IOOS_archive_CSW_search.ipynb

{
 "cells": [
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "# IOOS Archive CSW search"
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": true
   },
   "cell_type": "code",
   "source": "from owslib.csw import CatalogueServiceWeb\nfrom owslib import fes\nimport numpy as np\nfrom pprint import pprint\nfrom datetime import datetime, timedelta",
   "execution_count": 1,
   "outputs": []
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": true
   },
   "cell_type": "code",
   "source": "endpoint = 'https://data.nodc.noaa.gov/geoportal/csw'",
   "execution_count": 2,
   "outputs": []
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": false
   },
   "cell_type": "code",
   "source": "csw = CatalogueServiceWeb(endpoint,timeout=60)\nprint(csw.version)",
   "execution_count": 3,
   "outputs": [
    {
     "output_type": "stream",
     "text": "2.0.2\n",
     "name": "stdout"
    }
   ]
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": false
   },
   "cell_type": "code",
   "source": "csw.get_operation_by_name('GetRecords').constraints",
   "execution_count": 4,
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "[Constraint: SupportedCommonQueryables - ['Subject', 'Title', 'Abstract', 'AnyText', 'Format', 'Identifier', 'Modified', 'Type', 'BoundingBox'],\n Constraint: SupportedISOQueryables - ['apiso:Subject', 'apiso:Title', 'apiso:Abstract', 'apiso:AnyText', 'apiso:Format', 'apiso:Identifier', 'apiso:Modified', 'apiso:Type', 'apiso:BoundingBox', 'apiso:CRS.Authority', 'apiso:CRS.ID', 'apiso:CRS.Version', 'apiso:RevisionDate', 'apiso:AlternateTitle', 'apiso:CreationDate', 'apiso:PublicationDate', 'apiso:OrganizationName', 'apiso:HasSecurityConstraints', 'apiso:Language', 'apiso:ResourceIdentifier', 'apiso:ParentIdentifier', 'apiso:KeywordType', 'apiso:TopicCategory', 'apiso:ResourceLanguage', 'apiso:GeographicDescriptionCode', 'apiso:Denominator', 'apiso:DistanceValue', 'apiso:DistanceUOM', 'apiso:TempExtent_begin', 'apiso:TempExtent_end', 'apiso:ServiceType', 'apiso:ServiceTypeVersion', 'apiso:Operation', 'apiso:OperatesOn', 'apiso:OperatesOnIdentifier', 'apiso:OperatesOnName', 'apiso:CouplingType'],\n Constraint: AdditionalQueryables - ['apiso:Degree', 'apiso:AccessConstraints', 'apiso:OtherConstraints', 'apiso:Classification', 'apiso:ConditionApplyingToAccessAndUse', 'apiso:Lineage', 'apiso:ResponsiblePartyRole', 'apiso:SpecificationTitle', 'apiso:SpecificationDate', 'apiso:SpecificationDateType']]"
     },
     "metadata": {},
     "execution_count": 4
    }
   ]
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": true
   },
   "cell_type": "code",
   "source": "def fes_date_filter(start, stop, constraint='overlaps'):\n    \"\"\"\n    Take datetime-like objects and returns a fes filter for date range\n    (begin and end inclusive).\n    NOTE: Truncates the minutes!!!\n    Examples\n    --------\n    >>> from datetime import datetime, timedelta\n    >>> stop = datetime(2010, 1, 1, 12, 30, 59).replace(tzinfo=pytz.utc)\n    >>> start = stop - timedelta(days=7)\n    >>> begin, end = fes_date_filter(start, stop, constraint='overlaps')\n    >>> begin.literal, end.literal\n    ('2010-01-01 12:00', '2009-12-25 12:00')\n    >>> begin.propertyoperator, end.propertyoperator\n    ('ogc:PropertyIsLessThanOrEqualTo', 'ogc:PropertyIsGreaterThanOrEqualTo')\n    >>> begin, end = fes_date_filter(start, stop, constraint='within')\n    >>> begin.literal, end.literal\n    ('2009-12-25 12:00', '2010-01-01 12:00')\n    >>> begin.propertyoperator, end.propertyoperator\n    ('ogc:PropertyIsGreaterThanOrEqualTo', 'ogc:PropertyIsLessThanOrEqualTo')\n    \"\"\"\n    start = start.strftime('%Y-%m-%d %H:00')\n    stop = stop.strftime('%Y-%m-%d %H:00')\n    if constraint == 'overlaps':\n        propertyname = 'apiso:TempExtent_begin'\n        begin = fes.PropertyIsLessThanOrEqualTo(propertyname=propertyname,\n                                                literal=stop)\n        propertyname = 'apiso:TempExtent_end'\n        end = fes.PropertyIsGreaterThanOrEqualTo(propertyname=propertyname,\n                                                 literal=start)\n    elif constraint == 'within':\n        propertyname = 'apiso:TempExtent_begin'\n        begin = fes.PropertyIsGreaterThanOrEqualTo(propertyname=propertyname,\n                                                   literal=start)\n        propertyname = 'apiso:TempExtent_end'\n        end = fes.PropertyIsLessThanOrEqualTo(propertyname=propertyname,\n                                              literal=stop)\n    else:\n        raise NameError('Unrecognized constraint {}'.format(constraint))\n    return begin, end",
   "execution_count": 5,
   "outputs": []
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": false
   },
   "cell_type": "code",
   "source": "#filter 1: search only IOOS data\n# IOOS data may be found by using this free text search (from Yuanjie Li <yuanjie.li@noaa.gov>)\nval = 'Integrated Ocean Observing System Data Assembly Centers Data Stewardship Program'\nfilter1 = fes.PropertyIsLike(propertyname='apiso:AnyText',literal=('*%s*' % val),\n                        escapeChar='\\\\',wildCard='*',singleChar='?')\n\n#filter 2: search only Glider data\nval = 'Glider'\nfilter2 = fes.PropertyIsLike(propertyname='apiso:AnyText',literal=('*%s*' % val),\n                        escapeChar='\\\\',wildCard='*',singleChar='?')\n\n# filter 3: Search only specific region \nbbox = [-87.40, 34.25, -63.70, 66.70]    # [lon_min, lat_min, lon_max, lat_max]\nbbox_filter = fes.BBox(bbox,crs='urn:ogc:def:crs:OGC:1.3:CRS84')\n\n# filter 4: search only in last year\nnow = datetime.utcnow()\nstart, stop = now - timedelta(days=(365)), now + timedelta(days=(0))\nbegin, end = fes_date_filter(start, stop)\n\n# Add all the filters to list:\nfilter_list = [fes.And([filter1, filter2, bbox_filter, begin, end])]\n\n# Query CSW\ncsw.getrecords2(constraints=filter_list,maxrecords=15,esn='full')\n\nprint(len(csw.records.keys()))\nfor rec in list(csw.records.keys()):\n    print(csw.records[rec].title)",
   "execution_count": 6,
   "outputs": [
    {
     "output_type": "stream",
     "text": "9\nPhysical trajectory profile data from glider sp066 deployed by Woods Hole Oceanographic Institution in the East Coast - US/Canada from 2015-12-17 to 2016-04-06 (NCEI Accession 0153789)\nPhysical trajectory profile data from glider blue deployed by University of Massachusetts; University of Massachusetts - Dartmouth in the Mid-Atlantic Bight from 2016-05-18 to 2016-06-06 (NCEI Accession 0153544)\nPhysical trajectory profile data from glider ru30 deployed by Rutgers, the State University of New Jersey in the Mid-Atlantic Bight from 2016-09-01 to 2016-09-26 (NCEI Accession 0156681)\nPhysical trajectory profile data from glider whoi_406 deployed by Woods Hole Oceanographic Institution in the Mid-Atlantic Bight from 2016-09-02 to 2016-09-20 (NCEI Accession 0156640)\nPhysical trajectory profile data from glider ru28 deployed by Rutgers, the State University of New Jersey in the Mid-Atlantic Bight from 2016-09-01 to 2016-09-22 (NCEI Accession 0156659)\nPhysical trajectory profile data from glider ru32 deployed by Rutgers, the State University of New Jersey in the Mid-Atlantic Bight from 2016-08-10 to 2016-08-19 (NCEI Accession 0156390)\nPhysical trajectory profile data from glider ud_134 deployed by University of Delaware in the Mid-Atlantic Bight from 2016-08-30 to 2016-09-07 (NCEI Accession 0156682)\nPhysical trajectory profile data from glider ru28 deployed by Rutgers, the State University of New Jersey in the Mid-Atlantic Bight from 2016-07-14 to 2016-07-20 (NCEI Accession 0156658)\nPhysical trajectory profile data from glider blue deployed by University of Massachusetts; University of Massachusetts - Dartmouth in the Mid-Atlantic Bight from 2016-08-18 to 2016-08-22 (NCEI Accession 0156657)\n",
     "name": "stdout"
    }
   ]
  },
  {
   "metadata": {
    "trusted": true,
    "collapsed": true
   },
   "cell_type": "code",
   "source": "",
   "execution_count": null,
   "outputs": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "name": "conda-env-IOOS3-py",
   "display_name": "Python [conda env:IOOS3]",
   "language": "python"
  },
  "language_info": {
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "nbconvert_exporter": "python",
   "version": "3.5.2",
   "codemirror_mode": {
    "version": 3,
    "name": "ipython"
   },
   "pygments_lexer": "ipython3",
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}