Nothing
library(polmineR)
use("polmineR")
use(pkg = "RcppCWB", corpus = "REUTERS")
testthat::context("as.markdown")
test_that(
"as.markdown",
{
# Running these tests on CRAN would be too limiting for development of
# markdown pkg
skip_on_cran()
# p <- partition("REUTERS", id = "127", regex = TRUE)
# y <- as.markdown(p)
# if (.Platform$OS.type == "windows") y <- iconv(y, from = encoding(), to = "UTF-8")
#
# refdoc <- system.file(package = "polmineR", "fulltext", "reuters_as_markdown.txt")
# benchmark <- readLines(refdoc, warn = FALSE, encoding = "UTF-8")
# benchmark <- paste(benchmark, collapse = "\n")
# benchmark <- paste(benchmark, "\n", sep = "")
#
# expect_identical(nchar(y), nchar(benchmark))
# expect_identical(y, benchmark)
#
# y <- as.character(html(p))
# y <- gsub("\n\n", "\n", y)
#
# refdoc <- system.file(package = "polmineR", "fulltext", "reuters_as_html.html")
# benchmark <- readLines(refdoc, warn = FALSE, encoding = "UTF-8")
# benchmark <- paste(benchmark, collapse = "\n")
# benchmark <- gsub("\n\n", "\n", benchmark)
# benchmark <- paste(benchmark, "\n", sep = "")
# expect_identical(nchar(y), nchar(benchmark))
# expect_identical(y, benchmark)
# }
# )
#
#
# test_that(
# "plpr_partition",
# {
# skip_on_cran()
# m <- partition("GERMAPARLMINI", date = "2009-10-28", speaker = "Merkel", regex = TRUE) %>%
# as.speeches(s_attribute_name = "speaker", s_attribute_date = "date", gap = 100) %>%
# .[[2]]
# y <- as.markdown(m)
# if (.Platform$OS.type == "windows") y <- iconv(y, from = encoding(), to = "UTF-8")
#
# refdoc <- system.file(package = "polmineR", "fulltext", "merkel_as_markdown.txt")
# benchmark <- readLines(refdoc, warn = FALSE, encoding = "UTF-8")
# benchmark <- paste(benchmark, collapse = "\n")
# expect_identical(nchar(y), nchar(benchmark))
# expect_identical(y, benchmark)
#
# y <- as.character(read(m, beautify = TRUE))
# y <- gsub("\n\n", "\n", y)
#
# refdoc <- system.file(package = "polmineR", "fulltext", "merkel_as_html.html")
# benchmark <- readLines(refdoc, warn = FALSE, encoding = "UTF-8")
# benchmark <- paste(benchmark, collapse = "\n")
# benchmark <- paste(benchmark, "\n", sep = "")
# benchmark <- gsub("\n\n", "\n", benchmark)
# expect_identical(nchar(y), nchar(benchmark))
# expect_identical(y, benchmark)
}
)
test_that(
"read",
{
skip_on_cran()
# a <- corpus("REUTERS") %>%
# subset(places = "argentina") %>%
# html(meta = "places") %>%
# highlight(lightgreen = "higher") %>%
# tooltips(list(lightgreen = "Further information"))
#
# b <- corpus("REUTERS") %>%
# subset(places = "argentina") %>%
# read(
# meta = "places",
# highlight = list(lightgreen = "higher"),
# tooltips = list(lightgreen = "Further information")
# )
#
# expect_identical(a, b)
#
# y <- gsub("\n\n", "\n", b)
#
# refdoc <- system.file(package = "polmineR", "fulltext", "reading_reuters.html")
# benchmark <- readLines(refdoc, warn = FALSE, encoding = "UTF-8")
# benchmark <- paste(benchmark, collapse = "\n")
# benchmark <- paste(benchmark, "\n", sep = "")
# benchmark <- gsub("\n\n", "\n", benchmark)
# expect_identical(nchar(y), nchar(benchmark))
# expect_identical(as.character(y), benchmark)
}
)
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.