normalize: do not walk up

this is for safety as symlink might otherwise be stripped

Author: scivision

+stdlib/+java/normalize.m

@@ -1,55 +1,34 @@
-function n = normalize(p)
+%% NATIVE.NORMALIZE normalize path
+% 2-3x as fast as non-native backends
+
+function n = normalize(apath)
 arguments
-  p (1,1) string
+  apath (1,1) string
 end
 
-n = stdlib.posix(p);
-
-uncslash = ispc() && startsWith(n, "//");
-
-% use split to remove /../ and /./ and duplicated /
-parts = split(n, '/');
+parts = split(apath, ["/", filesep]);
 i0 = 1;
-if strncmp(n, "/", 1)
-  n = "/";
-elseif ispc() && strlength(n) >= 2 && ~stdlib.strempty(stdlib.root_name(p))
+if startsWith(apath, "/" | filesep)
+  n = extractBefore(apath, 2);
+elseif ispc() && strlength(apath) >= 2 && ~stdlib.strempty(stdlib.root_name(apath))
   n = parts(1);
   i0 = 2;
 else
   n = "";
 end
 
 for i = i0:length(parts)
-  if parts(i) == ".."
-    if n == ""
-      n = parts(i);
-    elseif endsWith(n, "..")
-      n = n + "/" + parts(i);
-    else
-      j = strfind(n, "/");
-      if isempty(j)
-        n = "";
-      else
-        n = n{1}(1:j(end)-1);
-      end
-    end
-  elseif all(parts(i) ~= [".", ""])
+  if ~ismember(parts(i), [".", ""])
     if n == ""
       n = parts(i);
-    elseif n == "/"
+    elseif ismember(n, ["/", filesep])
       n = n + parts(i);
     else
-      n = n + "/" + parts(i);
+      n = n + filesep + parts(i);
     end
   end
 end
 
-if uncslash
-  n = strcat("/", n);
-end
-
-n = fullfile(n);
-
 if stdlib.strempty(n)
   n = ".";
 end

+stdlib/+native/normalize.m

@@ -0,0 +1,22 @@
+function n = normalize(apath)
+
+n = string.empty;
+
+exe = stdlib.perl_exe();
+if stdlib.strempty(exe)
+  return
+end
+
+cmd = sprintf('"%s" -MFile::Spec -e "print File::Spec->canonpath( ''%s'' )"', exe, apath);
+
+[s, r] = system(cmd);
+if s == 0
+  n = string(r);
+  if strlength(n) == 0
+    n = ".";
+  end
+else
+  n = string.empty;
+end
+
+end

+stdlib/+perl/normalize.m

@@ -1,14 +1,9 @@
 function n = normalize(file)
-arguments
-  file (1,1) string
-end
 
 try
-  n = string(py.os.path.normpath(file));
-  n = strip(n, 'right', '/');
-  if ispc()
-    n = strip(n, 'right', filesep);
-  end
+  n = string(py.str(py.pathlib.Path(file)));
 catch
   n = "";
 end
+
+end

+stdlib/+python/normalize.m

@@ -1,7 +1,7 @@
 %% NORMALIZE remove redundant elements of path
 %
-% normalize(p) remove redundant elements of path p
-% path need not exist, normalized path is returned
+% normalize(p) remove redundant elements of path p. Does not walk up ".." to be safe.
+% Path need not exist. Does not access physical filesystem.
 %
 %%% Inputs
 % * p: path to normalize
@@ -13,7 +13,7 @@
 function [n, b] = normalize(file, backend)
 arguments
   file string
-  backend (1,:) string = ["native", "java", "python"]
+  backend (1,:) string = ["native", "python", "perl"]
 end
 
 o = stdlib.Backend(mfilename(), backend);

+stdlib/normalize.m

@@ -1,7 +1,14 @@
-function n = normalize(p)
+function n = normalize(file)
 
-% https://docs.oracle.com/en/java/javase/21/docs/api/java.base/java/nio/file/Path.html#normalize()
-o = javaPathObject(p).normalize();
-n = stdlib.drop_slash(jPosix(o));
+n = string(javaPathObject(file).normalize());
+
+if stdlib.strempty(n)
+  n = ".";
+else
+  n = strip(n, 'right', '/');
+  if ispc()
+    n = strip(n, 'right', filesep);
+  end
+end
 
 end

example/+javafun/normalize.m

@@ -2,7 +2,7 @@
     BenchmarkNormalize < matlab.perftest.TestCase
 
 properties
-P = './../asdf/../../././//'
+P = './a/././//'
 end
 
 properties(TestParameter)

example/BenchmarkNormalize.m

@@ -1,9 +1,9 @@
 function [r, s] =  BenchmarkNormalizeRun()
 tname = "BenchmarkNormalize";
 
-r.same = run_bench(tname);
-s.exist = sampleSummary(r.same);
-disp(sortrows(s.exist, "Median"))
+r = run_bench(tname);
+s = sampleSummary(r);
+disp(sortrows(s, "Median"))
 end
 
 

example/BenchmarkNormalizeRun.m

@@ -14,14 +14,9 @@
 
 p = {
   {"", "."}, ...
-  {"a/..", "."}, ...
+  {"a/..", "a/.."}, ...
   {"/a/b/", fullfile("/a", "b")},  ...
   {"a/b/", fullfile("a", "b")}, ...
-  {"a/../c", "c"}, ...
-  {"a/b/../c", fullfile("a", "c")}, ...
-  {"a/b/../../c", "c"}, ...
-  {"a/b/../../c/..", "."}, ...
-  {"a/b/../../c/../..", ".."}, ...
   {"a////b", fullfile("a", "b")}, ...
   {".a", ".a"}, ...
   {"..a", "..a"}, ...
@@ -31,9 +26,6 @@
   {"../a", fullfile("..", "a")}
 };
 
-if ispc
-  p{end+1} = {"//a/b/", "\\a\b"};
-end
 end
 
 function d = init_drop_slash()