search_solr_tools 3.1.2

data/lib/search_solr_tools/selectors/nodc_iso.rb

@@ -0,0 +1,107 @@
+require 'search_solr_tools'
+
+module SearchSolrTools
+  module Selectors
+    # The hash contains keys that should map to the fields in the solr schema,
+    # the keys are called selectors and are in charge of selecting the nodes
+    # from the ISO document, applying the default value if none of the xpaths
+    # resolved to a value and formatting the field.  xpaths and multivalue are
+    # required, default_value and format are optional
+    NODC = {
+      authoritative_id: {
+        xpaths: ['.//gmd:fileIdentifier/gco:CharacterString'],
+        multivalue: false
+      },
+      title: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:title/gco:CharacterString'],
+        multivalue: false
+      },
+      summary: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:abstract/gco:CharacterString'],
+        multivalue: false
+      },
+      data_centers: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::DATA_CENTER_NAMES[:NODC][:long_name]],
+        multivalue: false
+      },
+      authors: {
+        xpaths: [".//gmd:CI_ResponsibleParty[./gmd:role/gmd:CI_RoleCode[@codeListValue='principalInvestigator']]/gmd:individualName/gco:CharacterString"],
+        multivalue: true
+      },
+      keywords: {
+        xpaths: ['.//gmd:descriptiveKeywords/gmd:MD_Keywords/gmd:keyword/gco:CharacterString',
+                 './/gmd:descriptiveKeywords/gmd:MD_Keywords/gmd:keyword/gmx:Anchor'],
+        multivalue: true
+      },
+      last_revision_date: {
+        xpaths: ['.//gmd:dateStamp/gco:Date'],
+        default_values: [Helpers::SolrFormat.date_str(DateTime.now)], # formats the date into ISO8601 as in http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/DateField.html
+        multivalue: false,
+        format: Helpers::SolrFormat::DATE
+      },
+      dataset_url: {
+        xpaths: ['.//gmd:MD_DigitalTransferOptions/gmd:onLine/gmd:CI_OnlineResource[contains(./gmd:protocol/gco:CharacterString/text(),"ftp")]/gmd:linkage/gmd:URL',
+                 './/gmd:MD_DigitalTransferOptions/gmd:onLine/gmd:CI_OnlineResource[contains(./gmd:protocol/gco:CharacterString/text(),"FTP")]/gmd:linkage/gmd:URL'],
+        multivalue: false
+      },
+      spatial_coverages: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_DISPLAY
+      },
+      spatial: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_INDEX
+      },
+      spatial_area: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: false,
+        reduce: Helpers::IsoToSolrFormat::MAX_SPATIAL_AREA,
+        format: Helpers::IsoToSolrFormat::SPATIAL_AREA
+      },
+      temporal_coverages: {
+        xpaths: ['.//gmd:EX_TemporalExtent'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DISPLAY_STRING_FORMATTED
+      },
+      temporal_duration: {
+        xpaths: ['.//gmd:EX_TemporalExtent'],
+        multivalue: false,
+        reduce: Helpers::SolrFormat::REDUCE_TEMPORAL_DURATION,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DURATION
+      },
+      temporal: {
+        xpaths: ['.//gmd:EX_TemporalExtent'],
+        multivalue: true,
+        format:  Helpers::IsoToSolrFormat::TEMPORAL_INDEX_STRING
+      },
+      sensors: {
+        xpaths: ['.//gmi:acquisitionInformation/gmi:MI_AcquisitionInformation/gmi:instrument/gmi:MI_Instrument/gmi:citation/gmd:CI_Citation/gmd:title/gco:CharacterString'],
+        multivalue: true
+      },
+      source: {
+        xpaths: [''],
+        default_values: ['ADE'],
+        multivalue: false
+      },
+      facet_data_center: {
+        xpaths: [''],
+        default_values: ["#{Helpers::SolrFormat::DATA_CENTER_NAMES[:NODC][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:NODC][:short_name]}"],
+        multivalue: false
+      },
+      facet_spatial_scope: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::FACET_SPATIAL_SCOPE
+      },
+      facet_temporal_duration: {
+        xpaths: ['.//gmd:EX_TemporalExtent'],
+        default_values: [Helpers::SolrFormat::NOT_SPECIFIED],
+        format: Helpers::IsoToSolrFormat::FACET_TEMPORAL_DURATION,
+        multivalue: true
+      }
+    }
+  end
+end

data/lib/search_solr_tools/selectors/pdc_iso.rb

@@ -0,0 +1,108 @@
+require 'search_solr_tools'
+
+module SearchSolrTools
+  module Selectors
+    # The hash contains keys that should map to the fields in the solr schema,
+    # the keys are called selectors and are in charge of selecting the nodes
+    # from the ISO document, applying the default value if none of the xpaths
+    # resolved to a value and formatting the field. xpaths and multivalue are
+    # required, default_value, format, and reduce are optional.
+    #
+    # reduce takes the formatted result of multiple nodes and produces a single
+    #   result. This is for fields that are not multivalued, but their value
+    #   should consider information from all the nodes (for example, storing
+    #   only the maximum duration from multiple temporal coverage fields, taking
+    #   the sum of multiple spatial areas)
+    PDC = {
+      authoritative_id: {
+        xpaths: ['.//oai:header/oai:identifier'],
+        multivalue: false
+      },
+      title: {
+        xpaths: ['.//gmd:citation//gmd:title/gco:CharacterString'],
+        multivalue: false
+      },
+      summary: {
+        xpaths: ['.//gmd:abstract/gco:CharacterString'],
+        multivalue: false
+      },
+      data_centers: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::DATA_CENTER_NAMES[:PDC][:long_name]],
+        multivalue: false
+      },
+      authors: {
+        xpaths: ['.//gmd:identificationInfo//gmd:citedResponsibleParty//gmd:individualName/gco:CharacterString'],
+        multivalue: true
+      },
+      keywords: {
+        xpaths: ['.//gmd:descriptiveKeywords//gmd:keyword/gco:CharacterString'],
+        multivalue: true
+      },
+      last_revision_date: {
+        xpaths: ['.//oai:header/oai:datestamp'],
+        default_values: [Helpers::SolrFormat.date_str(DateTime.now)], # formats the date into ISO8601 as in http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/DateField.html
+        multivalue: false,
+        format: Helpers::SolrFormat::DATE
+      },
+      dataset_url: {
+        xpaths: ['.//gmd:dataSetURI/gco:CharacterString'],
+        multivalue: false,
+        format: Helpers::SolrFormat::HTTP_URL_FORMAT
+      },
+      spatial_coverages: {
+        xpaths: ['.//gmd:extent//gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_DISPLAY
+      },
+      spatial: {
+        xpaths: ['.//gmd:extent//gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_INDEX
+      },
+      spatial_area: {
+        xpaths: ['.//gmd:extent//gmd:EX_GeographicBoundingBox'],
+        multivalue: false,
+        reduce: Helpers::IsoToSolrFormat::MAX_SPATIAL_AREA,
+        format: Helpers::IsoToSolrFormat::SPATIAL_AREA
+      },
+      temporal: {
+        xpaths: ['.//gmd:EX_TemporalExtent/gmd:extent'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_INDEX_STRING
+      },
+      temporal_coverages: {
+        xpaths: ['.//gmd:EX_TemporalExtent/gmd:extent'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DISPLAY_STRING
+      },
+      temporal_duration: {
+        xpaths: ['.//gmd:EX_TemporalExtent/gmd:extent'],
+        multivalue: false,
+        reduce: Helpers::SolrFormat::REDUCE_TEMPORAL_DURATION,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DURATION
+      },
+      source: {
+        xpaths: [''],
+        default_values: ['ADE'],
+        multivalue: false
+      },
+      facet_data_center: {
+        xpaths: [''],
+        default_values: ["#{Helpers::SolrFormat::DATA_CENTER_NAMES[:PDC][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:PDC][:short_name]}"],
+        multivalue: false
+      },
+      facet_spatial_scope: {
+        xpaths: ['.//gmd:extent//gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::FACET_SPATIAL_SCOPE
+      },
+      facet_temporal_duration: {
+        xpaths: ['.//gmd:EX_TemporalExtent/gmd:extent'],
+        default_values: [Helpers::SolrFormat::NOT_SPECIFIED],
+        format: Helpers::IsoToSolrFormat::FACET_TEMPORAL_DURATION,
+        multivalue: true
+      }
+    }
+  end
+end

data/lib/search_solr_tools/selectors/rda.rb

@@ -0,0 +1,106 @@
+require 'search_solr_tools'
+
+module SearchSolrTools
+  module Selectors
+    # The hash contains keys that should map to the fields in the solr schema,
+    # the keys are called selectors and are in charge of selecting the nodes
+    # from the ISO document, applying the default value if none of the xpaths
+    # resolved to a value and formatting the field.  xpaths and multivalue are
+    # required, default_value and format are optional.
+    RDA = {
+      authoritative_id: {
+        xpaths: ['.//oai:header/oai:identifier'],
+        multivalue: false
+      },
+      title: {
+        xpaths: ['.//dif:Entry_Title'],
+        multivalue: false
+      },
+      summary: {
+        xpaths: ['.//dif:Summary'],
+        multivalue: false
+      },
+      data_centers: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::DATA_CENTER_NAMES[:RDA][:long_name]],
+        multivalue: false
+      },
+      authors: {
+        xpaths: [''],
+        multivalue: true
+      },
+      keywords: {
+        xpaths: [
+          './/dif:Parameters/dif:Category',
+          './/dif:Parameters/dif:Topic',
+          './/dif:Parameters/dif:Term',
+          './/dif:Parameters/dif:Variable_Level_1'
+        ].reverse,
+        multivalue: true
+      },
+      last_revision_date: {
+        xpaths: ['.//dif:Last_DIF_Revision_Date'],
+        default_values: [Helpers::SolrFormat.date_str(DateTime.now)], # formats the date into ISO8601 as in http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/DateField.html
+        multivalue: false,
+        format: Helpers::SolrFormat::DATE
+      },
+      dataset_url: {
+        xpaths: ['.//dif:Related_URL/dif:URL'],
+        multivalue: false
+      },
+      spatial_coverages: {
+        xpaths: ['.//dif:Spatial_Coverage'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_DISPLAY
+      },
+      spatial: {
+        xpaths: ['.//dif:Spatial_Coverage'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_INDEX
+      },
+      spatial_area: {
+        xpaths: ['.//dif:Spatial_Coverage'],
+        multivalue: false,
+        reduce: Helpers::IsoToSolrFormat::MAX_SPATIAL_AREA,
+        format: Helpers::IsoToSolrFormat::SPATIAL_AREA
+      },
+      temporal: {
+        xpaths: ['.//dif:Temporal_Coverage'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_INDEX_STRING
+      },
+      temporal_coverages: {
+        xpaths: ['.//dif:Temporal_Coverage'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DISPLAY_STRING
+      },
+      temporal_duration: {
+        xpaths: ['.//dif:Temporal_Coverage'],
+        multivalue: false,
+        reduce: Helpers::SolrFormat::REDUCE_TEMPORAL_DURATION,
+        format: Helpers::IsoToSolrFormat::TEMPORAL_DURATION
+      },
+      source: {
+        xpaths: [''],
+        default_values: ['ADE'],
+        multivalue: false
+      },
+      facet_data_center: {
+        xpaths: [''],
+        default_values: ["#{Helpers::SolrFormat::DATA_CENTER_NAMES[:RDA][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:RDA][:short_name]}"],
+        multivalue: false
+      },
+      facet_spatial_scope: {
+        xpaths: ['.//dif:Spatial_Coverage'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::FACET_SPATIAL_SCOPE
+      },
+      facet_temporal_duration: {
+        xpaths: ['.//dif:Temporal_Coverage'],
+        default_values: [Helpers::SolrFormat::NOT_SPECIFIED],
+        format: Helpers::IsoToSolrFormat::FACET_TEMPORAL_DURATION,
+        multivalue: true
+      }
+    }
+  end
+end

data/lib/search_solr_tools/selectors/tdar_opensearch.rb

@@ -0,0 +1,89 @@
+require 'search_solr_tools'
+
+module SearchSolrTools
+  module Selectors
+    # The hash contains keys that should map to the fields in the solr schema,
+    # the keys are called selectors and are in charge of selecting the nodes
+    # from the ISO document, applying the default value if none of the xpaths
+    # resolved to a value and formatting the field. xpaths and multivalue are
+    # required, default_value and format are optional
+    TDAR = {
+      authoritative_id: {
+        xpaths: ['.//atom:link/@href'],
+        multivalue: false,
+        format: proc do |node|
+          'TDAR-' << node.text.split('/')[4] || ''
+        end
+      },
+      title: {
+        xpaths: ['.//atom:title'],
+        multivalue: false
+      },
+      summary: {
+        xpaths: ['.//atom:summary'],
+        multivalue: false
+      },
+      data_centers: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::DATA_CENTER_NAMES[:TDAR][:long_name]],
+        multivalue: false
+      },
+      authors: {
+        xpaths: ['.//atom:author/atom:name'],
+        multivalue: true
+      },
+      keywords: {
+        xpaths: [''],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::KEYWORDS
+      },
+      last_revision_date: {
+        xpaths: ['.//atom:updated'],
+        default_values: [Helpers::SolrFormat.date_str(DateTime.now)], # formats the date into ISO8601 as in http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/DateField.html
+        multivalue: false,
+        format: Helpers::SolrFormat::DATE
+      },
+      dataset_url: {
+        xpaths: ['.//atom:link/@href'],
+        multivalue: false
+      },
+      spatial_coverages: {
+        xpaths: ['.//georss:box'],
+        multivalue: true,
+        format: Helpers::TdarFormat::SPATIAL_DISPLAY
+      },
+      spatial: {
+        xpaths: ['.//georss:box'],
+        multivalue: true,
+        format: Helpers::TdarFormat::SPATIAL_INDEX
+      },
+      spatial_area: {
+        xpaths: ['.//georss:box'],
+        multivalue: false,
+        reduce: Helpers::TdarFormat::MAX_SPATIAL_AREA,
+        format: Helpers::TdarFormat::SPATIAL_AREA
+      },
+      source: {
+        xpaths: [''],
+        default_values: ['ADE'],
+        multivalue: false
+      },
+      facet_data_center: {
+        xpaths: [''],
+        default_values: ["#{Helpers::SolrFormat::DATA_CENTER_NAMES[:TDAR][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:TDAR][:short_name]}"],
+        multivalue: false
+      },
+      facet_spatial_scope: {
+        xpaths: ['.//georss:box'],
+        multivalue: true,
+        format: Helpers::TdarFormat::FACET_SPATIAL_SCOPE
+      },
+      facet_temporal_duration: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::NOT_SPECIFIED],
+        format: Helpers::IsoToSolrFormat::FACET_TEMPORAL_DURATION,
+        multivalue: true
+      }
+    }
+  end
+end

data/lib/search_solr_tools/selectors/usgs_iso.rb

@@ -0,0 +1,105 @@
+require 'search_solr_tools'
+
+module SearchSolrTools
+  module Selectors
+    # The hash contains keys that should map to the fields in the solr schema,
+    # the keys are called selectors and are in charge of selecting the nodes
+    # from the ISO document, applying the default value if none of the xpaths
+    # resolved to a value and formatting the field. xpaths and multivalue are
+    # required, default_value and format are optional
+    USGS = {
+      authoritative_id: {
+        xpaths: ['.//gmd:fileIdentifier/gco:CharacterString'],
+        multivalue: false
+      },
+      title: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:title/gco:CharacterString'],
+        multivalue: false
+      },
+      summary: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:abstract/gco:CharacterString'],
+        multivalue: false
+      },
+      data_centers: {
+        xpaths: [''],
+        default_values: [Helpers::SolrFormat::DATA_CENTER_NAMES[:USGS][:long_name]],
+        multivalue: false
+      },
+      authors: {
+        xpaths: [".//gmd:contact/gmd:CI_ResponsibleParty[./gmd:role/gmd:CI_RoleCode[@codeListValue='originator']]/gmd:organisationName/gco:CharacterString"],
+        multivalue: true
+      },
+      keywords: {
+        xpaths: ['.//gmd:descriptiveKeywords/gmd:MD_Keywords/gmd:keyword/gco:CharacterString'],
+        multivalue: true
+      },
+      last_revision_date: {
+        xpaths: ['.//gmd:dateStamp/gco:DateTime'],
+        default_values: [Helpers::SolrFormat.date_str(DateTime.now)], # formats the date into ISO8601 as in http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/DateField.html
+        multivalue: false,
+        format: Helpers::SolrFormat::DATE
+      },
+      dataset_url: {
+        xpaths: ['.//gmd:transferOptions/gmd:MD_DigitalTransferOptions/gmd:onLine/gmd:CI_OnlineResource[contains(./gmd:name/gco:CharacterString/text(),"Summary")]/gmd:linkage/gmd:URL'],
+        multivalue: false
+      },
+      spatial_coverages: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_DISPLAY
+      },
+      spatial: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::SPATIAL_INDEX
+      },
+      spatial_area: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: false,
+        reduce: Helpers::IsoToSolrFormat::MAX_SPATIAL_AREA,
+        format: Helpers::IsoToSolrFormat::SPATIAL_AREA
+      },
+      temporal: {
+        xpaths: [".//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:date/gmd:CI_Date[./gmd:dateType/gmd:CI_DateTypeCode[@codeListValue='Time Period']]/gmd:date"],
+        multivalue: true,
+        format: Helpers::UsgsFormat::TEMPORAL_INDEX_STRING
+      },
+      temporal_coverages: {
+        xpaths: [".//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:date/gmd:CI_Date[./gmd:dateType/gmd:CI_DateTypeCode[@codeListValue='Time Period']]/gmd:date"],
+        multivalue: true,
+        format: Helpers::UsgsFormat::TEMPORAL_DISPLAY_STRING
+      },
+      temporal_duration: {
+        xpaths: [".//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:date/gmd:CI_Date[./gmd:dateType/gmd:CI_DateTypeCode[@codeListValue='Time Period']]/gmd:date"],
+        multivalue: false,
+        reduce: Helpers::SolrFormat::REDUCE_TEMPORAL_DURATION,
+        format: Helpers::UsgsFormat::TEMPORAL_DURATION
+      },
+      sensors: {
+        xpaths: [''],
+        multivalue: true
+      },
+      source: {
+        xpaths: [''],
+        default_values: ['ADE'],
+        multivalue: false
+      },
+      facet_data_center: {
+        xpaths: [''],
+        default_values: ["#{Helpers::SolrFormat::DATA_CENTER_NAMES[:USGS][:long_name]} | #{Helpers::SolrFormat::DATA_CENTER_NAMES[:USGS][:short_name]}"],
+        multivalue: false
+      },
+      facet_spatial_scope: {
+        xpaths: ['.//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:extent/gmd:EX_Extent/gmd:geographicElement/gmd:EX_GeographicBoundingBox'],
+        multivalue: true,
+        format: Helpers::IsoToSolrFormat::FACET_SPATIAL_SCOPE
+      },
+      facet_temporal_duration: {
+        xpaths: [".//gmd:identificationInfo/gmd:MD_DataIdentification/gmd:citation/gmd:CI_Citation/gmd:date/gmd:CI_Date[./gmd:dateType/gmd:CI_DateTypeCode[@codeListValue='Time Period']]/gmd:date"],
+        default_values: [Helpers::SolrFormat::NOT_SPECIFIED],
+        format: Helpers::UsgsFormat::FACET_TEMPORAL_DURATION,
+        multivalue: true
+      }
+    }
+  end
+end