Cache busting

docs/index.rst

@@ -213,6 +213,10 @@ uploading assets to S3.
                             upload to S3. SHA-1 file hashes are used to compute
                             file changes. You can delete `.file-hashes` from 
                             your S3 bucket to force all files to upload again.
+`S3_CACHE_BUSTING`          Append a query string with a file hash to all static URLs
+                            to force a fetch when file contents change. This is
+                            useful when used with long expiry headers while
+                            still retaining the ability to change static files.
 `S3_CACHE_CONTROL`          **Deprecated**. Please use `S3_HEADERS` instead.
 `S3_USE_CACHE_CONTROL`      **Deprecated**. Please use `S3_HEADERS` instead.
 =========================== ===================================================

flask_s3.py

@@ -5,7 +5,7 @@
 from collections import defaultdict
 
 from flask import url_for as flask_url_for
-from flask import current_app
+from flask import current_app, request
 from boto.s3.connection import S3Connection
 from boto.s3 import connect_to_region
 from boto.exception import S3CreateError, S3ResponseError
@@ -14,6 +14,9 @@
 logger = logging.getLogger('flask_s3')
 
 
+file_hashes = {}
+
+
 def hash_file(filename):
     """
     Generate a hash for the contents of a file
@@ -55,7 +58,34 @@ def url_for(endpoint, **values):
         if app.config['S3_CDN_DOMAIN']:
             bucket_path = '%s' % app.config['S3_CDN_DOMAIN']
         urls = app.url_map.bind(bucket_path, url_scheme=scheme)
-        return urls.build(endpoint, values=values, force_external=True)
+        url = urls.build(endpoint, values=values, force_external=True)
+
+        if app.config.get('S3_CACHE_BUSTING', False):
+            # We maintain a dictionary of file hashes to use as query parameters
+            # after filenames to force a cache miss when the files contents have
+            # changed. These hashes are calculated once per server restart when
+            # the file is first fetched
+
+            if url not in file_hashes:
+                filename = values['filename']
+                blueprints = app.blueprints.values()
+
+                if endpoint == 'static':
+                    filename = os.path.join(app.static_folder, filename)
+                else:
+                    for blueprint in blueprints:
+                        if endpoint == "{}.static".format(blueprint.name):
+                            filename = os.path.join(blueprint.static_folder,
+                                    filename)
+
+                if os.path.exists(filename):
+                    file_hashes[url] = hash_file(filename)
+
+            file_hash = file_hashes.get(url, None)
+            if file_hash and "?" not in url:
+                return "{}?{}".format(url, file_hash[-6:])
+
+        return url
     return flask_url_for(endpoint, **values)
 
 
@@ -285,7 +315,8 @@ def init_app(self, app):
                     ('S3_CDN_DOMAIN', ''),
                     ('S3_USE_CACHE_CONTROL', False),
                     ('S3_HEADERS', {}),
-                    ('S3_ONLY_MODIFIED', False)]
+                    ('S3_ONLY_MODIFIED', False),
+                    ('S3_CACH_BUSTING', False)]
 
         for k, v in defaults:
             app.config.setdefault(k, v)

tests/test_flask_static.py

@@ -5,7 +5,7 @@
 import os
 
 from mock import Mock, patch, call
-from flask import Flask, render_template_string, Blueprint
+from flask import Flask, render_template_string, Blueprint, url_for
 
 import flask_s3
 from flask_s3 import FlaskS3
@@ -124,6 +124,42 @@ def test_url_for_cdn_domain(self):
         exp = 'https://foo.cloudfront.net/static/bah.js'
         self.assertEquals(self.client_get(ufs).data, exp)
 
+    def test_url_with_hash(self):
+        self.app.config['S3_CACHE_BUSTING'] = True
+
+        static_folder = tempfile.mkdtemp()
+        self.app.static_folder = static_folder
+
+        static_url_loc = static_folder
+        filename = os.path.join(static_folder, "foo.png")
+
+        # Write random data into file
+        with open(filename, 'wb') as f:
+            f.write(os.urandom(1024))
+
+        with self.app.test_request_context("/"):
+            ufs = "{{url_for('static', filename='foo.png')}}"
+            exp = "https://foo.s3.amazonaws.com/static/foo.png"
+            url = self.client_get(ufs).data
+
+            self.assertTrue(url.startswith(exp))
+            # We have a query string
+            self.assertTrue("?" in url) 
+
+            again = self.client_get(ufs).data
+
+            # We get the same query string when fetching the same file
+            self.assertEquals(again, url)
+
+            # Change the contents of the file
+            with open(filename, 'wb') as f:
+                f.write(os.urandom(1025))
+
+            # Clear hashes (would be done on restart)
+            flask_s3.file_hashes = {}
+
+            changed = self.client_get(ufs).data
+            self.assertNotEquals(url, changed)
 
 
 class S3Tests(unittest.TestCase):