Merge pull request #192 from eWert-Online/fix-185

Glob: Make double asterisk match zero or more directories

CHANGES.md

@@ -4,6 +4,8 @@ Unreleased
 * Add the `[:alpha:]` character class in `Re.Perl` (#169)
 * Double asterisk (`**`) in `Re.Glob` (#172)
   Like `*` but also match `/` characters when `pathname` is set.
+* Double asterisk should match 0 or more directories unless in trailing
+  position. (#192, fixes #185)
 
 1.9.0 (05-Apr-2019)
 -------------------

lib/glob.ml

@@ -48,6 +48,26 @@ let of_string ~double_asterisk s : t =
     r
   in
 
+  (**
+   [read_ahead pattern] will attempt to read [pattern] and will return [true] if it was successful.
+   If it fails, it will return [false] and not increment the read index.
+  *)
+  let read_ahead pattern =
+    let pattern_len = String.length pattern in
+    (* if the pattern we are looking for exeeds the remaining length of s, return false immediately *)
+    if !i + pattern_len >= l then
+      false
+    else
+      try
+        for j = 0 to pattern_len - 1 do
+          let found = not (eos ()) && s.[!i + j] = pattern.[j] in
+          if not found then raise_no_trace Exit;
+        done;
+        i := !i + pattern_len;
+        true
+      with | Exit  -> false
+  in
+
   let char () =
     ignore (read '\\' : bool);
     if eos () then raise Parse_error;
@@ -75,7 +95,9 @@ let of_string ~double_asterisk s : t =
   in
 
   let piece () =
-    if read '*'
+    if double_asterisk && read_ahead "/**" && not (eos ())
+    then ManyMany
+    else if read '*'
     then if double_asterisk && read '*'
       then ManyMany
       else Many

lib_test/test_glob.ml

@@ -98,9 +98,20 @@ let _ =
   assert (re_match    (glob ~anchored "foo/**/bar") "foo/far/oof/bar");
   assert (re_match    (glob ~anchored "foo/**bar") "foo/far/oofbar");
   assert (re_match    (glob ~anchored "foo/**bar") "foo/bar");
+  assert (re_match    (glob ~anchored "foo/**bar") "foo/foobar");
   assert (re_match    (glob ~anchored "/**") "//foo");
   assert (re_match    (glob ~anchored "**") "foo//bar");
 
+  assert (re_match    (glob ~anchored "foo/bar/**/*.ml") "foo/bar/baz/foobar.ml");
+  assert (re_match    (glob ~anchored "foo/bar/**/*.ml") "foo/bar/foobar.ml");
+
+  assert (re_match    (glob ~anchored "foo/**/bar/**/baz") "foo/bar/baz");
+  assert (re_match    (glob ~anchored "foo/**/bar/**/baz") "foo/bar/x/y/z/baz");
+  assert (re_match    (glob ~anchored "foo/**/bar/**/baz") "foo/x/y/z/bar/baz");
+  assert (re_match    (glob ~anchored "foo/**/bar/**/baz") "foo/bar/x/bar/x/baz");
+  assert (re_mismatch (glob ~anchored "foo/**/bar/**/baz") "foo/bar/../x/baz");
+  assert (re_mismatch (glob ~anchored "foo/**/bar/**/baz") "foo/bar/./x/baz");
+
   (* Interaction with [~period] *)
   let period = true in
   assert (re_mismatch (glob ~anchored ~period "**") ".foobar");