tools/gcget

   1 #!/usr/bin/env python
   2
   3 #
   4 # gcget -- screen scrape Geocaching.com's annoying web interface
   5 #          aka SHOW ME THE CACHE!!!
   6 #
   7 # Copyright 2007, Evan Battaglia
   8 # Distributed under the terms of the GPL v2.
   9 #
  10 #
  11 # requires module mechanize
  12 #
  13
  14 # DEFAULT USERNAME AND PASSWORD: THESE ARE OVERWRITTEN BY COMMAND-LINE OPTIONS
  15
  16 USER="username"
  17 PASS="password"
  18
  19 # docs needed!
  20 # this has some extra args in:
  21 # gcget lat,lon maxnumgcs [maxdist] [threshold]
  22 # threshold -- if find more than this # of geocaches, don't get ANY,
  23 # instead give warning and quit
  24
  25 import sys
  26 import getopt
  27
  28 def help():
  29    print """gcget v0.1
  30 This program is free software, distributed under the terms of the GNU GPL v2.
  31
  32 Usage: gcget [-u username] [-p password] lat,lon maxnumberofgcs [maxdistance] [threshold]
  33 Downloads up to maxnumberofgcs at a distance of up to maxdistance from lat,lon.
  34 If we number of geocaches within maxdistance is above threshold, don't download any
  35   geocaches, just give a warning and quit.
  36
  37 If username and password are not given, will use default values hard-coded in script.
  38
  39 Happy caching!!!
  40 """
  41
  42 #
  43 # PARSE OPTIONS: USERNAME AND PASSWORD
  44 #
  45 #
  46 #
  47 try:
  48   opts, args = getopt.gnu_getopt(sys.argv[1:], "u:p:d", ["help"])
  49 except getopt.GetoptError:
  50   # print help information and exit:
  51   help()
  52   sys.exit(2)
  53
  54 DEBUG = False
  55
  56 for o, a in opts:
  57   if o == "-p":
  58     PASS = a
  59   if o == "-u":
  60     USER = a
  61   if o == "--help" or o == "-h":
  62     help()
  63     sys.exit()
  64   if o == "-d":
  65     DEBUG = True
  66
  67 if len(args) < 2:
  68   help()
  69   sys.exit()
  70
  71 #########################
  72
  73 ll = args[0].split(",")
  74 lat = ll[0]
  75 lon = ll[1]
  76
  77 if len(args) >= 3:
  78   maxdist = args[2]
  79 else:
  80   maxdist = "999"
  81
  82 if len(args) >= 4:
  83   threshold = int(args[3])
  84 else:
  85   threshold = 1000000;
  86
  87 # rounds up to multiples of 20. 20
  88 n = int((int(args[1])+19)/20)
  89
  90 import re
  91 from mechanize import Browser
  92 import ClientForm
  93
  94 # get magic number for "Next" button.
  95 # this is normally 16 (link hidden is $ctl16), unless there are less than 10 pages of results,
  96 # in which case it will be less (e.g. 09 for 3 pages of results)
  97 def getmagicnumber(b):
  98   for i in range(16,0,-1):
  99     if re.compile("pgrBottom.ctl%02d" % i).search(b.response().get_data()):
 100       return i
 101   return None
 102
 103 b=Browser()
 104 b.open("http://geocaching.com/seek/")
 105 b.follow_link(text="Log in")
 106 b.select_form(nr=0)
 107 b["myUsername"] = USER
 108 b["myPassword"] = PASS
 109 b.submit()
 110
 111 magicnumber = 0 # the ctl number of Next. get only once
 112
 113 try: b.select_form("form4")
 114 except: pass
 115 try:
 116   b.select_form("form4")
 117 except:
 118   b.select_form("form4")
 119   print >> sys.stderr, "Invalid username/password"
 120   if DEBUG:
 121     f=open("gcget.badlogin.html","w")
 122     f.write(b.response().get_data())
 123     f.close()
 124     print >> sys.stderr, "Dumping last HTML page recieved into gcget.badlogin.html"
 125   sys.exit()
 126
 127 b["origin_lat"] = lat
 128 b["origin_long"] = lon
 129 b["dist"] = maxdist
 130 b.submit()
 131
 132 thresholdre = re.compile("Total Records: <b>([0-9]*)</b>")
 133 m = thresholdre.search(b.response().get_data())
 134 if m:
 135   if int(m.group(1)) > threshold:
 136     sys.stderr.write("THRESHOLD %d > %d\n" % (int(m.group(1)), threshold))
 137     sys.exit(4)
 138   else:
 139     records = int(m.group(1))
 140     sys.stderr.write("ok found %d, getting min(%d,%d) gcs\n" % (int(m.group(1)), int(records), int(args[1])))
 141 else:
 142   print "can't find total records"
 143   sys.exit(0)
 144
 145 pages = 0
 146 # (records+19)/20 is the max pages
 147 for ii in range(min(n,(records+19)/20)):
 148   try:
 149     b.select_form(nr=0)
 150     b['CID'] = [i.name for i in b.find_control('CID').items]
 151     b.submit()
 152   except:
 153     break
 154
 155   # only print one header, start of xml file
 156   lines = b.response().get_data().split("\n")
 157   if ii == 0:
 158     print "\n".join(lines[0:2])
 159
 160   # core
 161   print "\n".join(lines[2:-1])
 162
 163   print "</waypoint>"
 164
 165   pages += 1
 166   sys.stderr.write("i")
 167   sys.stderr.flush()
 168
 169   b.back()
 170
 171   if not magicnumber:
 172     magicnumber = getmagicnumber(b)
 173     if not magicnumber:
 174 #      print "couldn't find magic number!" # why does this happen?
 175       break
 176
 177   b.select_form(nr=0)
 178   [f for f in b.forms()][0].new_control("hidden", "ctl00$ContentBody$pgrBottom$ctl%02d" % magicnumber, {})
 179   b.submit()
 180
 181 sys.stderr.write("\n")
 182
 183 if pages:
 184   print "</loc>"
 185
 186 #  f=open("delmeNOW","w")
 187 #  f.write(b.response().get_data())
 188 #  f.close()
 189