more filtering

morituri/common/path.py

@@ -28,17 +28,31 @@ class PathFilter(object):
     I filter path components for safe storage on file systems.
     """
 
-    def __init__(self, slashes=True, fat=True, special=True):
+    def __init__(self, slashes=True, quotes=True, fat=True, special=False):
         """
         @param slashes: whether to convert slashes to dashes
-        @parm  fat:     whether to strip characters illegal on FAT filesystems
+        @param quotes:  whether to normalize quotes
+        @param fat:     whether to strip characters illegal on FAT filesystems
+        @param special: whether to strip special characters
         """
         self._slashes = slashes
+        self._quotes = quotes
         self._fat = fat
         self._special = special
 
     def filter(self, path):
         if self._slashes:
-            path = re.sub(r'[/]', '-', path, re.UNICODE)
+            path = re.sub(r'[/\\]', '-', path, re.UNICODE)
+
+        if self._quotes:
+            path = re.sub(ur'[\u2019]', "'", path, re.UNICODE)
+
+        if self._special:
+            # replace separators with a hyphen
+            path = re.sub(r'[:\|]', '-', path, re.UNICODE)
+            path = re.sub(r'[\*\?&!\'\"\$\(\)`{}\[\]<>]', '_', path, re.UNICODE)
+
+        if self._fat:
+            path = re.sub(r'[:\*\?"<>|"]', '_', path, re.UNICODE)
 
         return path

morituri/common/program.py

@@ -67,11 +67,14 @@ class Program(log.Loggable):
 
         d = {}
 
-        for key in ['fat', 'special']:
+        for key, default in {
+            'fat': True,
+            'special': False
+        }.items():
             value = None
             value = self._config.getboolean('main', 'path_filter_'+ key)
             if value is None:
-                value = True
+                value = default
 
             d[key] = value
 

morituri/test/test_common_path.py

@@ -9,8 +9,17 @@ from morituri.test import common
 class FilterTestCase(common.TestCase):
 
     def setUp(self):
-        self._filter = path.PathFilter()
+        self._filter = path.PathFilter(special=True)
 
     def testSlash(self):
         part = u'A Charm/A Blade'
         self.assertEquals(self._filter.filter(part), u'A Charm-A Blade')
+
+    def testFat(self):
+        part = u'A Word: F**k you?'
+        self.assertEquals(self._filter.filter(part), u'A Word- F__k you_')
+
+    def testSpecial(self):
+        part = u'<<< $&*!\' "()`{}[]spaceship>>>'
+        self.assertEquals(self._filter.filter(part),
+               u'___ _____ ________spaceship___')

morituri/test/test_program_cdparanoia.py

@@ -69,7 +69,7 @@ class VersionTestCase(common.TestCase):
         self.failUnless(v)
         # of the form III 10.2
         # make sure it ends with a digit
-        self.failUnless(int(v[-1]))
+        self.failUnless(int(v[-1]), v)
 
 
 class AnalyzeFileTask(cdparanoia.AnalyzeTask):