path: root/bladictionary.py

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

import sys
import shlex
import optparse
from lxml import etree
import sqlite3
import requests
from definition import Definition

VERSION = "3.0"

def get_xml(word, word_dict):
	api_url = "http://services.aonaware.com/DictService/DictService.asmx/DefineInDict?dictId="+word_dict+"&word="+word

	try:
		xml = etree.parse(api_url)
	except IOError:
		print ("Error: Could not access the Dictionary service.")
		return

	#Root element tag is WordDefinition, which is the same as the element which contains the definition
	#That's akward, so let's just
	for element in xml.iter("{http://services.aonaware.com/webservices/}Definitions"):
		root = element
		break;

	return root

def parse_xml(xml):

	#Only worrying about one definition for now, add multiples later
	for element in xml.iter("{http://services.aonaware.com/webservices/}WordDefinition"):
		definition = element.text
		break;

	definition_lines = definition.split("\n")

	items = []
	synonyms = []
	antonyms = []
	id = 0
	syn_line = False
	ant_line = False
	word = definition_lines[0]
	for line in definition_lines[1:]:
		extra_parts = line.split("[")
		line_parts = extra_parts[0].split(":")

		#Work out if this is a line that includes identifying information (word type or id)
		#This is done by checking the indenting - if the first 8 characters are spaces, it's not
		#It's a bit hacky, maybe change this later

		id_line = False
		for char in line_parts[0][0:7]:
			if char != " ":
				id_line = True

		if id_line:
			if id > 0:
				#Add a previous line to the array of definitions, if there is one
				item = Definition(word, id, "wn", word_type, definition, [], synonyms, antonyms)
				items.append(item)
				synonyms = []
				antonyms = []
				syn_line = False
				ant_line = False

			id_parts = line_parts[0].strip().split(" ")

			if id_parts[0] == "[also":
				#This line comes at the end, don't worry about it for now
				break;

			if id_parts[0].isdigit() != True:
				word_type = id_parts[0]
				if len(id_parts) < 2:
					id = 1
				else :
					id = id_parts[1]
				definition = line_parts[1].strip() + " "
			else:
				id = id_parts[0]
				definition = line_parts[1].strip() + " "

			for word in line_parts[2:]:
				definition += word + " "
		elif syn_line:
			for part in line_parts:
				extra_words = part.strip().split(" ")
				for syn in extra_words:
					synonyms.append(syn.strip("{},] "))
		elif ant_line:
			for part in line_parts:
				extra_words = part.strip().split(" ")
				for ant in extra_words:
					antonyms.append(syn.strip("{},] "))
		else:
			definition += line_parts[0].strip() + " "
			if len(line_parts) > 1:
				for part in line_parts[1:-1]:
					definition += part + ":"
				definition += line_parts[-1]

		if len(extra_parts) > 0:
			for part in extra_parts:
				extra_words = part.split(" ")
				if extra_words[0] == "syn:":
					for syn in extra_words[1:]:
						synonyms.append(syn.strip("{},] "))
						syn_line = True
				elif extra_words[0] == "ant:":
					for ant in extra_words[1:]:
						antonyms.append(ant.strip("{},] "))
						ant_line = True

	item = Definition(word, id, "wn", word_type, definition, [], synonyms, antonyms)
	items.append(item)

	return items

def parse_args():
	args = sys.argv[1:]

	#If there are no args, try stdin
	if not args:
		args = shlex.split(sys.stdin.read())

	parser = optparse.OptionParser( usage = "!define <term> [type] [dictionary] | types:  noun, verb, adjective, adverb | dictionaries: wordnet, oed, foldoc, urban" )
	parser.add_option( "-v", "--version", action = "store_true", help = "Print the version number" )
	parser.add_option( "-c", "--channel", action = "store", help = "The IRC channel of the request")
	options, args = parser.parse_args( args )

	types = ["n", "noun", "v", "verb", "adj", "adjective", "adv", "adverb", "tech"]
	dicts = ["wn", "wordnet", "oed", "db", "foldoc", "urban"]

	word = ""
	word_type = ""
	word_dict = ""
	page_num = 1
	stop_id = 0

	#See if the dictionary or type have been specified
	if len(args) > 1:
		cur_id = 1
		for arg in args[1:]:
			if arg in types:
				word_type = arg
				if stop_id == 0:
					stop_id = cur_id
			elif arg in dicts:
				word_dict = arg
				if stop_id == 0:
					stop_id = cur_id
			elif len(arg) < 3 and arg.isdigit():
				page_num = int(arg)
				if stop_id == 0:
					stop_id = cur_id

			cur_id +=1

	#If no type or dictionary has been set, use all the args
	if stop_id == 0:
		stop_id = len(args)

	#Read everything before type/dict as the desired word
	if len(args) > 0:
		cur_id = 0
		for arg in args[:stop_id]:
			word += args[cur_id] + " "
			cur_id += 1

		word = word.strip()
	elif not options.version:
		print ("You must specify a word to define")
		sys.exit(1)

	return word, word_type, word_dict, page_num, options

#Get definitions from the OED text file
def parse_oed(word):

	#For some reason some definitions have — in front of the word type. We'll convert it to the standard format later
	#Note, that's not a -, it's some other unicode dash.
	#types = ["n.", "—n.", "v.", "—v.", "adj.", "—adj.", "adv.", "—adv."]
	types = {'n.' : 'n', '—n.' : 'n', 'v.' : 'v', '—v.' : 'v', 'adj.' : 'adj', '—adj.' : 'adj', 'adv.' : 'adv', '—adv.' : 'adv'}

	file = open("dictionaries/oed.txt")
	word_line = ""
	word_len = len(word.split(" "))
	#Find the definition based on the first word on each line
	#TODO: Fix for multi word definitions
	for line in file:
		word_parts = line.split(" ")
		word_part = ""

		#Read the appropriate number of words depending on how many were specified
		for part in word_parts[0:word_len]:
			word_part += part + " "
		word_part = word_part.strip()
		#Ignore case

		if word_part.lower() == word.lower():
			word_line = line
			break;

	if word_line == "":
		return

	prev_part = ""
	items = []
	id = 0
	definition = ""
	word_type = ""
	for part in line.split(" ")[word_len:]:
		#If we've reached a word type, add the current data to the list
		if part in types.keys():
			#Don't add empty definitions
			if int(id) > 0 and definition and definition != " ":
				item = Definition(word, id, "oed", word_type, definition, [], [], [])
				items.append(item)
				definition = ""

			#Convert the word type to the simple version
			word_type = types[part]

		elif part.isdigit() and prev_part is not None:
			#Ignore anything before the first definition
			if part != "1":
				item = Definition(word, id, "oed", word_type, definition, [], [], [])
				items.append(item)
			definition = ""
			id = part
		else:
			definition += part.strip() + " "

		prev_part = part

	#Words with one definition don't have an id
	if id == 0:
		id = 1

	#Add the last item
	item = Definition(word, id, "oed", word_type, definition, [], [], [])
	items.append(item)

	#Sort the items by type
	tmp_items = []
	simple_types = ["n", "v", "adj", "adv"]
	for simple_type in simple_types:
		for tmp_item in items:
			if tmp_item.word_type == simple_type:
				tmp_items.append(tmp_item)

	items = tmp_items
	return items

#Gets all matching WordNet definitions from the database
def get_sql(word, dictionary = "wn"):

	items = []
	types = ["n", "v", "adj", "adv", "tech"]

	if dictionary == "wordnet" or dictionary == "wn":
		dict_id = 1
	elif dictionary == "foldoc" or dictionary == "tech":
		dict_id = 2
	else:
		print ("Invalid dictionary")
		sys.exit(1)

	con = sqlite3.connect('dictionaries/wordnet.db');
	con.row_factory = sqlite3.Row

	with con:

		cur = con.cursor()
		cur.execute("SELECT * from definitions where word = ? and dictionary_id = ? ORDER BY type_id,sub_id;", [word, dict_id])

		rows = cur.fetchall()

		# If no results were found, try a case insensitive search
		if rows is None or len(rows) == 0:
			cur.execute("SELECT * from definitions where word = ? COLLATE NOCASE and dictionary_id = ? ORDER BY type_id,sub_id;", [word, dict_id])
			rows = cur.fetchall()

		for row in rows:

			id = row['id']
			word = row['word']
			sub_id = row['sub_id']
			type = types[row['type_id']-1]
			definition = row['definition']
			synset_id = row['synset_id']
			dictionary_id = row['dictionary_id']

			cur.execute("SELECT * from uses where definition_id = ?", [id])
			use_rows = cur.fetchall()
			uses = []

			for use in use_rows:
				uses.append(use['quote'])

			cur.execute("SELECT name from dictionaries where id = ?", [dictionary_id])
			dict_row = cur.fetchone()
			if dict_row is not None:
				dictionary = dict_row[0]
			else:
				dictionary = "wn"

			item = Definition(word, sub_id, dictionary, type,definition, uses, synset_id, [])
			items.append(item)

	return items



def parse_urban(word):

	word = word.replace(" ", "+")

	try:
		r = requests.get("http://api.urbandictionary.com/v0/define?term=" + word)
		json = r.json()
	except:
		#Should probably print an error, but then that would break printing other definitions, so let's not
		return

	items = []
	id = 1
	for json_item in json['list']:
		if json_item['definition']:
			item = Definition(word, id, "urban", "urban", json_item['definition'], [json_item['example'].replace("\r\n", " ")], [], [])
			items.append(item)
			id += 1

	return items

def main():

	word, word_type, word_dict, page_num, options = parse_args()

	if options.version:
		print (VERSION)
		sys.exit( 0 )

	if word_dict == "oed":
		items = parse_oed(word)
	elif word_dict == "wn":
		xml = get_xml(word, word_dict)

		if xml is None:
			print ("Error finding definitions for " + word)
			sys.exit( 1 )

		items = parse_xml(xml)
	elif word_dict == "foldoc" or word_type == "tech":
		items = get_sql(word, "foldoc")
	elif word_dict == "urban":
		items = parse_urban(word)
	else:
		if word_dict is None or word_dict == "":
			word_dict = "wn"
			items = get_sql(word, "wn")

			foldoc_items = get_sql(word, "foldoc")
			if foldoc_items is not None and len(foldoc_items) > 0:
				items += foldoc_items

			urban_items = parse_urban(word)
			if urban_items is not None and len(urban_items) > 0:
				items += urban_items

	if items is None or len(items) == 0:
		print ("No definitions found for "+word)
		sys.exit( 1 )

	line_length = 0
	num_more = 0
	suppress_print = False

	#Length limit for IRC messages
	#RFC specifies message length is 510. This includes protocol stuff.
	#37 is for the user/hostname of Bratchbot, then take off the length of the channel name
        #Actually this isn't right and 412 seems to work
	if options.channel:
		max_length = 410 - len(options.channel)
	else:
		max_length = 400

	types = ["n", "v", "adj", "adv", "tech", "urban"]
	type_id = 0
	all_types = word_type == ""

	#Normally we try and display 1 of each type if no type is specified
	#But if the word only has one type of definition, display all of them
	if all_types:
		word_types = []
		for item in items:
			if item.word_type not in word_types:
				word_types.append(item.word_type)
		if len(word_types) == 1:
			word_type = word_types[0]
			all_types = False

	for item in items:
		#If no type is specified, we display one of each
		if all_types:
			found_type = False
			cur_id = type_id
			if int(item.id) < page_num:
				if item.id != 0:
					continue
				elif item.id == 0 and page_num > 1:
					continue
			#Definitions should be ordered by type, so loop through all the types until we find a match
			for cur_type in types[type_id:]:

				if item.word_type == cur_type:
					word_type = cur_type
					type_id = cur_id
					found_type = True

				cur_id += 1

			#If there were no matches, stick with the current type for now
			if not found_type:

				if type_id < len(types):
					word_type = types[type_id]

		if word_type == item.word_type:
			#Keep track of how many we haven't printed
			if suppress_print:
				num_more+=1
			else:
				# Ignore anything after a line break as this breaks output
				item_definition = item.definition.split("\n")[0]
				if int(item.id) > 0:
					definition = item.word_type + " " + str(item.id) + ": " + item_definition
				else:
					definition = item.word_type + ": " + item_definition
				definition = definition.strip()
				if definition[-1] != "." and definition[-1] != " " and len(item.uses) == 0 :
					definition += ". "
				elif definition[-1] == "." :
					definition += " "
				#Print usage examples if they exist
				if len(item.uses) > 0:
					definition = definition.rstrip(". ")
					definition += "; \""+item.uses[0]+"\" "

				if not all_types and line_length + len(definition) < (max_length * (page_num -1)):
					line_length += len(definition) +1
					continue
				#Once we've reached the maximum length, stop printing any more
				elif line_length + len(definition) +1 > (max_length * page_num):
					#If we haven't printed anything so far, it's just one really long definition
					if line_length == 0:
						#So truncate it
						definition = definition[:max_length-3] + "..."
						print (definition),
					else:
						num_more += 1
					suppress_print = True
				else:
					print (definition),

				line_length += len(definition) + 1

			#Once we've printed one word of any given type, move on to the next type
			if all_types:
				type_id += 1

		elif all_types and item.word_type is not word_type:
			num_more+=1

	if (suppress_print or all_types) and num_more > 0:
		print ("(" + str(num_more) + " more)")

if __name__ == "__main__":
	main()