Merge pull request #23 from jonls/regex-matches

Support regular expression matches for cache_rules

README.rst

@@ -31,6 +31,9 @@ The configuration is stored in a YAML file like this:
       - match: "/assets/*"
         maxage: 30 days
 
+      - match_regexp: '^assets/image-\d{3}-.*\.png$'
+        maxage: 90 days
+
       - match: "/css/*"
         maxage: 30 days
 
@@ -81,7 +84,7 @@ Configuration file
     ``arn:aws:s3:::example.com`` and ``arn:aws:s3:::example.com/*``.
 
 **s3_reduced_redundancy**
-    An optional boolean to indicate whether the files should be uploaded 
+    An optional boolean to indicate whether the files should be uploaded
     to `reduced redundancy`_ storage.
 
 **cloudfront_distribution_id**
@@ -91,7 +94,9 @@ Configuration file
 
 **cache_rules**
     A list of rules to determine the cache configuration of the uploaded files.
-    The ``match`` key specifies a pattern that the rule applies to. Only the
+    The ``match`` key specifies a pattern that the rule applies to. This uses
+    glob-style matching (with ``*`` and ``?``). Matching can also be performed
+    with regular expressions by using ``match_regexp``. Only the
     first rule to match a given key will be used. The ``maxage`` key
     specifies the time to cache the file. The value should be either a number
     of seconds or a string like ``30 days``, ``5 minutes, 30 seconds``, etc.

s3_deploy/config.py

@@ -71,7 +71,16 @@ def resolve_cache_rules(key_name, rules):
     """Returns the value of the Cache-Control header after applying rules."""
 
     for rule in rules:
-        if match_key(rule['match'], key_name):
+        has_match = 'match' in rule
+        has_match_regexp = 'match_regexp' in rule
+        if has_match == has_match_regexp:
+            raise ValueError(
+                'Cache rule must have either match or match_regexp key'
+            )
+
+        pattern = rule['match'] if has_match else rule['match_regexp']
+
+        if match_key(pattern, key_name, regexp=has_match_regexp):
             cache_control = None
             if 'cache_control' in rule:
                 cache_control = rule['cache_control']

s3_deploy/filematch.py

@@ -30,10 +30,13 @@ def compile_re(pattern):
     return prefix + ''.join(re_pattern) + '$'
 
 
-def match_key(pattern, key):
-    """Match key to a glob (gitignore-style) pattern."""
+def match_key(pattern, key, regexp=False):
+    """Match key to a glob (gitignore-style) or regexp pattern."""
 
     if pattern == '':
         raise ValueError('Empty pattern is invalid')
 
-    return bool(re.search(compile_re(pattern), key))
+    if not regexp:
+        pattern = compile_re(pattern)
+
+    return bool(re.search(pattern, key))

s3_deploy/tests/test_config.py

@@ -58,6 +58,12 @@ def test_resolve_empty(self):
         cache = config.resolve_cache_rules('test', [])
         self.assertIsNone(cache)
 
+    def test_resolve_no_match_key(self):
+        with self.assertRaises(ValueError):
+            config.resolve_cache_rules('test', [
+                {'maxage': 3000},
+            ])
+
     def test_resolve_catch_all_to_cache_control(self):
         cache = config.resolve_cache_rules('test', [
             {'match': '*', 'cache_control': 'public, maxage=500'}
@@ -103,6 +109,18 @@ def test_resolve_second_rule(self):
         ])
         self.assertEqual(cache, 'max-age=200')
 
+    def test_resolve_multiple_match_keys(self):
+        with self.assertRaises(ValueError):
+            config.resolve_cache_rules('test', [
+                {'match': 'test*', 'match_regexp': 'test.*'},
+            ])
+
+    def test_resolve_regexp_match_key(self):
+        cache = config.resolve_cache_rules('test', [
+            {'match_regexp': 't[es]{2}(t|a)$', 'maxage': 100},
+        ])
+        self.assertEqual(cache, 'max-age=100')
+
 
 class LoadConfigFileTest(unittest.TestCase):
     def setUp(self):

s3_deploy/tests/test_filematch.py

@@ -6,64 +6,101 @@
 
 class FileMatchTest(unittest.TestCase):
     def test_match_exact_ok(self):
-        self.assertTrue(filematch.match_key('/index.html', 'index.html'))
+        self.assertTrue(filematch.match_key(
+            '/index.html', 'index.html', regexp=False))
 
     def test_match_exact_fail_different_name(self):
-        self.assertFalse(filematch.match_key('/index.html', 'image.png'))
+        self.assertFalse(filematch.match_key(
+            '/index.html', 'image.png', regexp=False))
 
     def test_match_exact_fail_different_path(self):
-        self.assertFalse(filematch.match_key('/index.html', 'dir/index.html'))
+        self.assertFalse(filematch.match_key(
+            '/index.html', 'dir/index.html', regexp=False))
 
     def test_match_exact_with_path_ok(self):
         self.assertTrue(filematch.match_key(
-            '/dir/index.html', 'dir/index.html'))
+            '/dir/index.html', 'dir/index.html', regexp=False))
 
     def test_match_exact_with_path_fail_different_name(self):
         self.assertFalse(filematch.match_key(
-            '/dir/index.html', 'dir/image.png'))
+            '/dir/index.html', 'dir/image.png', regexp=False))
 
     def test_match_exact_with_path_fail_different_path(self):
         self.assertFalse(filematch.match_key(
-            '/dir/index.html', 'altdir/image.png'))
+            '/dir/index.html', 'altdir/image.png', regexp=False))
 
     def test_match_base_ok_simple(self):
-        self.assertTrue(filematch.match_key('index.html', 'index.html'))
+        self.assertTrue(filematch.match_key(
+            'index.html', 'index.html', regexp=False))
 
     def test_match_base_ok_path(self):
-        self.assertTrue(filematch.match_key('index.html', 'dir/index.html'))
+        self.assertTrue(filematch.match_key(
+            'index.html', 'dir/index.html', regexp=False))
 
     def test_match_base_fail_different_name(self):
-        self.assertFalse(filematch.match_key('ok.html', 'book.html'))
+        self.assertFalse(filematch.match_key(
+            'ok.html', 'book.html', regexp=False))
 
     def test_match_base_fail_different_path(self):
-        self.assertFalse(filematch.match_key('ok.html', 'dir/book.html'))
+        self.assertFalse(filematch.match_key(
+            'ok.html', 'dir/book.html', regexp=False))
 
     def test_match_variable_base_ok(self):
-        self.assertTrue(filematch.match_key('*.html', 'index.html'))
+        self.assertTrue(filematch.match_key(
+            '*.html', 'index.html', regexp=False))
 
     def test_match_variable_base_within_ok(self):
-        self.assertTrue(filematch.match_key('image-*.png', 'image-999.png'))
+        self.assertTrue(filematch.match_key(
+            'image-*.png', 'image-999.png', regexp=False))
 
     def test_match_variable_with_path_ok(self):
-        self.assertTrue(filematch.match_key('images/*.png', 'images/999.png'))
+        self.assertTrue(filematch.match_key(
+            'images/*.png', 'images/999.png', regexp=False))
 
     def test_match_variable_with_path_fail(self):
-        self.assertFalse(filematch.match_key('images/*.png', 'images/README'))
+        self.assertFalse(filematch.match_key(
+            'images/*.png', 'images/README', regexp=False))
 
     def test_match_multilevel_fail(self):
         self.assertFalse(filematch.match_key(
-            '/dir/*.html', 'dir/second/file.html'))
+            '/dir/*.html', 'dir/second/file.html', regexp=False))
 
     def test_match_one_ok(self):
-        self.assertTrue(filematch.match_key('image?.png', 'image1.png'))
+        self.assertTrue(filematch.match_key(
+            'image?.png', 'image1.png', regexp=False))
 
     def test_match_one_fail(self):
-        self.assertFalse(filematch.match_key('image?.png', 'image10.png'))
+        self.assertFalse(filematch.match_key(
+            'image?.png', 'image10.png', regexp=False))
 
     def test_match_empty_on_file(self):
         with self.assertRaises(ValueError):
-            filematch.match_key('', 'index.html')
+            filematch.match_key('', 'index.html', regexp=False)
 
     def test_match_empty_on_empty(self):
         with self.assertRaises(ValueError):
-            filematch.match_key('', '')
+            filematch.match_key('', '', regexp=False)
+
+    def test_match_regex_dollar_suffix_ok(self):
+        self.assertTrue(filematch.match_key(
+            'image-\d+\.png$', 'image-999.png', regexp=True))
+
+    def test_match_regex_dollar_suffix_fail(self):
+        self.assertFalse(filematch.match_key(
+            'image-\d+\.png$', 'image-name.png', regexp=True))
+
+    def test_match_regex_caret_prefix_ok(self):
+        self.assertTrue(filematch.match_key(
+            '^dir\/index\.html', 'dir/index.html', regexp=True))
+
+    def test_match_regex_caret_prefix_fail(self):
+        self.assertFalse(filematch.match_key(
+            '^dir\/index\.html', 'altdir/index.html', regexp=True))
+
+    def test_match_regex_caret_and_dollar_ok(self):
+        self.assertTrue(filematch.match_key(
+            '^dir\/index\.html$', 'dir/index.html', regexp=True))
+
+    def test_match_regex_caret_and_dollar_fail(self):
+        self.assertFalse(filematch.match_key(
+            '^dir\/index\.html$', 'dir/index.htm', regexp=True))