Changeset - 023f7873ef59
[Not reviewed]
beta
0 3 0
Marcin Kuzminski - 13 years ago 2012-11-23 11:54:14
marcin@python-works.com
added caching layer into RSS/ATOM feeds
- updated code for new LimitedDiffContainers
- invalidate rss/atom cache keys on push
-
3 files changed with 59 insertions and 14 deletions:
0 comments (0 inline, 0 general)
rhodecode/controllers/feed.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    rhodecode.controllers.feed
 
    ~~~~~~~~~~~~~~~~~~~~~~~~~~
 

	
 
    Feed controller for rhodecode
 

	
 
    :created_on: Apr 23, 2010
 
    :author: marcink
 
    :copyright: (C) 2010-2012 Marcin Kuzminski <marcin@python-works.com>
 
    :license: GPLv3, see COPYING for more details.
 
"""
 
# This program is free software: you can redistribute it and/or modify
 
# it under the terms of the GNU General Public License as published by
 
# the Free Software Foundation, either version 3 of the License, or
 
# (at your option) any later version.
 
#
 
# This program is distributed in the hope that it will be useful,
 
# but WITHOUT ANY WARRANTY; without even the implied warranty of
 
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 
# GNU General Public License for more details.
 
#
 
# You should have received a copy of the GNU General Public License
 
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
 

	
 
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__)
 

	
 

	
 
class FeedController(BaseRepoController):
 

	
 
    @LoginRequired(api_access=True)
 
    @HasRepoPermissionAnyDecorator('repository.read', 'repository.write',
 
                                   'repository.admin')
 
    def __before__(self):
 
        super(FeedController, self).__before__()
 
        #common values for feeds
 
        self.description = _('Changes on %s repository')
 
        self.title = self.title = _('%s %s feed') % (c.rhodecode_name, '%s')
 
        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" % (
 
            h.shorter(cs.message, 160)
 
        )
 

	
 
    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:
 
            desc_msg.append('branch: %s<br/>' % cs.branch)
 
        if h.is_hg(c.rhodecode_repo):
 
            for book in cs.bookmarks:
 
                desc_msg.append('bookmark: %s<br/>' % book)
 
        for tag in cs.tags:
 
            desc_msg.append('tag: %s<br/>' % tag)
 
        # rev link
 
        _url = url('changeset_home', repo_name=cs.repository.name,
 
                   revision=cs.raw_id, qualified=True)
 
        desc_msg.append('changesest: <a href="%s">%s</a>' % (_url, cs.raw_id[:8]))
 

	
 
        desc_msg.append('<pre>')
 
        desc_msg.append(cs.message)
 
        desc_msg.append('\n')
 
        desc_msg.extend(self.__changes(cs))
 
        desc_msg.append('</pre>')
 
        return desc_msg
 

	
 
    def atom(self, repo_name):
 
        """Produce an atom-1.0 feed via feedgenerator module"""
 

	
 
        @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"""
 

	
 
        @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)
 

	
rhodecode/lib/utils2.py
Show inline comments
 
@@ -234,96 +234,108 @@ def safe_unicode(str_, from_encoding=Non
 

	
 
def safe_str(unicode_, to_encoding=None):
 
    """
 
    safe str function. Does few trick to turn unicode_ into string
 

	
 
    In case of UnicodeEncodeError we try to return it with encoding detected
 
    by chardet library if it fails fallback to string with errors replaced
 

	
 
    :param unicode_: unicode to encode
 
    :rtype: str
 
    :returns: str object
 
    """
 

	
 
    # if it's not basestr cast to str
 
    if not isinstance(unicode_, basestring):
 
        return str(unicode_)
 

	
 
    if isinstance(unicode_, str):
 
        return unicode_
 

	
 
    if not to_encoding:
 
        import rhodecode
 
        DEFAULT_ENCODINGS = aslist(rhodecode.CONFIG.get('default_encoding',
 
                                                        'utf8'), sep=',')
 
        to_encoding = DEFAULT_ENCODINGS
 

	
 
    if not isinstance(to_encoding, (list, tuple)):
 
        to_encoding = [to_encoding]
 

	
 
    for enc in to_encoding:
 
        try:
 
            return unicode_.encode(enc)
 
        except UnicodeEncodeError:
 
            pass
 

	
 
    try:
 
        import chardet
 
        encoding = chardet.detect(unicode_)['encoding']
 
        if encoding is None:
 
            raise UnicodeEncodeError()
 

	
 
        return unicode_.encode(encoding)
 
    except (ImportError, UnicodeEncodeError):
 
        return unicode_.encode(to_encoding[0], 'replace')
 

	
 
    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
 
    file based sqlite databases. This prevents errors on sqlite. This only
 
    applies to sqlalchemy versions < 0.7.0
 

	
 
    """
 
    import sqlalchemy
 
    from sqlalchemy import engine_from_config as efc
 
    import logging
 

	
 
    if int(sqlalchemy.__version__.split('.')[1]) < 7:
 

	
 
        # This solution should work for sqlalchemy < 0.7.0, and should use
 
        # proxy=TimerProxy() for execution time profiling
 

	
 
        from sqlalchemy.pool import NullPool
 
        url = configuration[prefix + 'url']
 

	
 
        if url.startswith('sqlite'):
 
            kwargs.update({'poolclass': NullPool})
 
        return efc(configuration, prefix, **kwargs)
 
    else:
 
        import time
 
        from sqlalchemy import event
 
        from sqlalchemy.engine import Engine
 

	
 
        log = logging.getLogger('sqlalchemy.engine')
 
        BLACK, RED, GREEN, YELLOW, BLUE, MAGENTA, CYAN, WHITE = xrange(30, 38)
 
        engine = efc(configuration, prefix, **kwargs)
 

	
 
        def color_sql(sql):
 
            COLOR_SEQ = "\033[1;%dm"
 
            COLOR_SQL = YELLOW
 
            normal = '\x1b[0m'
 
            return ''.join([COLOR_SEQ % COLOR_SQL, sql, normal])
 

	
 
        if configuration['debug']:
 
            #attach events only for debug configuration
 

	
 
            def before_cursor_execute(conn, cursor, statement,
 
                                    parameters, context, executemany):
 
                context._query_start_time = time.time()
 
                log.info(color_sql(">>>>> STARTING QUERY >>>>>"))
 

	
 
            def after_cursor_execute(conn, cursor, statement,
 
                                    parameters, context, executemany):
 
                total = time.time() - context._query_start_time
rhodecode/model/db.py
Show inline comments
 
# -*- coding: utf-8 -*-
 
"""
 
    rhodecode.model.db
 
    ~~~~~~~~~~~~~~~~~~
 

	
 
    Database Models for RhodeCode
 

	
 
    :created_on: Apr 08, 2010
 
    :author: marcink
 
    :copyright: (C) 2010-2012 Marcin Kuzminski <marcin@python-works.com>
 
    :license: GPLv3, see COPYING for more details.
 
"""
 
# This program is free software: you can redistribute it and/or modify
 
# it under the terms of the GNU General Public License as published by
 
# the Free Software Foundation, either version 3 of the License, or
 
# (at your option) any later version.
 
#
 
# This program is distributed in the hope that it will be useful,
 
# but WITHOUT ANY WARRANTY; without even the implied warranty of
 
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 
# GNU General Public License for more details.
 
#
 
# You should have received a copy of the GNU General Public License
 
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
 

	
 
import os
 
import logging
 
import datetime
 
import traceback
 
import hashlib
 
import time
 
from collections import defaultdict
 

	
 
from sqlalchemy import *
 
from sqlalchemy.ext.hybrid import hybrid_property
 
from sqlalchemy.orm import relationship, joinedload, class_mapper, validates
 
from sqlalchemy.exc import DatabaseError
 
from beaker.cache import cache_region, region_invalidate
 
from webob.exc import HTTPNotFound
 

	
 
from pylons.i18n.translation import lazy_ugettext as _
 

	
 
from rhodecode.lib.vcs import get_backend
 
from rhodecode.lib.vcs.utils.helpers import get_scm
 
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
 

	
 
from rhodecode.model.meta import Base, Session
 

	
 
URL_SEP = '/'
 
log = logging.getLogger(__name__)
 

	
 
#==============================================================================
 
# BASE CLASSES
 
#==============================================================================
 

	
 
_hash_key = lambda k: hashlib.md5(safe_str(k)).hexdigest()
 

	
 

	
 
class BaseModel(object):
 
    """
 
    Base Model for all classess
 
    """
 

	
 
    @classmethod
 
    def _get_keys(cls):
 
        """return column names for this model """
 
        return class_mapper(cls).c.keys()
 

	
 
    def get_dict(self):
 
        """
 
        return dict with keys and values corresponding
 
        to this model data """
 

	
 
        d = {}
 
        for k in self._get_keys():
 
            d[k] = getattr(self, k)
 

	
 
        # also use __json__() if present to get additional fields
 
        _json_attr = getattr(self, '__json__', None)
 
        if _json_attr:
 
            # update with attributes from __json__
 
            if callable(_json_attr):
 
                _json_attr = _json_attr()
 
            for k, val in _json_attr.iteritems():
 
                d[k] = val
 
        return d
 

	
 
    def get_appstruct(self):
 
        """return list with keys and values tupples corresponding
 
        to this model data """
 

	
 
@@ -896,96 +896,97 @@ class Repository(Base, BaseModel):
 
        """
 
        Returns statuses for this repository
 

	
 
        :param revisions: list of revisions to get statuses for
 
        :type revisions: list
 
        """
 

	
 
        statuses = ChangesetStatus.query()\
 
            .filter(ChangesetStatus.repo == self)\
 
            .filter(ChangesetStatus.version == 0)
 
        if revisions:
 
            statuses = statuses.filter(ChangesetStatus.revision.in_(revisions))
 
        grouped = {}
 

	
 
        #maybe we have open new pullrequest without a status ?
 
        stat = ChangesetStatus.STATUS_UNDER_REVIEW
 
        status_lbl = ChangesetStatus.get_status_lbl(stat)
 
        for pr in PullRequest.query().filter(PullRequest.org_repo == self).all():
 
            for rev in pr.revisions:
 
                pr_id = pr.pull_request_id
 
                pr_repo = pr.other_repo.repo_name
 
                grouped[rev] = [stat, status_lbl, pr_id, pr_repo]
 

	
 
        for stat in statuses.all():
 
            pr_id = pr_repo = None
 
            if stat.pull_request:
 
                pr_id = stat.pull_request.pull_request_id
 
                pr_repo = stat.pull_request.other_repo.repo_name
 
            grouped[stat.revision] = [str(stat.status), stat.status_lbl,
 
                                      pr_id, pr_repo]
 
        return grouped
 

	
 
    #==========================================================================
 
    # SCM CACHE INSTANCE
 
    #==========================================================================
 

	
 
    @property
 
    def invalidate(self):
 
        return CacheInvalidation.invalidate(self.repo_name)
 

	
 
    def set_invalidate(self):
 
        """
 
        set a cache for invalidation for this instance
 
        """
 
        CacheInvalidation.set_invalidate(self.repo_name)
 

	
 
    @LazyProperty
 
    def scm_instance(self):
 
        return self.scm_instance_cached()
 
        return self.__get_instance()
 

	
 
    def scm_instance_cached(self, cache_map=None):
 
        @cache_region('long_term')
 
        def _c(repo_name):
 
            return self.__get_instance()
 
        rn = self.repo_name
 
        log.debug('Getting cached instance of repo')
 

	
 
        if cache_map:
 
            # get using prefilled cache_map
 
            invalidate_repo = cache_map[self.repo_name]
 
            if invalidate_repo:
 
                invalidate_repo = (None if invalidate_repo.cache_active
 
                                   else invalidate_repo)
 
        else:
 
            # get from invalidate
 
            invalidate_repo = self.invalidate
 

	
 
        if invalidate_repo is not None:
 
            region_invalidate(_c, None, rn)
 
            # update our cache
 
            CacheInvalidation.set_valid(invalidate_repo.cache_key)
 
        return _c(rn)
 

	
 
    def __get_instance(self):
 
        repo_full_path = self.repo_full_path
 
        try:
 
            alias = get_scm(repo_full_path)[0]
 
            log.debug('Creating instance of %s repository' % alias)
 
            backend = get_backend(alias)
 
        except VCSError:
 
            log.error(traceback.format_exc())
 
            log.error('Perhaps this repository is in db and not in '
 
                      'filesystem run rescan repositories with '
 
                      '"destroy old data " option from admin panel')
 
            return
 

	
 
        if alias == 'hg':
 

	
 
            repo = backend(safe_str(repo_full_path), create=False,
 
                           baseui=self._ui)
 
            # skip hidden web repository
 
            if repo._get_hidden():
 
                return
 
        else:
 
            repo = backend(repo_full_path, create=False)
 

	
 
@@ -1395,97 +1396,101 @@ class UserFollowing(Base, BaseModel):
 
        return cls.query().filter(cls.follows_repo_id == repo_id)
 

	
 

	
 
class CacheInvalidation(Base, BaseModel):
 
    __tablename__ = 'cache_invalidation'
 
    __table_args__ = (
 
        UniqueConstraint('cache_key'),
 
        Index('key_idx', 'cache_key'),
 
        {'extend_existing': True, 'mysql_engine': 'InnoDB',
 
         'mysql_charset': 'utf8'},
 
    )
 
    cache_id = Column("cache_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
 
    cache_key = Column("cache_key", String(255, convert_unicode=False, assert_unicode=None), nullable=True, unique=None, default=None)
 
    cache_args = Column("cache_args", String(255, convert_unicode=False, assert_unicode=None), nullable=True, unique=None, default=None)
 
    cache_active = Column("cache_active", Boolean(), nullable=True, unique=None, default=False)
 

	
 
    def __init__(self, cache_key, cache_args=''):
 
        self.cache_key = cache_key
 
        self.cache_args = cache_args
 
        self.cache_active = False
 

	
 
    def __unicode__(self):
 
        return u"<%s('%s:%s')>" % (self.__class__.__name__,
 
                                  self.cache_id, self.cache_key)
 

	
 
    @property
 
    def prefix(self):
 
        _split = self.cache_key.split(self.cache_args, 1)
 
        if _split and len(_split) == 2:
 
            return _split[0]
 
        return ''
 

	
 
    @classmethod
 
    def clear_cache(cls):
 
        cls.query().delete()
 

	
 
    @classmethod
 
    def _get_key(cls, key):
 
        """
 
        Wrapper for generating a key, together with a prefix
 

	
 
        :param key:
 
        """
 
        import rhodecode
 
        prefix = ''
 
        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):
 
        return cls.query().filter(cls.cache_key == key).scalar()
 

	
 
    @classmethod
 
    def _get_or_create_key(cls, key, prefix, org_key, commit=True):
 
        inv_obj = Session().query(cls).filter(cls.cache_key == key).scalar()
 
        if not inv_obj:
 
            try:
 
                inv_obj = CacheInvalidation(key, org_key)
 
                Session().add(inv_obj)
 
                if commit:
 
                    Session().commit()
 
            except Exception:
 
                log.error(traceback.format_exc())
 
                Session().rollback()
 
        return inv_obj
 

	
 
    @classmethod
 
    def invalidate(cls, key):
 
        """
 
        Returns Invalidation object if this given key should be invalidated
 
        None otherwise. `cache_active = False` means that this cache
 
        state is not valid and needs to be invalidated
 

	
 
        :param key:
 
        """
 

	
 
        key, _prefix, _org_key = cls._get_key(key)
 
        inv = cls._get_or_create_key(key, _prefix, _org_key)
 

	
 
        if inv and inv.cache_active is False:
 
            return inv
 

	
 
    @classmethod
 
    def set_invalidate(cls, key):
 
        """
 
        Mark this Cache key for invalidation
 

	
 
        :param key:
 
        """
 

	
 
        key, _prefix, _org_key = cls._get_key(key)
 
        inv_objs = Session().query(cls).filter(cls.cache_args == _org_key).all()
 
        log.debug('marking %s key[s] %s for invalidation' % (len(inv_objs),
 
                                                             _org_key))
 
        try:
0 comments (0 inline, 0 general)