test_that("natural language claims are classified without requiring a reference row", {
  p <- classify_stat_claim("Iran's unemployment rate was 8.1 percent in 2025.")
  expect_s3_class(p, "stat_claim_profile")
  expect_equal(p$domain, "labour")
  expect_equal(p$indicator, "unemployment rate")
  expect_equal(p$geo_code, "IRN")
  expect_equal(p$time, "2025")
  expect_equal(p$value, 8.1)
  expect_equal(p$unit, "percent")
})

test_that("Persian digits, geography, indicator and units are classified", {
  p <- classify_stat_claim("نرخ بیکاری ایران در سال ۲۰۲۵ برابر با ۸٫۱ درصد بود.")
  expect_equal(p$domain, "labour")
  expect_equal(p$indicator, "unemployment rate")
  expect_equal(p$geo_code, "IRN")
  expect_equal(p$time, "2025")
  expect_equal(p$value, 8.1)
  expect_equal(p$unit, "percent")
})

test_that("latest intent is retained instead of inventing a year", {
  p <- classify_stat_claim("According to WHO, Iran's latest life expectancy is 77.3 years.")
  expect_equal(p$domain, "health")
  expect_equal(p$indicator, "life expectancy at birth")
  expect_equal(p$geo_code, "IRN")
  expect_equal(p$time_mode, "latest")
  expect_true(is.na(p$time))
  expect_equal(p$source_id, "who")
  expect_equal(p$unit, "years")
})



test_that("omitted year defaults to the latest available time and records that the choice was inferred", {
  p <- classify_stat_claim("Iran's population was 91 million.")
  expect_equal(p$indicator, "total population")
  expect_equal(p$geo_code, "IRN")
  expect_equal(p$time_mode, "latest")
  expect_true(is.na(p$time))
  expect_true(isTRUE(p$time_inferred))
})

test_that("verification with no year selects the latest matching official observation", {
  official <- data.frame(
    value = c(90000000, 91000000),
    indicator = c("total population", "total population"),
    country = c("Iran", "Iran"),
    year = c(2023, 2025),
    unit = c("persons", "persons"),
    source = c("Example NSO", "Example NSO"),
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Iran's population was 91 million.", evidence = official)
  expect_equal(x$status, "SUPPORTED")
  expect_equal(x$reference$time, "2025")
  expect_true(isTRUE(x$claim$time_inferred))
  expect_match(x$message, "No reference year was specified")
})

test_that("implicit latest is resolved within the matching indicator context", {
  official <- data.frame(
    value = c(91000000, 30000000),
    indicator = c("total population", "labour force total"),
    country = c("Iran", "Iran"),
    year = c(2025, 2026),
    unit = c("persons", "persons"),
    source = c("Example NSO", "Example NSO"),
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Iran's population was 91 million.", evidence = official)
  expect_equal(x$status, "SUPPORTED")
  expect_equal(x$reference$indicator, "total population")
  expect_equal(x$reference$time, "2025")
})

test_that("magnitudes are normalized for population-style claims", {
  p <- classify_stat_claim("Iran's population reached 91 million in 2025.")
  expect_equal(p$indicator, "total population")
  expect_equal(p$value, 91000000)
  expect_equal(p$value_scale, 1e6)
  expect_equal(p$unit, "persons")
})

test_that("offline evidence verification distinguishes supported and mismatch", {
  official <- data.frame(
    value = c(7.8, 8.2),
    indicator = c("unemployment rate", "unemployment rate"),
    country = c("Iran", "Iran"),
    year = c(2025, 2024),
    unit = c("percent", "percent"),
    source = c("Example NSO", "Example NSO"),
    stringsAsFactors = FALSE
  )
  a <- verify_claim("Iran's unemployment rate was 7.8 percent in 2025.", evidence = official)
  b <- verify_claim("Iran's unemployment rate was 8.1 percent in 2025.", evidence = official)
  expect_s3_class(a, "stat_claim_verification")
  expect_equal(a$status, "SUPPORTED")
  expect_equal(a$reference$time, "2025")
  expect_equal(b$status, "MISMATCH")
  expect_equal(b$reference$value, 7.8)
})

test_that("latest verification selects the most recent matching official row", {
  official <- data.frame(
    value = c(76.2, 76.8),
    indicator = c("life expectancy at birth", "life expectancy at birth"),
    country = c("Iran", "Iran"),
    year = c(2023, 2024),
    unit = c("years", "years"),
    source = c("Example health authority", "Example health authority"),
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Iran's latest life expectancy was 76.8 years.", evidence = official)
  expect_equal(x$status, "SUPPORTED")
  expect_equal(x$reference$time, "2024")
})

test_that("provider suggestions are domain-aware", {
  p <- classify_stat_claim("Iran's life expectancy was 77 years in 2025.")
  plan <- suggest_stat_providers(p)
  expect_equal(plan$id[1], "who")
  expect_true("worldbank" %in% plan$id)
})

test_that("age ranges and rate denominators are not mistaken for claim values", {
  p <- classify_stat_claim("Youth unemployment among people aged 15-24 was 20.5 percent in Iran in 2025.")
  expect_equal(p$value, 20.5)
  expect_equal(p$dimensions$age_group, "15-24")
  q <- classify_stat_claim("Maternal mortality in Iran was 18 per 100,000 live births in 2023.")
  expect_equal(q$value, 18)
  expect_equal(q$unit, "per 100,000")
})

test_that("requested sex dimension is preferred over aggregate or other-sex rows", {
  official <- data.frame(
    value = c(77.0, 78.4, 75.6),
    indicator = rep("life expectancy at birth", 3),
    country = rep("Iran", 3),
    year = rep(2024, 3),
    unit = rep("years", 3),
    source = rep("Example health authority", 3),
    sex = c("Both sexes", "Female", "Male"),
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Female life expectancy in Iran was 78.4 years in 2024.", evidence = official)
  expect_equal(x$status, "SUPPORTED")
  expect_equal(x$reference$value, 78.4)
})

test_that("WHO acronym detection does not confuse ordinary word who", {
  p <- classify_stat_claim("People who were unemployed in Iran represented 8.1 percent in 2025.")
  expect_true(is.na(p$source_id))
  q <- classify_stat_claim("According to WHO, Iran's life expectancy was 77 years in 2024.")
  expect_equal(q$source_id, "who")
})

test_that("an explicitly named unsupported auto-source is not silently replaced", {
  x <- verify_claim("According to IMF, Iran's inflation rate was 30 percent in 2025.")
  expect_equal(x$status, "NOT_FOUND")
  expect_match(x$message, "IMF")
})

test_that("matching number with contradictory known unit is not accepted as evidence", {
  official <- data.frame(
    value = 10,
    indicator = "unemployment rate",
    country = "Iran",
    year = 2025,
    unit = "persons",
    source = "Example NSO",
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Iran's unemployment rate was 10 percent in 2025.", evidence = official)
  expect_true(x$status %in% c("NOT_FOUND", "AMBIGUOUS"))
})


test_that("automatic provider routing does not require a named source", {
  p <- classify_stat_claim("Life expectancy in Iran was 77 years in 2023.")
  route <- .sf_resolve_provider_request(p, "auto", require_named_source = TRUE)
  expect_equal(route$mode, "auto")
  expect_true("who" %in% route$ids)
  expect_true("worldbank" %in% route$ids)
})

test_that("a named supported source is respected by automatic routing", {
  p <- classify_stat_claim("According to WHO, life expectancy in Iran was 77 years in 2023.")
  route <- .sf_resolve_provider_request(p, "auto", require_named_source = TRUE)
  expect_equal(route$mode, "named")
  expect_equal(route$ids, "who")
})

test_that("all routing exposes every full automatic-verification provider", {
  p <- classify_stat_claim("Life expectancy in Iran was 77 years in 2023.")
  route <- .sf_resolve_provider_request(p, "all", require_named_source = TRUE)
  expect_equal(route$mode, "all")
  expect_true(all(c("worldbank", "who", "unsd_sdg", "eurostat") %in% route$ids))
})

test_that("explicit routing cannot silently contradict a named source", {
  x <- verify_claim(
    "According to WHO, life expectancy in Iran was 77 years in 2023.",
    providers = "worldbank"
  )
  expect_equal(x$status, "NOT_FOUND")
  expect_equal(x$provider_mode, "explicit")
  expect_match(x$message, "exclude that source")
})

test_that("offline evidence records its source-selection mode", {
  official <- data.frame(
    value = 77,
    indicator = "life expectancy at birth",
    country = "Iran",
    year = 2023,
    unit = "years",
    source = "Example health authority",
    stringsAsFactors = FALSE
  )
  x <- verify_claim("Life expectancy in Iran was 77 years in 2023.", evidence = official)
  expect_equal(x$status, "SUPPORTED")
  expect_equal(x$provider_mode, "evidence")
  expect_equal(x$reference$source, "Example health authority")
})