Merge pull request #11204 from DougGregor/lex-multiline-nlescape-last-4.0

[4.0] [SE-0182][Lexer] Diagnose escaped newline at the end of the last line in multiline string

Author: DougGregor

include/swift/AST/DiagnosticsParse.def

@@ -142,6 +142,8 @@ NOTE(lex_multiline_string_indent_should_match_here,none,
       "should match %select{space|tab}0 here", (unsigned))
 NOTE(lex_multiline_string_indent_change_line,none,
       "change indentation of %select{this line|these lines}0 to match closing delimiter", (bool))
+ERROR(lex_escaped_newline_at_lastline,none,
+      "escaped newline at the last line is not allowed", ())
 
 ERROR(lex_invalid_character,none,
        "invalid character in source file", ())

lib/Parse/Lexer.cpp

@@ -1424,13 +1424,28 @@ getMultilineTrailingIndent(const Token &Str, DiagnosticEngine *Diags) {
       continue;
     case '\n':
     case '\r': {
-      auto bytes = start + 1;
-      auto length = end-(start+1);
-      
-      auto bytesLoc = Lexer::getSourceLoc(bytes);
-      auto string = StringRef(bytes, length);
-      
-      return std::make_tuple(string, bytesLoc);
+      start++;
+      auto startLoc = Lexer::getSourceLoc(start);
+      auto string = StringRef(start, end - start);
+
+      // Disallow escaped newline in the last line.
+      if (Diags) {
+        auto *Ptr = start - 1;
+        if (*Ptr == '\n') --Ptr;
+        if (*Ptr == '\r') --Ptr;
+        auto *LineEnd = Ptr + 1;
+        while (Ptr > begin && (*Ptr == ' ' || *Ptr == '\t')) --Ptr;
+        if (*Ptr == '\\') {
+          auto escapeLoc = Lexer::getSourceLoc(Ptr);
+          bool invalid = true;
+          while (*--Ptr == '\\') invalid = !invalid;
+          if (invalid)
+            Diags->diagnose(escapeLoc, diag::lex_escaped_newline_at_lastline)
+              .fixItRemoveChars(escapeLoc, Lexer::getSourceLoc(LineEnd));
+        }
+      }
+
+      return std::make_tuple(string, startLoc);
     }
     default:
       sawNonWhitespace = true;

test/Parse/multiline_errors.swift

@@ -28,6 +28,21 @@ _ = """
       // expected-note@-1{{should match tab here}}
       // expected-note@-3{{change indentation of this line to match closing delimiter}} {{1-1=	}}
 
+_ = """
+    \(42
+)
+    """ // expected-error@-1{{insufficient indentation of line in multi-line string literal}}
+        // expected-note@-1{{should match space here}}
+        // expected-note@-3{{change indentation of this line to match closing delimiter}} {{1-1=    }}
+
+_ = """
+    Foo
+\
+    Bar 
+    """ // expected-error@-2{{insufficient indentation of line in multi-line string literal}}
+        // expected-note@-1{{should match space here}}
+        // expected-note@-4{{change indentation of this line to match closing delimiter}} {{1-1=    }}
+
 // a tab is not the same as multiple spaces for de-indentation
 _ = """
   Thirteen
@@ -126,3 +141,46 @@ _ = """
   line two
   """
   // expected-error@-3 {{invalid escape sequence in literal}}
+
+_ = """
+  line one
+  line two\
+  """
+  // expected-error@-2 {{escaped newline at the last line is not allowed}} {{11-12=}}
+
+_ = """
+  \\\	   
+  """
+  // expected-error@-2 {{escaped newline at the last line is not allowed}} {{5-10=}}
+
+_ = """
+  \(42)\		
+  """
+  // expected-error@-2 {{escaped newline at the last line is not allowed}} {{8-11=}}
+
+_ = """
+  foo\
+  """
+  // expected-error@-2 {{escaped newline at the last line is not allowed}} {{6-7=}}
+
+_ = """
+  foo\  """
+  // expected-error@-1 {{escaped newline at the last line is not allowed}} {{6-7=}}
+
+_ = """
+  foo\
+  """ // OK because LF + CR is two new lines.
+
+_ = """
+\
+  """
+  // expected-error@-2 {{escaped newline at the last line is not allowed}} {{1-2=}}
+  // expected-error@-3{{insufficient indentation of line in multi-line string literal}}
+  // expected-note@-3{{should match space here}}
+  // expected-note@-5{{change indentation of this line to match closing delimiter}} {{1-1=  }}
+
+_ = """\
+  """
+  // FIXME: Bad diagnostics
+  // expected-error@-3 {{multi-line string literal content must begin on a new line}}
+  // expected-error@-4 {{escaped newline at the last line is not allowed}} {{8-9=}}

test/Parse/multiline_normalize_newline.swift

@@ -0,0 +1,70 @@
+// RUN: %target-swift-frontend -dump-parse %s 2>&1 | %FileCheck %s
+
+// CR
+_ = """"""
+//CHECK: string_literal_expr {{.*}} value=""
+
+_ = """  test  """
+//CHECK: string_literal_expr {{.*}} value="test"
+
+// CR+LF
+_ = """
+    """
+//CHECK: string_literal_expr {{.*}} value=""
+
+_ = """
+  test
+  """
+//CHECK: string_literal_expr {{.*}} value="test"
+
+// CR+LF
+_ = """
+    """
+//CHECK: string_literal_expr {{.*}} value=""
+_ = """
+  test
+  test
+  """
+//CHECK: string_literal_expr {{.*}} value="test\ntest"
+
+// LF+CR
+_ = """
+    foo
+    foo
+    """
+//CHECK: string_literal_expr {{.*}} value="\nfoo\n\nfoo\n"
+
+// LF+CR+LF
+_ = """
+
+    foo
+
+    foo
+
+    """
+//CHECK: string_literal_expr {{.*}} value="\nfoo\n\nfoo\n"
+
+// Mixed no-indent.
+_ = """
+<LF
+<LF<CR
+<CR+LF
+"""
+//CHECK: string_literal_expr {{.*}} value="<LF\n<LF\n<CR\n<CR+LF"
+
+// Mixed indent.
+_ = """
+	 <LF
+	 <LF	 <CR
+	 <CR+LF
+	 """
+//CHECK: string_literal_expr {{.*}} value="<LF\n<LF\n<CR\n<CR+LF"
+
+// Empty line CR, CR+LF, LF.
+_ = """
+   foo
+
+
+   bar
+   """
+//CHECK: string_literal_expr {{.*}} value="foo\n\n\n\nbar"

test/Parse/multiline_string.swift

@@ -82,6 +82,11 @@ _ = """
 """
 // CHECK: "\\"
 
+_ = """
+  \\
+  """
+// CHECK: "\\"
+
 _ = """
 
   ABC
@@ -122,7 +127,7 @@ _ = """
     \("""
       substring2 \          
       substring3
-      """)\
+      """)
     """) \
   whitepsace
   """