def Main():
'''Wrapper.'''
try:
#
# Collecting data from UNHCR.
#
print '%s Collecting data from UNHCR.' % item('bullet')
data = Fetch.Fetch()
#
# Patching data.
# Epoch time doesn't seem to be 1970.
#
print '%s Patching data.' % item('bullet')
# pdata = Patch.Epoch(data)
pdata = Patch.Date(data)
#
# Storing data in database.
#
print '%s Storing records in database.' % item('bullet')
CleanTable('monthly_arrivals_by_country')
StoreRecords(pdata, 'monthly_arrivals_by_country')
print '%s Collected data from UNHCR successfully.' % item('success')
scraperwiki.status('ok')
except Exception as e:
print '%s UNHCR Collector failed.' % item('error')
scraperwiki.status('error', 'Collection failed.')
def Main():
'''Wrapper.'''
try:
#
# Collecting data from UNHCR.
#
print '%s Collecting data from UNHCR.' % item('bullet')
data = Fetch.Fetch()
#
# Patching data.
# Epoch time doesn't seem to be 1970.
#
print '%s Patching data.' % item('bullet')
# pdata = Patch.Epoch(data)
pdata = Patch.Date(data)
#
# Storing data in database.
#
print '%s Storing records in database.' % item('bullet')
CleanTable('monthly_arrivals_by_country')
StoreRecords(pdata, 'monthly_arrivals_by_country')
print '%s Collected data from UNHCR successfully.' % item('success')
scraperwiki.status('ok')
except Exception as e:
print '%s UNHCR Collector failed.' % item('error')
scraperwiki.status('error', 'Collection failed.')
def collect():
'''Scrapes and stores data in database.'''
#
# Collecting URLs from OpenNepal.
#
print '%s Collecting dataset URLs from OpenNepal.' % item('bullet')
urls = []
for page in range(0, 1):
data = Scraper.ScrapeURLs(page=page)
urls += data
#
# Storing URLs.
#
CleanTable('opennepal_urls')
StoreRecords(urls, 'opennepal_urls')
#
# Scrape content.
#
errors = 0
content = []
print '%s Scraping datasets.' % item('bullet')
for url in urls:
try:
c = Scraper.ScrapeContent(url=url['url'])
content.append(c)
except Exception as e:
errors += 1
print '%s Error scraping dataset: %s' % (item('error'), url['url'])
print e
print '%s There were a total of %s error(s) scraping data.' % (item('warn'), str(errors))
#
# Storing content.
#
CleanTable('opennepal_content')
StoreRecords(content, 'opennepal_content')
print '%s Collected data from OpenNepal successfully.\n' % item('success')
return content
def export_json(data):
'''Exports scraped data to JSONs in disk.'''
print '%s Exporting datasets JSON to disk.' % item('bullet')
#
# Default directory.
#
d = os.path.split(os.path.dirname(os.path.realpath(__file__)))[0]
data_dir = os.path.join(d, 'data')
#
# Calling JSON generators
# to the default dir.
#
ExportDatasets(data, data_dir)
ExportResources(data, data_dir)
print '%s Successfully exported JSON files.\n' % item('success')
def patch(data):
'''Patching data.'''
print '%s Patching data.' % item('bullet')
out = []
for record in data:
record['id'] = Patcher.Slug(record['title'])
record['dataset_date'] = Patcher.Date(record['date'])
out.append(record)
#
# Storing content.
#
CleanTable('opennepal_content')
StoreRecords(out, 'opennepal_content')
print '%s Patched data successfully.\n' % item('success')
return out
def main(development=False):
'''Wrapper.'''
data = collect()
try:
#
# Either collect data or use
# previously collected data from
# database.
#
if development is False:
data = collect()
pdata = patch(data)
else:
cursor = scraperwiki.sqlite.execute('SELECT * FROM opennepal_content')
pdata = []
for record in cursor['data']:
pdata.append(dict(zip(cursor['keys'], record)))
#
# Create static JSON files.
#
export_json(data=pdata)
scraperwiki.status('ok')
#
# Send notification if scraper fails.
#
except Exception as e:
print '%s OpenNepal Scraper failed.' % item('error')
print e
scraperwiki.status('error', 'Collection failed.')
os.system("mail -s 'OpenNepal: Scraper failed.' [email protected]")
