Exemplo n.º 1
0
    def _rewrite_hybrid_join(self, node: saldag.HybridJoin):
        # TODO
        suffix = "rand"

        # in left parents' children, replace self with first primitive operator
        # in expanded subdag
        shuffled_a = sal.shuffle(node.left_parent, "shuffled_a")
        shuffled_a.is_mpc = True
        node.left_parent.children.remove(node)

        # same for right parent
        shuffled_b = sal.shuffle(node.right_parent, "shuffled_b")
        shuffled_b.is_mpc = True
        node.right_parent.children.remove(node)

        persisted_b = sal._persist(shuffled_b, "persisted_b")
        persisted_b.is_mpc = True
        persisted_a = sal._persist(shuffled_a, "persisted_a")
        persisted_a.is_mpc = True

        keys_a_closed = sal.project(shuffled_a, "keys_a_closed", ["a"])
        keys_a_closed.is_mpc = True
        keys_b_closed = sal.project(shuffled_b, "keys_b_closed", ["c"])
        keys_b_closed.is_mpc = True

        keys_a = sal._open(keys_a_closed, "keys_a", 1)
        keys_a.is_mpc = True
        keys_b = sal._open(keys_b_closed, "keys_b", 1)
        keys_b.is_mpc = True

        indexed_a = sal.index(keys_a, "indexed_a", "index_a")
        indexed_a.is_mpc = False

        indexed_b = sal.index(keys_b, "indexed_b", "index_b")
        indexed_b.is_mpc = False

        joined_indices = sal.join(indexed_a, indexed_b, "joined_indices",
                                  ["a"], ["c"])
        joined_indices.is_mpc = False

        indices_only = sal.project(joined_indices, "indices_only",
                                   ["index_a", "index_b"])
        indices_only.is_mpc = False

        # TODO: update stored_with to use union of parent out_rel stored_with sets
        indices_closed = sal._close(indices_only, "indices_closed", set([1,
                                                                         2]))
        indices_closed.is_mpc = True

        joined = sal._index_join(persisted_a, persisted_b, "joined", ["a"],
                                 ["c"], indices_closed)
        joined.is_mpc = True

        # replace self with leaf of expanded subdag in each child node
        for child in node.get_sorted_children():
            child.replace_parent(node, joined)
        # add former children to children of leaf
        joined.children = node.children
Exemplo n.º 2
0
    def hybrid_agg(in1):

        shuffled = sal.shuffle(in1, "shuffled")
        shuffled.out_rel.storedWith = set([1, 2, 3])
        shuffled.isMPC = True

        persisted = sal._persist(shuffled, "persisted")
        persisted.out_rel.storedWith = set([1, 2, 3])
        persisted.isMPC = True

        keysclosed = sal.project(shuffled, "keysclosed", ["b"])
        keysclosed.out_rel.storedWith = set([1, 2, 3])
        keysclosed.isMPC = True

        keys = sal._open(keysclosed, "keys", 1)
        keys.isMPC = True

        indexed = sal.index(keys, "indexed", "rowIndex")
        indexed.isMPC = False
        indexed.out_rel.storedWith = set([1])

        distinctKeys = sal.distinct(keys, "distinctKeys", ["b"])
        distinctKeys.isMPC = False
        distinctKeys.out_rel.storedWith = set([1])

        # TODO: hack to get keys stored
        # need to fix later!
        fakeDistinctKeys = sal.project(distinctKeys, "distinctKeys", ["b"])
        fakeDistinctKeys.isMPC = False
        fakeDistinctKeys.out_rel.storedWith = set([1])

        indexedDistinct = sal.index(distinctKeys, "indexedDistinct",
                                    "keyIndex")
        indexedDistinct.isMPC = False
        indexedDistinct.out_rel.storedWith = set([1])

        joinedindeces = sal.join(indexed, indexedDistinct, "joinedindeces",
                                 ["b"], ["b"])
        joinedindeces.isMPC = False
        joinedindeces.out_rel.storedWith = set([1])

        # TODO: could project row indeces away too
        indecesonly = sal.project(joinedindeces, "indecesonly",
                                  ["rowIndex", "keyIndex"])
        indecesonly.isMPC = False
        indecesonly.out_rel.storedWith = set([1])

        closedDistinct = sal._close(distinctKeys, "closedDistinct",
                                    set([1, 2, 3]))
        closedDistinct.isMPC = True
        closedLookup = sal._close(indecesonly, "closedLookup", set([1, 2, 3]))
        closedLookup.isMPC = True

        agg = sal.index_aggregate(persisted, "agg", ["b"], "d", "+", "d",
                                  closedLookup, closedDistinct)
        agg.isMPC = True
        sal._open(agg, "aggopened", 1)
Exemplo n.º 3
0
        def protocol():
            inpts = setup()
            in_1 = inpts[0]

            ind = sal.index(in_1, "index_1", "index")
            out = sal.collect(ind, 1)

            return set([in_1])
Exemplo n.º 4
0
        def protocol():
            inpts = setup()
            in_1 = inpts[0]

            ind = sal.index(in_1, "ind")
            out = sal.collect(ind, 1)

            return {in_1}
Exemplo n.º 5
0
        def protocol():
            inpts = setup()
            in_1 = inpts[0]

            ind = cc.index(in_1, "index_1", "index")
            out = cc.collect(ind, 1)

            return {in_1}
Exemplo n.º 6
0
    def protocol():

        inpts = setup()
        in_1 = inpts

        indexed = sal.index(in_1, 'index_1', 'index')
        out = sal.collect(indexed, 1)

        return set([in_1])
Exemplo n.º 7
0
        def hybrid_agg(in1):

            shuffled = sal.shuffle(in1, "shuffled")
            shuffled.out_rel.stored_with = set([1, 2, 3])
            shuffled.is_mpc = True

            persisted = sal._persist(shuffled, "persisted")
            persisted.out_rel.stored_with = set([1, 2, 3])
            persisted.is_mpc = True

            keys_closed = sal.project(shuffled, "keys_closed", ["b"])
            keys_closed.out_rel.stored_with = set([1, 2, 3])
            keys_closed.is_mpc = True

            keys = sal._open(keys_closed, "keys", 1)
            keys.is_mpc = True

            indexed = sal.index(keys, "indexed", "rowIndex")
            indexed.is_mpc = False
            indexed.out_rel.stored_with = set([1])

            sorted_by_key = sal.sort_by(indexed, "sorted_by_key", "b")
            sorted_by_key.is_mpc = False
            sorted_by_key.out_rel.stored_with = set([1])

            eq_flags = sal._comp_neighs(sorted_by_key, "eq_flags", "b")
            eq_flags.is_mpc = False
            eq_flags.out_rel.stored_with = set([1])

            # TODO: should be a persist op
            sorted_by_key_stored = sal.project(
                sorted_by_key, "sorted_by_key_stored", ["rowIndex", "b"])
            sorted_by_key_stored.is_mpc = False
            sorted_by_key_stored.out_rel.stored_with = set([1])

            closed_eq_flags = sal._close(eq_flags, "closed_eq_flags", set([1, 2, 3]))
            closed_eq_flags.is_mpc = True
            closed_sorted_by_key = sal._close(
                sorted_by_key_stored, "closed_sorted_by_key", set([1, 2, 3]))
            closed_sorted_by_key.is_mpc = True

            agg = sal.index_aggregate(
                persisted, "agg", ["b"], "d", "+", "d", closed_eq_flags, closed_sorted_by_key)
            agg.is_mpc = True
            sal._open(agg, "ssnopened", 1)
Exemplo n.º 8
0
    def hybrid_agg(in1):
        shuffled = sal.shuffle(in1, "shuffled")
        shuffled.out_rel.storedWith = {1, 2, 3}
        shuffled.isMPC = True

        persisted = sal._persist(shuffled, "persisted")
        persisted.out_rel.storedWith = {1, 2, 3}
        persisted.isMPC = True

        keysclosed = sal.project(shuffled, "keysclosed", ["b"])
        keysclosed.out_rel.storedWith = {1, 2, 3}
        keysclosed.isMPC = True

        keys = sal._open(keysclosed, "keys", 1)
        keys.isMPC = True

        indexed = sal.index(keys, "indexed", "rowIndex")
        indexed.isMPC = False
        indexed.out_rel.storedWith = {1}

        sortedByKey = sal.sort_by(indexed, "sortedByKey", "b")
        sortedByKey.isMPC = False
        sortedByKey.out_rel.storedWith = {1}

        eqFlags = sal._comp_neighs(sortedByKey, "eqFlags", "b")
        eqFlags.isMPC = False
        eqFlags.out_rel.storedWith = {1}

        # TODO: should be a persist op
        sortedByKeyStored = sal.project(sortedByKey, "sortedByKeyStored",
                                        ["rowIndex", "b"])
        sortedByKeyStored.isMPC = False
        sortedByKeyStored.out_rel.storedWith = {1}

        closedEqFlags = sal._close(eqFlags, "closedEqFlags", {1, 2, 3})
        closedEqFlags.isMPC = True
        closedSortedByKey = sal._close(sortedByKeyStored, "closedSortedByKey",
                                       {1, 2, 3})
        closedSortedByKey.isMPC = True

        agg = sal.index_aggregate(persisted, "agg", ["b"], "d", "+", "d",
                                  closedEqFlags, closedSortedByKey)
        agg.isMPC = True
        sal._open(agg, "ssnopened", 1)
Exemplo n.º 9
0
    def protocol():

        # define inputs
        colsInA = [
            defCol("a", "INTEGER", [1]),
            defCol("b", "INTEGER", [1]),
        ]
        in1 = sal.create("in1", colsInA, set([1]))
        in1.isMPC = False

        proja = sal.project(in1, "proja", ["a", "b"])
        proja.isMPC = False
        proja.out_rel.storedWith = set([1])

        # define inputs
        colsInB = [
            defCol("a", "INTEGER", [2]),
            defCol("b", "INTEGER", [2]),
        ]
        in2 = sal.create("in2", colsInB, set([2]))
        in2.isMPC = False

        projb = sal.project(in2, "projb", ["a", "b"])
        projb.isMPC = False
        projb.out_rel.storedWith = set([2])

        # define inputs
        colsInC = [
            defCol("a", "INTEGER", [3]),
            defCol("b", "INTEGER", [3]),
        ]
        in3 = sal.create("in3", colsInC, set([3]))
        in3.isMPC = False

        projc = sal.project(in3, "projc", ["a", "b"])
        projc.isMPC = False
        projc.out_rel.storedWith = set([3])

        clA = sal._close(proja, "clA", set([1, 2, 3]))
        clA.isMPC = True

        clB = sal._close(projb, "clB", set([1, 2, 3]))
        clB.isMPC = True

        clC = sal._close(projc, "clC", set([1, 2, 3]))
        clC.isMPC = True

        comb = sal.concat([clA, clB, clC], "comb")
        comb.out_rel.storedWith = set([1, 2, 3])
        comb.isMPC = True

        shuffled = sal.shuffle(comb, "shuffled")
        shuffled.out_rel.storedWith = set([1, 2, 3])
        shuffled.isMPC = True

        persisted = sal._persist(shuffled, "persisted")
        persisted.out_rel.storedWith = set([1, 2, 3])
        persisted.isMPC = True

        keysclosed = sal.project(shuffled, "keysclosed", ["a"])
        keysclosed.out_rel.storedWith = set([1, 2, 3])
        keysclosed.isMPC = True

        keys = sal._open(keysclosed, "keys", 1)
        keys.isMPC = True

        indexed = sal.index(keys, "indexed", "rowIndex")
        indexed.isMPC = False
        indexed.out_rel.storedWith = set([1])

        sortedByKey = sal.sort_by(indexed, "sortedByKey", "a")
        sortedByKey.isMPC = False
        sortedByKey.out_rel.storedWith = set([1])

        eqFlags = sal._comp_neighs(sortedByKey, "eqFlags", "a")
        eqFlags.isMPC = False
        eqFlags.out_rel.storedWith = set([1])

        # TODO: hack to get keys stored
        # need to fix later!
        sortedByKey = sal.project(sortedByKey, "sortedByKey",
                                  ["rowIndex", "a"])
        sortedByKey.isMPC = False
        sortedByKey.out_rel.storedWith = set([1])

        closedEqFlags = sal._close(eqFlags, "closedEqFlags", set([1, 2, 3]))
        closedEqFlags.isMPC = True
        closedSortedByKey = sal._close(sortedByKey, "closedSortedByKey",
                                       set([1, 2, 3]))
        closedSortedByKey.isMPC = True

        agg = sal.index_aggregate(persisted, "agg", ["a"], "b", "+", "b",
                                  closedEqFlags, closedSortedByKey)
        agg.out_rel.storedWith = set([1, 2, 3])
        agg.isMPC = True

        sal._open(agg, "opened", 1)

        # create condag
        return set([in1, in2, in3])
Exemplo n.º 10
0
    def protocol():

        # define inputs
        colsInA = [
            defCol("a", "INTEGER", [1]),
            defCol("b", "INTEGER", [1]),
        ]
        in1 = sal.create("in1", colsInA, set([1]))
        in1.isMPC = False

        proja = sal.project(in1, "proja", ["a", "b"])
        proja.isMPC = False
        proja.out_rel.storedWith = set([1])

        colsInB = [
            defCol("c", "INTEGER", [1], [2]),
            defCol("d", "INTEGER", [2])
        ]
        in2 = sal.create("in2", colsInB, set([2]))
        in2.isMPC = False

        projb = sal.project(in2, "projb", ["c", "d"])
        projb.isMPC = False
        projb.out_rel.storedWith = set([2])

        clA = sal._close(proja, "clA", set([1, 2, 3]))
        clA.isMPC = True
        clB = sal._close(projb, "clB", set([1, 2, 3]))
        clB.isMPC = True

        persistedA = sal._persist(clA, "persistedA")
        persistedB = sal._persist(clB, "persistedB")

        keysaclosed = sal.project(clA, "keysaclosed", ["a"])
        keysaclosed.out_rel.storedWith = set([1, 2, 3])
        keysaclosed.isMPC = True
        keysbclosed = sal.project(clB, "keysbclosed", ["c"])
        keysbclosed.isMPC = True
        keysbclosed.out_rel.storedWith = set([1, 2, 3])

        keysa = sal._open(keysaclosed, "keysa", 1)
        keysa.isMPC = True
        keysb = sal._open(keysbclosed, "keysb", 1)
        keysb.isMPC = True

        indexedA = sal.index(keysa, "indexedA", "indexA")
        indexedA.isMPC = False
        indexedA.out_rel.storedWith = set([1])
        indexedB = sal.index(keysb, "indexedB", "indexB")
        indexedB.isMPC = False
        indexedB.out_rel.storedWith = set([1])

        joinedindeces = sal.join(
            indexedA, indexedB, "joinedindeces", ["a"], ["c"])
        joinedindeces.isMPC = False
        joinedindeces.out_rel.storedWith = set([1])

        indecesonly = sal.project(
            joinedindeces, "indecesonly", ["indexA", "indexB"])
        indecesonly.isMPC = False
        indecesonly.out_rel.storedWith = set([1])

        indecesclosed = sal._close(
            indecesonly, "indecesclosed", set([1, 2, 3]))
        indecesclosed.isMPC = True

        joined = sal._index_join(persistedA, persistedB, "joined",
                                 ["a"], ["c"], indecesclosed)
        joined.isMPC = True

        sal._open(joined, "opened", 1)

        # create condag
        return set([in1, in2])
Exemplo n.º 11
0
    def hybrid_join():
        # define inputs
        colsInA = [
            defCol("a", "INTEGER", [1]),
            defCol("b", "INTEGER", [1]),
        ]
        in1 = sal.create("govreg", colsInA, {1})
        in1.isMPC = False

        proja = sal.project(in1, "proja", ["a", "b"])
        proja.isMPC = False
        proja.out_rel.storedWith = {1}

        colsInB = [
            defCol("c", "INTEGER", [1], [2]),
            defCol("d", "INTEGER", [2])
        ]
        in2 = sal.create("company0", colsInB, {2})
        in2.isMPC = False

        projb = sal.project(in2, "projb", ["c", "d"])
        projb.isMPC = False
        projb.out_rel.storedWith = {2}

        colsInC = [
            defCol("c", "INTEGER", [1], [3]),
            defCol("d", "INTEGER", [3])
        ]
        in3 = sal.create("company1", colsInC, {3})
        in3.isMPC = False

        projc = sal.project(in3, "projc", ["c", "d"])
        projc.isMPC = False
        projc.out_rel.storedWith = {3}

        clA = sal._close(proja, "clA", {1, 2, 3})
        clA.isMPC = True
        clB = sal._close(projb, "clB", {1, 2, 3})
        clB.isMPC = True
        clC = sal._close(projc, "clC", {1, 2, 3})
        clC.isMPC = True

        rightClosed = sal.concat([clB, clC], "clD")
        rightClosed.isMPC = True
        rightClosed.out_rel.storedWith = {1, 2, 3}

        shuffledA = sal.shuffle(clA, "shuffledA")
        shuffledA.isMPC = True
        persistedA = sal._persist(shuffledA, "persistedA")
        persistedA.isMPC = True
        shuffledB = sal.shuffle(rightClosed, "shuffledB")
        shuffledB.isMPC = True
        persistedB = sal._persist(shuffledB, "persistedB")
        persistedB.isMPC = True

        keysaclosed = sal.project(shuffledA, "keysaclosed", ["a"])
        keysaclosed.out_rel.storedWith = {1, 2, 3}
        keysaclosed.isMPC = True
        keysbclosed = sal.project(shuffledB, "keysbclosed", ["c"])
        keysbclosed.isMPC = True
        keysbclosed.out_rel.storedWith = {1, 2, 3}

        keysa = sal._open(keysaclosed, "keysa", 1)
        keysa.isMPC = True
        keysb = sal._open(keysbclosed, "keysb", 1)
        keysb.isMPC = True

        indexedA = sal.index(keysa, "indexedA", "indexA")
        indexedA.isMPC = False
        indexedA.out_rel.storedWith = {1}
        indexedB = sal.index(keysb, "indexedB", "indexB")
        indexedB.isMPC = False
        indexedB.out_rel.storedWith = {1}

        joinedindeces = sal.join(indexedA, indexedB, "joinedindeces", ["a"],
                                 ["c"])
        joinedindeces.isMPC = False
        joinedindeces.out_rel.storedWith = {1}

        indecesonly = sal.project(joinedindeces, "indecesonly",
                                  ["indexA", "indexB"])
        indecesonly.isMPC = False
        indecesonly.out_rel.storedWith = {1}

        indecesclosed = sal._close(indecesonly, "indecesclosed", {1, 2, 3})
        indecesclosed.isMPC = True

        joined = sal._index_join(persistedA, persistedB, "joined", ["a"],
                                 ["c"], indecesclosed)
        joined.isMPC = True

        return joined, {in1, in2, in3}
Exemplo n.º 12
0
        def hybrid_join():

            # define inputs
            cols_in_a = [
                defCol("a", "INTEGER", [1]),
                defCol("b", "INTEGER", [1]),
            ]
            in_1 = sal.create("govreg", cols_in_a, set([1]))
            in_1.is_mpc = False

            proj_a = sal.project(in_1, "proj_a", ["a", "b"])
            proj_a.is_mpc = False
            proj_a.out_rel.stored_with = set([1])

            cols_in_b = [
                defCol("c", "INTEGER", [1], [2]),
                defCol("d", "INTEGER", [2])
            ]
            in_2 = sal.create("company0", cols_in_b, set([2]))
            in_2.is_mpc = False

            proj_b = sal.project(in_2, "proj_b", ["c", "d"])
            proj_b.is_mpc = False
            proj_b.out_rel.stored_with = set([2])

            cols_in_c = [
                defCol("c", "INTEGER", [1], [3]),
                defCol("d", "INTEGER", [3])
            ]
            in_3 = sal.create("company1", cols_in_c, set([3]))
            in_3.is_mpc = False

            proj_c = sal.project(in_3, "proj_c", ["c", "d"])
            proj_c.is_mpc = False
            proj_c.out_rel.stored_with = set([3])

            cl_a = sal._close(proj_a, "cl_a", set([1, 2, 3]))
            cl_a.is_mpc = True
            cl_b = sal._close(proj_b, "cl_b", set([1, 2, 3]))
            cl_b.is_mpc = True
            cl_c = sal._close(proj_c, "cl_c", set([1, 2, 3]))
            cl_c.is_mpc = True

            right_closed = sal.concat([cl_b, cl_c], "clD")
            right_closed.is_mpc = True
            right_closed.out_rel.stored_with = set([1, 2, 3])

            shuffled_a = sal.shuffle(cl_a, "shuffled_a")
            shuffled_a.is_mpc = True
            persisted_a = sal._persist(shuffled_a, "persisted_a")
            persisted_a.is_mpc = True
            shuffled_b = sal.shuffle(right_closed, "shuffled_b")
            shuffled_b.is_mpc = True
            persisted_b = sal._persist(shuffled_b, "persisted_b")
            persisted_b.is_mpc = True

            keys_a_closed = sal.project(shuffled_a, "keys_a_closed", ["a"])
            keys_a_closed.out_rel.stored_with = set([1, 2, 3])
            keys_a_closed.is_mpc = True
            keys_b_closed = sal.project(shuffled_b, "keys_b_closed", ["c"])
            keys_b_closed.is_mpc = True
            keys_b_closed.out_rel.stored_with = set([1, 2, 3])

            keys_a = sal._open(keys_a_closed, "keys_a", 1)
            keys_a.is_mpc = True
            keys_b = sal._open(keys_b_closed, "keys_b", 1)
            keys_b.is_mpc = True

            indexed_a = sal.index(keys_a, "indexed_a", "index_a")
            indexed_a.is_mpc = False
            indexed_a.out_rel.stored_with = set([1])
            indexed_b = sal.index(keys_b, "indexed_b", "index_b")
            indexed_b.is_mpc = False
            indexed_b.out_rel.stored_with = set([1])

            joined_indeces = sal.join(
                indexed_a, indexed_b, "joined_indeces", ["a"], ["c"])
            joined_indeces.is_mpc = False
            joined_indeces.out_rel.stored_with = set([1])

            indeces_only = sal.project(
                joined_indeces, "indeces_only", ["index_a", "index_b"])
            indeces_only.is_mpc = False
            indeces_only.out_rel.stored_with = set([1])

            indeces_closed = sal._close(
                indeces_only, "indeces_closed", set([1, 2, 3]))
            indeces_closed.is_mpc = True

            joined = sal._index_join(persisted_a, persisted_b, "joined", [
                "a"], ["c"], indeces_closed)
            joined.is_mpc = True

            return joined, set([in_1, in_2, in_3])