Merge pull request #125 from NYPL-discovery/pb/add-standard-number-se…

…arch Add standard number search
NYPL · Oct 16, 2018 · 3ef4359 · 3ef4359
2 parents 4adcc9c + 7a0fc9f
commit 3ef4359
Show file tree

Hide file tree

Showing 40 changed files with 12,749 additions and 683 deletions.
diff --git a/data/contexts/resource.json b/data/contexts/resource.json
@@ -36,6 +36,7 @@
     "idBarcode": "dcterms:identifier",
     "idBnum": "dcterms:identifier",
     "identifier": "dcterms:identifier",
+    "identifierStatus": "bf:identifierStatus",
     "idCallnum": "dcterms:identifier",
     "idCatnyp": "dcterms:identifier",
     "idDcc": "dcterms:identifier",

diff --git a/lib/jsonld_serializers.js b/lib/jsonld_serializers.js
@@ -75,20 +75,18 @@ class JsonLdItemSerializer extends JsonLdSerializer {
       if (Array.isArray(v)) return v.map(formatVal)
       else if (v && typeof v === 'object') {
         var formatted = v
-        // Reassign id to @id
-        if (formatted.id) {
-          formatted = Object.assign({ '@id': v.id }, formatted)
-          delete formatted.id
-        }
+        // Reassign id to @id, type to @type, value to @value
+        ; ['id', 'type', 'value'].forEach((specialProp) => {
+          if (formatted[specialProp]) {
+            formatted[`@${specialProp}`] = formatted[specialProp]
+            delete formatted[specialProp]
+          }
+        })
         // TODO need to correct this in the indexer, not here
         if (formatted.label) {
           formatted.prefLabel = formatted.label
           delete formatted.label
         }
-        if (formatted.type) {
-          formatted['@type'] = formatted.type
-          delete formatted.type
-        }
         return formatted
       } else return v
     }
@@ -219,8 +217,6 @@ class ResourceSerializer extends JsonLdItemSerializer {
   statements () {
     var stmts = JsonLdItemSerializer.prototype.statements.call(this)
 
-    stmts.identifier = stmts.identifier.map(this._ensureIdentifierIsUrnStyle)
-
     if (this.body.parentUri) stmts.memberOf = R.flatten([util.eachValue(this.body.parentUri, (id) => ({ '@type': 'nypl:Resource', '@id': `res:${id}` }))])
 
     // Parse all contributor_(aut|ill|...) statements:
@@ -284,8 +280,6 @@ class ItemResourceSerializer extends JsonLdItemSerializer {
   statements () {
     var stmts = JsonLdItemSerializer.prototype.statements.call(this)
 
-    stmts.identifier = stmts.identifier.map(this._ensureIdentifierIsUrnStyle)
-
     if (stmts.identifier) {
       // Add idNyplSourceId convenience property by parsing identifiers that match urn:[source]:[id]
       this.body.identifier

diff --git a/lib/resources.js b/lib/resources.js
@@ -66,6 +66,9 @@ const SEARCH_SCOPES = {
   },
   callnumber: {
     fields: ['shelfMark']
+  },
+  standard_number: {
+    fields: ['shelfMark', 'identifierV2.value', 'uri', 'identifier', 'items.identifierV2.value']
   }
 }
 
@@ -470,33 +473,80 @@ const escapeQuery = function (str) {
 }
 
 /**
- *  Given GET params, returns a plainobject suitable for use in a ES query.
+ * For a set of parsed request params, returns a plainobject representing the
+ * keyword aspects of the ES query.
+ *
+ * @param {object} params - A hash of request params including `search_scope`, `q`
+ *
+ * @return {object} ES query object suitable to be POST'd to ES endpoint
  */
-const buildElasticQuery = function (params) {
+const buildElasticQueryForKeywords = function (params) {
   // Fill these with our top-level clauses:
-  var shoulds = []
-
-  // clean up params
-  ;['q'].forEach(function (param) {
-    if (params[param]) {
-      params[param] = params[param].replace(/date:/g, 'dateStartYear:')
-      params[param] = params[param].replace(/location:/g, 'locations:')
-      params[param] = params[param].replace(/subject:/g, 'subjectLiteral:')
+  const should = []
+
+  // We have an array of fields to match.
+  // Seperate the root-level fields from nested fields by building an object like this:
+  //   {
+  //     _root: [ 'fieldName1', 'fieldName2' ],
+  //     nestedName1: { 'nestedName1.nestedProperty1', 'nestedName1.nestedProperty2' }
+  //   }
+  const fieldMap = SEARCH_SCOPES[params.search_scope].fields.reduce((map, fieldName) => {
+    // Most fields will be matched at root level:
+    let nestedName = '_root'
+    // Any field starting with the following is a nested field:
+    if (['items'].indexOf(fieldName.split('.').shift()) >= 0) {
+      nestedName = fieldName.split('.').shift()
+    }
+    if (!map[nestedName]) map[nestedName] = []
+    map[nestedName].push(fieldName)
+    return map
+  }, { _root: [] })
+
+  should.push({
+    'query_string': {
+      'fields': fieldMap._root,
+      'query': escapeQuery(params.q),
+      'default_operator': 'AND'
     }
   })
 
-  if (params.q) {
-    shoulds.push({
-      'query_string': {
-        'fields': SEARCH_SCOPES[params.search_scope].fields,
-        // default_field: defaultField,
-        'query': escapeQuery(params.q),
-        'default_operator': 'AND'
-      }
+  // Add nested queries (if any) to things that *should* match:
+  Object.keys(fieldMap)
+    .filter((nestedName) => nestedName !== '_root')
+    .forEach((nestedName) => {
+      should.push({
+        nested: {
+          path: nestedName,
+          query: {
+            query_string: {
+              fields: fieldMap[nestedName],
+              query: escapeQuery(params.q),
+              default_operator: 'AND'
+            }
+          }
+        }
+      })
     })
+
+  const query = {}
+  if (should.length > 0) {
+    query.bool = { should }
   }
 
+  return query
+}
+
+/**
+ * For a set of parsed request params, returns a plainobject representing the
+ * filter aspects of the ES query.
+ *
+ * @param {object} params - A hash of request params including `filters`
+ *
+ * @return {object} ES query object suitable to be POST'd to ES endpoint
+ */
+const buildElasticQueryForFilters = function (params) {
   var filterClausesWithPaths = []
+
   if (params.filters) {
     filterClausesWithPaths = Object.keys(params.filters).map((prop) => {
       var config = FILTER_CONFIG[prop]
@@ -527,26 +577,19 @@ const buildElasticQuery = function (params) {
     })
   }
 
-  // Build ES query:
-  var query = {}
-
   // Gather root (not nested) filters:
   let rootFilterClauses = filterClausesWithPaths
     .filter((clauseWithPath) => !clauseWithPath.path)
     .map((clauseWithPath) => clauseWithPath.clause)
 
-  if (shoulds.length + rootFilterClauses.length > 0) {
-    query.bool = {}
-  }
-  if (shoulds.length > 0) {
-    query.bool.should = shoulds
-  }
+  const query = {}
 
   // Add nested filters:
-  filterClausesWithPaths.filter((clauseWithPath) => clauseWithPath.path)
+  filterClausesWithPaths
+    // Nested filters have a `path` property:
+    .filter((clauseWithPath) => clauseWithPath.path)
     .forEach((clauseWithPath) => {
-      if (!query.nested) query.nested = {}
-
+      // TODO: Note we seem to lack support for applying multiple nested filters
       query.nested = {
         path: clauseWithPath.path,
         query: {
@@ -558,7 +601,48 @@ const buildElasticQuery = function (params) {
     })
 
   if (rootFilterClauses.length > 0) {
-    query.bool.filter = rootFilterClauses
+    query.bool = { filter: rootFilterClauses }
+  }
+
+  return query
+}
+
+/**
+ *  Given GET params, returns a plainobject suitable for use in a ES query.
+ *
+ * @param {object} params - A hash of request params including `filters`,
+ * `search_scope`, `q`
+ *
+ * @return {object} ES query object suitable to be POST'd to ES endpoint
+ */
+const buildElasticQuery = function (params) {
+  // Build ES query:
+  var query = {}
+
+  // clean up params
+  ;['q'].forEach(function (param) {
+    if (params[param]) {
+      params[param] = params[param].replace(/date:/g, 'dateStartYear:')
+      params[param] = params[param].replace(/location:/g, 'locations:')
+      params[param] = params[param].replace(/subject:/g, 'subjectLiteral:')
+    }
+  })
+
+  if (params.q) {
+    // Merge keyword-specific ES query into the query we're building:
+    query.bool = buildElasticQueryForKeywords(params).bool
+  }
+
+  if (params.filters) {
+    // Merge filter-specific ES query into the query we're building:
+    const filterQuery = buildElasticQueryForFilters(params)
+    if (filterQuery.bool) {
+      if (!query.bool) query.bool = {}
+      query.bool.filter = filterQuery.bool.filter
+    }
+    if (filterQuery.nested) {
+      query.nested = filterQuery.nested
+    }
   }
 
   return query

diff --git a/swagger.v0.1.json b/swagger.v0.1.json
@@ -22,8 +22,8 @@
         "tags": [
           "discovery"
         ],
-        "summary": "Search resources",
-        "description": "Match by keyword\n\n> `/resources?q=war peace`\n\nFilters are applied using a `filters` param that expects this syntax on the query string:\n\n> /resources?filters[property1]=value1&filters[property2]=value2\n\nWhere `property*` is one of: \n\n> 'owner', 'subjectLiteral', 'holdingLocation', 'deliveryLocation', 'language', 'materialType', 'mediaType', 'carrierType', 'publisher', 'contributor', 'creator', 'issuance', 'createdYear', 'dateAfter', or 'dateBefore'.\n\nSee [the app README for more examples](https://github.com/NYPL-discovery/discovery-api/blob/master/README.md)\n",
+        "summary": "Search resources (bibs and items)",
+        "description": "Match by keyword\n> `/resources?q=war peace`\n\nFilters are applied using a `filters` param that expects this syntax on the query string:\n\n> `/resources?filters[property1]=value1&filters[property2]=value2`\n\nWhere `property*` is one of: \n\n> 'owner', 'subjectLiteral', 'holdingLocation', 'deliveryLocation', 'language', 'materialType', 'mediaType', 'carrierType', 'publisher', 'contributor', 'creator', 'issuance', 'createdYear', 'dateAfter', or 'dateBefore'.\n\nSee [the app README for more examples](https://github.com/NYPL-discovery/discovery-api/blob/master/README.md)\n",
         "parameters": [
           {
             "name": "q",
@@ -38,6 +38,7 @@
             "description": "Page number to return",
             "required": false,
             "type": "integer",
+            "default": 1,
             "minimum": 1
           },
           {
@@ -67,7 +68,7 @@
           {
             "name": "sort_direction",
             "in": "query",
-            "description": "Override the default direction for the current sort (asc, desc). Default depends on the field.",
+            "description": "Override the default direction for the current sort (asc, desc). Default depends on the field. (title defaults to asc, date defaults to desc, creator defaults to asc, relevance is fixed desc)",
             "required": false,
             "type": "string",
             "enum": [
@@ -78,7 +79,7 @@
           {
             "name": "search_scope",
             "in": "query",
-            "description": "Specify what (group of) fields to match against (all, title, contributor, subject, series, callnumber)",
+            "description": "Specify what (group of) fields to match against (all, title, contributor, subject, series, callnumber, standard_number). See https://github.com/NYPL-discovery/discovery-api/blob/master/lib/resources.js to review specific fields matched in each scope and how they're boosted.",
             "required": false,
             "type": "string",
             "enum": [
@@ -87,7 +88,8 @@
               "contributor",
               "subject",
               "series",
-              "callnumber"
+              "callnumber",
+              "standard_number"
             ],
             "default": "all"
           },

diff --git a/swagger.v0.1.yml b/swagger.v0.1.yml
@@ -15,19 +15,18 @@ paths:
     get:
       tags:
         - discovery
-      summary: Search resources
+      summary: Search resources (bibs and items)
       description: >
         Match by keyword
 
-
         > `/resources?q=war peace`
 
 
         Filters are applied using a `filters` param that expects this syntax on
         the query string:
 
 
-        > /resources?filters[property1]=value1&filters[property2]=value2
+        > `/resources?filters[property1]=value1&filters[property2]=value2`
 
 
         Where `property*` is one of: 
@@ -55,6 +54,7 @@ paths:
           description: Page number to return
           required: false
           type: integer
+          default: 1
           minimum: 1
         - name: per_page
           in: query
@@ -79,7 +79,8 @@ paths:
           in: query
           description: >-
             Override the default direction for the current sort (asc, desc).
-            Default depends on the field.
+            Default depends on the field. (title defaults to asc, date defaults
+            to desc, creator defaults to asc, relevance is fixed desc)
           required: false
           type: string
           enum:
@@ -89,7 +90,9 @@ paths:
           in: query
           description: >-
             Specify what (group of) fields to match against (all, title,
-            contributor, subject, series, callnumber)
+            contributor, subject, series, callnumber, standard_number). See
+            https://github.com/NYPL-discovery/discovery-api/blob/master/lib/resources.js
+            to review specific fields matched in each scope and how they're boosted.
           required: false
           type: string
           enum:
@@ -99,6 +102,7 @@ paths:
             - subject
             - series
             - callnumber
+            - standard_number
           default: all
         - name: 'filters[*]'
           in: query

diff --git a/test/delivery-locations-resolver.test.js b/test/delivery-locations-resolver.test.js
@@ -63,10 +63,6 @@ const scholarRooms = [
     id: 'loc:mala',
     label: 'Schwarzman Building - Allen Scholar Room'
   },
-  {
-    id: 'loc:maln',
-    label: 'Schwarzman Building - Noma Scholar Room'
-  },
   {
     id: 'loc:malw',
     label: 'Schwarzman Building - Wertheim Scholar Room'