misc/pythonScraper/scraper.py

#!/usr/bin/python

# Requireed python modules: 
# * Mechanize
# * BeatifulSoup
# apt-get install python-mechanize python-beautifulsoup beep


from mechanize import Browser
from mechanize import _response
from BeautifulSoup import BeautifulSoup

import time
import sys
import os

savedVal1 = 0

class CheckDevice:

        def __init__(self): #constructor
                self.savedVal1 = 0

        def beep(self):
                os.system("beep -l 200 -r 2")

        def printTime(self):
                os.system("date");

        def checkVars(self):
                br = Browser()
                try:
                        br.open("http://t-hoerup.dk/test.html")

                        soup = BeautifulSoup( br.response().read() )
                
                        rows = soup.findAll('tr')

                        for row in rows:
                                key = row.contents[1].string
                                val = row.contents[3].string
#                               print "key ", key
#                               print "val ", val

                                if (key == "Key1"):
                                        if (val != self.savedVal1):
                                                self.beep()
                                                self.printTime()
                                                print "Val1 has changed %s -> %s\n" % ( str(self.savedVal1), str(val) )
                                                self.savedVal1 = val
                                                                                

                except:
                        print "Error, probably a 404 : ", sys.exc_info()
        
#               print "----------------------------- done-------------------------"


def main():

        cd = CheckDevice()

        #dumb continous main-loop abort with ctrl+c and ignore the error output
        while (1):
                cd.checkVars()
                time.sleep(1) #seconds


#Finally, launch main function
main()
1	#!/usr/bin/python
2
3	# Requireed python modules:
4	# * Mechanize
5	# * BeatifulSoup
6	# apt-get install python-mechanize python-beautifulsoup beep
7
8
9
10	from mechanize import Browser
11	from mechanize import _response
12	from BeautifulSoup import BeautifulSoup
13
14	import time
15	import sys
16	import os
17
18	savedVal1 = 0
19
20	class CheckDevice:
21
22	def __init__(self): #constructor
23	self.savedVal1 = 0
24
25	def beep(self):
26	os.system("beep -l 200 -r 2")
27
28	def printTime(self):
29	os.system("date");
30
31	def checkVars(self):
32	br = Browser()
33	try:
34	br.open("http://t-hoerup.dk/test.html")
35
36	soup = BeautifulSoup( br.response().read() )
37
38	rows = soup.findAll('tr')
39
40	for row in rows:
41	key = row.contents[1].string
42	val = row.contents[3].string
43	# print "key ", key
44	# print "val ", val
45
46	if (key == "Key1"):
47	if (val != self.savedVal1):
48	self.beep()
49	self.printTime()
50	print "Val1 has changed %s -> %s\n" % ( str(self.savedVal1), str(val) )
51	self.savedVal1 = val
52
53
54
55	except:
56	print "Error, probably a 404 : ", sys.exc_info()
57
58	# print "----------------------------- done-------------------------"
59
60
61
62
63	def main():
64
65	cd = CheckDevice()
66
67	#dumb continous main-loop abort with ctrl+c and ignore the error output
68	while (1):
69	cd.checkVars()
70	time.sleep(1) #seconds
71
72
73	#Finally, launch main function
74	main()