Changeset - 1828eb7fa688
[Not reviewed]
beta
0 3 0
Marcin Kuzminski - 13 years ago 2012-06-02 18:01:56
marcin@python-works.com
#469 added --update-only option to whoosh to re-index only given list
of repos in index
3 files changed with 41 insertions and 16 deletions:
0 comments (0 inline, 0 general)
docs/changelog.rst
Show inline comments
 
@@ -11,24 +11,26 @@ Changelog
 
:branch: beta
 

	
 
news
 
++++
 
 
 
- new codereview system
 
- email map, allowing users to have multiple email addresses mapped into
 
  their accounts
 
- changed setup-app into setup-rhodecode and added default options to it.
 
- new git repos are created as bare now by default
 
- #464 added links to groups in permission box
 
- #465 mentions autocomplete inside comments boxes
 
- #469 added --update-only option to whoosh to re-index only given list
 
  of repos in index 
 

	
 
fixes
 
+++++
 

	
 
- improved translations
 
- fixes issue #455 Creating an archive generates an exception on Windows
 
- fixes #448 Download ZIP archive keeps file in /tmp open and results 
 
  in out of disk space
 
- fixes issue #454 Search results under Windows include proceeding
 
  backslash
 
- fixed issue #450. Rhodecode no longer will crash when bad revision is
 
  present in journal data.
rhodecode/lib/indexers/__init__.py
Show inline comments
 
@@ -84,51 +84,61 @@ class MakeIndex(BasePasterCommand):
 

	
 
    def command(self):
 
        logging.config.fileConfig(self.path_to_ini_file)
 
        from pylons import config
 
        add_cache(config)
 
        engine = engine_from_config(config, 'sqlalchemy.db1.')
 
        init_model(engine)
 
        index_location = config['index_dir']
 
        repo_location = self.options.repo_location \
 
            if self.options.repo_location else RepoModel().repos_path
 
        repo_list = map(strip, self.options.repo_list.split(',')) \
 
            if self.options.repo_list else None
 
        repo_update_list = map(strip, self.options.repo_update_list.split(',')) \
 
            if self.options.repo_update_list else None
 
        load_rcextensions(config['here'])
 
        #======================================================================
 
        # WHOOSH DAEMON
 
        #======================================================================
 
        from rhodecode.lib.pidlock import LockHeld, DaemonLock
 
        from rhodecode.lib.indexers.daemon import WhooshIndexingDaemon
 
        try:
 
            l = DaemonLock(file_=jn(dn(dn(index_location)), 'make_index.lock'))
 
            WhooshIndexingDaemon(index_location=index_location,
 
                                 repo_location=repo_location,
 
                                 repo_list=repo_list,)\
 
                                 repo_list=repo_list,
 
                                 repo_update_list=repo_update_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 OPTIONAL",
 
                          )
 
        self.parser.add_option('--index-only',
 
                          action='store',
 
                          dest='repo_list',
 
                          help="Specifies a comma separated list of repositores "
 
                                "to build index on OPTIONAL",
 
                                "to build index on. If not given all repositories "
 
                                "are scanned for indexing. OPTIONAL",
 
                          )
 
        self.parser.add_option('--update-only',
 
                          action='store',
 
                          dest='repo_update_list',
 
                          help="Specifies a comma separated list of repositores "
 
                                "to re-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)
 

	
 

	
 
class WhooshResultWrapper(object):
 
    def __init__(self, search_type, searcher, matcher, highlight_items,
 
                 repo_location):
rhodecode/lib/indexers/daemon.py
Show inline comments
 
@@ -44,48 +44,59 @@ from rhodecode.lib.utils2 import safe_un
 
from rhodecode.lib.indexers import SCHEMA, IDX_NAME
 

	
 
from rhodecode.lib.vcs.exceptions import ChangesetError, RepositoryError, \
 
    NodeDoesNotExistError
 

	
 
from whoosh.index import create_in, open_dir
 

	
 
log = logging.getLogger('whoosh_indexer')
 

	
 

	
 
class WhooshIndexingDaemon(object):
 
    """
 
    Daemon for atomic jobs
 
    Daemon for atomic indexing jobs
 
    """
 

	
 
    def __init__(self, indexname=IDX_NAME, index_location=None,
 
                 repo_location=None, sa=None, repo_list=None):
 
                 repo_location=None, sa=None, repo_list=None,
 
                 repo_update_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)
 

	
 
        #filter repo list
 
        if repo_list:
 
            filtered_repo_paths = {}
 
            self.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.filtered_repo_paths[repo_name] = repo
 

	
 
            self.repo_paths = self.filtered_repo_paths
 

	
 
            self.repo_paths = filtered_repo_paths
 
        #filter update repo list
 
        self.filtered_repo_update_paths = {}
 
        if repo_update_list:
 
            self.filtered_repo_update_paths = {}
 
            for repo_name, repo in self.repo_paths.items():
 
                if repo_name in repo_update_list:
 
                    self.filtered_repo_update_paths[repo_name] = repo
 
            self.repo_paths = self.filtered_repo_update_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
 

	
 
    def get_paths(self, repo):
 
        """
 
        recursive walk in root dir and return a set of all path in that dir
 
        based on repository walk function
 
@@ -163,60 +174,62 @@ class WhooshIndexingDaemon(object):
 
                i, iwc = self.add_doc(writer, idx_path, repo, repo_name)
 
                i_cnt += i
 
                iwc_cnt += iwc
 
            log.debug('added %s files %s with content for repo %s' % (
 
                         i_cnt + iwc_cnt, iwc_cnt, repo.path)
 
            )
 

	
 
        log.debug('>> COMMITING CHANGES <<')
 
        writer.commit(merge=True)
 
        log.debug('>>> FINISHED BUILDING INDEX <<<')
 

	
 
    def update_index(self):
 
        log.debug(('STARTING INCREMENTAL INDEXING UPDATE FOR EXTENSIONS %s '
 
                   'AND REPOS %s') % (INDEX_EXTENSIONS, self.repo_paths))
 
        log.debug((u'STARTING INCREMENTAL INDEXING UPDATE FOR EXTENSIONS %s '
 
                   'AND REPOS %s') % (INDEX_EXTENSIONS, self.repo_paths.keys()))
 

	
 
        idx = open_dir(self.index_location, indexname=self.indexname)
 
        # The set of all paths in the index
 
        indexed_paths = set()
 
        # The set of all paths we need to re-index
 
        to_index = set()
 

	
 
        reader = idx.reader()
 
        writer = idx.writer()
 

	
 
        # Loop over the stored fields in the index
 
        for fields in reader.all_stored_fields():
 
            indexed_path = fields['path']
 
            indexed_repo_path = fields['repository']
 
            indexed_paths.add(indexed_path)
 

	
 
            repo = self.repo_paths[fields['repository']]
 
            if not indexed_repo_path in self.filtered_repo_update_paths:
 
                continue
 

	
 
            repo = self.repo_paths[indexed_repo_path]
 

	
 
            try:
 
                node = self.get_node(repo, indexed_path)
 
            except (ChangesetError, NodeDoesNotExistError):
 
                # This file was deleted since it was indexed
 
                log.debug('removing from index %s' % indexed_path)
 
                writer.delete_by_term('path', indexed_path)
 

	
 
            else:
 
                # Check if this file was changed since it was indexed
 
                indexed_time = fields['modtime']
 
                mtime = self.get_node_mtime(node)
 
                if mtime > indexed_time:
 
                    # The file has changed, delete it and add it to the list of
 
                    # files to reindex
 
                    log.debug('adding to reindex list %s' % indexed_path)
 
                    writer.delete_by_term('path', indexed_path)
 
                    to_index.add(indexed_path)
 
            except (ChangesetError, NodeDoesNotExistError):
 
                # This file was deleted since it was indexed
 
                log.debug('removing from index %s' % indexed_path)
 
                writer.delete_by_term('path', indexed_path)
 

	
 
        # Loop over the files in the filesystem
 
        # Assume we have a function that gathers the filenames of the
 
        # documents to be indexed
 
        ri_cnt = riwc_cnt = 0
 
        for repo_name, repo in self.repo_paths.items():
 
            for path in self.get_paths(repo):
 
                path = safe_unicode(path)
 
                if path in to_index or path not in indexed_paths:
 
                    # This is either a file that's changed, or a new file
 
                    # that wasn't indexed before. So index it!
 
                    i, iwc = self.add_doc(writer, path, repo, repo_name)
0 comments (0 inline, 0 general)