Exclude LaTeX figures and tables from word count

DESCRIPTION

@@ -2,20 +2,23 @@ Package: wordcountaddin
 Type: Package
 Title: Word counts and readability statistics in R markdown documents
 Version: 0.3.0.9000
-Authors@R: c(person("Ben", "Marwick", 
+Authors@R: c(person("Ben", "Marwick",
                   email = "[email protected]",
                   role = c("aut", "cre")),
-            person("JooYoung", "Seo", 
-                  email = "[email protected]", 
+            person("JooYoung", "Seo",
+                  email = "[email protected]",
                   role = "ctb"),
-            person("Henrik", "Bengtsson", 
-                  email = "[email protected]", 
-                  role = "ctb"))
+            person("Henrik", "Bengtsson",
+                  email = "[email protected]",
+                  role = "ctb"),
+            person("Florian S.", "Schaffner",
+                  email = "[email protected]",
+                  role = "ctb"),)
 Maintainer: Ben Marwick <[email protected]>
-Description: An addin for RStudio that will count the words and characters 
+Description: An addin for RStudio that will count the words and characters
     in a plain text document. It is designed for use with RMarkdown
-    documents and will exclude YAML header content, code chunks and inline 
-    code from the counts. It also computes readability statistics so you can 
+    documents and will exclude YAML header content, code chunks and inline
+    code from the counts. It also computes readability statistics so you can
     get an idea of how easy or difficult your text is to read.
 License: MIT + file LICENSE
 LazyData: TRUE

R/hello.R

@@ -180,9 +180,13 @@ prep_text <- function(text){
   # don't include html tags
   text <- gsub("<.+?>|</.+?>", "", text)
 
-    # don't include percent signs because they trip up stringi
+  # don't include percent signs because they trip up stringi
   text <- gsub("%", "", text)
 
+  # don't include figures and tables inserted using plain LaTeX code
+  text <- gsub("\\\\begin\\{figure\\}(.*?)\\\\end\\{figure\\}", "", text)
+  text <- gsub("\\\\begin\\{table\\}(.*?)\\\\end\\{table\\}", "", text)
+
   # don't include LaTeX \eggs{ham}
   # how to do? problem with capturing \x
 

tests/testthat/test_wordcountaddin.R

@@ -213,6 +213,17 @@ test_that("Word count is correct for text with % sign", {
                "|Word count      |9         |9             |")
 })
 
+
+test_that("Word count is correct for text with figures included using LaTeX code", {
+  # test for escaping the percent sign in plain text
+  text_with_figures <- "One \\begin{figure} \\caption{text} \\label{text} \\includegraphics[width=\\textwidth]{figure.png} \\end{figure} Two \\begin{figure} \\caption{text} \\label{text} \\includegraphics[width=\\textwidth]{figure.png} \\end{figure} Three"
+
+  text_stats_percent_chr_out <- text_stats_chr(text_with_figures)
+  expect_equal(text_stats_percent_chr_out[3],
+               "|Word count      |3         |3             |")
+})
+
+
 test_that("Word count is a single integer for a Rmd file when using word_count", {
   # test that we can word count on a file
   the_rmd_word_count <- word_count(filename = test_path("test_wordcountaddin.Rmd"))