadata.
x$title <- meta$title[match(x$series_id, meta$series_id)]
# Complete CL_INDICATOR records, including provider-specific fields:
meta$raw_metadata[[1L]]

series accepts digit strings or safe positive integer IDs. Canonical calls remove duplicate IDs in first-seen order, preserve provider row order within each requested series, and fail atomically on an invalid response or failed request. language accepts "es" and "en"; database accepts "BIE" and "BISE". Geography is text so codes such as "00" retain leading zeros.

Data you receive

The first seven columns match the shared banxicoR convention:

Column	R type	Meaning
provider	character	"inegi"
series_id	character	Indicator ID
title	character	Response title, or NA when absent
period	character	Original API period text
date	Date	Validated interval start, or NA
value	double	Numeric observation, or NA_real_
value_raw	character	Original value text; JSON null becomes NA_character_

Additional character columns retain geography, frequency, unit, unit_multiplier, topic, source, note, last_updated, series_status, obs_status, obs_exception, obs_source, obs_note, database, and language. Empty responses return zero rows with the same types. No hidden metadata lookup fills titles. inegi_metadata() explicitly retrieves CL_INDICATOR and rejects missing, duplicate, or unrequested response IDs. Its raw_metadata list column retains each complete parsed provider record. Catalog results have character catalog, id, description, database, and language columns.

Raw periods, values, flags, and codes remain available. Missing or nonnumeric values remain rows with numeric NA; unexpected nonnumeric values without an exception flag warn. Values are never interpolated or rescaled by UNIT_MULT. Annual frequency codes 1/2/3 map YYYY or YYYY/01 to January 1; quarterly code 6 maps YYYY/01–YYYY/04 to quarter starts; monthly code 8 maps YYYY/01–YYYY/12 to month starts. Unknown, invalid, or ambiguous periods, including biweekly YYYY/MM, retain their text and receive missing dates. Date bounds are inclusive, accept scalar Date or valid ISO YYYY-MM-DD strings, and filter interval starts locally. A bounded response containing any undated period errors; request without bounds to inspect those periods. inegi_latest() uses the API’s recent-observation flag.

Existing code and paired providers

All 28 baseline exports remain. Existing inegi_series(series_id, token, ...) and inegi_series_multiple(series_id, token, names, ...) retain positional arguments and frame/list/compact schemas; both now default to INEGI_TOKEN. Legacy series calls use English; legacy catalogs use English/BIE and still require their token argument. The legacy frame has date, date_shortcut, values, and notes; metadata lists have metadata then data. Supported annual character-year and quarterly/monthly Date displays remain. Unsupported periods pass through with a warning; as_tt = TRUE requires valid, unambiguous Date output. Request URLs in call_unmasked and meta_call are redacted. Multiple-series output now follows requested order. incat_topic() now places descriptions in topic_description and IDs in topic, repairing the upstream swap while retaining column order. Shortcut IDs are preserved; their current economic meaning has not been verified.

old_shape <- inegi_series(series_id = "216064")
old_multi <- inegi_series_multiple(c("216064", "216097"),
  names = c("series_a", "series_b"))

# With both local packages installed and both provider tokens configured:
i <- inegiR::inegi_data("216064", start_date = "2025-01-01",
  end_date = "2025-12-31", language = "es")
b <- banxicoR::banxico_data("SF43718", start_date = "2025-01-01",
  end_date = "2025-12-31", language = "es")
common <- c("provider", "series_id", "title", "period", "date", "value", "value_raw")
combined <- rbind(i[common], b[common])
inegiR::inegi_latest("216064")
banxicoR::banxico_latest("SF43718")
inegiR::inegi_metadata("216064")
banxicoR::banxico_metadata("SF43718")

The shared interface does not imply equivalent indicators, frequencies, units, metadata layouts, or ordering. See the shared contract.

Verification limits

The tests use synthetic JSON, mocked HTTP, and synthetic tokens. They cover the indicator/catalog repair without external API access. Authenticated live INEGI behavior has not been verified. DENUE, SAKBE/routing, mnemonic XML, derived indicators, and current economic definitions are outside this repair’s validation scope. Build, install, test, and check results are in verification.