Add CRLF support to LineIterator.

[• rafael]

Summary

The MRI scripts have to work with CRLF, and in general it is probably a good idea to support this in a core utility like LineIterator.

[chandlerc]

So, I'm OK with this... but I wonder what it does to the performance of the line iterator. Hopefully we never care.

I've made a minor suggestion below that should at least make one of the cases quite a bit simpler.

lib/Support/LineIterator.cpp

63–65

This suggests that skipLineEnd should return whether it did indeed skip a line end?

[Eugene.Zelenko]

Committed in r221153.

lib/Support/LineIterator.cpp

 //===- LineIterator.cpp - Implementation of line iteration ----------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 
 #include "llvm/Support/LineIterator.h"
 #include "llvm/Support/MemoryBuffer.h"
 
 using namespace llvm;
 
+static bool isAtLineEnd(const char *P) {
+  if (*P == '\n')
+    return true;
+  if (*P == '\r' && *(P + 1) == '\n')
+    return true;
+  return false;
+}
+
+static bool skipIfAtLineEnd(const char *&P) {
+  if (*P == '\n') {
+    ++P;
+    return true;
+  }
+  if (*P == '\r' && *(P + 1) == '\n') {
+    P += 2;
+    return true;
+  }
+  return false;
+}
+
 line_iterator::line_iterator(const MemoryBuffer &Buffer, bool SkipBlanks,
                              char CommentMarker)
     : Buffer(Buffer.getBufferSize() ? &Buffer : nullptr),
       CommentMarker(CommentMarker), SkipBlanks(SkipBlanks), LineNumber(1),
       CurrentLine(Buffer.getBufferSize() ? Buffer.getBufferStart() : nullptr,
                   0) {
   // Ensure that if we are constructed on a non-empty memory buffer that it is
   // a null terminated buffer.
   if (Buffer.getBufferSize()) {
     assert(Buffer.getBufferEnd()[0] == '\0');
     // Make sure we don't skip a leading newline if we're keeping blanks
-    if (SkipBlanks || *Buffer.getBufferStart() != '\n')
+    if (SkipBlanks || !isAtLineEnd(Buffer.getBufferStart()))
       advance();
   }
 }
 
 void line_iterator::advance() {
   assert(Buffer && "Cannot advance past the end!");
 
   const char *Pos = CurrentLine.end();
-  assert(Pos == Buffer->getBufferStart() || *Pos == '\n' || *Pos == '\0');
+  assert(Pos == Buffer->getBufferStart() || isAtLineEnd(Pos) || *Pos == '\0');
 
-  if (*Pos == '\n') {
-    ++Pos;
+  if (skipIfAtLineEnd(Pos))
     ++LineNumber;
-  }
-  if (!SkipBlanks && *Pos == '\n') {
+  if (!SkipBlanks && isAtLineEnd(Pos)) {
     // Nothing to do for a blank line.
   } else if (CommentMarker == '\0') {
     // If we're not stripping comments, this is simpler.
-    size_t Blanks = 0;
-    while (Pos[Blanks] == '\n')
-      ++Blanks;
-    Pos += Blanks;
-    LineNumber += Blanks;
+    while (skipIfAtLineEnd(Pos))
+      ++LineNumber;
   } else {
     // Skip comments and count line numbers, which is a bit more complex.
     for (;;) {
-      if (*Pos == '\n' && !SkipBlanks)
+      if (isAtLineEnd(Pos) && !SkipBlanks)
         break;
       if (*Pos == CommentMarker)
         do {
           ++Pos;
-        } while (*Pos != '\0' && *Pos != '\n');
-      if (*Pos != '\n')
+        } while (*Pos != '\0' && !isAtLineEnd(Pos));
+      if (!skipIfAtLineEnd(Pos))
         break;
-      ++Pos;
       ++LineNumber;
     }
   }
 
   if (*Pos == '\0') {
     // We've hit the end of the buffer, reset ourselves to the end state.
     Buffer = nullptr;
     CurrentLine = StringRef();
     return;
   }
 
   // Measure the line.
   size_t Length = 0;
-  while (Pos[Length] != '\0' && Pos[Length] != '\n') {
+  while (Pos[Length] != '\0' && !isAtLineEnd(&Pos[Length])) {
     ++Length;
   }
 
   CurrentLine = StringRef(Pos, Length);
 }

test/Object/Inputs/mri-crlf.mri

+; this file intentionally has crlf line endings
+end

test/Object/mri-crlf.test

+; RUN: llvm-ar -M  < %S/Inputs/mri-crlf.mri