mex: parent 10x faster

Author: scivision

+stdlib/parent.m

@@ -1,5 +1,11 @@
 %% PARENT parent directory of path
+% optional: mex
 %
+%% Examples:
+% parent("a/b/c") == "a/b"
+% parent("a/b/c/") == "a/b"
+%
+% MEX is about 10x faster than plain Matlab for this function
 
 function p = parent(pth)
 arguments

buildfile.m

@@ -97,6 +97,7 @@ function publishTask(context)
 srcs = {
 ["src/is_admin.cpp", "src/admin_fs.cpp"] ...
 "src/is_char_device.cpp", ...
+["src/parent.cpp", "src/parent_fs.cpp", normal], ...
 "src/relative_to.cpp", ...
 "src/proximate_to.cpp", ...
 ["src/is_wsl.cpp", linux], ...

example/bench_parent.m

@@ -1,13 +1,7 @@
 f = mfilename("fullpath") + ".m";
-%f = tempname;
 
-fno = @() stdlib.parent(f, false);
-fjava = @() stdlib.parent(f, true);
+fno = @() stdlib.parent(f);
 
 t_no = timeit(fno);
-t_java = timeit(fjava);
 
 disp("No Java: " + t_no + " s")
-disp("Java: " + t_java + " s")
-
-disp("Java is " + t_no/t_java + " times faster")

octave_build.m

@@ -15,6 +15,7 @@ function octave_build(overwrite)
 fullfile(d, "is_wsl.cpp"), ...
 fullfile(d, "is_rosetta.cpp"), ...
 fullfile(d, "is_admin.cpp"), ...
+fullfile(d, "parent.cpp"), ...
 fullfile(d, "proximate_to.cpp"), ...
 fullfile(d, "relative_to.cpp"), ...
 };
@@ -24,11 +25,13 @@ function octave_build(overwrite)
 for s = srcs
   src = s{1};
   [~, n] = fileparts(src);
+  bin = fullfile(t, [n, ".oct"]);
 
-  if ~overwrite && isfile(fullfile(t, [n, ".oct"]))
+  if ~overwrite && stdlib.get_modtime(src) < stdlib.get_modtime(bin)
     continue
   end
 
-  disp(["mkoctfile: ", src, " => ", n, ".oct"])
-  mkoctfile(src, ["-I", inc], "--output", fullfile(t, n))
+  disp(["mkoctfile: ", src, " => ", bin])
+  delete(bin)
+  mkoctfile(src, ["-I", inc], "--output", bin)
 end

src/ffilesystem.h

@@ -16,5 +16,8 @@ bool fs_is_symlink(std::string_view);
 bool fs_create_symlink(std::string_view, std::string_view);
 std::string fs_read_symlink(std::string_view);
 
+std::string fs_parent(std::string_view);
+std::string fs_root_name(std::string_view);
+
 bool fs_win32_is_symlink(std::string_view);
 std::string fs_shortname(std::string_view);

src/normalize.cpp

@@ -10,8 +10,6 @@
 std::string fs_drop_slash(std::string_view in)
 {
   // drop all trailing "/" and duplicated internal "/"
-  if (fs_is_url(in))
-    return {};
 
   std::filesystem::path p(in);
 

src/octave/parent.cpp

@@ -0,0 +1,21 @@
+#include <octave/oct.h>
+
+#include <string>
+#include <filesystem>
+
+#include "pure.cpp"
+#include "normalize.cpp"
+
+
+DEFUN_DLD (parent, args, nargout,
+           "get parent directory")
+{
+  if (args.length() != 1){
+    octave_stdout << "Oct: One input required\n";
+    return octave_value("");
+  }
+
+  std::string out = std::filesystem::path(fs_drop_slash(args(0).string_value())).parent_path().generic_string();
+
+  return octave_value(out);
+}

src/parent.cpp

@@ -0,0 +1,44 @@
+#include "mex.hpp"
+#include "mexAdapter.hpp"
+
+#include <string>
+
+#include <vector>
+#include <memory>
+
+#include <filesystem>
+
+#include "ffilesystem.h"
+
+
+class MexFunction : public matlab::mex::Function {
+public:
+  void operator()(matlab::mex::ArgumentList outputs, matlab::mex::ArgumentList inputs) {
+// boilerplate engine & ArrayFactory setup
+    std::shared_ptr<matlab::engine::MATLABEngine> matlabEng = getEngine();
+
+    matlab::data::ArrayFactory factory;
+// wrangle inputs
+    std::string in;
+
+    if (inputs.size() != 1) {
+      matlabEng->feval(u"error", 0,
+        std::vector<matlab::data::Array>({ factory.createScalar("Mex: One input required") }));
+    }
+    if ((inputs[0].getType() == matlab::data::ArrayType::MATLAB_STRING && inputs[0].getNumberOfElements() == 1)){
+        matlab::data::TypedArray<matlab::data::MATLABString> stringArr = inputs[0];
+        in = stringArr[0];
+    } else if (inputs[0].getType() == matlab::data::ArrayType::CHAR){
+        matlab::data::CharArray charArr = inputs[0];
+        in.assign(charArr.begin(), charArr.end());
+    } else {
+        matlabEng->feval(u"error", 0,
+        std::vector<matlab::data::Array>({ factory.createScalar("Mex: First input must be a scalar string or char vector") }));
+    }
+
+// actual function algorithm / computation
+    std::string out = fs_parent(in);
+
+    outputs[0] = factory.createScalar(out);
+  }
+};

src/parent_fs.cpp

@@ -0,0 +1,27 @@
+#include <string>
+#include <string_view>
+
+#include <filesystem>
+
+#include "ffilesystem.h"
+
+std::string fs_parent(std::string_view in)
+{
+  std::string out;
+  if(in.empty())
+    out = ".";
+  else
+    out = std::filesystem::path(fs_drop_slash(in)).parent_path().generic_string();
+
+  // handle "/" and other no parent cases
+  if (out.empty()){
+    if (!in.empty() && in.front() == '/')
+      out = "/";
+    else
+      out = ".";
+  }
+
+  // make x: x:/
+  if (fs_is_windows() && out.length() == 2 && !fs_root_name(out).empty())
+    out.push_back('/');
+}

src/pure.cpp

@@ -46,6 +46,11 @@ std::string fs_as_posix(std::string_view path)
   return p.generic_string();
 }
 
+std::string fs_root_name(std::string_view path)
+{
+  return std::filesystem::path(path).root_name().generic_string();
+}
+
 
 bool fs_is_url(std::string_view path)
 {