scripts/parse/smprl.py

#! /usr/bin/python
# -*- coding: utf-8 -*-

"""	
	This module defines classes related to SPMRL constituency and dependency
	annotation format.
	
	Version 1.0												(02-Jun-2015)
	- SPMRLConstTree and SPMRLConstNode are added.
"""

import constparse


#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬
# SPMRLConstTree
#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬

class SPMRLConstTree(constparse.ConstTree):
	'''
	Class for SPMRL-format constituency tree
	'''
	
	def __init__(self):
		'''
		Constructor
		'''
		
		constparse.ConstTree.__init__(self)
		
	
	
	def _createRoot(self):
		'''
		Creates and returns the root node
		'''
		
		return SPMRLConstNode()
		
	
	
	def removeFeatures(self):
		'''
		Removes feature annotation from the syntactic label.
		
		Feature annotations are surrounded by ## and follow the syntactic 
		label (e.g. (CLS-SUJ##lem=il|cpos=CL|g=m|n=p|p=1|s=suj## Nous)).
		'''
		
		self.root.removeFeatures(pflgAllSubtree = True)
		
	
	



#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬
# SPMRLConstNode
#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬

class SPMRLConstNode(constparse.ConstNode):
	'''
	Class for SPMRL-format constituency tree node
	'''
	
	def __init__(self):
		'''
		Constructor
		'''
		
		constparse.ConstNode.__init__(self)
		
	
	
	def _getNewNode(self):
		'''
		Creates and returns a node
		'''
		
		return SPMRLConstNode()
		
	
	
	def getFeatures(self):
		'''
		Returns the feature annotation attached to the label
		'''
		
		vGluePos = self.getLabel().find('##')
		
		if vGluePos != -1:
			return self.getLabel()[vGluePos + 2 : -2]
		else:
			return ''
		
	
	
	def removeFeatures(self, pflgAllSubtree = False):
		'''
		Removes feature annotation from the syntactic label.
		
		Feature annotations are surrounded by ## and follow the syntactic 
		label (e.g. (CLS-SUJ##lem=il|cpos=CL|g=m|n=p|p=1|s=suj## Nous)).
		
		The removal can be done to only the current node or the entire subtree
		under the node.
		'''
		
		vGluePos = self.getLabel().find('##')
		
		if vGluePos != -1:
			self.synTag = self.getLabel()[ : vGluePos]
		
		if pflgAllSubtree:
			for vChild in self.getChildren():
				vChild.removeFeatures(pflgAllSubtree = pflgAllSubtree)
		
	
	


#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬
# SPMRLConstParseLoader
#¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬

class SPMRLConstParseLoader(constparse.ConstParseLoader):
	'''
	Constituency parse loader
	
	see the parent class for details
	'''
	
	def __init__(self):
		'''
		Constructor
		'''
		
		# list of ConstParses
		self.parses = None
		
	
	
	def _createNewTree(self):
		'''
		Creates and returns a new constituency tree
		'''
		
		return SPMRLConstTree()