schema_packer.py

#-*- coding:utf-8 -*-

import datetime

from collections import defaultdict

XML_NAMESPACE = 'http://www.w3.org/XML/1998/namespace'


class SchemaPacker():
    
    def __init__(self):
        # packed_subentry_id -> packed_schema_id -> (Subentry, Schema) (any that matches)
        self.unique_schemas = defaultdict(lambda: defaultdict(lambda: None))
        # packed_subentry -> id
        self.free_subentry_id = 0
        self.unique_subentry_id = defaultdict(lambda: -1)
        # packed_schema -> id
        self.free_schema_id = 0
        self.unique_schema_id = defaultdict(lambda: -1)
        # subentry -> schema -> (packed_subentry, packed_schema)
        self.packed_schemas = defaultdict(lambda: defaultdict(lambda: None))


    def _pack_subentry(self, subentry):

        sie = subentry.inherent_sie.name
        aspect = ''
        if subentry.aspect is not None:
            aspect = subentry.aspect.name
        negativity = ''
        if subentry.aspect is not None:
            negativity = subentry.negativity.name
        predicativity = subentry.predicativity.name
        return (sie, aspect, negativity, predicativity)
            
        
    def _pack_schema(self, schema):

        opinion = 'unk'
        if schema.opinion.key is not None:
            opinion = schema.opinion.key
        positions = []
        for position in schema.positions.all():
            positions.append(position.id)
        positions = tuple(sorted(positions))
        return (opinion, positions)
        
        
    def _pack(self, subentry, schema):

        if self.packed_schemas[subentry][schema] is None:
            
            self.packed_schemas[subentry][schema] = (self._pack_subentry(subentry), self._pack_schema(schema))
        
        return self.packed_schemas[subentry][schema]
    
    
    def add(self, subentry, schema):
        
        packed_subentry, packed_schema = self._pack(subentry, schema)

        if self.unique_subentry_id[packed_subentry] == -1:
            self.unique_subentry_id[packed_subentry] = self.free_subentry_id
            self.free_subentry_id += 1
        packed_subentry_id = self.unique_subentry_id[packed_subentry]
        if self.unique_schema_id[packed_schema] == -1:
            self.unique_schema_id[packed_schema] = self.free_schema_id
            self.free_schema_id += 1
        packed_schema_id = self.unique_schema_id[packed_schema]
        self.unique_schemas[packed_subentry_id][packed_schema_id] = (subentry, schema)
        self.packed_schemas[subentry][schema] = (packed_subentry, packed_schema)

        
    def get_ids(self, subentry, schema):
        
        packed_subentry, packed_schema = self.packed_schemas[subentry][schema]
        
        return (self.unique_subentry_id[packed_subentry], self.unique_schema_id[packed_schema])