modelverse/state/modelverse_state/rdf.py

from modelverse_state import status
import sys
from collections import defaultdict
import os
import rdflib
import json

import cPickle as pickle

# Work around Python 2 where a 'big integer' automatically becomes a long
if sys.version > '3': # pragma: no cover
    integer_types = (int,)
    primitive_types = (int, float, str, bool)
else: # pragma: no cover
    integer_types = (int, long)
    primitive_types = (int, long, float, str, bool, unicode)
complex_primitives = frozenset(["if", "while", "assign", "call", "break", "continue", "return","resolve","access", "constant", "input", "output", "declare", "global"])

def instance_to_string(value):
    return value["value"]

def string_to_instance(value):
    return {'value': value}

class ModelverseState(object):
    def __init__(self, bootfile = None):
        self.graph = rdflib.Graph()
        self.mv = rdflib.Namespace("http://modelverse.mv/#")
        self.graph.bind("MV", self.mv)

        self.parse(bootfile)
        self.root = 0

        self.GC = True
        self.to_delete = set()

    def parse(self, filename):
        triplestore = filename + ".nt"
        try:
            if os.path.getmtime(triplestore) > os.path.getmtime(filename):
                self.graph.parse(triplestore, format="nt")
            else:
                raise Exception("Invalid triplestore")
        except Exception as e:
            # We have to parse the file and create the pickle
            symbols = {}

            def resolve(symb):
                try:
                    return int(symb)
                except:
                    if symb[0] == "?":
                        derefs = symb[1:].split("/")
                        v, _ = self.read_dict(symbols["root"], "__hierarchy")
                        for deref in derefs:
                            v, _ = self.read_dict(v, deref)
                        return v
                    else:
                        return symbols[symb]

            with open(filename, 'r') as f:
                for line in f:
                    element_type, constructor = line.split(None, 1)
                    name, values = constructor.split("(", 1)
                    name = name.split()[0]
                    values, _ = values.rsplit(")", 1)

                    if element_type == "Node":
                        if values == "":
                            symbols[name], status = self.create_node()
                        else:
                            value = values
                            if value in complex_primitives:
                                value = string_to_instance(value)
                            else:
                                value = eval(value)
                            symbols[name], status = self.create_nodevalue(value)
                    elif element_type == "Edge":
                        values = [v.split()[0] for v in values.split(",")]
                        symbols[name], status = self.create_edge(resolve(values[0]), resolve(values[1]))
                    else:
                        raise Exception("Unknown element type: %s" % element_type)

                    if status != 100:
                        raise Exception("Failed to process line for reason %s: %s" % (status, line))

            # Creation successful, now also create a pickle
            self.graph.serialize(triplestore, format="nt")
            #TODO this loses information about the root!
            return symbols["root"]

    def read_root(self):
        return (self.root, status.SUCCESS)

    def create_node(self):
        return (rdflib.BNode(), status.SUCCESS)

    def create_edge(self, source, target):
        if not isinstance(source, rdflib.BNode):
            return (None, status.FAIL_CE_SOURCE)
        if not isinstance(target, rdflib.BNode):
            return (None, status.FAIL_CE_TARGET)
        edge = rdflib.BNode()
        self.graph.add((edge, self.mv.hasSource, source))
        self.graph.add((edge, self.mv.hasTarget, target))
        return (edge, status.SUCCESS)
        
    def is_valid_datavalue(self, value):
        if isinstance(value, dict):
            if "value" in value and value["value"] in complex_primitives:
                return True
            else:
                return False
        elif not isinstance(value, primitive_types):
            return False
        elif isinstance(value, integer_types) and not (-2**63 <= value <= 2**64 - 1):
            return False
        return True

    def create_nodevalue(self, value):
        if not self.is_valid_datavalue(value):
            return (None, status.FAIL_CNV_OOB)
        node = rdflib.BNode()
        self.graph.add((node, self.mv.hasValue, rdflib.Literal(json.dumps(value))))
        return (node, status.SUCCESS)

    def create_dict(self, source, data, destination):
        if not isinstance(source, rdflib.BNode):
            return (None, status.FAIL_CDICT_SOURCE)
        if not isinstance(target, rdflib.BNode):
            return (None, status.FAIL_CDICT_TARGET)
        if not self.is_valid_datavalue(data):
            return (None, status.FAIL_CDICT_OOB)

        n = self.create_nodevalue(data)[0]
        e = self.create_edge(source, destination)[0]
        t = self.create_edge(e, n)
        return (None, status.SUCCESS)

    def read_value(self, node):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RV_UNKNOWN)
        result = self.graph.query(
            """
            SELECT ?value
            WHERE {
                _:%s MV:hasValue ?value .
            }
            """ % node)
        if len(result) == 0:
            return (None, status.FAIL_RV_NO_VALUE)
        return (result[0], status.SUCCESS)

    def read_outgoing(self, elem):
        if not isinstance(elem, rdflib.BNode):
            return (None, status.FAIL_RO_UNKNOWN)
        result = self.graph.query(
            """
            SELECT ?link
            WHERE {
                _:%s MV:hasTarget ?link .
            }
            """ % elem)
        return (list(result), status.SUCCESS)

    def read_incoming(self, elem):
        if not isinstance(elem, rdflib.BNode):
            return (None, status.FAIL_RI_UNKNOWN)
        result = self.graph.query(
            """
            SELECT ?link
            WHERE {
                _:%s MV:hasSource ?link .
            }
            """ % elem)
        return (list(result), status.SUCCESS)

    def read_edge(self, edge):
        result = self.graph.query(
            """
            SELECT ?source, ?target
            WHERE {
                _:%s MV:hasSource ?source ;
                     MV:hasTarget ?target .
            }
            """ % edge)
        if len(result) == 0:
            return ([None, None], status.FAIL_RE_UNKNOWN)
        else:
            return (list(result), status.SUCCESS)

    def read_dict(self, node, value):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RDICT_UNKNOWN)
        if not self.is_valid_datavalue(value):
            return (None, status.FAIL_RDICT_OOB)

        q = """
            SELECT ?value_node
            WHERE {
                ?main_edge MV:hasSource _:%s ;
                           MV:hasTarget ?value_node .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget ?attr_node .
                ?attr_node MV:hasValue '%s' .
            }
            """ % (node, json.dumps(value))
        print(q)
        result = self.graph.query(q)
        if len(result) == 0:
            return (None, status.FAIL_RDICT_NOT_FOUND)
        return (list(result)[0][0], status.SUCCESS)

    def read_dict_keys(self, node):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RDICT_UNKNOWN)

        result = self.graph.query(
            """
            SELECT ?key
            WHERE {
                ?main_edge MV:hasSource _:%s .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget ?key .
            }
            """)
        return (list(result), status.SUCCESS)

    def read_dict_edge(self, node, value):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RDICTE_UNKNOWN)
        if not self.is_valid_datavalue(value):
            return (None, status.FAIL_RDICTE_OOB)
        result = self.graph.query(
            """
            SELECT ?main_edge
            WHERE {
                ?main_edge MV:hasSource _:%s ;
                           MV:hasTarget ?value_node .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget ?attr_node .
                ?attr_node MV:hasValue '%s' .
            }
            """ % (node, value))
        if len(result) == 0:
            return (None, status.FAIL_RDICTE_NOT_FOUND)
        return (result[0], status.SUCCESS)

    def read_dict_node(self, node, value_node):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RDICTN_UNKNOWN)
        result = self.graph.query(
            """
            SELECT ?value
            WHERE {
                ?main_edge MV:hasSource _:%s ;
                           MV:hasTarget ?value_node .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget _:%s .
                ?value_node MV:hasValue ?value .
            }
            """ % (node, value))
        if len(result) == 0:
            return (None, status.FAIL_RDICTN_NOT_FOUND)
        return (result[0], status.SUCCESS)

    def read_dict_node_edge(self, node, value_node):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RDICTNE_UNKNOWN)
        result = self.graph.query(
            """
            SELECT ?main_edge
            WHERE {
                ?main_edge MV:hasSource _:%s .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget _:%s .
            }
            """ % (node, value))
        if len(result) == 0:
            return (None, status.FAIL_RDICTNE_NOT_FOUND)
        return (result[0], status.SUCCESS)

    def read_reverse_dict(self, node, value):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_RRDICT_UNKNOWN)
        if not self.is_valid_datavalue(value):
            return (None, status.FAIL_RRDICT_OOB)
        result = self.graph.query(
            """
            SELECT ?main_edge
            WHERE {
                ?main_edge MV:hasTarget _:%s .
                ?attr_edge MV:hasSource ?main_edge ;
                           MV:hasTarget ?value_node .
                ?value_node MV:hasValue '%s' .
            }
            """ % (node, value))

        return (list(result), status.SUCCESS)

    def delete_node(self, node):
        if node == self.root:
            return (None, status.FAIL_DN_UNKNOWN)
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_DN_UNKNOWN)
        # Remove its value if it exists
        self.graph.remove((node, None, None))

        # Get all edges connecting this
        result = self.graph.query(
            """
            SELECT ?edge
            WHERE {
                { ?edge MV:hasTarget _:%s . }
                UNION
                { ?edge MV:hasSource _:%s . }
            }
            """)
        # ... and remove them
        for e in result:
            self.delete_edge(e)

        return (None, status.SUCCESS)

    def delete_edge(self, edge):
        if not isinstance(node, rdflib.BNode):
            return (None, status.FAIL_DN_UNKNOWN)
        # Remove its links
        self.graph.remove((node, None, None))

        # Get all edges connecting this
        result = self.graph.query(
            """
            SELECT ?edge
            WHERE {
                { ?edge MV:hasTarget _:%s . }
                UNION
                { ?edge MV:hasSource _:%s . }
            }
            """)
        # ... and remove them
        for e in result:
            self.delete_edge(e)

        return (None, status.SUCCESS)

    def garbage_collect(self):  
        pass

    def purge(self):
        pass