def testGetInclusizeSubsets(self):
expectSet = ((frozenset(("A",)),
frozenset(("C",)),
frozenset(("A","C",))))
subsets = Subsets(SubsetsTests.testSet)
iss = subsets.getInclusiveSubsets(frozenset(("A", "C",)))
self.assertEqual(iss, expectSet)
def testGetSubsets(self):
expectSet = ((frozenset(("A",)),
frozenset(("B",)),
frozenset(("C",)),
frozenset(("A","B",)),
frozenset(("A","C",)),
frozenset(("B","C",)),
frozenset(("A","B","C"))))
subsets = Subsets(SubsetsTests.testSet)
ss = subsets.getSubsets()
self.assertEqual(ss, expectSet)
def __init__(self, subsets=None, isInclusive=False):
"""create new Venn. If subsets is None, a private Subsets object is
created,"""
# is this a standard venn or inclusive?
self.isInclusive = isInclusive
self.subsets = subsets
if self.subsets is None:
self.subsets = Subsets()
# Tables mappings set name to items and items to set names
self.nameToItems = SetDict()
self.itemToNames = SetDict()
# Venn table, dict index by name, of items (lazy build)
self.venn = None
def __init__(self, subsets=None, isInclusive=False):
"""create new Venn. If subsets is None, a private Subsets object is
created,"""
# is this a standard venn or inclusive?
self.isInclusive = isInclusive
self.subsets = subsets
if self.subsets is None:
self.subsets = Subsets()
# Tables mappings set name to items and items to set names
self.nameToItems = SetDict()
self.itemToNames = SetDict()
# Venn table, dict index by name, of items (lazy build)
self.venn = None
class Venn(object):
"""Generate Venn diagram set intersections. Each set has
list of ids associated with it that are shared between sets.
"""
def __init__(self, subsets=None, isInclusive=False):
"""create new Venn. If subsets is None, a private Subsets object is
created,"""
# is this a standard venn or inclusive?
self.isInclusive = isInclusive
self.subsets = subsets
if self.subsets is None:
self.subsets = Subsets()
# Tables mappings set name to items and items to set names
self.nameToItems = SetDict()
self.itemToNames = SetDict()
# Venn table, dict index by name, of items (lazy build)
self.venn = None
def addItem(self, setName, item):
"add a single item from a named set"
self.subsets.add(setName)
self.nameToItems.add(setName, item)
self.itemToNames.add(item, setName)
self.venn = None
def addItems(self, setName, items):
"add items from a named set"
self.subsets.add(setName)
for item in items:
self.nameToItems.add(setName, item)
self.itemToNames.add(item, setName)
self.venn = None
def getNumItems(self):
return len(self.itemToNames)
def _buildVenn(self):
"build Venn table"
self.venn = SetDict(self.subsets.getSubsets())
for item in list(self.itemToNames.keys()):
nameSet = frozenset(self.itemToNames[item])
self.venn.add(nameSet, item)
def _buildInclusive(self):
"build as inclusive subsets"
self.venn = SetDict(self.subsets.getSubsets())
for item in list(self.itemToNames.keys()):
setName = frozenset(self.itemToNames[item])
for iss in self.subsets.getInclusiveSubsets(setName):
self.venn.add(iss, item)
def _update(self):
"build venn or inclusive venn, if it doesn't exists"
if self.venn is None:
if self.isInclusive:
self._buildInclusive()
else:
self._buildVenn()
def getSubsetIds(self, subset):
"get ids for the specified subset"
self._update()
ids = self.venn.get(subset)
if ids is None:
ids = []
return ids
def getSubsetCounts(self, subset):
"get counts for the specified subset"
return len(self.getSubsetIds(subset))
def getTotalCounts(self):
"get total of counts for all subsets (meaningless on inclusive)"
t = 0
for subset in self.subsets.getSubsets():
t += self.getSubsetCounts(subset)
return t
@staticmethod
def formatSubsetName(subset,
subsetNameSeparator=" ",
setNameFormatter=str):
return subsetNameSeparator.join(
sorted([setNameFormatter(s) for s in subset]))
def writeCounts(self, fh, subsetNameSeparator=" ", setNameFormatter=str):
"write TSV of subset counts to an open file"
fileOps.prRowv(fh, "subset", "count")
for subset in self.subsets.getSubsets():
fileOps.prRowv(
fh,
self.formatSubsetName(subset, subsetNameSeparator,
setNameFormatter),
self.getSubsetCounts(subset))
def writeSets(self, fh, subsetNameSeparator=" ", setNameFormatter=str):
"write TSV of subsets and ids to an open file"
fileOps.prRowv(fh, "subset", "ids")
for subset in self.subsets.getSubsets():
fileOps.prRowv(
fh,
self.formatSubsetName(subset, subsetNameSeparator,
setNameFormatter),
self.getSubsetCounts(subset))
class Venn(object):
"""Generate Venn diagram set intersections. Each set has
list of ids associated with it that are shared between sets.
"""
def __init__(self, subsets=None, isInclusive=False):
"""create new Venn. If subsets is None, a private Subsets object is
created,"""
# is this a standard venn or inclusive?
self.isInclusive = isInclusive
self.subsets = subsets
if self.subsets is None:
self.subsets = Subsets()
# Tables mappings set name to items and items to set names
self.nameToItems = SetDict()
self.itemToNames = SetDict()
# Venn table, dict index by name, of items (lazy build)
self.venn = None
def addItem(self, setName, item):
"add a single item from a named set"
self.subsets.add(setName)
self.nameToItems.add(setName, item)
self.itemToNames.add(item, setName)
self.venn = None
def addItems(self, setName, items):
"add items from a named set"
self.subsets.add(setName)
for item in items:
self.nameToItems.add(setName, item)
self.itemToNames.add(item, setName)
self.venn = None
def getNumItems(self):
return len(self.itemToNames)
def __buildVenn(self):
"build Venn table"
self.venn = SetDict(self.subsets.getSubsets())
for item in self.itemToNames.iterkeys():
nameSet = frozenset(self.itemToNames[item])
self.venn.add(nameSet, item)
def __buildInclusive(self):
"build as inclusive subsets"
self.venn = SetDict(self.subsets.getSubsets())
for item in self.itemToNames.iterkeys():
setName = frozenset(self.itemToNames[item])
for iss in self.subsets.getInclusiveSubsets(setName):
self.venn.add(iss, item)
def __update(self):
"build venn or inclusive venn, if it doesn't exists"
if self.venn is None:
if self.isInclusive:
self.__buildInclusive()
else:
self.__buildVenn()
def getSubsetIds(self, subset):
"get ids for the specified subset"
self.__update()
ids = self.venn.get(subset)
if ids is None:
ids = []
return ids
def getSubsetCounts(self, subset):
"get counts for the specified subset"
return len(self.getSubsetIds(subset))
def getTotalCounts(self):
"get total of counts for all subsets (meaningless on inclusive)"
t = 0
for subset in self.subsets.getSubsets():
t += self.getSubsetCounts(subset)
return t
@staticmethod
def formatSubsetName(subset, subsetNameSeparator=" ", setNameFormatter=str):
return subsetNameSeparator.join(sorted([setNameFormatter(s) for s in subset]))
def writeCounts(self, fh, subsetNameSeparator=" ", setNameFormatter=str):
"write TSV of subset counts to an open file"
fileOps.prRowv(fh, "subset", "count")
for subset in self.subsets.getSubsets():
fileOps.prRowv(fh, self.formatSubsetName(subset, subsetNameSeparator, setNameFormatter), self.getSubsetCounts(subset))
def writeSets(self, fh, subsetNameSeparator=" ", setNameFormatter=str):
"write TSV of subsets and ids to an open file"
fileOps.prRowv(fh, "subset", "ids")
for subset in self.subsets.getSubsets():
fileOps.prRowv(fh, self.formatSubsetName(subset, subsetNameSeparator, setNameFormatter), self.getSubsetCounts(subset))