def convertStringtoPattern(xcur):
## need to integrate the 'virtual level'
lRes=[]
for elt in xcur:
if isinstance(elt,list):
lRes.extend([convertStringtoPattern(elt)])
else:
try:
float(elt)
f= featureObject()
f.setName("x")
f.setType(featureObject.NUMERICAL)
f.setValue(elt)
f.setObjectName(elt)
f.setWeight(1)
f.setTH(self.THNUMERICAL)
except:
f= featureObject()
f.setName("f")
f.setType(featureObject.EDITDISTANCE)
f.setValue(elt)
f.setTH(100.0)
lRes.append(f)
return lRes
def getSetOfVInfoFeatures(self, TH, lAttributes, myObject):
"""
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
for attr in lAttributes:
name = attr[0].getName()
value = attr[0].getValue()
feature = featureObject()
feature.setName(name)
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(feature.NUMERICAL)
self.addFeature(feature)
if self.getSetofFeatures() == []:
feature = featureObject()
feature.setName('EMPTY')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(True)
feature.setType(featureObject.BOOLEAN)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfListedAttributes(self, TH, lAttributes, myObject):
"""
move to XMLObjectClass ??
Generate a set of features:
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
lHisto = {}
for elt in self.getAllNamedObjects(myObject):
for attr in lAttributes:
try:
lHisto[attr]
except KeyError:
lHisto[attr] = {}
if elt.hasAttribute(attr):
try:
lHisto[attr][round(float(
elt.getAttribute(attr)))].append(elt)
except KeyError:
lHisto[attr][round(float(
elt.getAttribute(attr)))] = [elt]
if lHisto != {}:
for attr in lAttributes:
for value in lHisto[attr]:
if len(lHisto[attr][value]) > 0.1:
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName(attr)
# feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(ftype)
self.addFeature(feature)
if 'virtual' in lAttributes:
ftype = featureObject.BOOLEAN
feature = featureObject()
feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(self.getAttribute('virtual'))
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfFeaturesBB(self, TH, lAttributes, myObject):
"""
features: BB X Y H W
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
#build BB
if self.getBB() is None:
self.addBoundingBox()
x, y, h, w = self.getBB()
feature = featureObject()
feature.setName('lm')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self.getName())
feature.setValue(round(x))
feature.setType(feature.NUMERICAL)
self.addFeature(feature)
feature = featureObject()
feature.setName('rm')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self.getName())
feature.setValue(round(x + w))
feature.setType(feature.NUMERICAL)
self.addFeature(feature)
if self.getSetofFeatures() == []:
feature = featureObject()
feature.setName('EMPTY')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self.getName())
feature.setValue(True)
feature.setType(featureObject.BOOLEAN)
self.addFeature(feature)
return self.getSetofFeatures()
def getContentAnaphoraAttributes(self,TH,lAttributes,myObject):
"""
textual content + position wrt parent
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
lHisto = {}
lHisto['position']={}
for elt in self.getAllNamedObjects(myObject):
## if elt is first in elt.getParent()
position= elt.getParent().getObjects().index(elt)
if position == len(elt.getParent().getObjects())-1:position=-1
try:lHisto['position'][str(position)].append(elt)
except KeyError : lHisto['position'][str(position)]= [elt]
for attr in lAttributes:
try:lHisto[attr]
except KeyError:lHisto[attr] = {}
if elt.hasAttribute(attr):
try:
try:lHisto[attr][round(float(elt.getAttribute(attr)))].append(elt)
except KeyError: lHisto[attr][round(float(elt.getAttribute(attr)))] = [elt]
except TypeError:pass
elif attr == 'text':
try:lHisto[attr][elt.getContent()].append(elt)
except KeyError: lHisto[attr][elt.getContent()] = [elt]
for attr in lAttributes:
for value in lHisto[attr]:
if len(lHisto[attr][value]) > 0.1: # 0.1: keep all!
if attr not in ['position','text']:
ftype= featureObject.NUMERICAL
else:
ftype= featureObject.EDITDISTANCE
feature = featureObject()
feature.setName(attr)
# feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfFeaturesPageSize(self, TH, lAttributes, myObject):
"""
features: BB X Y H W
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
feature = featureObject()
feature.setName('h')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(float(self.getAttribute('height'))))
feature.setType(feature.NUMERICAL)
self.addFeature(feature)
feature = featureObject()
feature.setName('w')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(float(self.getAttribute('width'))))
feature.setType(feature.NUMERICAL)
self.addFeature(feature)
if self.getSetofFeatures() == []:
feature = featureObject()
feature.setName('EMPTY')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(True)
feature.setType(featureObject.BOOLEAN)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfFeaturesXPos(self, TH, lAttr, myObject):
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName('x')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(self.getX()))
feature.setType(ftype)
self.addFeature(feature)
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName('x2')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(self.getX() + self.getWidth()))
feature.setType(ftype)
self.addFeature(feature)
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName('xc')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(self.getX() + self.getWidth() / 2))
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfListedAttributes(self, TH, lAttributes, myObject):
"""
Generate a set of features: X start of the lines
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
lHisto = {}
for elt in self.getAllNamedObjects(myObject):
for attr in lAttributes:
try:
lHisto[attr]
except KeyError:
lHisto[attr] = {}
if elt.hasAttribute(attr):
# if elt.getWidth() >500:
# print elt.getName(),attr, elt.getAttribute(attr) #, elt.getNode()
try:
try:
lHisto[attr][round(float(
elt.getAttribute(attr)))].append(elt)
except KeyError:
lHisto[attr][round(float(
elt.getAttribute(attr)))] = [elt]
except TypeError:
pass
for attr in lAttributes:
for value in lHisto[attr]:
# print attr, value, lHisto[attr][value]
if len(lHisto[attr][value]) > 0.1:
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName(attr)
# feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(ftype)
self.addFeature(feature)
if 'text' in lAttributes:
if len(self.getContent()):
ftype = featureObject.EDITDISTANCE
feature = featureObject()
# feature.setName('content')
feature.setName('f')
feature.setTH(90)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(self.getContent().split()[0])
feature.setType(ftype)
self.addFeature(feature)
if 'tokens' in lAttributes:
if len(self.getContent()):
for token in self.getContent().split():
if len(token) > 4:
ftype = featureObject.EDITDISTANCE
feature = featureObject()
feature.setName('token')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(token.lower())
feature.setType(ftype)
self.addFeature(feature)
if 'xc' in lAttributes:
ftype = featureObject.NUMERICAL
feature = featureObject()
# feature.setName('xc')
feature.setName('xc')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(round(self.getX() + self.getWidth() / 2))
feature.setType(ftype)
self.addFeature(feature)
#
if 'virtual' in lAttributes:
ftype = featureObject.BOOLEAN
feature = featureObject()
feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(self.getAttribute('virtual'))
feature.setType(ftype)
self.addFeature(feature)
if 'bl' in lAttributes:
for inext in self.next:
ftype = featureObject.NUMERICAL
feature = featureObject()
baseline = self.getBaseline()
nbl = inext.getBaseline()
if baseline and nbl:
feature.setName('bl')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
# avg of baseline?
avg1 = baseline.getY() + (baseline.getY2() -
baseline.getY()) / 2
avg2 = nbl.getY() + (nbl.getY2() - nbl.getY()) / 2
feature.setValue(round(abs(avg2 - avg1)))
feature.setType(ftype)
self.addFeature(feature)
if 'linegrid' in lAttributes:
#lgridlist.append((ystart,rowH, y1,yoverlap))
for ystart, rowh, _, _ in self.lgridlist:
ftype = featureObject.BOOLEAN
feature = featureObject()
feature.setName('linegrid%s' % rowh)
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(ystart)
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfListedAttributes(self, TH, lAttributes, myObject):
"""
Generate a set of features: X start of the lines
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
if 'virtual' in lAttributes:
ftype = featureObject.BOOLEAN
feature = featureObject()
feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(self.getAttribute('virtual'))
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()
lHisto = {}
for elt in self.getAllNamedObjects(myObject):
if float(elt.getAttribute('width')) > 0:
for attr in lAttributes:
try:
lHisto[attr]
except KeyError:
lHisto[attr] = {}
if elt.hasAttribute(attr):
try:
lHisto[attr][round(float(
elt.getAttribute(attr)))].append(elt)
except:
lHisto[attr][round(float(
elt.getAttribute(attr)))] = [elt]
for attr in lAttributes:
for value in lHisto[attr]:
if len(lHisto[attr][value]) > 0.1:
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName(attr)
l = sum(x.getHeight() for x in lHisto[attr][value])
feature.setWeight(l)
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(ftype)
self.addFeature(feature)
if self.getSetofFeatures() == []:
feature = featureObject()
feature.setName('EMPTY')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(True)
feature.setType(featureObject.BOOLEAN)
self.addFeature(feature)
return self.getSetofFeatures()
def getSetOfListedAttributes(self, TH, lAttributes, myObject):
"""
move to XMLObjectClass ??
Generate a set of features:
"""
from spm.feature import featureObject
if self._lBasicFeatures is None:
self._lBasicFeatures = []
# needed to keep canonical values!
elif self.getSetofFeatures() != []:
return self.getSetofFeatures()
lHisto = {}
for elt in self.getAllNamedObjects(myObject):
for attr in lAttributes:
try:
lHisto[attr]
except KeyError:
lHisto[attr] = {}
if elt.hasAttribute(attr):
try:
try:
lHisto[attr][round(float(
elt.getAttribute(attr)))].append(elt)
except KeyError:
lHisto[attr][round(float(
elt.getAttribute(attr)))] = [elt]
except TypeError:
pass
# empty object
if lHisto == {}:
return self.getSetofFeatures()
for attr in lAttributes:
for value in lHisto[attr]:
# print attr, value, lHisto[attr][value]
if len(lHisto[attr][value]) > 0.1:
ftype = featureObject.NUMERICAL
feature = featureObject()
feature.setName(attr)
# feature.setName('f')
feature.setTH(TH)
for o in lHisto[attr][value]:
feature.addNode(o)
feature.setObjectName(self)
feature.setValue(value)
feature.setType(ftype)
self.addFeature(feature)
if 'tokens' in lAttributes:
if len(self.getContent()):
for token in self.getContent().split():
if len(token) > 4:
ftype = featureObject.EDITDISTANCE
feature = featureObject()
feature.setName('token')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(token.lower())
feature.setType(ftype)
self.addFeature(feature)
#
if 'virtual' in lAttributes:
ftype = featureObject.BOOLEAN
feature = featureObject()
feature.setName('f')
feature.setTH(TH)
feature.addNode(self)
feature.setObjectName(self)
feature.setValue(self.getAttribute('virtual'))
feature.setType(ftype)
self.addFeature(feature)
return self.getSetofFeatures()