Changeset - 1fed3c9161bb
[Not reviewed]
beta
0 3 0
Marcin Kuzminski - 15 years ago 2010-12-29 13:54:03
marcin@python-works.com
fixes #90 + docs update
3 files changed with 44 insertions and 15 deletions:
0 comments (0 inline, 0 general)
docs/setup.rst
Show inline comments
 
@@ -51,39 +51,48 @@ You are ready to use rhodecode, to run i
 
  users and repositories
 
  
 
    
 
Setting up Whoosh full text search
 
----------------------------------
 

	
 
Index for whoosh can be build starting from version 1.1 using paster command
 
passing repo locations to index, as well as Your config file that stores
 
whoosh index files locations. There is possible to pass `-f` to the options
 
Starting from version 1.1 whoosh index can be build using paster command.
 
You have to specify the config file that stores location of index, and
 
location of repositories (`--repo-location`). Starting from version 1.2 it is 
 
also possible to specify a comma separated list of repositories (`--index-only`)
 
to build index only on chooses repositories skipping any other found in repos
 
location
 

	
 
There is possible also to pass `-f` to the options
 
to enable full index rebuild. Without that indexing will run always in in
 
incremental mode.
 

	
 
::
 
incremental mode::
 

	
 
 paster make-index production.ini --repo-location=<location for repos> 
 

	
 
for full index rebuild You can use
 

	
 

	
 
::
 
for full index rebuild You can use::
 

	
 
 paster make-index production.ini -f --repo-location=<location for repos>
 

	
 
- For full text search You can either put crontab entry for
 

	
 
building index just for chosen repositories is possible with such command::
 
 
 
 paster make-index production.ini --repo-location=<location for repos> --index-only=vcs,rhodecode
 

	
 
This command can be run even from crontab in order to do periodical 
 
index builds and keep Your index always up to date. An example entry might 
 
look like this
 

	
 
In order to do periodical index builds and keep Your index always up to date.
 
It's recommended to do a crontab entry for incremental indexing. 
 
An example entry might look like this
 

	
 
::
 
 
 
 /path/to/python/bin/paster /path/to/rhodecode/production.ini --repo-location=<location for repos> 
 
  
 
When using incremental(default) mode whoosh will check last modification date 
 
When using incremental (default) mode whoosh will check last modification date 
 
of each file and add it to reindex if newer file is available. Also indexing 
 
daemon checks for removed files and removes them from index. 
 

	
 
Sometime You might want to rebuild index from scratch. You can do that using 
 
the `-f` flag passed to paster command or, in admin panel You can check 
 
`build from scratch` flag.
rhodecode/lib/indexers/__init__.py
Show inline comments
 
@@ -3,12 +3,14 @@ import sys
 
import traceback
 
from os.path import dirname as dn, join as jn
 

	
 
#to get the rhodecode import
 
sys.path.append(dn(dn(dn(os.path.realpath(__file__)))))
 

	
 
from string import strip
 

	
 
from rhodecode.model import init_model
 
from rhodecode.model.scm import ScmModel
 
from rhodecode.config.environment import load_environment
 
from rhodecode.lib.utils import BasePasterCommand, Command, add_cache
 

	
 
from shutil import rmtree
 
@@ -68,33 +70,41 @@ class MakeIndex(BasePasterCommand):
 
        add_cache(config)
 
        engine = engine_from_config(config, 'sqlalchemy.db1.')
 
        init_model(engine)
 

	
 
        index_location = config['index_dir']
 
        repo_location = self.options.repo_location
 
        repo_list = map(strip, self.options.repo_list.split(','))
 

	
 
        #======================================================================
 
        # WHOOSH DAEMON
 
        #======================================================================
 
        from rhodecode.lib.pidlock import LockHeld, DaemonLock
 
        from rhodecode.lib.indexers.daemon import WhooshIndexingDaemon
 
        try:
 
            l = DaemonLock()
 
            WhooshIndexingDaemon(index_location=index_location,
 
                                 repo_location=repo_location)\
 
                                 repo_location=repo_location,
 
                                 repo_list=repo_list)\
 
                .run(full_index=self.options.full_index)
 
            l.release()
 
        except LockHeld:
 
            sys.exit(1)
 

	
 
    def update_parser(self):
 
        self.parser.add_option('--repo-location',
 
                          action='store',
 
                          dest='repo_location',
 
                          help="Specifies repositories location to index REQUIRED",
 
                          )
 
        self.parser.add_option('--index-only',
 
                          action='store',
 
                          dest='repo_list',
 
                          help="Specifies a comma separated list of repositores "
 
                                "to build index on OPTIONAL",
 
                          )
 
        self.parser.add_option('-f',
 
                          action='store_true',
 
                          dest='full_index',
 
                          help="Specifies that index should be made full i.e"
 
                                " destroy old and build from scratch",
 
                          default=False)
rhodecode/lib/indexers/daemon.py
Show inline comments
 
@@ -67,24 +67,34 @@ log.addHandler(ch)
 
class WhooshIndexingDaemon(object):
 
    """
 
    Deamon for atomic jobs
 
    """
 

	
 
    def __init__(self, indexname='HG_INDEX', index_location=None,
 
                 repo_location=None, sa=None):
 
                 repo_location=None, sa=None, repo_list=None):
 
        self.indexname = indexname
 

	
 
        self.index_location = index_location
 
        if not index_location:
 
            raise Exception('You have to provide index location')
 

	
 
        self.repo_location = repo_location
 
        if not repo_location:
 
            raise Exception('You have to provide repositories location')
 

	
 
        self.repo_paths = ScmModel(sa).repo_scan(self.repo_location, None)
 

	
 
        if repo_list:
 
            filtered_repo_paths = {}
 
            for repo_name, repo in self.repo_paths.items():
 
                if repo_name in repo_list:
 
                    filtered_repo_paths[repo.name] = repo
 

	
 
            self.repo_paths = filtered_repo_paths
 

	
 

	
 
        self.initial = False
 
        if not os.path.isdir(self.index_location):
 
            os.makedirs(self.index_location)
 
            log.info('Cannot run incremental index since it does not'
 
                     ' yet exist running full build')
 
            self.initial = True
 
@@ -151,14 +161,14 @@ class WhooshIndexingDaemon(object):
 

	
 
        if not os.path.exists(self.index_location):
 
            os.mkdir(self.index_location)
 

	
 
        idx = create_in(self.index_location, SCHEMA, indexname=IDX_NAME)
 
        writer = idx.writer()
 
        print self.repo_paths.values()
 
        for cnt, repo in enumerate(self.repo_paths.values()):
 

	
 
        for repo in self.repo_paths.values():
 
            log.debug('building index @ %s' % repo.path)
 

	
 
            for idx_path in self.get_paths(repo):
 
                self.add_doc(writer, idx_path, repo)
 

	
 
        log.debug('>> COMMITING CHANGES <<')
0 comments (0 inline, 0 general)