def run(url):
soup = get_soup(url)
jobs_list = soup.find_all('h4')
for job_entry in jobs_list:
globals.job_title = job_entry.a.text
globals.info_link = job_entry.a['href']
globals.job_summary = globals.info_link
listing_soup = get_soup(globals.info_link)
if listing_soup.body.find_all('p', string="Job Type: Full-time"):
globals.full_or_part = 'Full-time'
elif listing_soup.body.find_all('p', string="Job Type: Part-time"):
globals.full_or_part = 'Part-time'
date_text = listing_soup.body.find_all(
'span', {'class': 'subtitle'})[0].text.split()
month_string = date_text[2]
day = int(date_text[3][0:len(date_text[3]) - 1])
year = int(date_text[4])
month = month_to_num(month_string)
globals.job_post_date = datetime(year, month, day)
update_db(organization)
def run(url):
soup = get_soup(url)
job_grid = soup.find('div', {'class': 'wpjb-job-list'})
job_class = Job(organization, "")
job_class.organization_id = organization_id
insert_count = 0
for job_div in job_grid.find_all('div', {'class': 'wpjb-col-main'}):
major_line = job_div.find('div', {'class': 'wpjb-line-major'})
job_class.title = major_line.a.text
job_class.info_link = major_line.a['href']
job_class.full_or_part = major_line.find('span', {
'class': 'wpjb-sub-title'
}).text.strip()
minor_line = job_div.find('div', {'class': 'wpjb-line-minor'})
job_class.location = minor_line.find('span', {
'class': 'wpjb-job_location'
}).text.strip()
date = minor_line.find('span', {
'class': 'wpjb-job_created_at'
}).text.strip().split(', ')
month = month_to_num(date[0])
day = int(date[1])
if month <= datetime.now().month:
year = datetime.now().year
else:
year = datetime.now().year - 1
job_class.post_date = datetime(year, month, day)
insert_count += job_insert(job_class)
return insert_count
def run(url):
soup = get_soup(url)
jobs_list = soup.find_all('h4')
job_class = Job(organization, "")
job_class.organization_id = organization_id
insert_count = 0
for job_entry in jobs_list:
job_class.title = job_entry.a.text
job_class.info_link = job_entry.a['href']
listing_soup = get_soup(job_class.info_link)
if listing_soup.body.find_all('p', string="Job Type: Full-time"):
job_class.full_or_part = 'Full-time'
elif listing_soup.body.find_all('p', string="Job Type: Part-time"):
job_class.full_or_part = 'Part-time'
date_text = listing_soup.body.find_all(
'span', {'class': 'subtitle'})[0].text.split()
month_string = date_text[2]
day = int(date_text[3][0:len(date_text[3]) - 1])
year = int(date_text[4])
month = month_to_num(month_string)
job_class.post_date = datetime(year, month, day)
insert_count += job_insert(job_class)
return insert_count
def run(url):
soup = get_soup(url)
jobs_container = soup.find('dl', {'class': 'employment-opportunities'})
job_class = Job(organization, "")
job_class.info_link = url
job_class.organization_id = organization_id
insert_count = 0
for job_listing in jobs_container.find_all('dt'):
job_heading = job_listing.h3.text.split(' Posted ')
job_class.title = job_heading[0]
job_class.summary = job_listing.p.text
date = job_heading[1].split(' ')
month = month_to_num(date[0])
day = int(date[1][0:len(date[1]) - 1])
year = int(date[2])
job_class.post_date = datetime(year, month, day)
insert_count += job_insert(job_class)
return insert_count
info_link = html_element.a['href']
job_summary = info_link
listing_soup = get_soup(info_link)
if listing_soup.body.find_all('p', string="Job Type: Full-time"):
full_or_part = 'Full-time'
elif listing_soup.body.find_all('p', string="Job Type: Part-time"):
full_or_part = 'Part-time'
date_text = listing_soup.body.find_all(
'span', {'class': 'subtitle'})[0].text.split()
month_string = date_text[2]
day = int(date_text[3][0:len(date_text[3]) - 1])
year = int(date_text[4])
month = month_to_num(month_string)
job_post_date = datetime(year, month, day)
update_db(organization)
reset_vars()
reset_vars()
# Antelope Valley Domestic Violence Council (Valley Oasis)
organization = "Antelope Valley Domestic Violence Council (Valley Oasis)"
soup = get_soup("http://www.valleyoasis.org/job-opportunities.html")
for html_element in soup.find("div", {
"itemtype": "http://schema.org/WebPage"