tests/testthat/test-as.dictionary.R

test_that("as.dictionary works", {
    df <- data.frame(
        word = letters[1:6],
        sentiment = c("neg", "neg", "neutral", "neutral", "pos", "pos"),
        lexicon = rep("madeup", 6),
        stringsAsFactors = FALSE
    )
    lis <- list(neg = c("a", "b"), 
                neutral = c("c", "d"), 
                pos = c("e", "f"))
    expect_equal(
        dictionary(lis),
        as.dictionary(df, format = "tidytext")
    )
    df <- data.frame(
        word = letters[1:6],
        sentiment = c("neg", "neg", "neutral", "neutral", "pos", "pos"),
        lexicon = rep("madeup", 6),
        stringsAsFactors = TRUE
    )
    expect_equal(
        dictionary(lis),
        as.dictionary(df, format = "tidytext")
    )
    expect_equal(
        dictionary(lis),
        as.dictionary(lis)
    )
})

test_that("as.dictionary warnings and errors", {
    df <- data.frame(
        word = letters[1:6],
        sentiment2 = c("neg", "neg", "neutral", "neutral", "pos", "pos"),
        lexicon = rep("madeup", 6)
    )
    expect_error(
        as.dictionary(df, format = "tidytext"),
        "data\\.frame must contain word and sentiment columns"
    )
    df <- data.frame(
        word = letters[1:6],
        sentiment = c("neg", "neg", "neutral", "neutral", "pos", "pos"),
        lexicon = c(rep("madeup", 3), rep("other", 3))
    )
    expect_warning(
        as.dictionary(df, format = "tidytext"),
        "you may be mixing different dictionaries"
    )
    df <- data.frame(
        word = letters[1:6],
        sentiment = NA,
        lexicon = "madeup"
    )
    expect_error(
        as.dictionary(df, format = "tidytext"),
        "sentiment values are missing"
    )
})


test_that("as.dictionary function works for tidytext sentiment", {
    skip_if_not_installed("tidytext")
    data(sentiments, package = "tidytext")
    expect_true(
        is.dictionary(as.dictionary(sentiments))
    )
})


test_that("as.dictionary function passes additional arguments to dictionary", {
    df <- data.frame(
        word = c("A", "B"),
        sentiment = c("pos", "neg")
    )
    
    dict <- as.dictionary(df, tolower = FALSE)
    
    expect_equal(
        dict[["pos"]], "A"
    )
})

test_that("options for tidytext only currently supported", {
    df <- data.frame(
        word = c("A", "B"),
        sentiment = c("pos", "neg")
    )
    expect_error(
        as.dictionary(df, format = "koRpus"),
        "'arg' should be (one of )*[“\"]tidytext[”\"]"
    )
})

test_that("as.dictionary.list works", {
    l <- list(Word = c("A", "B"),
              Sentiment = c("Pos Word", "Neg_Word"))
    expect_identical(
        dictionary(l),
        as.dictionary(l)
    )
    expect_identical(
        dictionary(l, tolower = FALSE),
        as.dictionary(l, tolower = FALSE)
    )
    expect_identical(
        meta(as.dictionary(l, separator = " "), type = "all")$object$separator,
        " "
    )
    expect_identical(
        meta(as.dictionary(l, separator = "_"), type = "all")$object$separator,
        "_"
    )
})
quanteda/quanteda documentation built on April 15, 2024, 7:59 a.m.