def __init__(self, name, type, Model): """ Input : - name : 'ooshop', 'monoprix' ... - Model : database entity model for exemple : MonoprixProduct """ self.name = name self.type = type self.Model = Model self.dir_path = IndexController.BASE_PATH + '/' +self.type+'/'+self.name self.service = SessionServer(self.dir_path)
class IndexController(object):
"""
Index controller, this method can:
- build complete index from database
- perform a query against index
"""
BASE_PATH = '/tmp/dalliz/matcher'
METHOD = 'tfidf'
MAX_RESULTS = 20
def __init__(self, name, type, Model):
"""
Input :
- name : 'ooshop', 'monoprix' ...
- Model : database entity model for exemple : MonoprixProduct
"""
self.name = name
self.type = type
self.Model = Model
self.dir_path = IndexController.BASE_PATH + '/' +self.type+'/'+self.name
self.service = SessionServer(self.dir_path)
def build_all_index(self):
"""
Building the index from scratch.
"""
reg = re.compile(r'\b[a-zA-Z]{3,}\b')
documents = self.Model.objects.all()
documents = [{'id': d.id, 'tokens': reg.findall(unaccent(d.name.lower()))} for d in documents]
self.service.train(documents, method=IndexController.METHOD)
self.service.index(documents)
def query(self, document ):
"""
Querying index with a document
Input :
-document : {'tokens': [list of words]}
"""
return self.service.find_similar(document, max_results = IndexController.MAX_RESULTS)
def get_documents(self, datetime = None):
"""
This method returns documents filtered by updated time lesser thant the
provided datetime.
"""
reg = re.compile(r'\b[a-zA-Z]{3,}\b')
documents = self.Model.objects.all()
if datetime is not None:
documents = documents.filter(created__gte=datetime)
return [{'id': d.id, 'tokens': reg.findall(unaccent(d.name.lower()))} for d in documents]
def add_documents(self, documents):
"""
Adding documents to the index, if document id already exists in index, it will override it
Input :
- documents = [{'id':,'tokens':..}]
"""
try:
self.service.index(documents)
except AttributeError, e:
self.service.train(documents, method=IndexController.METHOD)
