Tryag File Manager

/home/cluster1/data/bu01/1121861/html/TMAT/tmat/plugins/nahuatlPlugin.py

import sys
import os
from subprocess import Popen, PIPE
import re
from twisted.plugin import IPlugin
from tokenizer import TokenizerPlugin
from myparser import ParserPlugin
from zope.interface import Interface, Attribute, implements
from xml.dom.minidom import getDOMImplementation, parseString

class CommandLineTokenizer(object):
    implements(IPlugin, TokenizerPlugin)
    running = False
    displayName = "Command-Line (Local) Tokenizer"
    preferences = {"Initialization Command":"string",
                   "Start Command":"string",
                   "Stop Command":"string",
                   "Tokenize Command":"string",
                   "Get Version Command":"string",
		   "Uninitialize Command":"string"}
    values = {}
    def setValue(self,key, val):
	self.values[key]=val
    def initialize(self):
	try:
	    initCommand="%s\n" % self.values["Initialization Command"]
	except:
	    initCommand=""
	self.pipe=Popen(initCommand,shell=True,bufsize=0,stdin=PIPE,stdout=PIPE,stderr=PIPE)
    def tokenize(self, text):
	if not(self.running):
	    self.start()
	self.communicate("Tokenize Command")
	retText=[]
	## get all of the junk off the front
	line=""
	for inLine in [x for x in text.split("\n") if x.strip()]:
	    self.pipe.stdin.write("%s\n" % inLine.strip())
	    self.pipe.stdin.flush()
	    while not(line):
		line = self.pipe.stdout.readline().strip()
	    retText.append(line)
	    while line:
		line = self.pipe.stdout.readline().strip()
		retText.append(line)
	    retText=retText[:-1]
	retList=[]
	i=0
	for token in retText:
	    i=text.find(token,i)
	    retList.append((i,i+len(token),True))
            i += len(token)
	return retList
    def start(self):
	if not(self.pipe):
	    self.initialize()
	self.communicate("Start Command")
	self.running=True
    def stop(self):
	self.communicate("Stop Command")
	self.running=False
    def communicate(self, key,readback=True):
	self.pipe.stdin.write("%s\n" % self.values[key])
	self.pipe.stdin.flush()
	if readback:
	    return self.pipe.stdout.readline()
	else:
	    return 0
    def version(self):
	return self.communicate("Get Version Command")
    def uninitialize(self):
	self.communicate("Uninitialize Command",False)
        if self.pipe.poll():
	    self.pipe.stdin.close()
            pid=self.pipe.pid
            try:
                os.kill(pid,9)
            except:
                pass

class CommandLineParser(object):
    implements(IPlugin, ParserPlugin)
    running = False
    displayName = "Command Line Parser"
    preferences = {"Initialization Command":"string",
                   "Start Command":"string",
                   "Stop Command":"string",
                   "Parse Command":"string",
                   "Get Version Command":"string"}
    values = {}
    def setValue(self,key, val):
	self.values[key]=val
    def initialize(self):
	try:
	    initCommand="%s\n" % self.values["Initialization Command"]
	except:
	    initCommand=""
	self.pipe=Popen(initCommand,shell=True,bufsize=0,stdin=PIPE,stdout=PIPE,stderr=PIPE)
    def start(self):
	if not(self.pipe):
	    self.initialize()
	self.communicate("Start Command")
	self.running=True
    def stop(self):
	self.communicate("Stop Command")
	self.running=False
    def communicate(self, key):
	self.pipe.stdin.write("%s\n" % self.values[key])
	self.pipe.stdin.flush()
	return self.pipe.stdout.readline()
    def version(self):
	return self.communicate("Get Version Command")
    def uninitialize(self):
	self.communicate("Uninitialize Command")
    def parse(self, wordforms):
	impl=getDOMImplementation()
	myDom=impl.createDocument(None,"parserInput",None)
	topNode=myDom.documentElement

for word in wordforms:
	    wordNode=myDom.createElement("wordform")
	    wordNode.setAttribute("form",word)
	    topNode.appendChild(wordNode)
	self.input.write(myDom.toxml())
	outDom=parseString(self.output.read())
	topOut=outDom.getElementsByTagName("parserOutput")
	retDict={}
	for parseNode in topOut.getElementsByTagName("parses"):
	    word=parseNode.getAttribute("form")
	    retDict[word]=[]
	    for parse in parseNode.getElementsByTagName("parse"):
		morphemes=parse.GetAttribute("morphemes")
		glosses=parse.GetAttribute("glosses")
		retDict[word].append({"morpheme":morphemes,
				      "gloss":glosses})
	return retDict
    def initialize(self):
	try:
	    initCommand="%s\n" % self.values["Initialization Command"]
	except:
	    initCommand=""
	self.pipe=Popen(initCommand,shell=True,bufsize=0,stdin=PIPE,stdout=PIPE,stderr=PIPE)
    def parseToken(self, token):
 	return [(x, self.getGloss(x)) for x in self.getUnderlyingForms(token)]

x = CommandLineTokenizer()
y = CommandLineParser()