From 02011f1f5d3fae6eac0209b5faccc06079dd1b41 Mon Sep 17 00:00:00 2001
From: Michael Merickel <michael@merickel.org>
Date: Tue, 20 Oct 2015 00:32:14 -0500
Subject: first cut at removing default cache busters

---
 docs/api/static.rst                     |   9 ---
 docs/narr/assets.rst                    |  80 ++++++++++------------
 pyramid/config/views.py                 |  20 ++----
 pyramid/static.py                       |  84 +-----------------------
 pyramid/tests/test_config/test_views.py |  10 ---
 pyramid/tests/test_static.py            | 113 +-------------------------------
 6 files changed, 43 insertions(+), 273 deletions(-)

diff --git a/docs/api/static.rst b/docs/api/static.rst
index b6b279139..e6d6e4618 100644
--- a/docs/api/static.rst
+++ b/docs/api/static.rst
@@ -9,17 +9,8 @@
      :members:
      :inherited-members:
 
-  .. autoclass:: PathSegmentCacheBuster
-     :members:
-
   .. autoclass:: QueryStringCacheBuster
      :members:
 
-  .. autoclass:: PathSegmentMd5CacheBuster
-     :members:
-
-  .. autoclass:: QueryStringMd5CacheBuster
-     :members:
-
   .. autoclass:: QueryStringConstantCacheBuster
      :members:
diff --git a/docs/narr/assets.rst b/docs/narr/assets.rst
index 020794062..397e0258d 100644
--- a/docs/narr/assets.rst
+++ b/docs/narr/assets.rst
@@ -372,30 +372,38 @@ assets by passing the optional argument, ``cachebust`` to
 .. code-block:: python
    :linenos:
 
+   import time
+   from pyramid.static import QueryStringConstantCacheBuster
+
    # config is an instance of pyramid.config.Configurator
-   config.add_static_view(name='static', path='mypackage:folder/static',
-                          cachebust=True)
+   config.add_static_view(
+       name='static', path='mypackage:folder/static',
+       cachebust=QueryStringConstantCacheBuster(str(int(time.time()))),
+   )
 
 Setting the ``cachebust`` argument instructs :app:`Pyramid` to use a cache
-busting scheme which adds the md5 checksum for a static asset as a path segment
-in the asset's URL:
+busting scheme which adds the curent time for a static asset to the query
+string in the asset's URL:
 
 .. code-block:: python
    :linenos:
 
    js_url = request.static_url('mypackage:folder/static/js/myapp.js')
-   # Returns: 'http://www.example.com/static/c9658b3c0a314a1ca21e5988e662a09e/js/myapp.js'
+   # Returns: 'http://www.example.com/static/js/myapp.js?x=1445318121'
 
-When the asset changes, so will its md5 checksum, and therefore so will its
-URL.  Supplying the ``cachebust`` argument also causes the static view to set
-headers instructing clients to cache the asset for ten years, unless the
-``cache_max_age`` argument is also passed, in which case that value is used.
+When the web server restarts, the time constant will change and therefore so
+will its URL.  Supplying the ``cachebust`` argument also causes the static
+view to set headers instructing clients to cache the asset for ten years,
+unless the ``cache_max_age`` argument is also passed, in which case that
+value is used.
 
-.. note::
+.. warning::
 
-   md5 checksums are cached in RAM, so if you change a static resource without
-   restarting your application, you may still generate URLs with a stale md5
-   checksum.
+   Cache busting is an inherently complex topic as it integrates the asset
+   pipeline and the web application. It is expected and desired that
+   application authors will write their own cache buster implementations
+   conforming to the properties of their own asset pipelines. See
+   :ref:`custom_cache_busters` for information on writing your own.
 
 Disabling the Cache Buster
 ~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -406,40 +414,21 @@ configured cache busters without changing calls to
 ``PYRAMID_PREVENT_CACHEBUST`` environment variable or the
 ``pyramid.prevent_cachebust`` configuration value to a true value.
 
+.. _custom_cache_busters:
+
 Customizing the Cache Buster
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Revisiting from the previous section:
-
-.. code-block:: python
-   :linenos:
+The ``cachebust`` option to
+:meth:`~pyramid.config.Configurator.add_static_view` may be set to any object
+that implements the interface :class:`~pyramid.interfaces.ICacheBuster`.
 
-   # config is an instance of pyramid.config.Configurator
-   config.add_static_view(name='static', path='mypackage:folder/static',
-                          cachebust=True)
-
-Setting ``cachebust`` to ``True`` instructs :app:`Pyramid` to use a default
-cache busting implementation that should work for many situations.  The
-``cachebust`` may be set to any object that implements the interface
-:class:`~pyramid.interfaces.ICacheBuster`.  The above configuration is exactly
-equivalent to:
-
-.. code-block:: python
-   :linenos:
-
-   from pyramid.static import PathSegmentMd5CacheBuster
-
-   # config is an instance of pyramid.config.Configurator
-   config.add_static_view(name='static', path='mypackage:folder/static',
-                          cachebust=PathSegmentMd5CacheBuster())
-
-:app:`Pyramid` includes a handful of ready to use cache buster implementations:
-:class:`~pyramid.static.PathSegmentMd5CacheBuster`, which inserts an md5
-checksum token in the path portion of the asset's URL,
-:class:`~pyramid.static.QueryStringMd5CacheBuster`, which adds an md5 checksum
-token to the query string of the asset's URL, and
+:app:`Pyramid` ships with a very simplistic
 :class:`~pyramid.static.QueryStringConstantCacheBuster`, which adds an
-arbitrary token you provide to the query string of the asset's URL.
+arbitrary token you provide to the query string of the asset's URL. This
+is almost never what you want in production as it does not allow fine-grained
+busting of individual assets.
+
 
 In order to implement your own cache buster, you can write your own class from
 scratch which implements the :class:`~pyramid.interfaces.ICacheBuster`
@@ -456,15 +445,16 @@ the hash of the currently checked out code:
 
    import os
    import subprocess
-   from pyramid.static import PathSegmentCacheBuster
+   from pyramid.static import QueryStringCacheBuster
 
-   class GitCacheBuster(PathSegmentCacheBuster):
+   class GitCacheBuster(QueryStringCacheBuster):
        """
        Assuming your code is installed as a Git checkout, as opposed to an egg
        from an egg repository like PYPI, you can use this cachebuster to get
        the current commit's SHA1 to use as the cache bust token.
        """
-       def __init__(self):
+       def __init__(self, param='x'):
+           super(GitCacheBuster, self).__init__(param=param)
            here = os.path.dirname(os.path.abspath(__file__))
            self.sha1 = subprocess.check_output(
                ['git', 'rev-parse', 'HEAD'],
diff --git a/pyramid/config/views.py b/pyramid/config/views.py
index 34fc49c00..e386bc4e1 100644
--- a/pyramid/config/views.py
+++ b/pyramid/config/views.py
@@ -34,7 +34,6 @@ from pyramid.interfaces import (
     )
 
 from pyramid import renderers
-from pyramid.static import PathSegmentMd5CacheBuster
 
 from pyramid.compat import (
     string_types,
@@ -1862,14 +1861,12 @@ class ViewsConfiguratorMixin(object):
         The ``cachebust`` keyword argument may be set to cause
         :meth:`~pyramid.request.Request.static_url` to use cache busting when
         generating URLs. See :ref:`cache_busting` for general information
-        about cache busting.  The value of the ``cachebust`` argument may be
-        ``True``, in which case a default cache busting implementation is used.
-        The value of the ``cachebust`` argument may also be an object which
-        implements :class:`~pyramid.interfaces.ICacheBuster`.  See the
-        :mod:`~pyramid.static` module for some implementations.  If the
-        ``cachebust`` argument is provided, the default for ``cache_max_age``
-        is modified to be ten years.  ``cache_max_age`` may still be explicitly
-        provided to override this default.
+        about cache busting. The value of the ``cachebust`` argument must
+        be an object which implements
+        :class:`~pyramid.interfaces.ICacheBuster`.  If the ``cachebust``
+        argument is provided, the default for ``cache_max_age`` is modified
+        to be ten years.  ``cache_max_age`` may still be explicitly provided
+        to override this default.
 
         The ``permission`` keyword argument is used to specify the
         :term:`permission` required by a user to execute the static view.  By
@@ -1967,9 +1964,6 @@ def isexception(o):
 
 @implementer(IStaticURLInfo)
 class StaticURLInfo(object):
-    # Indirection for testing
-    _default_cachebust = PathSegmentMd5CacheBuster
-
     def _get_registrations(self, registry):
         try:
             reg = registry._static_url_registrations
@@ -2033,8 +2027,6 @@ class StaticURLInfo(object):
             cb = None
         else:
             cb = extra.pop('cachebust', None)
-        if cb is True:
-            cb = self._default_cachebust()
         if cb:
             def cachebust(subpath, kw):
                 subpath_tuple = tuple(subpath.split('/'))
diff --git a/pyramid/static.py b/pyramid/static.py
index cb78feb9b..2aff02c0c 100644
--- a/pyramid/static.py
+++ b/pyramid/static.py
@@ -1,5 +1,4 @@
 # -*- coding: utf-8 -*-
-import hashlib
 import os
 
 from os.path import (
@@ -27,7 +26,7 @@ from pyramid.httpexceptions import (
     HTTPMovedPermanently,
     )
 
-from pyramid.path import AssetResolver, caller_package
+from pyramid.path import caller_package
 from pyramid.response import FileResponse
 from pyramid.traversal import traversal_path_info
 
@@ -159,71 +158,6 @@ def _secure_path(path_tuple):
     encoded = slash.join(path_tuple) # will be unicode
     return encoded
 
-def _generate_md5(spec):
-    asset = AssetResolver(None).resolve(spec)
-    md5 = hashlib.md5()
-    with asset.stream() as stream:
-        for block in iter(lambda: stream.read(4096), b''):
-            md5.update(block)
-    return md5.hexdigest()
-
-class Md5AssetTokenGenerator(object):
-    """
-    A mixin class which provides an implementation of
-    :meth:`~pyramid.interfaces.ICacheBuster.target` which generates an md5
-    checksum token for an asset, caching it for subsequent calls.
-    """
-    def __init__(self):
-        self.token_cache = {}
-
-    def tokenize(self, pathspec):
-        # An astute observer will notice that this use of token_cache doesn't
-        # look particularly thread safe.  Basic read/write operations on Python
-        # dicts, however, are atomic, so simply accessing and writing values
-        # to the dict shouldn't cause a segfault or other catastrophic failure.
-        # (See: http://effbot.org/pyfaq/what-kinds-of-global-value-mutation-are-thread-safe.htm)
-        #
-        # We do have a race condition that could result in the same md5
-        # checksum getting computed twice or more times in parallel.  Since
-        # the program would still function just fine if this were to occur,
-        # the extra overhead of using locks to serialize access to the dict
-        # seems an unnecessary burden.
-        #
-        token = self.token_cache.get(pathspec)
-        if not token:
-            self.token_cache[pathspec] = token = _generate_md5(pathspec)
-        return token
-
-class PathSegmentCacheBuster(object):
-    """
-    An implementation of :class:`~pyramid.interfaces.ICacheBuster` which
-    inserts a token for cache busting in the path portion of an asset URL.
-
-    To use this class, subclass it and provide a ``tokenize`` method which
-    accepts a ``pathspec`` and returns a token.
-
-    .. versionadded:: 1.6
-    """
-    def pregenerate(self, pathspec, subpath, kw):
-        token = self.tokenize(pathspec)
-        return (token,) + subpath, kw
-
-    def match(self, subpath):
-        return subpath[1:]
-
-class PathSegmentMd5CacheBuster(PathSegmentCacheBuster,
-                                Md5AssetTokenGenerator):
-    """
-    An implementation of :class:`~pyramid.interfaces.ICacheBuster` which
-    inserts an md5 checksum token for cache busting in the path portion of an
-    asset URL.  Generated md5 checksums are cached in order to speed up
-    subsequent calls.
-
-    .. versionadded:: 1.6
-    """
-    def __init__(self):
-        super(PathSegmentMd5CacheBuster, self).__init__()
-
 class QueryStringCacheBuster(object):
     """
     An implementation of :class:`~pyramid.interfaces.ICacheBuster` which adds
@@ -249,22 +183,6 @@ class QueryStringCacheBuster(object):
             kw['_query'] = tuple(query) + ((self.param, token),)
         return subpath, kw
 
-class QueryStringMd5CacheBuster(QueryStringCacheBuster,
-                                Md5AssetTokenGenerator):
-    """
-    An implementation of :class:`~pyramid.interfaces.ICacheBuster` which adds
-    an md5 checksum token for cache busting in the query string of an asset
-    URL.  Generated md5 checksums are cached in order to speed up subsequent
-    calls.
-
-    The optional ``param`` argument determines the name of the parameter added
-    to the query string and defaults to ``'x'``.
-
-    .. versionadded:: 1.6
-    """
-    def __init__(self, param='x'):
-        super(QueryStringMd5CacheBuster, self).__init__(param=param)
-
 class QueryStringConstantCacheBuster(QueryStringCacheBuster):
     """
     An implementation of :class:`~pyramid.interfaces.ICacheBuster` which adds
diff --git a/pyramid/tests/test_config/test_views.py b/pyramid/tests/test_config/test_views.py
index 1c2d300a1..acfb81962 100644
--- a/pyramid/tests/test_config/test_views.py
+++ b/pyramid/tests/test_config/test_views.py
@@ -4104,16 +4104,6 @@ class TestStaticURLInfo(unittest.TestCase):
         self.assertEqual(config.view_kw['renderer'],
                          'mypackage:templates/index.pt')
 
-    def test_add_cachebust_default(self):
-        config = self._makeConfig()
-        inst = self._makeOne()
-        inst._default_cachebust = lambda: DummyCacheBuster('foo')
-        inst.add(config, 'view', 'mypackage:path', cachebust=True)
-        cachebust = config.registry._static_url_registrations[0][3]
-        subpath, kw = cachebust('some/path', {})
-        self.assertEqual(subpath, 'some/path')
-        self.assertEqual(kw['x'], 'foo')
-
     def test_add_cachebust_prevented(self):
         config = self._makeConfig()
         config.registry.settings['pyramid.prevent_cachebust'] = True
diff --git a/pyramid/tests/test_static.py b/pyramid/tests/test_static.py
index a3df74b44..7f50a0e43 100644
--- a/pyramid/tests/test_static.py
+++ b/pyramid/tests/test_static.py
@@ -368,118 +368,7 @@ class Test_static_view_use_subpath_True(unittest.TestCase):
         from pyramid.httpexceptions import HTTPNotFound
         self.assertRaises(HTTPNotFound, inst, context, request)
 
-class TestMd5AssetTokenGenerator(unittest.TestCase):
-    _fspath = None
-    _tmp = None
-
-    @property
-    def fspath(self):
-        if self._fspath:
-            return self._fspath
-
-        import os
-        import tempfile
-        self._tmp = tmp = tempfile.mkdtemp()
-        self._fspath = os.path.join(tmp, 'test.txt')
-        return self._fspath
-
-    def tearDown(self):
-        import shutil
-        if self._tmp:
-            shutil.rmtree(self._tmp)
-
-    def _makeOne(self):
-        from pyramid.static import Md5AssetTokenGenerator as cls
-        return cls()
-
-    def test_package_resource(self):
-        fut = self._makeOne().tokenize
-        expected = '76d653a3a044e2f4b38bb001d283e3d9'
-        token = fut('pyramid.tests:fixtures/static/index.html')
-        self.assertEqual(token, expected)
-
-    def test_filesystem_resource(self):
-        fut = self._makeOne().tokenize
-        expected = 'd5155f250bef0e9923e894dbc713c5dd'
-        with open(self.fspath, 'w') as f:
-            f.write("Are we rich yet?")
-        token = fut(self.fspath)
-        self.assertEqual(token, expected)
-
-    def test_cache(self):
-        fut = self._makeOne().tokenize
-        expected = 'd5155f250bef0e9923e894dbc713c5dd'
-        with open(self.fspath, 'w') as f:
-            f.write("Are we rich yet?")
-        token = fut(self.fspath)
-        self.assertEqual(token, expected)
-
-        # md5 shouldn't change because we've cached it
-        with open(self.fspath, 'w') as f:
-            f.write("Sorry for the convenience.")
-        token = fut(self.fspath)
-        self.assertEqual(token, expected)
-
-class TestPathSegmentMd5CacheBuster(unittest.TestCase):
-
-    def _makeOne(self):
-        from pyramid.static import PathSegmentMd5CacheBuster as cls
-        inst = cls()
-        inst.tokenize = lambda pathspec: 'foo'
-        return inst
-
-    def test_token(self):
-        fut = self._makeOne().tokenize
-        self.assertEqual(fut('whatever'), 'foo')
-
-    def test_pregenerate(self):
-        fut = self._makeOne().pregenerate
-        self.assertEqual(fut('foo', ('bar',), 'kw'), (('foo', 'bar'), 'kw'))
-
-    def test_match(self):
-        fut = self._makeOne().match
-        self.assertEqual(fut(('foo', 'bar')), ('bar',))
-
-class TestQueryStringMd5CacheBuster(unittest.TestCase):
-
-    def _makeOne(self, param=None):
-        from pyramid.static import QueryStringMd5CacheBuster as cls
-        if param:
-            inst = cls(param)
-        else:
-            inst = cls()
-        inst.tokenize = lambda pathspec: 'foo'
-        return inst
-
-    def test_token(self):
-        fut = self._makeOne().tokenize
-        self.assertEqual(fut('whatever'), 'foo')
-
-    def test_pregenerate(self):
-        fut = self._makeOne().pregenerate
-        self.assertEqual(
-            fut('foo', ('bar',), {}),
-            (('bar',), {'_query': {'x': 'foo'}}))
-
-    def test_pregenerate_change_param(self):
-        fut = self._makeOne('y').pregenerate
-        self.assertEqual(
-            fut('foo', ('bar',), {}),
-            (('bar',), {'_query': {'y': 'foo'}}))
-
-    def test_pregenerate_query_is_already_tuples(self):
-        fut = self._makeOne().pregenerate
-        self.assertEqual(
-            fut('foo', ('bar',), {'_query': [('a', 'b')]}),
-            (('bar',), {'_query': (('a', 'b'), ('x', 'foo'))}))
-
-    def test_pregenerate_query_is_tuple_of_tuples(self):
-        fut = self._makeOne().pregenerate
-        self.assertEqual(
-            fut('foo', ('bar',), {'_query': (('a', 'b'),)}),
-            (('bar',), {'_query': (('a', 'b'), ('x', 'foo'))}))
-
-class TestQueryStringConstantCacheBuster(TestQueryStringMd5CacheBuster):
+class TestQueryStringConstantCacheBuster(unittest.TestCase):
 
     def _makeOne(self, param=None):
         from pyramid.static import QueryStringConstantCacheBuster as cls
-- 
cgit v1.2.3


From 891d05499e54c067863a4fae9f2301ab772761dc Mon Sep 17 00:00:00 2001
From: Michael Merickel <michael@merickel.org>
Date: Thu, 12 Nov 2015 00:20:27 -0600
Subject: add manifest cache buster

---
 pyramid/static.py                     | 82 +++++++++++++++++++++++++++++++++++
 pyramid/tests/fixtures/manifest.json  |  4 ++
 pyramid/tests/fixtures/manifest2.json |  4 ++
 pyramid/tests/test_static.py          | 54 +++++++++++++++++++++++
 4 files changed, 144 insertions(+)
 create mode 100644 pyramid/tests/fixtures/manifest.json
 create mode 100644 pyramid/tests/fixtures/manifest2.json

diff --git a/pyramid/static.py b/pyramid/static.py
index 2aff02c0c..59f440c82 100644
--- a/pyramid/static.py
+++ b/pyramid/static.py
@@ -1,7 +1,9 @@
 # -*- coding: utf-8 -*-
+import json
 import os
 
 from os.path import (
+    getmtime,
     normcase,
     normpath,
     join,
@@ -202,3 +204,83 @@ class QueryStringConstantCacheBuster(QueryStringCacheBuster):
 
     def tokenize(self, pathspec):
         return self._token
+
+class ManifestCacheBuster(object):
+    """
+    An implementation of :class:`~pyramid.interfaces.ICacheBuster` which
+    uses a supplied manifest file to map an asset path to a cache-busted
+    version of the path.
+
+    The file is expected to conform to the following simple JSON format:
+
+    .. code-block:: json
+
+       {
+           "css/main.css": "css/main-678b7c80.css",
+           "images/background.png": "images/background-a8169106.png",
+       }
+
+    Specifically, it is a JSON-serialized dictionary where the keys are the
+    source asset paths used in calls to
+    :meth:`~pyramid.request.Request.static_url. For example::
+
+    .. code-block:: python
+
+       >>> request.static_url('myapp:static/css/main.css')
+       "http://www.example.com/static/css/main-678b7c80.css"
+
+    If a path is not found in the manifest it will pass through unchanged.
+
+    If ``reload`` is ``True`` then the manifest file will be reloaded when
+    changed. It is not recommended to leave this enabled in production.
+
+    If the manifest file cannot be found on disk it will be treated as
+    an empty mapping unless ``reload`` is ``False``.
+
+    The default implementation assumes the requested (possibly cache-busted)
+    path is the actual filename on disk. Subclasses may override the ``match``
+    method to alter this behavior. For example, to strip the cache busting
+    token from the path.
+
+    .. versionadded:: 1.6
+    """
+    exists = staticmethod(exists) # testing
+    getmtime = staticmethod(getmtime) # testing
+
+    def __init__(self, manifest_path, reload=False):
+        self.manifest_path = manifest_path
+        self.reload = reload
+
+        self._mtime = None
+        if not reload:
+            self._manifest = self.parse_manifest()
+
+    def parse_manifest(self):
+        """
+        Return a mapping parsed from the ``manifest_path``.
+
+        Subclasses may override this method to use something other than
+        ``json.loads``.
+
+        """
+        with open(self.manifest_path, 'rb') as fp:
+            content = fp.read().decode('utf-8')
+            return json.loads(content)
+
+    @property
+    def manifest(self):
+        """ The current manifest dictionary."""
+        if self.reload:
+            if not self.exists(self.manifest_path):
+                return {}
+            mtime = self.getmtime(self.manifest_path)
+            if self._mtime is None or mtime > self._mtime:
+                self._manifest = self.parse_manifest()
+                self._mtime = mtime
+        return self._manifest
+
+    def pregenerate(self, pathspec, subpath, kw):
+        path = '/'.join(subpath)
+        path = self.manifest.get(path, path)
+        new_subpath = path.split('/')
+        return (new_subpath, kw)
diff --git a/pyramid/tests/fixtures/manifest.json b/pyramid/tests/fixtures/manifest.json
new file mode 100644
index 000000000..0a43bc5e3
--- /dev/null
+++ b/pyramid/tests/fixtures/manifest.json
@@ -0,0 +1,4 @@
+{
+    "css/main.css": "css/main-test.css",
+    "images/background.png": "images/background-a8169106.png"
+}
diff --git a/pyramid/tests/fixtures/manifest2.json b/pyramid/tests/fixtures/manifest2.json
new file mode 100644
index 000000000..fd6b9a7bb
--- /dev/null
+++ b/pyramid/tests/fixtures/manifest2.json
@@ -0,0 +1,4 @@
+{
+    "css/main.css": "css/main-678b7c80.css",
+    "images/background.png": "images/background-a8169106.png"
+}
diff --git a/pyramid/tests/test_static.py b/pyramid/tests/test_static.py
index 7f50a0e43..ac30e9e50 100644
--- a/pyramid/tests/test_static.py
+++ b/pyramid/tests/test_static.py
@@ -1,6 +1,9 @@
 import datetime
+import os.path
 import unittest
 
+here = os.path.dirname(__file__)
+
 # 5 years from now (more or less)
 fiveyrsfuture = datetime.datetime.utcnow() + datetime.timedelta(5*365)
 
@@ -406,6 +409,57 @@ class TestQueryStringConstantCacheBuster(unittest.TestCase):
             fut('foo', ('bar',), {'_query': (('a', 'b'),)}),
             (('bar',), {'_query': (('a', 'b'), ('x', 'foo'))}))
 
+class TestManifestCacheBuster(unittest.TestCase):
+
+    def _makeOne(self, path, **kw):
+        from pyramid.static import ManifestCacheBuster as cls
+        return cls(path, **kw)
+
+    def test_it(self):
+        manifest_path = os.path.join(here, 'fixtures', 'manifest.json')
+        fut = self._makeOne(manifest_path).pregenerate
+        self.assertEqual(fut('foo', ('bar',), {}), (['bar'], {}))
+        self.assertEqual(
+            fut('foo', ('css', 'main.css'), {}),
+            (['css', 'main-test.css'], {}))
+
+    def test_reload(self):
+        manifest_path = os.path.join(here, 'fixtures', 'manifest.json')
+        new_manifest_path = os.path.join(here, 'fixtures', 'manifest2.json')
+        inst = self._makeOne('foo', reload=True)
+        inst.getmtime = lambda *args, **kwargs: 0
+        fut = inst.pregenerate
+
+        # test without a valid manifest
+        self.assertEqual(
+            fut('foo', ('css', 'main.css'), {}),
+            (['css', 'main.css'], {}))
+
+        # swap to a real manifest, setting mtime to 0
+        inst.manifest_path = manifest_path
+        self.assertEqual(
+            fut('foo', ('css', 'main.css'), {}),
+            (['css', 'main-test.css'], {}))
+
+        # ensure switching the path doesn't change the result
+        inst.manifest_path = new_manifest_path
+        self.assertEqual(
+            fut('foo', ('css', 'main.css'), {}),
+            (['css', 'main-test.css'], {}))
+
+        # update mtime, should cause a reload
+        inst.getmtime = lambda *args, **kwargs: 1
+        self.assertEqual(
+            fut('foo', ('css', 'main.css'), {}),
+            (['css', 'main-678b7c80.css'], {}))
+
+    def test_invalid_manifest(self):
+        self.assertRaises(IOError, lambda: self._makeOne('foo'))
+
+    def test_invalid_manifest_with_reload(self):
+        inst = self._makeOne('foo', reload=True)
+        self.assertEqual(inst.manifest, {})
+
 class DummyContext:
     pass
 
-- 
cgit v1.2.3


From 31f3d86d3bf5db3c4aa5085c7b7a4d6396f29931 Mon Sep 17 00:00:00 2001
From: Michael Merickel <michael@merickel.org>
Date: Thu, 12 Nov 2015 00:55:38 -0600
Subject: complete cache buster docs using manifest example

---
 docs/api/static.rst  |   3 +
 docs/glossary.rst    |   4 ++
 docs/narr/assets.rst | 183 ++++++++++++++++++++++-----------------------------
 3 files changed, 84 insertions(+), 106 deletions(-)

diff --git a/docs/api/static.rst b/docs/api/static.rst
index e6d6e4618..f3727e197 100644
--- a/docs/api/static.rst
+++ b/docs/api/static.rst
@@ -9,6 +9,9 @@
      :members:
      :inherited-members:
 
+  .. autoclass:: ManifestCacheBuster
+     :members:
+
   .. autoclass:: QueryStringCacheBuster
      :members:
 
diff --git a/docs/glossary.rst b/docs/glossary.rst
index 9c0ea8598..b4bb36421 100644
--- a/docs/glossary.rst
+++ b/docs/glossary.rst
@@ -1089,3 +1089,7 @@ Glossary
       data in a Redis database.  See 
       https://pypi.python.org/pypi/pyramid_redis_sessions for more information.
 
+   cache busting
+      A technique used when serving a cacheable static asset in order to force
+      a client to query the new version of the asset. See :ref:`cache_busting`
+      for more information.
diff --git a/docs/narr/assets.rst b/docs/narr/assets.rst
index 397e0258d..d36fa49c0 100644
--- a/docs/narr/assets.rst
+++ b/docs/narr/assets.rst
@@ -356,14 +356,14 @@ may change, and then you'll want the client to load a new copy of the asset.
 Under normal circumstances you'd just need to wait for the client's cached copy
 to expire before they get the new version of the static resource.
 
-A commonly used workaround to this problem is a technique known as "cache
-busting".  Cache busting schemes generally involve generating a URL for a
-static asset that changes when the static asset changes.  This way headers can
-be sent along with the static asset instructing the client to cache the asset
-for a very long time.  When a static asset is changed, the URL used to refer to
-it in a web page also changes, so the client sees it as a new resource and
-requests the asset, regardless of any caching policy set for the resource's old
-URL.
+A commonly used workaround to this problem is a technique known as
+:term:`cache busting`.  Cache busting schemes generally involve generating a
+URL for a static asset that changes when the static asset changes.  This way
+headers can be sent along with the static asset instructing the client to cache
+the asset for a very long time.  When a static asset is changed, the URL used
+to refer to it in a web page also changes, so the client sees it as a new
+resource and requests the asset, regardless of any caching policy set for the
+resource's old URL.
 
 :app:`Pyramid` can be configured to produce cache busting URLs for static
 assets by passing the optional argument, ``cachebust`` to
@@ -397,7 +397,7 @@ view to set headers instructing clients to cache the asset for ten years,
 unless the ``cache_max_age`` argument is also passed, in which case that
 value is used.
 
-.. warning::
+.. note::
 
    Cache busting is an inherently complex topic as it integrates the asset
    pipeline and the web application. It is expected and desired that
@@ -429,16 +429,14 @@ arbitrary token you provide to the query string of the asset's URL. This
 is almost never what you want in production as it does not allow fine-grained
 busting of individual assets.
 
-
 In order to implement your own cache buster, you can write your own class from
 scratch which implements the :class:`~pyramid.interfaces.ICacheBuster`
 interface.  Alternatively you may choose to subclass one of the existing
 implementations.  One of the most likely scenarios is you'd want to change the
-way the asset token is generated.  To do this just subclass either
-:class:`~pyramid.static.PathSegmentCacheBuster` or
+way the asset token is generated.  To do this just subclass
 :class:`~pyramid.static.QueryStringCacheBuster` and define a
-``tokenize(pathspec)`` method. Here is an example which just uses Git to get
-the hash of the currently checked out code:
+``tokenize(pathspec)`` method. Here is an example which uses Git to get
+the hash of the current commit:
 
 .. code-block:: python
    :linenos:
@@ -466,26 +464,60 @@ the hash of the currently checked out code:
 Choosing a Cache Buster
 ~~~~~~~~~~~~~~~~~~~~~~~
 
-The default cache buster implementation,
-:class:`~pyramid.static.PathSegmentMd5CacheBuster`, works very well assuming
-that you're using :app:`Pyramid` to serve your static assets.  The md5 checksum
-is fine grained enough that browsers should only request new versions of
-specific assets that have changed.  Many caching HTTP proxies will fail to
-cache a resource if the URL contains a query string.  In general, therefore,
-you should prefer a cache busting strategy which modifies the path segment to a
-strategy which adds a query string.
-
-It is possible, however, that your static assets are being served by another
-web server or externally on a CDN.  In these cases modifying the path segment
-for a static asset URL would cause the external service to fail to find the
-asset, causing your customer to get a 404.  In these cases you would need to
-fall back to a cache buster which adds a query string.  It is even possible
-that there isn't a copy of your static assets available to the :app:`Pyramid`
-application, so a cache busting implementation that generates md5 checksums
-would fail since it can't access the assets.  In such a case,
-:class:`~pyramid.static.QueryStringConstantCacheBuster` is a reasonable
-fallback.  The following code would set up a cachebuster that just uses the
-time at start up as a cachebust token:
+Many caching HTTP proxies will fail to cache a resource if the URL contains
+a query string.  Therefore, in general, you should prefer a cache busting
+strategy which modifies the path segment rather than methods which add a
+token to the query string.
+
+You will need to consider whether the :app:`Pyramid` application will be
+serving your static assets, whether you are using an external asset pipeline
+to handle rewriting urls internal to the css/javascript, and how fine-grained
+do you want the cache busting tokens to be.
+
+In many cases you will want to host the static assets on another web server
+or externally on a CDN. In these cases your :app:`Pyramid` application may not
+even have access to a copy of the static assets. In order to cache bust these
+assets you will need some information about them.
+
+If you are using an external asset pipeline to generate your static files you
+should consider using the :class:`~pyramid.static.ManifestCacheBuster`.
+This cache buster can load a standard JSON formatted file generated by your
+pipeline and use it to cache bust the assets. This has many performance
+advantages as :app:`Pyramid` does not need to look at the files to generate
+any cache busting tokens, but still supports fine-grained per-file tokens.
+
+Assuming an example ``manifest.json`` like:
+
+.. code-block:: json
+
+   {
+       "css/main.css": "css/main-678b7c80.css",
+       "images/background.png": "images/background-a8169106.png"
+   }
+
+The following code would set up a cachebuster:
+
+.. code-block:: python
+   :linenos:
+
+   from pyramid.path import AssetResolver
+   from pyramid.static import ManifestCacheBuster
+
+   resolver = AssetResolver()
+   manifest = resolver.resolve('myapp:static/manifest.json')
+   config.add_static_view(
+       name='http://mycdn.example.com/',
+       path='mypackage:static',
+       cachebust=ManifestCacheBuster(manifest.abspath()))
+
+A simpler approach is to use the
+:class:`~pyramid.static.QueryStringConstantCacheBuster` to generate a global
+token that will bust all of the assets at once. The advantage of this strategy
+is that it is simple and by using the query string there doesn't need to be
+any shared information between your application and the static assets.
+
+The following code would set up a cachebuster that just uses the time at
+start up as a cachebust token:
 
 .. code-block:: python
    :linenos:
@@ -496,7 +528,7 @@ time at start up as a cachebust token:
    config.add_static_view(
        name='http://mycdn.example.com/',
        path='mypackage:static',
-       cachebust=QueryStringConstantCacheBuster(str(time.time())))
+       cachebust=QueryStringConstantCacheBuster(str(int(time.time()))))
 
 .. index::
    single: static assets view
@@ -508,85 +540,24 @@ Often one needs to refer to images and other static assets inside CSS and
 JavaScript files. If cache busting is active, the final static asset URL is not
 available until the static assets have been assembled. These URLs cannot be
 handwritten. Thus, when having static asset references in CSS and JavaScript,
-one needs to perform one of the following tasks.
+one needs to perform one of the following tasks:
 
 * Process the files by using a precompiler which rewrites URLs to their final
-  cache busted form.
+  cache busted form. Then, you can use the
+  :class:`~pyramid.static.ManifestCacheBuster` to synchronize your asset
+  pipeline with :app:`Pyramid`, allowing the pipeline to have full control
+  over the final URLs of your assets.
 
 * Templatize JS and CSS, and call ``request.static_url()`` inside their
   template code.
 
 * Pass static URL references to CSS and JavaScript via other means.
 
-Below are some simple approaches for CSS and JS programming which consider
-asset cache busting. These approaches do not require additional tools or
-packages.
-
-Relative cache busted URLs in CSS
-+++++++++++++++++++++++++++++++++
-
-Consider a CSS file ``/static/theme/css/site.css`` which contains the following
-CSS code.
-
-.. code-block:: css
-
-    body {
-        background: url(/static/theme/img/background.jpg);
-    }
-
-Any changes to ``background.jpg`` would not appear to the visitor because the
-URL path is not cache busted as it is. Instead we would have to construct an
-URL to the background image with the default ``PathSegmentCacheBuster`` cache
-busting mechanism::
-
-    https://site/static/1eeb262c717/theme/img/background.jpg
-
-Every time the image is updated, the URL would need to be changed. It is not
-practical to write this non-human readable URL into a CSS file.
-
-However, the CSS file itself is cache busted and is located under the path for
-static assets. This lets us use relative references in our CSS to cache bust
-the image.
-
-.. code-block:: css
-
-    body {
-        background: url(../img/background.jpg);
-    }
-
-The browser would interpret this as having the CSS file hash in URL::
-
-    https://site/static/ab234b262c71/theme/css/../img/background.jpg
-
-The downside of this approach is that if the background image changes, one
-needs to bump the CSS file. The CSS file hash change signals the caches that
-the relative URL to the image in the CSS has been changed. When updating CSS
-and related image assets, updates usually happen hand in hand, so this does not
-add extra effort to theming workflow.
-
-Passing cache busted URLs to JavaScript
-+++++++++++++++++++++++++++++++++++++++
-
-For JavaScript, one can pass static asset URLs as function arguments or
-globals. The globals can be generated in page template code, having access to
-the ``request.static_url()`` function.
-
-Below is a simple example of passing a cached busted image URL in the Jinja2
-template language. Put the following code into the ``<head>`` section of the
-relevant page.
-
-.. code-block:: html
-
-    <script>
-        window.assets.backgroundImage =
-            "{{ '/theme/img/background.jpg'|static_url() }}";
-    </script>
-
-Then in your main ``site.js`` file, put the following code.
-
-.. code-block:: javascript
-
-    var image = new Image(window.assets.backgroundImage);
+If your CSS and JavaScript assets use URLs to reference other assets it is
+recommended that you implement an external asset pipeline that can rewrite the
+generated static files with new URLs containing cache busting tokens. The
+machinery inside :app:`Pyramid` will not help with this step as it has very
+little knowledge of the asset types your application may use.
 
 .. _advanced_static:
 
-- 
cgit v1.2.3