def transform_for_postajob(job):
"""
Cleans a job coming from My.jobs post-a-job. This should add any
required fields, and re-format any fields that are not coming in
in the required format.
inputs:
:job: A dictionary with the following fields: postajob.job.id (id),
city, company.id (company), country, country_short, date_new,
date_updated, description, guid, link, on_sites, state,
state_short, reqid, title, uid, and zipcode.
outputs:
A solr-ready job as a dictionary
"""
try:
company = Company.objects.get(id=job['company'])
except Company.DoesNotExist:
return None
job['date_new'] = _clean_time(job['date_new'])
job['date_updated'] = _clean_time(job['date_updated'])
solr_job = {'is_posted': True}
on_sites = job.get('on_sites', '0')
if not on_sites:
solr_job['on_sites'] = ''
else:
solr_job['on_sites'] = [str(x) for x in on_sites.split(',')]
solr_job['id'] = 'postajob.job.%s' % job['guid']
# This has to be seo.joblisting, otherwise the jobs won't be included
# in the search results.
solr_job['django_ct'] = 'seo.joblisting'
solr_job['django_id'] = 0
solr_job['city_slug'] = slugify(job['city'])
solr_job['country_short'] = job['country_short']
solr_job['date_updated_exact'] = job['date_updated']
solr_job['job_source_name'] = 'Post-a-Job'
solr_job['date_updated'] = job['date_updated']
solr_job['salted_date'] = DEJobFeed.date_salt(job['date_updated'])
solr_job['reqid'] = job['reqid']
solr_job['company_digital_strategies_customer'] = company.digital_strategies_customer
solr_job['guid'] = job['guid']
solr_job['uid'] = job['id']
solr_job['company_member'] = company.member
solr_job['city'] = job['city']
solr_job['date_new'] = job['date_new']
solr_job['country_exact'] = job['country']
solr_job['country_slug'] = slugify(job['country'])
solr_job['company_ac'] = company.name
solr_job['html_description'] = DEJobFeed.markdown_to_html(job['description'])
solr_job['state'] = job['state']
solr_job['country_ac'] = job['country']
solr_job['city_ac'] = job['city']
solr_job['state_short_exact'] = job['state_short']
solr_job['title_ac'] = job['title']
solr_job['company_canonical_microsite'] = company.canonical_microsite
solr_job['description'] = job['description']
solr_job['state_ac'] = job['state']
solr_job['company'] = company.name
solr_job['state_short'] = job['state_short']
solr_job['title_exact'] = job['title']
solr_job['link'] = job.get('link', '')
solr_job['apply_info'] = job.get('apply_info', '')
solr_job['company_enhanced'] = company.enhanced
solr_job['state_slug'] = slugify(job['state'])
solr_job['city_exact'] = job['city']
solr_job['title_slug'] = slugify(job['title'])
solr_job['state_exact'] = job['state']
solr_job['zipcode'] = job['zipcode']
solr_job['title'] = job['title']
solr_job['date_new_exact'] = job['date_new']
solr_job['country'] = job['country']
solr_job['company_exact'] = company.name
solr_job['company_canonical_microsite_exact'] = company.canonical_microsite
solr_job['date_added'] = solr_job['date_added_exact'] = datetime.datetime.now()
# Requires city, state_short, state, and country_short to be filled
# in on solr_job to work.
solr_job['location'] = DEJobFeed.location(solr_job)
# Requires city, state, location, and country to be filled in on
# solr_jobs.
solr_job['full_loc'] = DEJobFeed.full_loc(solr_job)
solr_job['full_loc_exact'] = solr_job['full_loc']
solr_job['company_slab'] = DEJobFeed.co_slab(company.name)
# Requires solr_job['country_short'], solr_job['state'], and
# solr_job['city'] to already be filled in.
solr_job['city_slab'] = DEJobFeed.city_slab(solr_job)
# Requires solr_job['country_short'] and solr_job['state'] to already be
# filled in.
solr_job['state_slab'] = DEJobFeed.state_slab(solr_job)
# Requires solr_job['country_short'] to already be filled in.
solr_job['country_slab'] = DEJobFeed.country_slab(solr_job)
# Requires solr_job['title'] to already be filled in.
solr_job['title_slab'] = DEJobFeed.title_slab(solr_job)
solr_job['location_exact'] = solr_job['location']
solr_job['state_slab_exact'] = solr_job['state_slab']
solr_job['company_slab_exact'] = solr_job['company_slab']
solr_job['country_slab_exact'] = solr_job['country_slab']
solr_job['city_slab_exact'] = solr_job['city_slab']
solr_job['title_slab_exact'] = solr_job['title_slab']
solr_job['text'] = " ".join([force_text((job.get(k)) or "None") for k in
text_fields])
return solr_job
def transform_for_postajob(job):
"""
Cleans a job coming from My.jobs post-a-job. This should add any
required fields, and re-format any fields that are not coming in
in the required format.
inputs:
:job: A dictionary with the following fields: postajob.job.id (id),
city, company.id (company), country, country_short, date_new,
date_updated, description, guid, link, on_sites, state,
state_short, reqid, title, uid, and zipcode.
outputs:
A solr-ready job as a dictionary
"""
try:
company = Company.objects.get(id=job['company'])
except Company.DoesNotExist:
return None
job['date_new'] = _clean_time(job['date_new'])
job['date_updated'] = _clean_time(job['date_updated'])
solr_job = {'is_posted': True}
on_sites = job.get('on_sites', '0')
if not on_sites:
solr_job['on_sites'] = ''
else:
solr_job['on_sites'] = [str(x) for x in on_sites.split(',')]
solr_job['id'] = 'postajob.job.%s' % job['guid']
# This has to be seo.joblisting, otherwise the jobs won't be included
# in the search results.
solr_job['django_ct'] = 'seo.joblisting'
solr_job['django_id'] = 0
solr_job['city_slug'] = slugify(job['city'])
solr_job['country_short'] = job['country_short'].upper()
solr_job['country_short_exact'] = job['country_short'].upper()
solr_job['date_updated_exact'] = job['date_updated']
solr_job['job_source_name'] = 'Post-a-Job'
solr_job['date_updated'] = job['date_updated']
solr_job['salted_date'] = DEJobFeed.date_salt(job['date_new'])
solr_job['reqid'] = job['reqid']
solr_job['company_digital_strategies_customer'] = company.digital_strategies_customer
solr_job['guid'] = job['guid']
solr_job['uid'] = job['id']
solr_job['company_member'] = company.member
solr_job['city'] = job['city']
solr_job['date_new'] = job['date_new']
solr_job['country_exact'] = job['country']
solr_job['country_slug'] = slugify(job['country'])
solr_job['company_ac'] = company.name
solr_job['html_description'] = DEJobFeed.markdown_to_html(job['description'])
solr_job['state'] = job['state']
solr_job['country_ac'] = job['country']
solr_job['city_ac'] = job['city']
solr_job['state_short_exact'] = job['state_short']
solr_job['title_ac'] = job['title']
solr_job['company_canonical_microsite'] = company.canonical_microsite
solr_job['description'] = job['description']
solr_job['state_ac'] = job['state']
solr_job['company'] = company.name
solr_job['state_short'] = job['state_short']
solr_job['title_exact'] = job['title']
solr_job['link'] = job.get('link', '')
solr_job['apply_info'] = job.get('apply_info', '')
solr_job['company_enhanced'] = company.enhanced
solr_job['state_slug'] = slugify(job['state'])
solr_job['city_exact'] = job['city']
solr_job['title_slug'] = slugify(job['title'])
solr_job['state_exact'] = job['state']
solr_job['zipcode'] = job['zipcode']
solr_job['title'] = job['title']
solr_job['date_new_exact'] = job['date_new']
solr_job['country'] = job['country']
solr_job['company_exact'] = company.name
solr_job['company_canonical_microsite_exact'] = company.canonical_microsite
solr_job['date_added'] = solr_job['date_added_exact'] = datetime.datetime.now()
# Requires city, state_short, state, and country_short to be filled
# in on solr_job to work.
solr_job['location'] = DEJobFeed.location(solr_job)
# Requires city, state, location, and country to be filled in on
# solr_jobs.
solr_job['full_loc'] = DEJobFeed.full_loc(solr_job)
solr_job['full_loc_exact'] = solr_job['full_loc']
solr_job['company_slab'] = DEJobFeed.co_slab(company.name)
# Requires solr_job['country_short'], solr_job['state'], and
# solr_job['city'] to already be filled in.
solr_job['city_slab'] = DEJobFeed.city_slab(solr_job)
# Requires solr_job['country_short'] and solr_job['state'] to already be
# filled in.
solr_job['state_slab'] = DEJobFeed.state_slab(solr_job)
# Requires solr_job['country_short'] to already be filled in.
solr_job['country_slab'] = DEJobFeed.country_slab(solr_job)
# Requires solr_job['title'] to already be filled in.
solr_job['title_slab'] = DEJobFeed.title_slab(solr_job)
solr_job['location_exact'] = solr_job['location']
solr_job['state_slab_exact'] = solr_job['state_slab']
solr_job['company_slab_exact'] = solr_job['company_slab']
solr_job['country_slab_exact'] = solr_job['country_slab']
solr_job['city_slab_exact'] = solr_job['city_slab']
solr_job['title_slab_exact'] = solr_job['title_slab']
solr_job['all_locations'] = [job['zipcode'], job['city'], job['state'], job['state_short'],
"%s, %s" % (job['city'], job['state']), job['country']]
solr_job['text'] = " ".join([force_text((job.get(k)) or "None") for k in
text_fields])
return solr_job
# Requires city, state_short, state, and country_short to be filled
# in on job to work.
job['location'] = DEJobFeed.location(job)
# Requires city, state, location, and country to be filled in on
# jobs.
job['full_loc'] = DEJobFeed.full_loc(job)
job['full_loc_exact'] = job['full_loc']
job['company_slab'] = DEJobFeed.co_slab(company.name)
# Requires job['country_short'], job['state'], and
# job['city'] to already be filled in.
job['city_slab'] = DEJobFeed.city_slab(job)
# Requires job['country_short'] and job['state'] to already be
# filled in.
job['state_slab'] = DEJobFeed.state_slab(job)
# Requires job['country_short'] to already be filled in.
job['country_slab'] = DEJobFeed.country_slab(job)
# Requires job['title'] to already be filled in.
job['title_slab'] = DEJobFeed.title_slab(job)
job['location_exact'] = job['location']
job['state_slab_exact'] = job['state_slab']
job['company_slab_exact'] = job['company_slab']
job['country_slab_exact'] = job['country_slab']
job['city_slab_exact'] = job['city_slab']
job['title_slab_exact'] = job['title_slab']
onets = [node.text for node in data.findall('.//JobCategoryCode')]
onets = set(DEJobFeed.clean_onet(onet) for onet in onets)
job['onet'] = job['onet_exact'] = list(onets)
# Standard Mocs
# Requires city, state_short, state, and country_short to be filled
# in on job to work.
job['location'] = DEJobFeed.location(job)
# Requires city, state, location, and country to be filled in on
# jobs.
job['full_loc'] = DEJobFeed.full_loc(job)
job['full_loc_exact'] = job['full_loc']
job['company_slab'] = DEJobFeed.co_slab(company.name)
# Requires job['country_short'], job['state'], and
# job['city'] to already be filled in.
job['city_slab'] = DEJobFeed.city_slab(job)
# Requires job['country_short'] and job['state'] to already be
# filled in.
job['state_slab'] = DEJobFeed.state_slab(job)
# Requires job['country_short'] to already be filled in.
job['country_slab'] = DEJobFeed.country_slab(job)
# Requires job['title'] to already be filled in.
job['title_slab'] = DEJobFeed.title_slab(job)
job['location_exact'] = job['location']
job['state_slab_exact'] = job['state_slab']
job['company_slab_exact'] = job['company_slab']
job['country_slab_exact'] = job['country_slab']
job['city_slab_exact'] = job['city_slab']
job['title_slab_exact'] = job['title_slab']
onets = [node.text for node in xml.findall('.//onet')]
onets = set(DEJobFeed.clean_onet(onet) for onet in onets)
job['onet'] = job['onet_exact'] = list(onets)
# Standard Mocs