#!/usr/bin/env python
# copyright Alain Tesio alain@onesite.org
# GPL license

# Tested with Python 2.1.1

# Last updated: Mon Jan 24 21:11:52 CET 2005

import sys,string,httplib,re,os

# Change these settings and create the cachedir
cachedir="/var/cache/say"
playcommand="play %s"


regexp1=re.compile("<a href=\"javascript:popWin\('(/cgi-bin/audio.pl\?[^']*)'")
regexp2=re.compile("<br>\* <A HREF=\"http://([^/]*)(/.*\.wav)\">Click here to listen")
server="www.m-w.com"

def error(err):
	sys.stderr.write("%s\n" % err)
	sys.exit(1)

def out(s):
	sys.stdout.write(s)

def wget(host,path):
	out("Getting '%s' at host '%s' ... " % (path,host))
	h=httplib.HTTP(host)
	h.putrequest('GET',path)
	h.endheaders()
	errcode, errmsg, headers = h.getreply()
	if errcode!=200:
		out("Error\n")
	f = h.getfile()
	data = f.read()
	f.close()
	out("size=%i\n" % len(data))
	return data

def wget_pattern(host,path,pattern):
	body=wget(host,path)
	for line in body.split("\n"):
		matchobject=pattern.search(line)
		if matchobject:
			return matchobject.groups()

def check_file_from_word(word):
	intermdir=word.upper()[:2]
	if len(intermdir)==1:
		intermdir+="A"
	wavdir="%s/%s" % (cachedir,intermdir)
	if not(os.path.isdir(wavdir)):
		os.mkdir(wavdir)
	localpath="%s/%s.wav" % (wavdir,word.upper())
	if not(os.path.isfile(localpath)):
		try:
			wavurl1=wget_pattern(server,"/cgi-bin/dictionary?book=Dictionary&va=%s" % word,regexp1)[0]
		except TypeError:
			error("Word '%s' not found" % word)
		try:
			(server2,url2)=wget_pattern(server,wavurl1,regexp2)
		except TypeError:
			error("Error looking up the second url")
		wavdata=wget(server2,url2)
		if len(wavdata)<20 or wavdata.find("was not found")!=-1:
			error("Wav file not found")
		wavfile=open(localpath,"w")
		wavfile.write(wavdata)
		wavfile.close()
		os.chmod(localpath,0644)
	return localpath

if len(sys.argv)==1:
	error("Usage: %s word" % sys.argv[0])
else:
	word=sys.argv[1]
	localpath=check_file_from_word(word)
	os.system(playcommand % localpath)