Add --match option (and tests)

Uses shell filename pattern

Author: chlunde

hardlink.py

@@ -51,6 +51,7 @@
 import sys
 import time
 import filecmp
+import fnmatch
 
 from optparse import OptionParser
 
@@ -226,6 +227,9 @@ def hardlink_identical_files(directories, filename, options):
         directories.append(filename)
     # Is it a regular file?
     elif stat.S_ISREG(stat_info.st_mode):
+        if options.match:
+            if not fnmatch.fnmatch(filename, options.match):
+                return
         # Create the hash for the file.
         file_hash = hash_value(stat_info.st_size, stat_info.st_mtime,
                                options.notimestamp or options.contentonly)
@@ -398,6 +402,10 @@ def parse_command_line():
                       help="Regular expression used to exclude files/dirs (may specify multiple times)",
                       action="append", dest="excludes", default=[],)
 
+    parser.add_option("-m", "--match",
+        help="Shell pattern used to match files", metavar="PATTERN",
+        action="store", dest="match", default=None,)
+
     (options, args) = parser.parse_args()
     if not args:
         parser.print_help()

tests.py

@@ -54,7 +54,6 @@ def setUp(self):
 
         self.verify_file_contents()
 
-
     def verify_file_contents(self):
         for filename, contents in self.testfs.items():
             with open(filename, "r") as f:
@@ -64,22 +63,20 @@ def verify_file_contents(self):
         # Bug?  Should hardlink to the file with most existing links?
         # self.assertEqual(get_inode("dir1/name1.ext"), get_inode("dir1/link"))
 
-
     def test_hardlink_tree_dryrun(self):
         sys.argv = ["hardlink.py", "-v", "0", "--no-stats", "--dry-run", self.root]
         hardlink.main()
 
         self.verify_file_contents()
 
-        self.assertEqual(os.lstat("dir1/name1.ext").st_nlink, 2) # Existing link
+        self.assertEqual(os.lstat("dir1/name1.ext").st_nlink, 2)  # Existing link
         self.assertEqual(os.lstat("dir1/name2.ext").st_nlink, 1)
         self.assertEqual(os.lstat("dir1/name3.ext").st_nlink, 1)
         self.assertEqual(os.lstat("dir2/name1.ext").st_nlink, 1)
         self.assertEqual(os.lstat("dir3/name1.ext").st_nlink, 1)
         self.assertEqual(os.lstat("dir3/name1.noext").st_nlink, 1)
         self.assertEqual(os.lstat("dir4/name1.ext").st_nlink, 1)
 
-
     def test_hardlink_tree(self):
         sys.argv = ["hardlink.py", "-v", "0", "--no-stats", self.root]
         hardlink.main()
@@ -94,7 +91,6 @@ def test_hardlink_tree(self):
 
         self.assertNotEqual(get_inode("dir1/name1.ext"), get_inode("dir4/name1.ext"))
 
-
     def test_hardlink_tree_filenames_equal(self):
         sys.argv = ["hardlink.py", "-v", "0", "--no-stats", "--filenames-equal", self.root]
         hardlink.main()
@@ -109,7 +105,6 @@ def test_hardlink_tree_filenames_equal(self):
 
         self.assertNotEqual(get_inode("dir1/name1.ext"), get_inode("dir4/name1.ext"))
 
-
     def test_hardlink_tree_exclude(self):
         sys.argv = ["hardlink.py", "-v", "0", "--no-stats", "--exclude", ".*noext$", self.root]
         hardlink.main()
@@ -124,7 +119,6 @@ def test_hardlink_tree_exclude(self):
 
         self.assertNotEqual(get_inode("dir1/name1.ext"), get_inode("dir4/name1.ext"))
 
-
     def test_hardlink_tree_timestamp_ignore(self):
         sys.argv = ["hardlink.py", "-v", "0", "--no-stats", "--timestamp-ignore", self.root]
         hardlink.main()
@@ -139,6 +133,20 @@ def test_hardlink_tree_timestamp_ignore(self):
 
         self.assertEqual(get_inode("dir1/name1.ext"), get_inode("dir4/name1.ext"))
 
+    def test_hardlink_tree_match(self):
+        sys.argv = ["hardlink.py", "-v", "0", "--no-stats", "--match", "*.ext", self.root]
+        hardlink.main()
+
+        self.verify_file_contents()
+
+        self.assertEqual(get_inode("dir1/name1.ext"), get_inode("dir1/name2.ext"))
+        self.assertEqual(get_inode("dir1/name1.ext"), get_inode("dir2/name1.ext"))
+        self.assertNotEqual(get_inode("dir1/name1.ext"), get_inode("dir3/name1.noext"))
+
+        self.assertEqual(get_inode("dir1/name3.ext"), get_inode("dir3/name1.ext"))
+
+        self.assertNotEqual(get_inode("dir1/name1.ext"), get_inode("dir4/name1.ext"))
+
 
 if __name__ == '__main__':
     unittest.main()