trim(): fix handling of indentation-only lines shorter than min_… (#164)

* Improve handling of lines containing only indentation

* Update NEWS

* Handle short indentation-only lines; fixes #163

* Add test for empty intermediate line handling

Co-authored-by: Jim Hester <james.f.hester@gmail.com>

NEWS.md

@@ -16,7 +16,7 @@
 * `single_quote()` `double_quote()` and `backtick()` all return `NA` for `NA`
   inputs (#135).
 
-* Improve `trim()`'s handling of lines containing only indentation (#162)
+* Improve `trim()`'s handling of lines containing only indentation (#162, #163)
 
 # glue 1.3.1
 

src/trim.c

@@ -1,7 +1,12 @@
 #include "Rinternals.h"
 #include <stdbool.h>
 #include <stdlib.h>
-#include <string.h> // for strlen()
+#include <string.h> // for strlen(), strchr(), strncpy()
+
+const char* skip(const char* s, const char* delim) {
+  while (strchr(delim, s[0])) s++;
+  return s;
+}
 
 SEXP trim_(SEXP x) {
   size_t len = LENGTH(x);
@@ -80,8 +85,21 @@ SEXP trim_(SEXP x) {
         i += 2;
         continue;
       } else if (new_line) {
-        if (i + min_indent < str_len && (xx[i] == ' ' || xx[i] == '\t')) {
-          i += min_indent;
+        const char* after = skip(xx + i, "\t ");
+        const size_t skipped = after - (xx + i);
+        /*
+         * if the line consists only of tabs and spaces, and if the line is
+         * shorter than min_indent, copy the entire line and proceed to the
+         * next
+         */
+        if (after[0] == '\n' && skipped < min_indent) {
+          strncpy(str + j, xx + i, skipped);
+          i += skipped;
+          j += skipped;
+        } else {
+          if (i + min_indent < str_len && (xx[i] == ' ' || xx[i] == '\t')) {
+            i += min_indent;
+          }
         }
         new_line = false;
         continue;

tests/testthat/test-trim.R

@@ -136,4 +136,23 @@ test_that("lines containing only indentation are handled properly", {
          \tc"),
     " \ta\nb\n \t\n \tc"
   )
+  # A line shorter than min_indent that contains only indentation should not be
+  # trimmed, removed, or prepended to the next line.
+  expect_identical(
+    trim("
+       \ta
+       \tb
+      \t
+       \tc"),
+    "a\nb\n      \t\nc"
+  )
+  # Ensure empty intermedite lines are handled properly
+  expect_identical(
+    trim("
+       \ta
+       \tb
+
+       \tc"),
+    "a\nb\n\nc"
+  )
 })