#!/bin/env python import sys # Provides the API to control the dictionary. global __debug global DBname global word_list __debug = True DBname = "dict.db" word_list = [] #strings which are tag in XML class dict: def __init__(self): import sqlite3 self.conn = sqlite3.connect(DBname, isolation_level=None) # Turn on autocommit mode # Set isolation_level to "IMMEDIATE" self.conn.isolation_level = "IMMEDIATE" self.cur = self.conn.cursor() self.numwords = -1 self.wordid_list = [] self.length = 0 def getnumwords(self, level = 0): if self.numwords == -1: if level == 0: self.cur.execute("SELECT COUNT(wordid) from las_word") else: self.cur.execute("SELECT COUNT(wordid) from las_word where length = ?", (level, )) self.numwords = self.cur.fetchone() return self.numwords def getrandomwordid(self, length=0, numwords = 1): if self.wordid_list == [] or self.length != length: if length == 0: self.cur.execute("SELECT wordid from las_word") else: self.length = length self.cur.execute("SELECT wordid from las_word where length = ?", (length, )) self.wordid_list = self.cur.fetchall() #count = self.wordid_list.count #count = len(self.wordid_list) import random randids = random.sample(self.wordid_list , numwords) return randids class word: def __init__(self, identifier=None, value= None): import sqlite3 self.conn = sqlite3.connect(DBname, isolation_level=None) # Turn on autocommit mode # Set isolation_level to "IMMEDIATE" self.conn.isolation_level = "IMMEDIATE" self.cur = self.conn.cursor() if identifier == "las_word_id": self.las_word_id = value self.cur.execute("SELECT * from las_word where laswid = ?", (value,)) elif identifier == "wordid": self.wordid = value self.cur.execute("SELECT * from las_word where wordid = ?", (value,)) elif identifier == "word": self.word = value self.cur.execute("SELECT * from las_word where lemma = ?", (value,)) elif identifier == None or value == None: self.las_word_id = None self.wordid = None self.word = None self.length = None return else: return "Invalid Usage" (laswid, wordid, lemma, length) = self.cur.fetchone() self.las_word_id = laswid self.wordid = wordid self.word = lemma self.length = length def getword(self): return self.word def getwordid(self): return self.wordid def getsynsetid(self): self.synsetid_list = [] self.cur.execute("SELECT * from las_sense where wordid = ?", (self.wordid,)) for (wordid, synsetid, rank) in self.cur: self.synsetid_list.append(synsetid) return self.synsetid_list def getdef(self): self.def_list = [] try: self.synsetid_list except AttributeError: self.getsynsetid() for synsetid in self.synsetid_list: self.cur.execute("SELECT * from las_synset where synsetid = ?", (synsetid,) ) for (synsetid, pos, definition) in self.cur: self.def_list.append( (pos, definition)) return self.def_list def getusage(self): try: self.synsetid_list except AttributeError: self.getsynsetid() self.usage_list = [] for synsetid in self.synsetid_list: self.cur.execute("SELECT * from las_sample where synsetid = ?", (synsetid,)) for (synsetid, sampleid, sample) in self.cur: self.usage_list.append( (sample)) return self.usage_list if __name__ == "__main__": k = dict() num_words = k.getnumwords() print num_words wordid = k.getrandomwordid(15) #will return word of length 15 l = word("wordid", wordid ) print l.getword() l.getsynsetid() print l.getdef() print l.getusage()