Mercurial > hg > horiscopes
comparison V4/runme.py @ 14:a0c217ee4168
current edits
author | DaveM |
---|---|
date | Sun, 04 Mar 2018 15:03:15 +0000 |
parents | b253748dbb11 |
children | 50a95089414d |
comparison
equal
deleted
inserted
replaced
13:b253748dbb11 | 14:a0c217ee4168 |
---|---|
12 import pickle | 12 import pickle |
13 from HTMLParser import HTMLParser | 13 from HTMLParser import HTMLParser |
14 # from lxml import html | 14 # from lxml import html |
15 from bs4 import BeautifulSoup | 15 from bs4 import BeautifulSoup |
16 | 16 |
17 def parsePage(horiscope, resp): | 17 def parsePage(resp): |
18 horiscope = syn.planetPositions() | 18 horiscope = syn.planetPositions() |
19 soup = BeautifulSoup(resp.content, 'lxml') | 19 soup = BeautifulSoup(resp.content, 'lxml') |
20 tcCell = soup.find_all('div', attrs={'class':'right-sedy-banner-svetlejsi'}) | 20 tcCell = soup.find_all('div', attrs={'class':'right-sedy-banner-svetlejsi'}) |
21 for cell in tcCell: | 21 for cell in tcCell: |
22 divList = cell.find_all('div') | 22 divList = cell.find_all('div') |
23 for i in range(len(divList)): | 23 for i in range(len(divList)): |
24 planetName = divList[i].getText().lower() | 24 planetName = divList[i].getText().lower() |
25 if planetName in planetPositions.planetNames: | 25 if planetName in planetPositions.planetNames: |
26 horiscope.planets[planetName].setLocation(divList[i+2].getText(),divList[i+4].getText()) | 26 horiscope.planets[planetName].setLocation(divList[i+2].getText(),divList[i+4].getText()) |
27 return horiscope | |
27 | 28 |
28 def makePeople(filename): | 29 def makePeople(filename): |
29 stream = csv.DictReader(open(filename,'rb')) | 30 stream = csv.DictReader(open(filename,'rb')) |
30 dictList = [] | 31 dictList = [] |
31 people = [] | 32 people = [] |
97 parseTries = 3 | 98 parseTries = 3 |
98 while parseTries > 0: | 99 while parseTries > 0: |
99 try: | 100 try: |
100 person.makePayload() | 101 person.makePayload() |
101 resp = requestURL(person.url,person.payload) | 102 resp = requestURL(person.url,person.payload) |
102 parsePage(person.horiscope,resp) | 103 person.horiscope = parsePage(resp) |
103 pdb.set_trace() | 104 pdb.set_trace() |
104 parseTries = 0 | 105 parseTries = 0 |
105 except: | 106 except: |
106 print sys.exc_info()[0] | 107 print sys.exc_info()[0] |
107 parseTries -= 1 | 108 parseTries -= 1 |