added caching layer into RSS/ATOM feeds

- updated code for new LimitedDiffContainers
- invalidate rss/atom cache keys on push
-

--HG--
branch : beta

3 files changed, 97 insertions, 52 deletions

diff --git a/rhodecode/controllers/feed.py b/rhodecode/controllers/feed.py
index c05bbb29..4b145553 100644
--- a/rhodecode/controllers/feed.py
+++ b/rhodecode/controllers/feed.py
@@ -28,12 +28,14 @@ import logging
 from pylons import url, response, tmpl_context as c
 from pylons.i18n.translation import _
 
+from beaker.cache import cache_region, region_invalidate
 from webhelpers.feedgenerator import Atom1Feed, Rss201rev2Feed
 
 from rhodecode.lib import helpers as h
 from rhodecode.lib.auth import LoginRequired, HasRepoPermissionAnyDecorator
 from rhodecode.lib.base import BaseRepoController
-from rhodecode.lib.diffs import DiffProcessor
+from rhodecode.lib.diffs import DiffProcessor, LimitedDiffContainer
+from rhodecode.model.db import CacheInvalidation
 
 log = logging.getLogger(__name__)
 
@@ -51,6 +53,9 @@ class FeedController(BaseRepoController):
         self.language = 'en-us'
         self.ttl = "5"
         self.feed_nr = 20
+        # we need to protect from parsing huge diffs here other way
+        # we can kill the server, 32*1024 chars is a reasonable limit
+        self.feed_diff_limit = 32 * 1024
 
     def _get_title(self, cs):
         return "%s" % (
@@ -59,26 +64,28 @@ class FeedController(BaseRepoController):
 
     def __changes(self, cs):
         changes = []
-        _diff = cs.diff()
-        # we need to protect from parsing huge diffs here other way
-        # we can kill the server, 32*1024 chars is a reasonable limit
-        HUGE_DIFF = 32 * 1024
-        if len(_diff) > HUGE_DIFF:
-            changes = ['\n ' + _('Changeset was too big and was cut off...')]
-            return changes
-        diffprocessor = DiffProcessor(_diff)
-        stats = diffprocessor.prepare(inline_diff=False)
-        for st in stats:
+        diff_processor = DiffProcessor(cs.diff(),
+                                       diff_limit=self.feed_diff_limit)
+        _parsed = diff_processor.prepare(inline_diff=False)
+        limited_diff = False
+        if isinstance(_parsed, LimitedDiffContainer):
+            limited_diff = True
+
+        for st in _parsed:
             st.update({'added': st['stats'][0],
                        'removed': st['stats'][1]})
             changes.append('\n %(operation)s %(filename)s '
                            '(%(added)s lines added, %(removed)s lines removed)'
                             % st)
+        if limited_diff:
+            changes = changes + ['\n ' +
+                                 _('Changeset was too big and was cut off...')]
         return changes
 
     def __get_desc(self, cs):
         desc_msg = []
-        desc_msg.append('%s %s %s:<br/>' % (cs.author, _('commited on'),
+        desc_msg.append('%s %s %s<br/>' % (h.person(cs.author),
+                                           _('commited on'),
                                            h.fmt_date(cs.date)))
         #branches, tags, bookmarks
         if cs.branch:
@@ -102,46 +109,67 @@ class FeedController(BaseRepoController):
 
     def atom(self, repo_name):
         """Produce an atom-1.0 feed via feedgenerator module"""
-        feed = Atom1Feed(
-             title=self.title % repo_name,
-             link=url('summary_home', repo_name=repo_name,
-                      qualified=True),
-             description=self.description % repo_name,
-             language=self.language,
-             ttl=self.ttl
-        )
-
-        for cs in reversed(list(c.rhodecode_repo[-self.feed_nr:])):
-            feed.add_item(title=self._get_title(cs),
-                          link=url('changeset_home', repo_name=repo_name,
-                                   revision=cs.raw_id, qualified=True),
-                          author_name=cs.author,
-                          description=''.join(self.__get_desc(cs)),
-                          pubdate=cs.date,
-                          )
 
-        response.content_type = feed.mime_type
-        return feed.writeString('utf-8')
+        @cache_region('long_term')
+        def _get_feed_from_cache(key):
+            feed = Atom1Feed(
+                 title=self.title % repo_name,
+                 link=url('summary_home', repo_name=repo_name,
+                          qualified=True),
+                 description=self.description % repo_name,
+                 language=self.language,
+                 ttl=self.ttl
+            )
+
+            for cs in reversed(list(c.rhodecode_repo[-self.feed_nr:])):
+                feed.add_item(title=self._get_title(cs),
+                              link=url('changeset_home', repo_name=repo_name,
+                                       revision=cs.raw_id, qualified=True),
+                              author_name=cs.author,
+                              description=''.join(self.__get_desc(cs)),
+                              pubdate=cs.date,
+                              )
+
+            response.content_type = feed.mime_type
+            return feed.writeString('utf-8')
+
+        key = repo_name + '_ATOM'
+        inv = CacheInvalidation.invalidate(key)
+        if inv is not None:
+            region_invalidate(_get_feed_from_cache, None, key)
+            CacheInvalidation.set_valid(inv.cache_key)
+        return _get_feed_from_cache(key)
 
     def rss(self, repo_name):
         """Produce an rss2 feed via feedgenerator module"""
-        feed = Rss201rev2Feed(
-            title=self.title % repo_name,
-            link=url('summary_home', repo_name=repo_name,
-                     qualified=True),
-            description=self.description % repo_name,
-            language=self.language,
-            ttl=self.ttl
-        )
 
-        for cs in reversed(list(c.rhodecode_repo[-self.feed_nr:])):
-            feed.add_item(title=self._get_title(cs),
-                          link=url('changeset_home', repo_name=repo_name,
-                                   revision=cs.raw_id, qualified=True),
-                          author_name=cs.author,
-                          description=''.join(self.__get_desc(cs)),
-                          pubdate=cs.date,
-                         )
-
-        response.content_type = feed.mime_type
-        return feed.writeString('utf-8')
+        @cache_region('long_term')
+        def _get_feed_from_cache(key):
+            feed = Rss201rev2Feed(
+                title=self.title % repo_name,
+                link=url('summary_home', repo_name=repo_name,
+                         qualified=True),
+                description=self.description % repo_name,
+                language=self.language,
+                ttl=self.ttl
+            )
+
+            for cs in reversed(list(c.rhodecode_repo[-self.feed_nr:])):
+                feed.add_item(title=self._get_title(cs),
+                              link=url('changeset_home', repo_name=repo_name,
+                                       revision=cs.raw_id, qualified=True),
+                              author_name=cs.author,
+                              description=''.join(self.__get_desc(cs)),
+                              pubdate=cs.date,
+                             )
+
+            response.content_type = feed.mime_type
+            return feed.writeString('utf-8')
+
+        key = repo_name + '_RSS'
+        inv = CacheInvalidation.invalidate(key)
+        if inv is not None:
+            region_invalidate(_get_feed_from_cache, None, key)
+            CacheInvalidation.set_valid(inv.cache_key)
+        return _get_feed_from_cache(key)
+
diff --git a/rhodecode/lib/utils2.py b/rhodecode/lib/utils2.py
index 5e36e670..1b2fb178 100644
--- a/rhodecode/lib/utils2.py
+++ b/rhodecode/lib/utils2.py
@@ -279,6 +279,18 @@ def safe_str(unicode_, to_encoding=None):
     return safe_str
 
 
+def remove_suffix(s, suffix):
+    if s.endswith(suffix):
+        s = s[:-1 * len(suffix)]
+    return s
+
+
+def remove_prefix(s, prefix):
+    if s.startswith(prefix):
+        s = s[:-1 * len(prefix)]
+    return s
+
+
 def engine_from_config(configuration, prefix='sqlalchemy.', **kwargs):
     """
     Custom engine_from_config functions that makes sure we use NullPool for
diff --git a/rhodecode/model/db.py b/rhodecode/model/db.py
index 22251271..74aff8fe 100755
--- a/rhodecode/model/db.py
+++ b/rhodecode/model/db.py
@@ -46,7 +46,7 @@ from rhodecode.lib.vcs.exceptions import VCSError
 from rhodecode.lib.vcs.utils.lazy import LazyProperty
 
 from rhodecode.lib.utils2 import str2bool, safe_str, get_changeset_safe, \
-    safe_unicode
+    safe_unicode, remove_suffix
 from rhodecode.lib.compat import json
 from rhodecode.lib.caching_query import FromCache
 
@@ -941,6 +941,7 @@ class Repository(Base, BaseModel):
 
     @LazyProperty
     def scm_instance(self):
+        return self.scm_instance_cached()
         return self.__get_instance()
 
     def scm_instance_cached(self, cache_map=None):
@@ -1440,7 +1441,11 @@ class CacheInvalidation(Base, BaseModel):
         iid = rhodecode.CONFIG.get('instance_id')
         if iid:
             prefix = iid
-        return "%s%s" % (prefix, key), prefix, key.rstrip('_README')
+        #remove specific suffixes like _README or _RSS
+        key = remove_suffix(key, '_README')
+        key = remove_suffix(key, '_RSS')
+        key = remove_suffix(key, '_ATOM')
+        return "%s%s" % (prefix, key), prefix, key
 
     @classmethod
     def get_by_key(cls, key):