diff: avoid spurious diffs when two distinct dir entries compare equal

Problem reported by Christoph Anton Mitterer in:
http://lists.gnu.org/archive/html/bug-diffutils/2010-08/msg00000.html

* NEWS: Mention this bug fix.
* src/dir.c (compare_names_for_qsort): Fall back on file_name_cmp
if two distinct entries in the same directory compare equal.
(diff_dirs): Prefer a file_name_cmp match when available.
* tests/Makefile.am (TESTS): New test colliding-file-names.
* tests/colliding-file-names: New file.

4 files changed, 65 insertions, 2 deletions

diff --git a/NEWS b/NEWS
index 27b9886..1c41e9e 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,11 @@ GNU diffutils NEWS                                    -*- outline -*-
 
 * Noteworthy changes in release ?.? (????-??-??) [?]
 
+** Bug fixes
+
+  diff no longer reports spurious differences merely because two entries
+  in the same directory have names that compare equal in the current
+  locale, or compare equal because --ignore-file-name-case was given.
 
 * Noteworthy changes in release 3.0 (2010-05-03) [stable]
 
diff --git a/src/dir.c b/src/dir.c
index 5b4eaec..1b078ab 100644
--- a/src/dir.c
+++ b/src/dir.c
@@ -166,14 +166,16 @@ compare_names (char const *name1, char const *name2)
 	  : file_name_cmp (name1, name2));
 }
 
-/* A wrapper for compare_names suitable as an argument for qsort.  */
+/* Compare names FILE1 and FILE2 when sorting a directory.
+   Prefer filtered comparison, breaking ties with file_name_cmp.  */
 
 static int
 compare_names_for_qsort (void const *file1, void const *file2)
 {
   char const *const *f1 = file1;
   char const *const *f2 = file2;
-  return compare_names (*f1, *f2);
+  int diff = compare_names (*f1, *f2);
+  return diff ? diff : file_name_cmp (*f1, *f2);
 }
 
 /* Compare the contents of two directories named in CMP.
@@ -253,6 +255,41 @@ diff_dirs (struct comparison const *cmp,
 	     pretend the "next name" in that dir is very large.  */
 	  int nameorder = (!*names[0] ? 1 : !*names[1] ? -1
 			   : compare_names (*names[0], *names[1]));
+
+	  /* Prefer a file_name_cmp match if available.  This algorithm is
+	     O(N**2), where N is the number of names in a directory
+	     that compare_names says are all equal, but in practice N
+	     is so small it's not worth tuning.  */
+	  if (nameorder == 0)
+	    {
+	      int raw_order = file_name_cmp (*names[0], *names[1]);
+	      if (raw_order != 0)
+		{
+		  int greater_side = raw_order < 0;
+		  int lesser_side = 1 - greater_side;
+		  char const **lesser = names[lesser_side];
+		  char const *greater_name = *names[greater_side];
+		  char const **p;
+
+		  for (p = lesser + 1;
+		       *p && compare_names (*p, greater_name) == 0;
+		       p++)
+		    {
+		      int cmp = file_name_cmp (*p, greater_name);
+		      if (0 <= cmp)
+			{
+			  if (cmp == 0)
+			    {
+			      memmove (lesser + 1, lesser,
+				       (char *) p - (char *) lesser);
+			      *lesser = greater_name;
+			    }
+			  break;
+			}
+		    }
+		}
+	    }
+
 	  int v1 = (*handle_file) (cmp,
 				   0 < nameorder ? 0 : *names[0]++,
 				   nameorder < 0 ? 0 : *names[1]++);
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 6a4858c..242d501 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -1,6 +1,7 @@
 TESTS = \
   basic \
   binary \
+  colliding-file-names \
   help-version	\
   function-line-vs-leading-space \
   label-vs-func	\
diff --git a/tests/colliding-file-names b/tests/colliding-file-names
new file mode 100644
index 0000000..f14dce7
--- /dev/null
+++ b/tests/colliding-file-names
@@ -0,0 +1,20 @@
+#!/bin/sh
+# Check that diff responds well if a directory has multiple file names
+# that compare equal.
+
+: ${srcdir=.}
+. "$srcdir/init.sh"; path_prepend_ ../src
+
+mkdir d1 d2 || fail=1
+
+for i in abc abC aBc aBC Abc AbC ABc ABC; do
+ echo $i >d1/$i || fail=1
+done
+
+for i in ABC ABc AbC Abc aBC aBc abC abc; do
+ echo $i >d2/$i || fail=1
+done
+
+diff -r --ignore-file-name-case d1 d2 || fail=1
+
+Exit $fail