Don't consider files that MATLAB wouldn't

In project mode, further exclude files that would not be considered
by MATLAB.

docs/cli.html

@@ -95,6 +95,58 @@ <h3>Project mode</h3>
 	directories/files you're responsible for.
       </div>
 
+      <div>
+	Please note that MISS_HIT will <i>not</i> consider files and
+	directories that would be invisible to MATLAB in project
+	mode. Consider for example a projet layout like this:
+<pre>
+  /
+    miss_hit.cfg
+    file1.m
+    foo/
+      file2.m
+    +bar/
+      file3.m
+</pre>
+      </div>
+
+      <div>
+	The configuration file contains the following information:
+	<pre>entrypoint "Potato" {}</pre>
+      </div>
+
+      <div>
+	When running MISS_HIT without project mode, all files are
+	considered. However, when running MISS_HIT in project mode,
+	only files <tt>file1.m</tt> and <tt>file3.m</tt> are
+	considered. File <tt>file2.m</tt> is not considered because
+	directory <tt>foo</tt> is not on the path, as declared in our
+	entrypoint.
+      </div>
+
+      <div>
+	If you want to include <tt>file2.m</tt> in your analysis, then
+	you have to include that directory in your entry point, like
+	so:
+<pre>
+entrypoint "Potato" {
+  paths {
+    ".",
+    "foo"
+  }
+}
+</pre>
+      </div>
+
+      <div>
+	This behaviour is consistent with how MATLAB deals with its
+	search path. For more information consult:
+	<ul>
+	  <li><a href="https://www.mathworks.com/help/matlab/matlab_env/files-and-folders-that-matlab-accesses.html"><img src="assets/external-link.svg"> Files and Folders that MATLAB Accesses</a></li>
+	  <li><a href="https://www.mathworks.com/help/matlab/matlab_oop/organizing-classes-in-folders.html"><img src="assets/external-link.svg"> Folders Containing Class Definitions</a></li>
+	</ul>
+      </div>
+
       <h3>Common options</h3>
       <div>
 	All MISS_HIT tools (except mh_diff) have these options. Some

miss_hit_core/command_line.py

@@ -221,30 +221,50 @@ def execute(mh, options, extra_options, back_end, process_slx=True):
                                       "not exist." %
                                       options.entry_point)
 
+            # Get PATH
+            item_list = n_ep.get_path()
+
             if options.debug_show_path:
                 print("Using the following PATH:")
-                for path in n_ep.get_path():
+                for path in item_list:
                     print("> %s" % os.path.relpath(path))
 
-            item_list = n_ep.get_path()
-
-            # If the user has supplied files/dirs to analyze, we only
-            # do that if they are part of _this_ entrypoint.
-            if options.files:
-                items_in_path = set()
-                for path_root in item_list:
-                    for path, _, files in os.walk(path_root):
-                        items_in_path.add(os.path.normpath(path))
-                        for f in files:
-                            items_in_path.add(
+            # Determine relevant files based on these
+            # directories. This is a bit more complex than
+            # "everything".
+            #
+            # See
+            # https://www.mathworks.com/help/matlab/matlab_env/files-and-folders-that-matlab-accesses.html
+
+            items_in_path = set()
+            files_in_path = set()
+            for path_root in item_list:
+                for path, dirs, files in os.walk(path_root):
+                    items_in_path.add(os.path.normpath(path))
+                    for f in files:
+                        if f.endswith(".m") or f.endswith(".slx"):
+                            files_in_path.add(
                                 os.path.normpath(os.path.join(path, f)))
+                    irrelevant_dirs = set(d for d in dirs
+                                          if not (d.startswith("+") or
+                                                  d.startswith("@")))
+                    for idir in irrelevant_dirs:
+                        dirs.remove(idir)
+            items_in_path |= files_in_path
 
+            if options.files:
+                # If the user has supplied files/dirs to analyze, we
+                # only do that if they are part of _this_ entrypoint.
                 item_list = list(options.files)
                 for item in item_list:
                     if os.path.abspath(item) not in items_in_path:
                         mh.command_line_error("'%s' is not part of "
                                               "entry point %s" %
                                               (item, options.entry_point))
+            else:
+                # Otherwise we look at all applicable files on the
+                # path.
+                item_list = list(sorted(files_in_path))
 
         else:
             # Without an entry point, we build a minimally sufficient

tests/projects/irrelevant/+good/irrelevant/bad.m

@@ -0,0 +1 @@
+if just ++ not ++ matlab

tests/projects/irrelevant/@thing/irrelevant/bad.m

@@ -0,0 +1 @@
+if just ++ not ++ matlab

tests/projects/irrelevant/README.md

@@ -0,0 +1,9 @@
+When using an entry-point, we should not pick up files from our path
+that MATLAB wouldn't pick up either.
+
+See:
+* https://www.mathworks.com/help/matlab/matlab_env/files-and-folders-that-matlab-accesses.html
+* https://www.mathworks.com/help/matlab/matlab_oop/organizing-classes-in-folders.html
+
+Specifically if a subdirectory not starting with + or @ is in a
+directory on the path, we do not consider it.

tests/projects/irrelevant/cmdline

@@ -0,0 +1 @@
+--entry-point=potato

tests/projects/irrelevant/irrelevant/bad.m

@@ -0,0 +1 @@
+if just ++ not ++ matlab

tests/projects/irrelevant/miss_hit.cfg

@@ -0,0 +1,3 @@
+project_root
+
+entrypoint "potato" {}

tests/projects/irrelevant/output.txt

@@ -0,0 +1,11 @@
+=== STYLE ===
+MISS_HIT Style Summary: 2 file(s) analysed, everything seems fine
+=== LINT ===
+MISS_HIT Lint Summary: 2 file(s) analysed, everything seems fine
+=== METRICS ===
+=== Code metric by file:
+
+* Code metrics for file /home/flosch/miss_hit/tests/projects/irrelevant/+good/good.m:
+
+* Code metrics for file /home/flosch/miss_hit/tests/projects/irrelevant/@thing/thing.m:
+MISS_HIT Metric Summary: 2 file(s) analysed, everything seems fine