Make path::normalize() to preserve ./, invalidate paths starting with \, / on Windows

7 files changed, 112 insertions, 58 deletions

diff --git a/butl/buildfile b/butl/buildfile
index 5c5aae3..6e259c8 100644
--- a/butl/buildfile
+++ b/butl/buildfile
@@ -21,6 +21,7 @@ lib{butl}:                               \
 {hxx     txx    }{ prefix-map          } \
 {hxx ixx     cxx}{ process             } \
 {hxx         cxx}{ sha256              } \
+{hxx            }{ small-vector        } \
 {hxx     txx    }{ string-table        } \
 {hxx         cxx}{ timestamp           } \
 {hxx         cxx}{ triplet             } \
diff --git a/butl/path b/butl/path
index 2a32c70..2e243ff 100644
--- a/butl/path
+++ b/butl/path
@@ -808,7 +808,11 @@ namespace butl
     // Normalize the path and return *this. Normalization involves collapsing
     // the '.' and '..'  directories if possible, collapsing multiple
     // directory separators, and converting all directory separators to the
-    // canonical form.
+    // canonical form. If cur_empty is true then collapse relative paths
+    // representing the current directory (for example, '.', './', 'foo/..')
+    // to an empty path. Otherwise convert it to the canonical form (./ on
+    // POSIX systems). Note that a non-empty path cannot become an empty one
+    // in the latter case.
     //
     // If actual is true, then for case-insensitive filesystems obtain the
     // actual spelling of the path. Only an absolute path can be actualized.
@@ -818,7 +822,7 @@ namespace butl
     // etc.) are returned in their actual spelling.
     //
     basic_path&
-    normalize (bool actual = false);
+    normalize (bool actual = false, bool cur_empty = false);
 
     // Make the path absolute using the current directory unless it is already
     // absolute. Return *this.
diff --git a/butl/path.txx b/butl/path.txx
index 5bcaea0..0cdaf5b 100644
--- a/butl/path.txx
+++ b/butl/path.txx
@@ -147,7 +147,7 @@ namespace butl
 
   template <typename C, typename K>
   basic_path<C, K>& basic_path<C, K>::
-  normalize (bool actual)
+  normalize (bool actual, bool cur_empty)
   {
     if (empty ())
       return *this;
@@ -273,12 +273,24 @@ namespace butl
         p += traits::directory_separator;
     }
 
-    if (tsep && (!p.empty () || abs)) // Distinguish "/"-empty and "."-empty.
+    if (tsep)
     {
       if (p.empty ())
       {
-        p += traits::directory_separator;
-        ts = -1;
+        // Distinguish "/"-empty and "."-empty.
+        //
+        if (abs)
+        {
+          p += traits::directory_separator;
+          ts = -1;
+        }
+        else if (!cur_empty) // Collapse to canonical current directory.
+        {
+          p = ".";
+          ts = 1; // Canonical separator is always first.
+        }
+        else // Collapse to empty path.
+          ts = 0;
       }
       else
         ts = 1; // Canonical separator is always first.
@@ -313,11 +325,11 @@ namespace butl
 
 #ifdef _WIN32
     // We do not support any special Windows path name notations like in C:abc,
-    // \\?\c:\abc, \\server\abc and \\?\UNC\server\abc (more about them at
-    // https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247.aspx).
+    // /, \, /abc, \abc, \\?\c:\abc, \\server\abc and \\?\UNC\server\abc (more
+    // about them in "Naming Files, Paths, and Namespaces" MSDN article).
     //
     if ((n > 2 && s[1] == ':' && s[2] != '\\' && s[2] != '/') ||
-        (n > 1 && s[0] == '\\' && s[1] == '\\'))
+        (n > 0 && (s[0] == '\\' || s[0] == '/')))
     {
       if (exact)
         return data_type ();
diff --git a/butl/process.cxx b/butl/process.cxx
index c75cfbd..cf4b26d 100644
--- a/butl/process.cxx
+++ b/butl/process.cxx
@@ -171,7 +171,7 @@ namespace butl
       ep = path (move (s)); // Move back into result.
 
       if (norm)
-        ep.normalize (); //@@ NORM
+        ep.normalize ();
 
       return exists (ep.string ().c_str ());
     };
@@ -212,10 +212,17 @@ namespace butl
       e = strchr (b, traits::path_separator);
 
       // Empty path (i.e., a double colon or a colon at the beginning or end
-      // of PATH) means search in the current dirrectory.
+      // of PATH) means search in the current dirrectory. Silently skip
+      // invalid paths.
       //
-      if (search (b, e != nullptr ? e - b : strlen (b)))
-        return r;
+      try
+      {
+        if (search (b, e != nullptr ? e - b : strlen (b)))
+          return r;
+      }
+      catch (const invalid_path&)
+      {
+      }
     }
 
     // If we were given a fallback, try that.
@@ -575,10 +582,17 @@ namespace butl
       e = strchr (b, traits::path_separator);
 
       // Empty path (i.e., a double colon or a colon at the beginning or end
-      // of PATH) means search in the current dirrectory.
+      // of PATH) means search in the current dirrectory. Silently skip
+      // invalid paths.
       //
-      if (search (b, e != nullptr ? e - b : strlen (b)))
-        return r;
+      try
+      {
+        if (search (b, e != nullptr ? e - b : strlen (b)))
+          return r;
+      }
+      catch (const invalid_path&)
+      {
+      }
     }
 
     // Finally, if we were given a fallback, try that. This case is similar to
diff --git a/tests/dir-iterator/buildfile b/tests/dir-iterator/buildfile
index 5c8736e..fe310ad 100644
--- a/tests/dir-iterator/buildfile
+++ b/tests/dir-iterator/buildfile
@@ -2,7 +2,6 @@
 # copyright : Copyright (c) 2014-2016 Code Synthesis Ltd
 # license   : MIT; see accompanying LICENSE file
 
-exe{driver}: cxx{driver} ../../butl/lib{butl}
-exe{driver}: test.arguments = $src_root
+exe{driver}: cxx{driver} ../../butl/lib{butl} test{testscript}
 
 include ../../butl/
diff --git a/tests/dir-iterator/testscript b/tests/dir-iterator/testscript
new file mode 100644
index 0000000..42dc1fa
--- /dev/null
+++ b/tests/dir-iterator/testscript
@@ -0,0 +1,16 @@
+# file      : tests/dir-iterator/testscript
+# copyright : Copyright (c) 2014-2016 Code Synthesis Ltd
+# license   : MIT; see accompanying LICENSE file
+
+test.options = -v
+
+: file
+:
+mkdir a;
+touch a/b;
+$* a 2>"reg     b"
+
+: dir
+:
+mkdir -p a/b;
+$* a 2>"dir     b"
diff --git a/tests/path/driver.cxx b/tests/path/driver.cxx
index fbcaf33..fe192ce 100644
--- a/tests/path/driver.cxx
+++ b/tests/path/driver.cxx
@@ -42,28 +42,24 @@ main ()
     return x.string () == s && x.representation () == r;
   };
 
+#ifndef _WIN32
   assert (test ("/", "/", "/"));
   assert (test ("//", "/", "/"));
   assert (test ("/tmp/foo", "/tmp/foo", "/tmp/foo"));
   assert (test ("/tmp/foo/", "/tmp/foo", "/tmp/foo/"));
   assert (test ("/tmp/foo//", "/tmp/foo", "/tmp/foo/"));
 
-#ifdef _WIN32
-  assert (test ("/\\", "/", "/"));
+  assert (dir_test ("/", "/", "/"));
+  assert (dir_test ("/tmp/foo/", "/tmp/foo", "/tmp/foo/"));
+  assert (dir_test ("tmp/foo", "tmp/foo", "tmp/foo/"));
+#else
   assert (test ("C:", "C:", "C:"));
   assert (test ("C:\\", "C:", "C:\\"));
   assert (test ("c:/", "c:", "c:/"));
   assert (test ("C:\\tmp\\foo\\", "C:\\tmp\\foo", "C:\\tmp\\foo\\"));
   assert (test ("C:\\tmp\\foo\\/\\", "C:\\tmp\\foo", "C:\\tmp\\foo\\"));
-#endif
 
-  assert (dir_test ("/", "/", "/"));
-  assert (dir_test ("/tmp/foo/", "/tmp/foo", "/tmp/foo/"));
-#ifndef _WIN32
-  assert (dir_test ("tmp/foo", "tmp/foo", "tmp/foo/"));
-#else
   assert (dir_test ("tmp\\foo", "tmp\\foo", "tmp\\foo\\"));
-
   assert (dir_test ("C:\\", "C:", "C:\\"));
   assert (dir_test ("C:\\tmp/foo\\", "C:\\tmp/foo", "C:\\tmp/foo\\"));
   assert (dir_test ("c:/tmp\\foo", "c:/tmp\\foo", "c:/tmp\\foo\\"));
@@ -131,13 +127,15 @@ main ()
 
   // base
   //
+  assert (path (".txt").base ().representation () == ".txt");
+  assert (path ("foo.txt.orig").base ().representation () == "foo.txt");
+
+#ifndef _WIN32
   assert (path ("/").base ().representation () == "/");
   assert (path ("/foo.txt").base ().representation () == "/foo");
   assert (path ("/foo.txt/").base ().representation () == "/foo/");
-  assert (path (".txt").base ().representation () == ".txt");
   assert (path ("/.txt").base ().representation () == "/.txt");
-  assert (path ("foo.txt.orig").base ().representation () == "foo.txt");
-#ifdef _WIN32
+#else
   assert (path ("C:").base ().representation () == "C:");
   assert (path ("C:\\foo.txt").base ().representation () == "C:\\foo");
   assert (path ("C:\\foo.txt\\").base ().representation () == "C:\\foo\\");
@@ -186,6 +184,7 @@ main ()
     assert (++i != p.rend () && *i == "foo");
     assert (++i == p.rend ());
   }
+#ifndef _WIN32
   {
     path p ("/foo/bar");
     path::iterator i (p.begin ());
@@ -202,7 +201,6 @@ main ()
     assert (++i != p.rend () && *i == "");
     assert (++i == p.rend ());
   }
-#ifndef _WIN32
   {
     path p ("/");
     path::iterator i (p.begin ());
@@ -244,16 +242,14 @@ main ()
       assert (test (++p.begin (), p.end ()) == "bar");
       assert (test (p.begin (), ++p.begin ()) == "foo/");
     }
+#ifndef _WIN32
     {
       path p ("/foo/bar");
       assert (test (p.begin (), p.end ()) == "/foo/bar");
       assert (test (++p.begin (), p.end ()) == "foo/bar");
       assert (test (++(++p.begin ()), p.end ()) == "bar");
 
-#ifndef _WIN32
       assert (test (p.begin (), ++p.begin ()) == "/");
-#endif
-
       assert (test (++p.begin (), ++(++p.begin ())) == "foo/");
       assert (test (++(++p.begin ()), ++(++(++p.begin ()))) == "bar");
     }
@@ -263,14 +259,10 @@ main ()
       assert (test (++p.begin (), p.end ()) == "foo/bar/");
       assert (test (++(++p.begin ()), p.end ()) == "bar/");
 
-#ifndef _WIN32
       assert (test (p.begin (), ++p.begin ()) == "/");
-#endif
-
       assert (test (++p.begin (), ++(++p.begin ())) == "foo/");
       assert (test (++(++p.begin ()), ++(++(++p.begin ()))) == "bar/");
     }
-#ifndef _WIN32
     {
       path p ("/");
       assert (test (p.begin (), p.end ()) == "/");
@@ -287,7 +279,6 @@ main ()
   assert ((path ("foo/") / path ("bar/")).representation () == "foo/bar/");
   assert ((path ("foo/") / path ()).representation () == "foo/");
 #else
-  assert ((path ("\\") / path ("tmp")).representation () == "\\tmp");
   assert ((path ("C:\\") / path ("tmp")).representation () == "C:\\tmp");
   assert ((path ("foo\\") / path ("bar")).representation () == "foo\\bar");
   assert ((path ("foo\\") / path ("bar\\")).representation () == "foo\\bar\\");
@@ -303,13 +294,16 @@ main ()
   assert (path ("..///foo").normalize ().representation () == "../foo");
   assert (path ("../../foo").normalize ().representation () == "../../foo");
   assert (path (".././foo").normalize ().representation () == "../foo");
-  assert (path (".").normalize ().representation () == "");
-  assert (path ("././").normalize ().representation () == "");
+  assert (path (".").normalize ().representation () == "./");
+  assert (path (".").normalize (false, true).representation () == "");
+  assert (path ("././").normalize ().representation () == "./");
+  assert (path ("././").normalize (false, true).representation () == "");
   assert (path ("./..").normalize ().representation () == "../");
   assert (path ("./../").normalize ().representation () == "../");
   assert (path ("../.").normalize ().representation () == "../");
   assert (path (".././").normalize ().representation () == "../");
-  assert (path ("foo/./..").normalize ().representation () == "");
+  assert (path ("foo/./..").normalize ().representation () == "./");
+  assert (path ("foo/./..").normalize (false, true).representation () == "");
   assert (path ("/foo/./..").normalize ().representation () == "/");
   assert (path ("/foo/./../").normalize ().representation () == "/");
   assert (path ("./foo").normalize ().representation () == "foo");
@@ -319,27 +313,34 @@ main ()
   assert (path ("..///foo").normalize ().representation () == "..\\foo");
   assert (path ("..\\../foo").normalize ().representation () == "..\\..\\foo");
   assert (path (".././foo").normalize ().representation () == "..\\foo");
-  assert (path (".").normalize ().representation () == "");
-  assert (path (".\\.\\").normalize ().representation () == "");
+  assert (path (".").normalize ().representation () == ".\\");
+  assert (path (".").normalize (false, true).representation () == "");
+  assert (path (".\\.\\").normalize ().representation () == ".\\");
+  assert (path (".\\.\\").normalize (false, true).representation () == "");
   assert (path ("./..").normalize ().representation () == "..\\");
   assert (path ("../.").normalize ().representation () == "..\\");
-  assert (path ("foo/./..").normalize ().representation () == "");
+  assert (path ("foo/./..").normalize ().representation () == ".\\");
+  assert (path ("foo/./..").normalize (false, true).representation () == "");
   assert (path ("C:/foo/./..").normalize ().representation () == "C:\\");
   assert (path ("C:/foo/./../").normalize ().representation () == "C:\\");
   assert (path ("./foo").normalize ().representation () == "foo");
   assert (path ("./foo\\").normalize ().representation () == "foo\\");
 
   assert (path ("C:\\").normalize ().representation () == "C:\\");
-  assert (path ("C:\\Foo12//Bar").normalize ().representation () == "C:\\Foo12\\Bar");
+
+  assert (path ("C:\\Foo12//Bar").normalize ().representation () ==
+          "C:\\Foo12\\Bar");
 #endif
 
   // comparison
   //
-  assert (path ("/") == path ("/"));
   assert (path ("./foo") == path ("./foo"));
   assert (path ("./foo/") == path ("./foo"));
   assert (path ("./boo") < path ("./foo"));
-#ifdef _WIN32
+
+#ifndef _WIN32
+  assert (path ("/") == path ("/"));
+#else
   assert (path (".\\foo") == path ("./FoO"));
   assert (path (".\\foo") == path ("./foo\\"));
   assert (path (".\\boo") < path (".\\Foo"));
@@ -371,19 +372,20 @@ main ()
     assert (test ("foo/bar", "foo"));
     assert (test ("foo/bar", "foo/"));
     assert (!test ("foo/bar", "bar"));
+
+#ifndef _WIN32
     assert (!test ("/foo-bar", "/foo"));
     assert (test ("/foo/bar", "/foo"));
     assert (test ("/foo/bar/baz", "/foo/bar"));
     assert (!test ("/foo/bar/baz", "/foo/baz"));
-#ifdef _WIN32
+    assert (test ("/", "/"));
+    assert (test ("/foo/bar/baz", "/"));
+#else
     assert (test ("c:", "c:"));
     assert (test ("c:", "c:\\"));
     assert (!test ("c:", "d:"));
     assert (test ("c:\\foo", "c:"));
     assert (test ("c:\\foo", "c:\\"));
-#else
-    assert (test ("/", "/"));
-    assert (test ("/foo/bar/baz", "/"));
 #endif
   }
 
@@ -400,19 +402,19 @@ main ()
     assert (test ("foo/bar", "bar"));
     assert (test ("foo/bar/", "bar/"));
     assert (!test ("foo/bar", "foo"));
+
+#ifndef _WIN32
+    assert (test ("/", "/"));
     assert (!test ("/foo-bar", "bar"));
     assert (test ("/foo/bar", "bar"));
     assert (test ("/foo/bar/baz", "bar/baz"));
     assert (!test ("/foo/bar/baz", "bar"));
-
-#ifdef _WIN32
+#else
     assert (test ("c:", "c:"));
     assert (test ("c:\\", "c:"));
     assert (!test ("d:", "c:"));
     assert (test ("c:\\foo", "foo"));
     assert (test ("c:\\foo\\", "foo\\"));
-#else
-    assert (test ("/", "/"));
 #endif
   }
 
@@ -424,11 +426,14 @@ main ()
       return path (p).leaf (path (d)).representation ();
     };
 
+#ifndef _WIN32
     assert (test ("/foo", "/") == "foo");
+    assert (test ("/foo/bar", "/foo/") == "bar");
+#endif
+
     //assert (test ("foo/bar", "foo") == "bar");
     assert (test ("foo/bar", "foo/") == "bar");
     assert (test ("foo/bar/", "foo/") == "bar/");
-    assert (test ("/foo/bar", "/foo/") == "bar");
   }
 
   // directory(path)
@@ -439,12 +444,15 @@ main ()
       return path (p).directory (path (l)).representation ();
     };
 
+#ifndef _WIN32
     assert (test ("/foo", "foo") == "/");
+    assert (test ("/foo/bar/baz", "bar/baz") == "/foo/");
+#endif
+
     assert (test ("foo/bar", "bar") == "foo/");
     assert (test ("foo/bar/", "bar/") == "foo/");
     assert (test ("foo/bar/", "bar") == "foo/");
     assert (test ("foo/bar/baz", "bar/baz") == "foo/");
-    assert (test ("/foo/bar/baz", "bar/baz") == "/foo/");
   }
 
   // relative