diff: ignore tabs consistently with expanding them

* src/io.c (find_and_hash_each_line):
* src/util.c (lines_differ):
Treat '\0', '\a', '\b', '\f', '\r', '\v' consistently with how
side.c treats them when expanding them, e.g., backspacing from
column 1 is a no-op when counting tab columns.
* tests/ignore-tab-expansion: New test.
* tests/Makefile.am (TESTS): Add it.

NEWS

@@ -19,6 +19,10 @@ GNU diffutils NEWS                                    -*- outline -*-
     cp /proc/cmdline t; cmp -s /proc/cmdline t || echo files differ
   [bug present since "the beginning"]
 
+  diff -E no longer mishandles some input lines containing '\a', '\b',
+  '\f', '\r', '\v', or '\0'.
+  [bug present since 2.8]
+
   diff -ly no longer mishandles non-ASCII input
   [bug#64461 introduced in 2.9]
 

src/io.c

@@ -292,7 +292,7 @@ find_and_hash_each_line (struct file_data *current)
         case IGNORE_TAB_EXPANSION_AND_TRAILING_SPACE:
         case IGNORE_TRAILING_SPACE:
           {
-	    intmax_t column = 0;
+	    intmax_t tab = 0, column = 0;
             for (unsigned char c; (c = *p++) != '\n'; )
               {
                 if (ig_white_space & IGNORE_TRAILING_SPACE
@@ -318,19 +318,29 @@ find_and_hash_each_line (struct file_data *current)
                   switch (c)
                     {
                     case '\b':
-		      column = (column ? column : tabsize) - 1;
+		      if (0 < column)
+			column--;
+		      else if (0 < tab)
+			{
+			  tab--;
+			  column = tabsize - 1;
+			}
                       break;
 
                     case '\t':
                       c = ' ';
                       repetitions = tabsize - column % tabsize;
+		      tab += column / tabsize + 1;
 		      column = 0;
                       break;
 
                     case '\r':
-                      column = 0;
+		      tab = column = 0;
                       break;
 
+		    case '\0': case '\a': case '\f': case '\v':
+		      break;
+
                     default:
                       column++;
                       break;

src/util.c

@@ -1212,14 +1212,38 @@ lines_differ (char const *s1, char const *s2)
           if (c1 != c2)
             break;
         }
-      if (c1 == '\n')
-        return false;
 
-      column++;
-      if (c1 == '\t' || column == tabsize)
+      switch (c1)
 	{
+	case '\n':
+	  return false;
+
+	case '\r':
+	  tab = column = 0;
+	  break;
+
+	case '\b':
+	  if (0 < column)
+	    column--;
+	  else if (0 < tab)
+	    {
+	      tab--;
+	      column = tabsize - 1;
+	    }
+	  break;
+
+	case '\0': case '\a': case '\f': case '\v':
+	  break;
+
+	default:
+	  column++;
+	  if (column < tabsize)
+	    break;
+	  FALLTHROUGH;
+	case '\t':
 	  tab++;
 	  column = 0;
+	  break;
 	}
     }
 

tests/Makefile.am

@@ -16,6 +16,7 @@ TESTS = \
   invalid-re	\
   function-line-vs-leading-space \
   ignore-matching-lines \
+  ignore-tab-expansion \
   label-vs-func	\
   large-subopt \
   new-file \

tests/ignore-tab-expansion

@@ -0,0 +1,16 @@
+#!/bin/sh
+# Test ignoring tab expansion.
+
+. "${srcdir=.}/init.sh"; path_prepend_ ../src
+
+fail=0
+
+for p in '\b' '\r' '\t ' '\n'; do
+  printf "$p"'\b\tx\n' >a || framework_failure_
+  printf "$p"'\b        x\n' >b || framework_failure
+
+  diff -E a b >out || fail=1
+  compare /dev/null out || fail=1
+done
+
+Exit $fail