def group(self, attrIdx, count=False, aggrAttrIdx=[], aggrFunc=[]):
#TODO: remove attributes of intermediary tables in attribute store
attrIdx,aggrAttrIdx,aggrFunc = util.makelist(attrIdx,aggrAttrIdx,aggrFunc)
assert len(aggrAttrIdx) == len(aggrFunc)
tmptable,idxmap = self.copy()
aggrAttrIdx = util.mapIdx(aggrAttrIdx,idxmap)
if count:
cntIdx = tmptable.addAttr(gsql.WEIGHT_ATTR_NAME,val=Value(val=1))
aggrAttrIdx.append(cntIdx)
aggrFunc.append('cnt')
# Find values for aggregation
agg = aggregator.Aggregator(aggrFunc)
aggCols = [tmptable.getColumn(idx) for idx in aggrAttrIdx]
# Find groups of rows, and corresponding list of aggregation attributes
tproj,_ = tmptable.project(attrIdx)
groups = {}
for i,row in enumerate(tproj.data):
key = tuple(row)
if not key in groups:
groups[key] = []
groups[key].append([col[i] for col in aggCols])
# groups[key] is a list of lists: each inner list is the list of
# aggregation values corresponding to this row
# Create final table
tfinal,_ = tmptable.project(attrIdx+aggrAttrIdx)
for key in groups:
aggvals = agg.calc(groups[key])
newrow = list(key) + aggvals
tfinal.data.append(newrow)
idxmap = dict(zip(attrIdx+aggrAttrIdx,tfinal.columns))
return tfinal,idxmap
def group(self, attrIdx, count=False, aggrAttrIdx=[], aggrFunc=[]):
#TODO: remove attributes of intermediary tables in attribute store
attrIdx, aggrAttrIdx, aggrFunc = util.makelist(attrIdx, aggrAttrIdx,
aggrFunc)
assert len(aggrAttrIdx) == len(aggrFunc)
tmptable, idxmap = self.copy()
aggrAttrIdx = util.mapIdx(aggrAttrIdx, idxmap)
if count:
cntIdx = tmptable.addAttr(gsql.WEIGHT_ATTR_NAME, val=Value(val=1))
aggrAttrIdx.append(cntIdx)
aggrFunc.append('cnt')
# Find values for aggregation
agg = aggregator.Aggregator(aggrFunc)
aggCols = [tmptable.getColumn(idx) for idx in aggrAttrIdx]
# Find groups of rows, and corresponding list of aggregation attributes
tproj, _ = tmptable.project(attrIdx)
groups = {}
for i, row in enumerate(tproj.data):
key = tuple(row)
if not key in groups:
groups[key] = []
groups[key].append([col[i] for col in aggCols])
# groups[key] is a list of lists: each inner list is the list of
# aggregation values corresponding to this row
# Create final table
tfinal, _ = tmptable.project(attrIdx + aggrAttrIdx)
for key in groups:
aggvals = agg.calc(groups[key])
newrow = list(key) + aggvals
tfinal.data.append(newrow)
idxmap = dict(zip(attrIdx + aggrAttrIdx, tfinal.columns))
return tfinal, idxmap
def link1(table, nodeDesc1, nodeDesc2, edgeDesc):
# Get attributes indexes
node1Idx = table.getIndex(nodeDesc1.idAttr)
node2Idx = table.getIndex(nodeDesc2.idAttr)
relationAttr1 = [attr for attr, _ in edgeDesc.relation]
relationAttr2 = [attr for _, attr in edgeDesc.relation]
node1RelIdx = table.getIndex(relationAttr1)
node2RelIdx = table.getIndex(relationAttr2)
node1DataIdx = table.getIndex(nodeDesc1.dataAttr)
node2DataIdx = table.getIndex(nodeDesc2.dataAttr)
# Create new graph and add nodes
g = graph.Graph(edgeDesc.type, False)
g.addNodes(table, node1Idx)
g.addNodes(table, node2Idx)
# Tranform table
assert not (node1Idx in node1RelIdx or node2Idx in node2RelIdx)
projlist1 = [node1Idx] + node1RelIdx + node1DataIdx
projlist2 = [node2Idx] + node2RelIdx + node2DataIdx
for cond in nodeDesc1.filter + nodeDesc2.filter:
cond.configureForTable(table)
t1, idxmap1 = table.select(nodeDesc1.filter, projlist1)
t2, idxmap2 = table.select(nodeDesc2.filter, projlist2)
t3, idxmap3 = t1.join(
t2,
zip(util.mapIdx(node1RelIdx, idxmap1),
util.mapIdx(node2RelIdx, idxmap2)))
if edgeDesc.threshold:
t4, idxmap4 = t3.group([
util.mapIdx(node1Idx, idxmap1, idxmap3),
util.mapIdx(node2Idx, idxmap2, idxmap3)
], True)
t5, idxmap5 = t4.select(
condition.Condition(WEIGHT_ATTR_NAME, ">=", edgeDesc.threshold))
idxmap = util.mergeIdxmap(idxmap4, idxmap5)
else:
t5, idxmap = t3.group([
util.mapIdx(node1Idx, idxmap1, idxmap3),
util.mapIdx(node2Idx, idxmap2, idxmap3)
], False)
# Add edges
g.addEdges(t5, util.mapIdx(node1Idx, idxmap1, idxmap3, idxmap),
util.mapIdx(node2Idx, idxmap2, idxmap3, idxmap))
pdb.set_trace()
return g
# Node description:
# id (attribute name)
# filter (condition on some attributes)
# attributes (attributes in the table, other than ID)
def link1(table,nodeDesc1,nodeDesc2,edgeDesc):
# Get attributes indexes
node1Idx = table.getIndex(nodeDesc1.idAttr)
node2Idx = table.getIndex(nodeDesc2.idAttr)
relationAttr1 = [attr for attr,_ in edgeDesc.relation]
relationAttr2 = [attr for _,attr in edgeDesc.relation]
node1RelIdx = table.getIndex(relationAttr1)
node2RelIdx = table.getIndex(relationAttr2)
node1DataIdx = table.getIndex(nodeDesc1.dataAttr)
node2DataIdx = table.getIndex(nodeDesc2.dataAttr)
# Create new graph and add nodes
g = graph.Graph(edgeDesc.type,False)
g.addNodes(table,node1Idx)
g.addNodes(table,node2Idx)
# Tranform table
assert not (node1Idx in node1RelIdx or node2Idx in node2RelIdx)
projlist1 = [node1Idx] + node1RelIdx + node1DataIdx
projlist2 = [node2Idx] + node2RelIdx + node2DataIdx
for cond in nodeDesc1.filter+nodeDesc2.filter:
cond.configureForTable(table)
t1,idxmap1 = table.select(nodeDesc1.filter, projlist1)
t2,idxmap2 = table.select(nodeDesc2.filter, projlist2)
t3,idxmap3 = t1.join(t2,zip(util.mapIdx(node1RelIdx,idxmap1),util.mapIdx(node2RelIdx,idxmap2)))
if edgeDesc.threshold:
t4,idxmap4 = t3.group([util.mapIdx(node1Idx,idxmap1,idxmap3),util.mapIdx(node2Idx,idxmap2,idxmap3)],True)
t5,idxmap5 = t4.select(condition.Condition(WEIGHT_ATTR_NAME,">=",edgeDesc.threshold))
idxmap = util.mergeIdxmap(idxmap4,idxmap5)
else:
t5,idxmap = t3.group([util.mapIdx(node1Idx,idxmap1,idxmap3),util.mapIdx(node2Idx,idxmap2,idxmap3)],False)
# Add edges
g.addEdges(t5,util.mapIdx(node1Idx,idxmap1,idxmap3,idxmap),util.mapIdx(node2Idx,idxmap2,idxmap3,idxmap))
pdb.set_trace()
return g
# Node description:
# id (attribute name)
# filter (condition on some attributes)
# attributes (attributes in the table, other than ID)