wrap lines containing magrittr operators such as %>% (#68)

first, change `%>%` to `%>       %` in mask_comments(), i.e., add a number of spaces before the closing %, so that deparse() can break the line after %; then remove the extra spaces in unmask_source()

close #54 and close #62

Co-authored-by: Yihui Xie <[email protected]>

NEWS

@@ -1,5 +1,9 @@
 		   CHANGES IN formatR VERSION 1.9
 
+NEW FEATURES
+
+  o Lines will be wrapped after operators `%>%`, `%T%`, `%$%`, and `%<>%` now
+  (thanks, @g4challenge #54, @jzelner #62, @edlee123 #68).
 
 		   CHANGES IN formatR VERSION 1.8
 

R/tidy.R

@@ -72,9 +72,12 @@ tidy_source = function(
     n2 = attr(regexpr('\n*$', one), 'match.length')
   }
   on.exit(.env$line_break <- NULL, add = TRUE)
-  if (comment) text = mask_comments(text, width.cutoff, blank, wrap)
+  # insert enough spaces into infix operators such as %>% so the lines can be
+  # broken after the operators
+  spaces = paste(rep(' ', max(10, width.cutoff)), collapse = '')
+  if (comment) text = mask_comments(text, width.cutoff, blank, wrap, spaces)
   text.mask = tidy_block(text, width.cutoff, arrow && length(grep('=', text)))
-  text.tidy = if (comment) unmask_source(text.mask) else text.mask
+  text.tidy = if (comment) unmask_source(text.mask, spaces) else text.mask
   text.tidy = reindent_lines(text.tidy, indent)
   if (brace.newline) text.tidy = move_leftbrace(text.tidy)
   # restore new lines in the beginning and end
@@ -104,7 +107,7 @@ tidy_block = function(text, width = getOption('width'), arrow = FALSE) {
 }
 
 # Restore the real source code from the masked text
-unmask_source = function(text.mask) {
+unmask_source = function(text.mask, spaces) {
   if (length(text.mask) == 0) return(text.mask)
   m = .env$line_break
   if (!is.null(m)) text.mask = gsub(m, '\n', text.mask)
@@ -123,6 +126,8 @@ unmask_source = function(text.mask) {
   # remove white spaces on blank lines
   text.mask = gsub(blank.comment2, '\\1\\2', text.mask)
   text.tidy = gsub(pat.comment, '', text.mask)
+  # restore infix operators such as %>%
+  text.tidy = gsub(paste0('(%)(', infix_ops, ')', spaces, '(%)\\s*(\n)'), '\\1\\2\\3\\4', text.tidy)
   # inline comments should be terminated by $ or \n
   text.tidy = gsub(paste(inline.comment, '(\n|$)', sep = ''), '  \\1\\2', text.tidy)
   # the rest of inline comments should be appended by \n

R/utils.R

@@ -18,7 +18,7 @@ replace_assignment = function(exp) {
 }
 
 ## mask comments to cheat R
-mask_comments = function(x, width, keep.blank.line, wrap = TRUE) {
+mask_comments = function(x, width, keep.blank.line, wrap = TRUE, spaces) {
   d = utils::getParseData(parse_source(x))
   if (nrow(d) == 0 || (n <- sum(d$terminal)) == 0) return(x)
   d = d[d$terminal, ]
@@ -69,10 +69,14 @@ mask_comments = function(x, width, keep.blank.line, wrap = TRUE) {
     if (blank[i] > 0)
       d.text[i] = paste(c(d.text[i], rep(blank.comment, blank[i])), collapse = '\n')
   }
+  # break lines after some infix operators such as %>%
+  d.text = gsub(paste0('^(%)(', infix_ops, ')(%)$'), paste0('\\1\\2', spaces, '\\3'), d.text)
 
   unlist(lapply(split(d.text, d.line), paste, collapse = ' '), use.names = FALSE)
 }
 
+infix_ops = '[>$]|T>|<>'
+
 # no blank lines before an 'else' statement!
 move_else = function(x) {
   blank = grepl('^\\s*$', x)

tests/testit/test-tidy.R

@@ -140,3 +140,22 @@ assert(
   'line breaks in strings are preserved instead of being replaced by \\n',
   tidy.res(x1) %==% x1
 )
+
+# tests for magrittr newlines
+
+x1 = '
+iris %>% group_by(Species) %>%
+summarize(meanlen = mean(Sepal.Length)) %$% arrange(meanlen) %>%meanlen
+'
+
+x2 = '
+iris %>%
+  group_by(Species) %>%
+  summarize(meanlen = mean(Sepal.Length)) %$%
+  arrange(meanlen) %>%
+  meanlen
+'
+
+assert('magrittr lines are wrapped after the pipes', {
+  (paste(tidy.res(x1, indent = 2), collapse = '\n') %==% x2)
+})