| import requests | import requests |
| import ckanapi | import ckanapi |
| # copy (select id,url,format,extras from resource where extras like '%"autoupdate": "active"%') TO STDOUT WITH CSV; | |
| # f759e4b6-723c-4863-8a26-1529d689cad8,http://data.gov.au/geoserver/geelong-roofprints-kml/wms?request=GetCapabilities,wms,"{""autoupdate"": ""active""}" | |
| import fileinput | |
| import csv | import csv |
| import sys | |
| import psycopg2 | |
| import json | |
| from subprocess import Popen, PIPE | |
| def updateresource(resource_id): | def updateresource(resource_id, dataset_id): |
| url = 'http://data.disclosurelo.gs' | print ' --- ' |
| api_key = '' | ckan = ckanapi.RemoteCKAN(api_url,api_key) |
| db_credentials = '' | |
| print resource_id | |
| ckan = ckanapi.RemoteCKAN('http://data.disclosurelo.gs') | |
| #ckan = ckanapi.RemoteCKAN('http://demo.ckan.org') | #ckan = ckanapi.RemoteCKAN('http://demo.ckan.org') |
| resource = ckan.action.resource_show(id=resource_id) | resource = ckan.action.resource_show(id=resource_id) |
| print 'updating '+resource['name']+'('+resource_id+', '+dataset_id+')' | |
| print resource | print resource |
| url = resource['url'] | url = resource['url'] |
| #last_modified= 'Mon, 24 Feb 2014 01:48:29 GMT' | #last_modified= 'Mon, 24 Feb 2014 01:48:29 GMT' |
| #etag='"1393206509.38-638"' | #etag='"1393206509.38-638"' |
| headers={} | headers={} |
| if 'etag' in resource: | if 'etag' in resource: |
| headers['If-None-Match'] = resource['etag'] | headers['If-None-Match'] = resource['etag'] |
| if 'file_last_modified' in resource: | if 'file_last_modified' in resource: |
| headers["If-Modified-Since"] = resource['file_last_modified'] | headers["If-Modified-Since"] = resource['file_last_modified'] |
| print headers | |
| r = requests.head(url, headers=headers) | r = requests.head(url, headers=headers) |
| if r.status_code == 304: | if r.status_code == 304: |
| print 'not modified' | print 'not modified' |
| return | return |
| else: | else: |
| print r.status_code | print r.status_code |
| print r.headers | print r.headers |
| if 'last-modified' in r.headers: | if 'last-modified' in r.headers: |
| resource['file_last_modified'] = r.headers['last-modified'] | resource['file_last_modified'] = r.headers['last-modified'] |
| if 'etag' in r.headers: | if 'etag' in r.headers: |
| resource['etag'] = r.headers['etag'] | resource['etag'] = r.headers['etag'] |
| #save updated resource | #save updated resource |
| # result = ckan.action.resource_update(id,resource) | print resource |
| if 'format' == 'shp': | result = ckan.call_action('resource_update',resource) |
| print "geoingest!" db_settings_json, api_url, api_key, resource['dataset_id'] | if resource['format'].lower() == 'shp' or resource['format'].lower() == 'kml': |
| print "geoingest!" | |
| pargs= ['dga-spatialingestor.py', db_settings_json, api_url, api_key, dataset_id] | |
| print pargs | |
| p = Popen(pargs)#, stdout=PIPE, stderr=PIPE) | |
| p.communicate() | |
| else: | else: |
| ckan.action.datapusher_submit(resource_id=id) | print "datapusher!" |
| # https://github.com/ckan/ckan/blob/master/ckanext/datapusher/logic/action.py#L19 | |
| ckan.action.datapusher_submit(resource_id=resource_id) | |
| if len(sys.argv) != 5: | if len(sys.argv) != 4: |
| print "spatial ingester. command line: postgis_url api_url api_key" | print "autoupdate ingester. command line: postgis_url api_url api_key" |
| sys.exit(errno.EACCES) | sys.exit(-1) |
| else: | else: |
| (path, db_settings_json, api_url, api_key) = sys.argv | (path, db_settings_json, api_url, api_key) = sys.argv |
| db_settings = json.loads(db_settings_json) | db_settings = json.loads(db_settings_json) |
| datastore_db_settings = dict(db_settings) | |
| datastore_db_settings['dbname'] = db_settings['datastore_dbname'] | |
| datastore_db_settings_json = json.dumps(datastore_db_settings) | |
| for line in fileinput.input(): | try: |
| row = csv.reader(line) | conn = psycopg2.connect(dbname=db_settings['dbname'], user=db_settings['user'], password=db_settings['password'], host=db_settings['host']) |
| updateresource(line.replace("\n","")) | except: |
| failure("I am unable to connect to the database.") | |
| # Open a cursor to perform database operations | |
| cur = conn.cursor() | |
| conn.set_isolation_level(0) | |
| cur.execute('select resource.id resource_id, package.id dataset_id from resource inner join resource_group on resource.resource_group_id = resource_group.id inner join package on resource_group.package_id = package.id where resource.extras like \'%"autoupdate": "active"%\';') | |
| row = cur.fetchone() | |
| while row is not None: | |
| updateresource(row[0],row[1]) | |
| # process | |
| row = cur.fetchone() | |
| cur.close() | |
| conn.close() | |