Optionally include empty last line when string ends with newline.

Author: toinehartman

src/org/rascalmpl/library/String.rsc

@@ -767,7 +767,7 @@ str trimFinalNewlines(str input, list[str] lineseps = newLineCharacters) {
 }
 
 @synopsis{Split a string in <text, newline> pairs for each line.}
-list[tuple[str, str]] separateLines(str input, list[str] lineseps = newLineCharacters) {
+list[tuple[str, str]] separateLines(str input, bool includeEmptyLastLine = false, list[str] lineseps = newLineCharacters) {
     orderedSeps = reverse(sort(lineseps, bySize));
 
     list[tuple[str, str]] lines = [];
@@ -781,8 +781,8 @@ list[tuple[str, str]] separateLines(str input, list[str] lineseps = newLineChara
     }
 
     // last line
-    if (str nl <- orderedSeps, nl == input[-size(nl)..]) {
-        lines += <input[next..next+size(nl)], "">;
+    if (next < size(input) || includeEmptyLastLine) {
+        lines += <input[next..], "">;
     }
 
     return lines;
@@ -793,8 +793,8 @@ str mergeLines(list[tuple[str, str]] lines)
     = ("" | it + line + sep | <line, sep> <- lines);
 
 @synopsis{Process the text of a string per line, maintaining the original newline characters.}
-str perLine(str input, str(str) lineFunc, list[str] lineseps = newLineCharacters)
-    = mergeLines([<lineFunc(l), nl> | <l, nl> <- separateLines(input, lineseps=lineseps)]);
+str perLine(str input, str(str) lineFunc, bool includeEmptyLastLine = false, list[str] lineseps = newLineCharacters)
+    = mergeLines([<lineFunc(l), nl> | <l, nl> <- separateLines(input, includeEmptyLastLine=includeEmptyLastLine, lineseps=lineseps)]);
 
 @synopsis{Trim trailing non-newline whitespace from each line in a multi-line string.}
 str trimTrailingWhitespace(str input, list[str] lineseps = newLineCharacters) {

src/org/rascalmpl/library/lang/rascal/tests/library/String.rsc

@@ -282,4 +282,18 @@ test bool trimFinalNewlineTestWhiteSpace() = trimFinalNewlines("a\n\n\nb\n\n ")
 test bool trimTrailingWhitespaceTest() = trimTrailingWhitespace("a  \nb\t\n  c  \n") == "a\nb\n  c\n";
 
 // perLine
-test bool perLineTest() = perLine("a\nb\r\nc\n\r\n", str(str line) { return line + "x"; }) == "ax\nbx\r\ncx\nx\r\nx";
+test bool perLineTest() = perLine("a\nb\r\nc\n\r\n", str(str line) { return line + "x"; }) == "ax\nbx\r\ncx\nx\r\n";
+
+// separateLines
+test bool separateLinesTestSimple() = separateLines("a\nb\r\nc\n\r\n") == [<"a", "\n">, <"b", "\r\n">, <"c", "\n">, <"", "\r\n">];
+test bool separateLinesTestSimpleWithLast() = separateLines("a\nb\r\nc\n\r\n", includeEmptyLastLine=true) == [<"a", "\n">, <"b", "\r\n">, <"c", "\n">, <"", "\r\n">, <"", "">];
+test bool separateLinesTestNoFinalNewline() = separateLines("a\nb\r\nc") == [<"a", "\n">, <"b", "\r\n">, <"c", "">];
+test bool separateLinesTestNoFinalNewlineNoEmpty() = separateLines("a\nb\r\nc", includeEmptyLastLine=true) == [<"a", "\n">, <"b", "\r\n">, <"c", "">];
+test bool separateLinesTestOnlyNewlines() = separateLines("\n\r\n\n\r\n") == [<"", "\n">, <"", "\r\n">, <"", "\n">, <"", "\r\n">];
+test bool separateLinesTestNoNewlines() = separateLines("abc") == [<"abc", "">];
+
+// substrings
+test bool substringsTestEmpty() = substrings("") == {};
+test bool substringsTestSingle() = substrings("a") == {};
+test bool substringsTestTwo() = substrings("ab") == {"a", "b"};
+test bool substringsTestThree() = substrings("abc") == {"a", "b", "c", "ab", "bc"};