Mercurial > ~darius > hgwebdir.cgi > scrape-vb
diff scrape-vb.py @ 6:9f3eb9a07966
Add config parser.
Add ability to print out an example config file.
Add -f option to read a saved page.
author | darius |
---|---|
date | Wed, 29 Aug 2007 07:37:59 +0000 |
parents | 275603a8e2ae |
children | bf896507faa9 |
line wrap: on
line diff
--- a/scrape-vb.py Tue Aug 28 02:58:50 2007 +0000 +++ b/scrape-vb.py Wed Aug 29 07:37:59 2007 +0000 @@ -6,7 +6,7 @@ # Prints out (and emails) when criteria match based on cost, # destination, etc # -# $Id: scrape-vb.py,v 1.4 2007/08/28 02:58:50 darius Exp $ +# $Id: scrape-vb.py,v 1.5 2007/08/29 07:37:59 darius Exp $ ############################################################################ # # Copyright (C) 2007 Daniel O'Connor. All rights reserved. @@ -34,8 +34,37 @@ # ############################################################################ -import os, re, BeautifulSoup, datetime, time, smtplib, sys, urllib, ConfigParser +import os, re, BeautifulSoup, datetime, time, smtplib, sys, urllib +import ConfigParser, optparse + +usage = '''%prog [options] +Reads configuration from ./scrape-vb.ini and ~/.scrape-vb.ini''' + +optparse = optparse.OptionParser(usage, version="$Id: scrape-vb.py,v 1.5 2007/08/29 07:37:59 darius Exp $") +optparse.add_option('-d', '--debug', action="store_true", default=False, + help="Disable mail sending, prints mail message to stdout") +optparse.add_option('-f', '--file', help="Do not fetch the page, use this file instead") +optparse.add_option('-e', '--example', action="store_true", default=False, + help="Print an example configuration file to stdout and exit") +(options, args) = optparse.parse_args() +if (options.example): + print '''[global] +mailsubj="Subject line for emails" +# The following 3 options are necessary before email will be sent +mailfrom=user@host.com +mailsend=True +mailhost=mail.server.com + +[user@host.com] +# All fields are optional +city1=Foo +city2=Bar +when=dd/mm/yy +maxcost=123 +''' + sys.exit(0) + parsetitle = re.compile('([a-z ]+) - ([a-z ]+) \$([0-9]+)', re.IGNORECASE) parsetper = re.compile('Travel Period: ([0-9]+/[0-9]+/[0-9]+) - ([0-9]+/[0-9]+/[0-9]+)', re.IGNORECASE) @@ -50,8 +79,10 @@ conf.read(conflist) try: - #f = open("vb-happyhour.html") - f = urllib.urlopen(conf.get('global', 'vburl')) + if (options.file != None): + f = open(options.file) + else: + f = urllib.urlopen(conf.get('global', 'vburl')) except IOError, e: print "Unable to fetch page - " + str(e) sys.exit(1) @@ -90,7 +121,7 @@ for email in conf.sections(): if (email == 'global'): continue - + # Stuff configuration into a dictionary for our convenience t = {'email' : email} for i in conf.items(email): t[i[0]] = i[1] @@ -128,6 +159,10 @@ mailfrom = conf.get('global', 'mailfrom') except ConfigParser.NoOptionError: mailsend = False + +if (options.debug == True and mailsend): + print "mailsend overridden due to debugging" + mailsend = False if (mailsend): server = smtplib.SMTP(mailhost)