In [1]:
from owslib.csw import CatalogueServiceWeb
from owslib.fes import SortBy, SortProperty
from owslib import fes
import datetime as dt
In [2]:
#csw = CatalogueServiceWeb('http://www.ngdc.noaa.gov/geoportal/csw',timeout=60) # NGDC Geoportal
csw = CatalogueServiceWeb('http://data.noaa.gov/csw',timeout=60)
In [3]:
csw.get_operation_by_name('GetRecords').constraints
Out[3]:
In [4]:
# adjust to match MaxRecordDefault of CSW, if would be cleaner if we pick this up Capabilities XML
# this issue will allow for this: https://github.com/geopython/OWSLib/issues/211
pagesize = 10
sort_property = 'dc:title' # a supported queryable of the CSW
sort_order = 'ASC' # should be 'ASC' or 'DESC'
In [5]:
sortby = SortBy([SortProperty(sort_property, sort_order)])
foo=sortby.properties
In [6]:
# hopefully something like this will be implemented in fes soon
def dateRange(start_date='1900-01-01',stop_date='2100-01-01',constraint='overlaps'):
if constraint == 'overlaps':
start = fes.PropertyIsLessThanOrEqualTo(propertyname='apiso:TempExtent_begin', literal=stop_date)
stop = fes.PropertyIsGreaterThanOrEqualTo(propertyname='apiso:TempExtent_end', literal=start_date)
elif constraint == 'within':
start = fes.PropertyIsGreaterThanOrEqualTo(propertyname='apiso:TempExtent_begin', literal=start_date)
stop = fes.PropertyIsLessThanOrEqualTo(propertyname='apiso:TempExtent_end', literal=stop_date)
return start,stop
In [7]:
val = 'salinity'
box=[-72.0, 41.0, -69.0, 43.0] # gulf of maine
In [8]:
# specific specific times (UTC) ...
# hurricane sandy
jd_start = dt.datetime(2012,10,26)
jd_stop = dt.datetime(2012,11,2)
# 2014 feb 10-15 storm
jd_start = dt.datetime(2014,2,10)
jd_stop = dt.datetime(2014,2,15)
# 2014 recent
jd_start = dt.datetime(2014,3,8)
jd_stop = dt.datetime(2014,3,11)
# 2014 recent
jd_start = dt.datetime(1988,1,1)
jd_stop = dt.datetime(1988,3,1)
# 2011
#jd_start = dt.datetime(2013,4,20)
#jd_stop = dt.datetime(2013,4,24)
# ... or relative to now
#jd_now = dt.datetime.utcnow()
#jd_start = jd_now - dt.timedelta(days=3)
#jd_stop = jd_now + dt.timedelta(days=3)
start_date = jd_start.strftime('%Y-%m-%d %H:00')
stop_date = jd_stop.strftime('%Y-%m-%d %H:00')
jd_start = dt.datetime.strptime(start_date,'%Y-%m-%d %H:%M')
jd_stop = dt.datetime.strptime(stop_date,'%Y-%m-%d %H:%M')
print start_date,'to',stop_date
In [9]:
start,stop = dateRange(start_date,stop_date)
filter1 = fes.PropertyIsLike(propertyname='apiso:AnyText',literal=('*%s*' % val),
escapeChar='\\',wildCard='*',singleChar='?')
bbox = fes.BBox(box,crs='urn:ogc:def:crs:OGC:1.3:CRS84')
#filter_list = [fes.And([ bbox, filter1, start,stop]) ]
In [10]:
filter_list = [fes.And([ bbox, filter1]) ]
csw.getrecords2(constraints=filter_list)
csw.results['matches']
Out[10]:
In [11]:
filter_list = [fes.And([ bbox, filter1, start,stop]) ]
csw.getrecords2(constraints=filter_list)
csw.results['matches']
Out[11]:
In [12]:
filter_list = [filter1]
csw.getrecords2(constraints=filter_list)
csw.results['matches']
Out[12]:
In [13]:
filter_list = [fes.And([ bbox, filter1]) ]
startposition = 0
maxrecords = 20
while True:
print 'getting records %d to %d' % (startposition, startposition+pagesize)
csw.getrecords2(constraints=filter_list,
startposition=startposition, maxrecords=pagesize, sortby=sortby)
# print csw.request
for rec,item in csw.records.iteritems():
print item.title
if csw.results['nextrecord'] == 0:
break
startposition += pagesize
if startposition >= maxrecords:
break