improved fixed form lexing for defined operators

3 files changed, 59 insertions, 23 deletions

diff --git a/include/flang/Parse/Lexer.h b/include/flang/Parse/Lexer.h
index f4a7ad647f..3b158a3d43 100644
--- a/include/flang/Parse/Lexer.h
+++ b/include/flang/Parse/Lexer.h
@@ -226,6 +226,10 @@ class Lexer {
   void LexFixedFormIdentifier(const fixedForm::KeywordMatcher &Matcher,
                               Token &Tok);
 
+  /// LexPossibleDefinedOperator - Tries to lex a defined operator
+  /// that begins with '.', returns true if sucessfull.
+  bool LexPossibleDefinedOperator(Token &Result);
+
   /// LexFORMATDescriptor - Lex a format desriptor.
   void LexFORMATDescriptor(Token &Result);
 
diff --git a/lib/Parse/Lexer.cpp b/lib/Parse/Lexer.cpp
index c314ab9989..083b38cfa1 100644
--- a/lib/Parse/Lexer.cpp
+++ b/lib/Parse/Lexer.cpp
@@ -882,7 +882,7 @@ void Lexer::FormDefinedOperatorTokenWithChars(Token &Result) {
   unsigned TokLen;
   llvm::StringRef FullOp;
 
-  if (!Text.IsInCurrentAtom(TokStart)) {
+  if (!Text.IsInCurrentAtom(TokStart) || Result.needsCleaning()) {
     llvm::SmallVector<llvm::StringRef, 2> Spelling;
     FormTokenWithChars(Result, tok::unknown);
     getSpelling(Result, Spelling);
@@ -1004,6 +1004,50 @@ void Lexer::LexFixedFormIdentifier(const fixedForm::KeywordMatcher &Matcher,
     Tok.setFlag(Token::NeedsCleaning);
 }
 
+bool Lexer::LexPossibleDefinedOperator(Token &Result) {
+  auto Char = getNextChar();
+  if (Features.FixedForm) {
+    while(isHorizontalWhitespace(Char)) {
+      Result.setFlag(Token::NeedsCleaning);
+      Char = getNextChar();
+    }
+  }
+  if (!isLetter(Char))
+    return false;
+  // Match [A-Za-z]*, we have already matched '.'.
+  if(Features.FixedForm) {
+    for(;;) {
+      Char = getNextChar();
+      if(!isLetter(Char)) {
+        if(!isHorizontalWhitespace(Char))
+          break;
+        Result.setFlag(Token::NeedsCleaning);
+      }
+    }
+  } else {
+    while (isLetter(Char))
+      Char = getNextChar();
+  }
+
+  if (Char != '.') {
+    Diags.Report(SourceLocation::getFromPointer(TokStart), diag::err_defined_operator_missing_end);
+    FormTokenWithChars(Result, tok::unknown);
+    return true;
+  }
+
+  // FIXME: fixed-form
+  Char = getNextChar();
+  if (Char == '_') {
+    // Parse the kind.
+    do {
+      Char = getNextChar();
+    } while (isIdentifierBody(Char) || isDecimalNumberBody(Char));
+  }
+
+  FormDefinedOperatorTokenWithChars(Result);
+  return true;
+}
+
 /// LexFormatDescriptor - Lex the remainder of a format descriptor.
 void Lexer::LexFORMATDescriptor(Token &Result) {
   // Match [_A-Za-z0-9]*, we have already matched [A-Za-z$]
@@ -1247,28 +1291,8 @@ void Lexer::LexTokenInternal(Token &Result, bool IsPeekAhead) {
     return LexTokenInternal(Result, IsPeekAhead);
 
   case '.':
-    Char = getNextChar();
-    if (isLetter(Char)) {
-      // Match [A-Za-z]*, we have already matched '.'.
-      while (isLetter(Char))
-        Char = getNextChar();
-
-      if (Char != '.') {
-        Diags.Report(SourceLocation::getFromPointer(TokStart),diag::err_defined_operator_missing_end);
-        FormTokenWithChars(Result, tok::unknown);
-        return;
-      }
-
-      Char = getNextChar();
-      if (Char == '_') {
-        // Parse the kind.
-        do {
-          Char = getNextChar();
-        } while (isIdentifierBody(Char) || isDecimalNumberBody(Char));
-      }
-
-      return FormDefinedOperatorTokenWithChars(Result);
-    }
+    if(LexPossibleDefinedOperator(Result))
+      return;
     // FALLTHROUGH
   case '0': case '1': case '2': case '3': case '4':
   case '5': case '6': case '7': case '8': case '9':
diff --git a/test/Lexer/fixedFormDefinedOperators.f b/test/Lexer/fixedFormDefinedOperators.f
new file mode 100644
index 0000000000..71b53735aa
--- /dev/null
+++ b/test/Lexer/fixedFormDefinedOperators.f
@@ -0,0 +1,8 @@
+      program hello
+C RUN: %flang -fsyntax-only %s
+      integer i
+      logical l
+      if(l . and. l) i = 2
+      if(l . a nd . i . e q. 0) then
+      end if
+      end