Changeset - 0736230c7d91
[Not reviewed]
beta
0 1 0
Marcin Kuzminski - 15 years ago 2010-12-29 01:18:13
marcin@python-works.com
#92 removed content of binary files for whoosh indexer
1 file changed with 2 insertions and 2 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/indexers/daemon.py
Show inline comments
 
@@ -75,98 +75,98 @@ class WhooshIndexingDaemon(object):
 

	
 
        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)
 
        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
 
        """
 
        index_paths_ = set()
 
        try:
 
            for topnode, dirs, files in repo.walk('/', 'tip'):
 
                for f in files:
 
                    index_paths_.add(jn(repo.path, f.path))
 
                for dir in dirs:
 
                    for f in files:
 
                        index_paths_.add(jn(repo.path, f.path))
 

	
 
        except RepositoryError, e:
 
            log.debug(traceback.format_exc())
 
            pass
 
        return index_paths_
 

	
 
    def get_node(self, repo, path):
 
        n_path = path[len(repo.path) + 1:]
 
        node = repo.get_changeset().get_node(n_path)
 
        return node
 

	
 
    def get_node_mtime(self, node):
 
        return mktime(node.last_changeset.date.timetuple())
 

	
 
    def add_doc(self, writer, path, repo):
 
        """Adding doc to writer this function itself fetches data from
 
        the instance of vcs backend"""
 
        node = self.get_node(repo, path)
 

	
 
        #we just index the content of chosen files
 
        if node.extension in INDEX_EXTENSIONS:
 
        #we just index the content of chosen files, and skip binary files
 
        if node.extension in INDEX_EXTENSIONS and not node.is_binary:
 

	
 
            u_content = node.content
 
            if not isinstance(u_content, unicode):
 
                log.warning('  >> %s Could not get this content as unicode '
 
                          'replacing with empty content', path)
 
                u_content = u''
 
            else:
 
                log.debug('    >> %s [WITH CONTENT]' % path)
 

	
 
        else:
 
            log.debug('    >> %s' % path)
 
            #just index file name without it's content
 
            u_content = u''
 

	
 
        writer.add_document(owner=unicode(repo.contact),
 
                        repository=safe_unicode(repo.name),
 
                        path=safe_unicode(path),
 
                        content=u_content,
 
                        modtime=self.get_node_mtime(node),
 
                        extension=node.extension)
 

	
 

	
 
    def build_index(self):
 
        if os.path.exists(self.index_location):
 
            log.debug('removing previous index')
 
            rmtree(self.index_location)
 

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

	
 

	
 
    def update_index(self):
 
        log.debug('STARTING INCREMENTAL INDEXING UPDATE')
 

	
 
        idx = open_dir(self.index_location, indexname=self.indexname)
0 comments (0 inline, 0 general)