@eeacms/volto-globalsearch 0.1.0

package/src/config/facets.js

@@ -0,0 +1,298 @@
+import {
+  booleanFacet,
+  dateRangeFacet,
+  fixedRangeFacet,
+  histogramFacet,
+  makeRange,
+  multiTermFacet,
+} from '@eeacms/search';
+import spatialWhitelist from './json/spatialWhitelist.json';
+import placesBlacklist from './json/placesBlacklist.json';
+import topicsBlacklist from './json/topicsBlacklist.json';
+import { getTodayWithTime } from '../utils';
+
+const facets = [
+  booleanFacet(() => ({
+    field: 'IncludeArchived',
+    label: 'Include archived content',
+    id: 'archived-facet',
+    showInFacetsList: false,
+    showInSecondaryFacetsList: true,
+    isFilter: true, // filters don't need facet options to show up
+
+    // we want this to be applied by default
+    // when the facet is checked, then apply the `on` key:
+    off: {
+      constant_score: {
+        filter: {
+          bool: {
+            should: [
+              { bool: { must_not: { exists: { field: 'expires' } } } },
+              // Functions should be supported in the buildFilters
+              { range: { expires: { gte: getTodayWithTime() } } },
+            ],
+          },
+        },
+      },
+    },
+    on: null,
+  })),
+  multiTermFacet({
+    field: 'moreLikeThis',
+    isFilterable: true,
+    isMulti: false,
+    label: 'More like this',
+    showInFacetsList: false,
+    filterListComponent: 'MoreLikeThisEntry',
+    factory: 'MoreLikeThis',
+    condition: 'like',
+    queryParams: {
+      fields: ['title', 'text'],
+      min_term_freq: 1,
+      max_query_terms: 12,
+    },
+
+    // registryConfig: 'MoreLikeThis',
+  }),
+  multiTermFacet({
+    field: 'topic',
+    isFilterable: true,
+    isMulti: true,
+    label: 'Topics',
+    blacklist: topicsBlacklist,
+    factory: 'MultiTermListFacet',
+    wrapper: 'ModalFacetWrapper',
+    show: 10000,
+    showAllOptions: true, // show all options (even if 0) in modal facet
+    // factory: 'sui.Facet',
+  }),
+  multiTermFacet({
+    field: 'spatial',
+    isFilterable: true,
+    isMulti: true,
+    label: 'Countries',
+    whitelist: spatialWhitelist,
+    wrapper: 'ModalFacetWrapper',
+    show: 10000,
+    factory: 'MultiTermListFacet',
+    iconsFamily: 'Countries',
+    enableExact: true,
+    sortOn: 'value',
+  }),
+  multiTermFacet({
+    field: 'op_cluster',
+    isFilterable: true,
+    isMulti: true,
+    label: 'Section',
+    wrapper: 'ModalFacetWrapper',
+    show: 10000,
+    factory: 'MultiTermListFacet',
+    showInFacetsList: false,
+  }),
+
+  multiTermFacet({
+    field: 'places',
+    isFilterable: true,
+    isMulti: true,
+    label: 'Regions/Places/Cities/Seas...',
+    blacklist: placesBlacklist,
+    wrapper: 'ModalFacetWrapper',
+    show: 10000,
+    factory: 'MultiTermListFacet',
+    showAllOptions: true, // show all options (even if 0) in modal facet
+  }),
+  multiTermFacet({
+    field: 'objectProvides',
+    isFilterable: false,
+    isMulti: true,
+    label: 'Content types',
+    iconsFamily: 'Content types',
+    //whitelist: objectProvidesWhitelist,
+    wrapper: 'ModalFacetWrapper',
+    factory: 'MultiTermListFacet',
+    optionsFilter: 'typesForClustersOptionsFilter',
+  }),
+  multiTermFacet({
+    field: 'cluster_name',
+    isFilterable: false,
+    isMulti: true,
+    label: 'Sources',
+    wrapper: 'ModalFacetWrapper',
+    factory: 'MultiTermListFacet',
+    iconsFamily: 'Sources',
+  }),
+  histogramFacet({
+    wrapper: 'ModalFacetWrapper',
+    field: 'year',
+    // isFilterable: false,
+    isMulti: true,
+    label: 'Publishing year',
+    // TODO: implement split in buckets
+    ranges: makeRange({
+      step: 1,
+      normalRange: [1994, 2022],
+      includeOutlierStart: false,
+      includeOutlierEnd: false,
+    }),
+    step: 10,
+    // [
+    //   {
+    //     to: 1900,
+    //   },
+    //   {
+    //     key: '2001-2010',
+    //     from: 2001,
+    //     to: 2010,
+    //   },
+    //   {
+    //     from: 2011,
+    //   },
+    // ]
+    // min_max_script:
+    //
+    //"def vals = doc['year']; if (vals.length == 0){return 2000} else {def ret = [];for (val in vals){def tmp_val = val.substring(0,4);ret.add(tmp_val.toLowerCase() == tmp_val.toUpperCase() ? Integer.parseInt(tmp_val) : 2000);}return ret;}",
+
+    aggs_script:
+      "def vals = doc['year']; if (vals.length == 0){return 2500} else {def ret = [];for (val in vals){def tmp_val = val.substring(0,4);ret.add(tmp_val.toLowerCase() == tmp_val.toUpperCase() ? Integer.parseInt(tmp_val) : 2500);}return ret;}",
+  }),
+
+  histogramFacet({
+    wrapper: 'ModalFacetWrapper',
+    field: 'time_coverage',
+    // isFilterable: false,
+    isMulti: true,
+    label: 'Time coverage',
+    // TODO: implement split in buckets
+    ranges: makeRange({
+      step: 10,
+      normalRange: [1700, 2210],
+      includeOutlierStart: false,
+      includeOutlierEnd: false,
+    }),
+    step: 10,
+    // [
+    //   {
+    //     to: 1900,
+    //   },
+    //   {
+    //     key: '2001-2010',
+    //     from: 2001,
+    //     to: 2010,
+    //   },
+    //   {
+    //     from: 2011,
+    //   },
+    // ]
+    // min_max_script:
+    //
+    //"def vals = doc['year']; if (vals.length == 0){return 2000} else {def ret = [];for (val in vals){def tmp_val = val.substring(0,4);ret.add(tmp_val.toLowerCase() == tmp_val.toUpperCase() ? Integer.parseInt(tmp_val) : 2000);}return ret;}",
+
+    aggs_script:
+      "def vals = doc['time_coverage']; if (vals.length == 0){return 2500} else {def ret = [];for (val in vals){def tmp_val = val.substring(0,4);ret.add(tmp_val.toLowerCase() == tmp_val.toUpperCase() ? Integer.parseInt(tmp_val) : 2500);}return ret;}",
+  }),
+
+  fixedRangeFacet({
+    wrapper: 'ModalFacetWrapper',
+    field: 'readingTime',
+    label: 'Reading time (minutes)',
+    rangeType: 'fixed',
+    isMulti: true,
+    ranges: [
+      { key: 'All' },
+      { from: 0, to: 4.99999, key: 'Short (<5 minutes)' },
+      { from: 5, to: 24.9999, key: 'Medium (5-25 minutes)' },
+      { from: 25, to: 10000, key: 'Large (25+ minutes)' },
+      //        { to: -0.0001, key: 'Unknown' },
+    ],
+    factory: 'ModalFixedRangeFacet',
+    default: {
+      values: [{ name: 'All', rangeType: 'fixed' }],
+      type: 'any',
+    },
+  }),
+  dateRangeFacet({
+    field: 'issued.date',
+    label: ' ',
+    isFilter: true, // filters don't need facet options to show up
+    showInFacetsList: false,
+    showInSecondaryFacetsList: true,
+    // rangeType: 'dateRange',
+    isMulti: false,
+    ranges: [
+      { key: 'All time' },
+      { key: 'Last week', from: 'now-1w', to: 'now' },
+      { key: 'Last month', from: 'now-1m', to: 'now' },
+      { key: 'Last 3 months', from: 'now-3m', to: 'now' },
+      { key: 'Last year', from: 'now-1y', to: 'now' },
+      { key: 'Last 2 years', from: 'now-2y', to: 'now' },
+    ],
+    factory: 'DropdownRangeFilter',
+    default: {
+      values: ['All time'],
+      type: 'any',
+    },
+  }),
+  multiTermFacet({
+    wrapper: 'ModalFacetWrapper',
+    field: 'language',
+    isFilterable: false,
+    isMulti: true,
+    label: 'Language',
+    default: {
+      values: ['en'],
+      type: 'any',
+    },
+    factory: 'MultiTermListFacet',
+    facetValues: [
+      'ar',
+      'sr',
+      'sq',
+      'bg',
+      'bs',
+      'cs',
+      'hr',
+      'da',
+      'nl',
+      'el',
+      'en',
+      'et',
+      'fi',
+      'fr',
+      'ga',
+      'de',
+      'hu',
+      'is',
+      'it',
+      'lv',
+      'lt',
+      'mk',
+      'mt',
+      'no',
+      'pl',
+      'pt',
+      'ro',
+      'ru',
+      'sh',
+      'sk',
+      'sl',
+      'es',
+      'sv',
+      'tr',
+    ],
+    sortOn: 'custom',
+    sortOnCustomLabel: 'Alphabetical',
+  }),
+];
+
+export default {
+  facets,
+};
+
+// default filter values
+// TODO: this needs to be removed and the configuration plugged into each of
+// the facets
+// readingTime: {
+//   value: { name: 'All', rangeType: 'fixed' },
+//   type: 'any',
+// },
+// },

package/src/config/filters.js

@@ -0,0 +1,15 @@
+import { getTodayWithTime } from '../utils';
+
+export default {
+  // filter values that are always added to the ES requests
+  permanentFilters: [
+    { term: { hasWorkflowState: 'published' } },
+    () => ({
+      constant_score: {
+        filter: {
+          range: { 'issued.date': { lte: getTodayWithTime() } },
+        },
+      },
+    }),
+  ],
+};

package/src/config/global-search-config.js

@@ -0,0 +1,118 @@
+import { build_runtime_mappings } from '../utils';
+import clusterParams, { clusters } from './clusters';
+
+import facets from './facets';
+import views from './views';
+import query from './query';
+import filters from './filters';
+import download from './download';
+import vocabs from './vocabulary';
+
+import objectProvidesWhitelist from './json/objectProvidesWhitelist.json';
+import spatialWhitelist from './json/spatialWhitelist.json';
+import placesBlacklist from './json/placesBlacklist.json';
+import typesWhitelist from './json/typesWhitelist.json';
+
+const globalSearchConfig = {
+  title: 'Global search and catalogue',
+  layoutComponent: 'FilterAsideLayout',
+  contentBodyComponent: 'FilterAsideContentView',
+  enableNLP: true, // enables NLP capabilities
+  facetsListComponent: 'VerticalCardsModalFacets',
+  runtime_mappings: build_runtime_mappings(clusters),
+  useSearchPhrases: false,
+  searchAsYouType: false,
+
+  ...vocabs,
+  ...facets,
+  ...views,
+  ...query,
+  ...filters,
+  ...download,
+  ...clusterParams,
+
+  sourceExcludedFields: ['embedding'], // don't need these in results
+  // 'fulltext',
+
+  // these are used to "clean" the values in the results.
+  field_filters: {
+    type: {
+      whitelist: typesWhitelist,
+    },
+    objectProvides: {
+      whitelist: objectProvidesWhitelist,
+    },
+    spatial: {
+      whitelist: spatialWhitelist,
+    },
+    places: {
+      blacklist: placesBlacklist,
+    },
+  },
+
+  highlight: {
+    queryParams: {
+      fragment_size: 200,
+      number_of_fragments: 3,
+    },
+    fields: ['description.highlight'],
+    queryBuilder: {
+      factory: 'highlightQueryBuilder',
+    },
+  },
+
+  sortOptions: [
+    {
+      name: 'Title a-z',
+      value: 'title.index',
+      direction: 'asc',
+    },
+    {
+      name: 'Title z-a',
+      value: 'title.index',
+      direction: 'desc',
+    },
+    {
+      name: 'Oldest',
+      value: 'issued.date',
+      direction: 'asc',
+    },
+    {
+      name: 'Newest',
+      value: 'issued.date',
+      direction: 'desc',
+    },
+  ],
+
+  defaultPromptQueries: [
+    'what is the status of our forests',
+    'what is PFAS?',
+    'how does PFAS get into human?',
+    'what is the cumulative surface area of the Natura 2000?',
+    'what percentage of europe population is connected to waste water treatement?',
+    'How is the status of fish species in Europe?',
+    'What is the BISE?',
+    'how much does transport contributes to GHG emissions in EU?',
+    'Which cities in Europe have the worst air quality?',
+    'What car is best for the environment?',
+    'What transport mode is best for the environment?',
+    'What EU legislation safeguards our water?',
+    'What is land accounting?',
+    'What countries had the highest land take in the EEA-39 between 2000 and 2018?',
+    'Why is urban sprawl bad?',
+    'what are controlled substances',
+    'what is plastic',
+    'What year did car manufacturers meet their binding emissions target?',
+    'What is the percentage of surface water bodies with less than good status?',
+    'What is the most common pollutant in water bodies in Europe?',
+    'Where can I access greenhouse gas data',
+    'What is the trend on greenhouse gas emissions from transport',
+    'what is the best transport mode',
+    'what are ecosystem services',
+    'bathing water quality',
+    'greenhouse gas emissions trends',
+    'waste generation trends',
+  ],
+};
+
+export default globalSearchConfig;

package/src/config/index.js

@@ -0,0 +1,92 @@
+import { suiFacet, mergeConfig } from '@eeacms/search';
+import { getGlobalsearchThumbUrl, getGlobalsearchIconUrl } from './../utils';
+import { typesForClustersOptionsFilter } from './clusters';
+
+// import WebsiteFilterListComponent from './../components/WebsiteFilterListComponent';
+
+import contentTypeNormalize from './json/contentTypeNormalize.json';
+import globalSearchConfig from './global-search-config.js';
+
+const getClientProxyAddress = () => {
+  const url = new URL(window.location);
+  url.pathname = '';
+  url.search = '';
+  return url.toString();
+};
+
+export default function install(config) {
+  // console.log(process.env.RAZZLE_ENV_CONFIG);
+  const envConfig = process.env.RAZZLE_ENV_CONFIG
+    ? JSON.parse(process.env.RAZZLE_ENV_CONFIG)
+    : globalSearchConfig;
+
+  const pjson = require('../../package.json');
+  envConfig.app_name = pjson.name;
+  envConfig.app_version = pjson.version;
+
+  config.searchui.globalsearch = {
+    ...mergeConfig(envConfig, config.searchui.default),
+    elastic_index: 'es',
+    host: process.env.RAZZLE_ES_PROXY_ADDR || 'http://localhost:3000',
+  };
+
+  if (typeof window !== 'undefined') {
+    config.searchui.globalsearch.host =
+      process.env.RAZZLE_ES_PROXY_ADDR || getClientProxyAddress();
+  }
+
+  config.searchui.standalone = {
+    ...mergeConfig(envConfig, config.searchui.default),
+    host: process.env.RAZZLE_ES_PROXY_ADDR,
+    elastic_index: 'es',
+    facets: [],
+    highlight: {},
+    sortOptions: [],
+    tableViewParams: {
+      columns: [
+        {
+          title: 'Title',
+          field: 'title',
+        },
+      ],
+    },
+    listingViewParams: {
+      titleField: 'title',
+      extraFields: [],
+      details: {
+        titleField: 'title',
+        extraFields: [],
+      },
+      sections: [],
+    },
+  };
+
+  config.resolve.getGlobalsearchIconUrl = getGlobalsearchIconUrl(
+    contentTypeNormalize,
+  );
+  config.resolve.getGlobalsearchThumbUrl = getGlobalsearchThumbUrl(
+    contentTypeNormalize,
+  );
+
+  config.resolve.typesForClustersOptionsFilter = typesForClustersOptionsFilter;
+  // config.resolve.WebsiteFilterListComponent = {
+  //   component: WebsiteFilterListComponent,
+  // };
+  // config.resolve.FreshnessRangeFacet = {
+  //   ...config.resolve.FixedRangeFacet,
+  //   component: (props) => <div>Mumu</div>,
+  // };
+
+  // config.resolve.LandingPage = { component: LandingPage };
+
+  config.searchui.minimal = mergeConfig(config.searchui.default, envConfig);
+  config.searchui.minimal.facets = [
+    suiFacet({ field: 'Sector' }),
+    suiFacet({
+      field: 'Origin_of_the_measure',
+      label: 'Origin of the measure',
+    }),
+  ];
+
+  return config;
+}

package/src/config/json/contentTypeNormalize.json

@@ -0,0 +1,56 @@
+{
+  "highlight": "highlight",
+  "press-release": "pressrelease",
+  "news": "highlight",
+  "news-item": "news-item",
+  "event": "event",
+  "promotion": "generic",
+  "article": "article",
+  "eco-tip": "ecotip",
+  "image": "generic",
+  "video": "cloudvideo",
+  "report": "report",
+  "publication": "report",
+  "dataset": "data",
+  "data-(rdf)": "data",
+  "data": "data",
+  "data-visualization": "davizvisualization",
+  "indicator-specification": "specification",
+  "indicator-factsheet": "assessment",
+  "indicator-assessment": "assessment",
+  "infographic": "interactive-data",
+  "briefing": "fiche",
+  "page": "document",
+  "link": "link",
+  "data-file": "datafile",
+  "assessment-part": "assessmentpart",
+  "file": "file",
+  "eea-job-vacancy": "eeavacancy",
+  "epub-file": "epubfile",
+  "external-data-reference": "externaldataspec",
+  "eyewitness-story": "generic",
+  "figure": "eeafigure",
+  "figure-file": "eeafigurefile",
+  "folder": "folder",
+  "gis-map-application": "gis-application",
+  "methodology-reference": "generic",
+  "organization": "organisation",
+  "organisation": "organisation",
+  "policy-question": "policyquestion",
+  "policy-document": "policydocumentreference",
+  "rationale-reference": "rationalereference",
+  "soer-key-fact": "soerkeyfact",
+  "soer-message": "soermessage",
+  "sparql": "sparql",
+  "data-table-via-sparql": "sparql",
+  "speech": "news-item",
+  "text": "document",
+  "work-item": "generic",
+  "collection---old-style ": "topic",
+  "Catalogue---listing": "topic",
+  "legislation-instrument": "legislation-instrument",
+  "legislation": "legislation-instrument",
+  "legislation-data-reporting": "legislation-instrument",
+  "reporting-obligation": "reporting-obligation",
+  "country-country": "country-profile"
+}

package/src/config/json/objectProvidesWhitelist.json

@@ -0,0 +1,56 @@
+[
+  "Article",
+  "EEA Job Vacancy",
+  "Briefing (all)",
+  "Briefing (Country)",
+  "News",
+  "Video",
+  "Web page",
+  "Map (interactive)",
+  "Infographic",
+  "Data receiver (Reportnet client)",
+  "Data visualisation",
+  "Data (at EEA)",
+  "Data table (via SPARQL)",
+  "Dashboard",
+  "Topic/subtopic site",
+  "Graph (image)",
+  "Map (image)",
+  "Publication / Report",
+  "Microdata (file)",
+  "Data provider (organisation)",
+  "Catalogue / listing",
+  "Indicator assessment",
+  "Data (external links)",
+  "Key message",
+  "Website main section",
+  "Organisation",
+  "Legislation Instrument",
+  "Legislation",
+  "Legislation (Data reporting)",
+  "References and links",
+  "Environmental question",
+  "Reporting obligation",
+  "Contract opportunity",
+  "Glossary term",
+  "Country profile",
+  "Country fact sheet",
+  "Webpage",
+  "File",
+  "Figure (chart/map)",
+  "Report",
+  "Indicator",
+  "Chart (interactive)",
+  "External data reference",
+  "Research and knowledge project",
+  "FAQ",
+  "Briefing",
+  "Information portal",
+  "Guidance",
+  "Case study",
+  "Tool",
+  "Topic page",
+  "Adaptation option",
+  "Collection",
+  "Data set"
+]