[GR-28734] Glob escaping

PullRequest: truffleruby/3430

Author: aardvark179

spec/ruby/core/dir/fixtures/common.rb

@@ -82,6 +82,7 @@ def self.mock_dir_files
 
         special/test{1}/file[1]
         special/{}/special
+        special/test\ +()[]{}/hello_world.erb
       ]
 
       platform_is_not :windows do

spec/ruby/core/dir/glob_spec.rb

@@ -79,6 +79,7 @@
       nested/
       nested/.dotsubir/
       special/
+      special/test\ +()[]{}/
       special/test{1}/
       special/{}/
       subdir_one/
@@ -130,6 +131,7 @@
       ./nested/
       ./nested/.dotsubir/
       ./special/
+      ./special/test\ +()[]{}/
       ./special/test{1}/
       ./special/{}/
       ./subdir_one/

spec/ruby/core/dir/shared/glob.rb

@@ -111,6 +111,10 @@
     it "matches files with backslashes in their name" do
       Dir.glob('special/\\\\{a,b}').should == ['special/\a']
     end
+
+    it "matches directory with special characters in their name in complex patterns" do
+      Dir.glob("special/test +()\\[\\]\\{\\}/hello_world{.{en},}{.{html},}{+{phone},}{.{erb},}").should == ['special/test +()[]{}/hello_world.erb']
+    end
   end
 
   it "matches regexp special ^" do
@@ -225,6 +229,7 @@
       dir/
       nested/
       special/
+      special/test\ +()[]{}/
       special/test{1}/
       special/{}/
       subdir_one/

src/main/ruby/truffleruby/core/dir_glob.rb

@@ -31,13 +31,15 @@
 
 class Dir
   module Glob
-    no_meta_chars = '[^*?\\[\\]{}\\\\]'
+    no_meta_chars = /[^*?\[\]{}\\]/
+    no_meta_chars_unescaped = /(?:#{no_meta_chars}|\\\\|\\\*|\\\?|\\\[|\\\]|\\\{|\\\})/
     NO_GLOB_META_CHARS = /\A#{no_meta_chars}+\z/
-    TRAILING_BRACES = /\A(#{no_meta_chars}+)(?:\{(#{no_meta_chars}*)\})?\z/
+    NO_GLOB_META_CHARS_UNESCAPED = /\A#{no_meta_chars_unescaped}+\z/
+    TRAILING_BRACES = /\A(#{no_meta_chars_unescaped}+)(?:\{(#{no_meta_chars_unescaped}*)\})?\z/
 
     class Node
       def initialize(nxt, flags)
-        @flags = flags
+        @flags = flags | File::FNM_EXTGLOB
         @next = nxt
         @separator = nil
       end
@@ -349,6 +351,10 @@ def process_entry(entry, entry_type, matches, parent, glob_base_dir)
       end
     end
 
+    def self.unescape(pattern)
+      pattern.gsub(/\\(.)/, '\\1')
+    end
+
     def self.path_split(str)
       start = 0
       ret = []
@@ -383,7 +389,10 @@ def self.path_split(str)
     end
 
     def self.single_compile(glob, flags=0)
-      if glob.getbyte(-1) != 47 && NO_GLOB_META_CHARS.match?(glob) # byte value 47 = ?/
+      escape = (flags & File::FNM_NOESCAPE) == 0
+      if escape && glob.getbyte(-1) != 47 && NO_GLOB_META_CHARS_UNESCAPED.match?(glob) # byte value 47 = ?/
+        return ConstantEntry.new nil, flags, unescape(glob)
+      elsif !escape && glob.getbyte(-1) != 47 && NO_GLOB_META_CHARS.match?(glob) # byte value 47 = ?/
         return ConstantEntry.new nil, flags, glob
       end
 
@@ -393,8 +402,10 @@ def self.single_compile(glob, flags=0)
         last = DirectoriesOnly.new nil, flags
       else
         file = parts.pop
-        if NO_GLOB_META_CHARS.match?(file)
+        if !escape && NO_GLOB_META_CHARS.match?(file)
           last = ConstantEntry.new nil, flags, file
+        elsif escape && NO_GLOB_META_CHARS_UNESCAPED.match?(file)
+          last = ConstantEntry.new nil, flags, unescape(file)
         elsif file == '*'
           last = AllNameEntryMatch.new nil, flags, file
         elsif file && file[0] == '*' && NO_GLOB_META_CHARS.match?(file[1..])
@@ -414,14 +425,25 @@ def self.single_compile(glob, flags=0)
           else
             last = RecursiveDirectories.new last, flags
           end
-        elsif NO_GLOB_META_CHARS.match?(dir)
+        elsif !escape && NO_GLOB_META_CHARS.match?(dir)
           while NO_GLOB_META_CHARS.match?(parts[-2])
             next_sep = parts.pop
             next_sect = parts.pop
 
             dir = next_sect << next_sep << dir
           end
 
+          last = ConstantDirectory.new last, flags, dir
+        elsif escape && NO_GLOB_META_CHARS_UNESCAPED.match?(dir)
+
+          dir = unescape(dir)
+          while NO_GLOB_META_CHARS_UNESCAPED.match?(parts[-2])
+            next_sep = parts.pop
+            next_sect = unescape(parts.pop)
+
+            dir = next_sect << next_sep << dir
+          end
+
           last = ConstantDirectory.new last, flags, dir
         elsif !dir.empty?
           last = DirectoryMatch.new last, flags, dir
@@ -466,9 +488,10 @@ def self.glob(base_dir, pattern, flags, matches)
         # only as a suffix.
 
         if braces = m[2]
-          stem = m[1]
+          stem = unescape(m[1])
 
           braces.split(',').each do |s|
+            s = unescape(s)
             path = "#{stem}#{s}"
             if Truffle::FileOperations.exist? path_join(base_dir, path)
               matches << path
@@ -480,6 +503,7 @@ def self.glob(base_dir, pattern, flags, matches)
             matches << stem if Truffle::FileOperations.exist? path_join(base_dir, stem)
           end
         else
+          pattern = unescape(pattern)
           matches << pattern if Truffle::FileOperations.exist?(path_join(base_dir, pattern))
         end