Merge 09ca7a2 into 0dd13d7

Docma-TU · Feb 26, 2019 · 30d3643 · 30d3643
2 parents 0dd13d7 + 09ca7a2
commit 30d3643
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 5 deletions.
diff --git a/R/as.textmeta.corpus.R b/R/as.textmeta.corpus.R
@@ -56,7 +56,13 @@ as.textmeta.corpus <- function(corpus, cols, dateFormat = "%Y-%m-%d", idCol = "i
     length(dateCol) == 1, length(titleCol) == 1, length(textCol) == 1,
     length(duplicateAction) == 1)
 
-  if(missing(cols)) cols <- colnames(corpus$documents)
+  # restructure corpus using 1.4.x format
+  corpus <- list(documents = data.frame(texts = quanteda::texts(corpus), 
+                                        quanteda::docvars(corpus),
+                                        stringsAsFactors = FALSE),
+                 metadata = quanteda::metacorpus(corpus))
+
+  if (missing(cols)) cols <- colnames(corpus$documents)
   cols <- setdiff(cols, c(idCol, dateCol, titleCol, textCol))
 
   if(!(idCol %in% colnames(corpus$documents))){

diff --git a/tests/testthat/test_as.textmeta.corpus.R b/tests/testthat/test_as.textmeta.corpus.R
@@ -18,8 +18,4 @@ test_that("as.textmeta.corpus", {
 
   obj <- as.textmeta.corpus(corp)
   expect_true(is.textmeta(obj))
-
-  corp$documents <- corp$documents[,-1]
-  obj <- as.textmeta.corpus(corp)
-  expect_true(is.textmeta(obj))
 })
diff --git a/vignettes/Vignette.Rmd b/vignettes/Vignette.Rmd
@@ -13,6 +13,7 @@ editor_options:
 vignette: >
   %\VignetteIndexEntry{Vignette tosca}
   %\VignetteEngine{knitr::rmarkdown}
+  %\VignetteEncoding{UTF-8}
 ---
 \newpage
 # Introduction