Pull request #1453: Index more owner data

Merge in CATS/clowder from index-more-owner-data to develop

* commit 'bc8c40aac0ab98cb97885d050f3e2d40463308d8':
  CHANGELOG updated.
  CHANGELOG updated.
  CHANGELOG updated.
  fix syntax help label
  support email field search
  add name and email to index

Author: lmarini

CHANGELOG.md

@@ -5,11 +5,15 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/)
 and this project adheres to [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
+**_Warning:_ This update modifies how information is stored in Elasticsearch for text based searching. To take advantage 
+of these changes a reindex of Elasticsearch is required. This can be started by an admin either from GUI or through the API.**
+
 ### Added
 - Added ability to delete extractor, both from api and GUI.
   [CATS-1044](https://opensource.ncsa.illinois.edu/jira/browse/CATS-1044)
 - API add tags endpoint now returns the added tags.
   [CATS-1053](https://opensource.ncsa.illinois.edu/jira/browse/CATS-1053)
+- Ability to search by creator name and email address for all resources.
 
 ### Fixed
 - Ability to delete tags on file page.

app/models/ElasticSearchObject.scala

@@ -11,6 +11,14 @@ case class ElasticsearchObject (
   resource: ResourceRef,
   name: String,
   creator: String,
+  /*
+  TODO
+    Do we still need the creator_id?
+    Do we need these in separate fields, or just include name & email under creator?
+    Should email addresses be handled in a separate way?
+   */
+  creator_name: String,
+  creator_email: String,
   created: Date,
   created_as: String = "",
   parent_of: List[String] = List.empty,
@@ -39,6 +47,8 @@ object ElasticsearchObject {
       "resource" -> JsString(eso.resource.toString),
       "name" -> JsString(eso.name),
       "creator" -> JsString(eso.creator),
+      "creator_name" -> JsString(eso.creator_name),
+      "creator_email" -> JsString(eso.creator_email),
       "created" -> JsString(eso.created.toString),
       "created_as" -> JsString(eso.created_as.toString),
       "parent_of" -> JsArray(eso.parent_of.toSeq.map( (p:String) => Json.toJson(p)): Seq[JsValue]),
@@ -60,6 +70,8 @@ object ElasticsearchObject {
       (json \ "resource").as[ResourceRef],
       (json \ "name").as[String],
       (json \ "creator").as[String],
+      (json \ "creator_name").as[String],
+      (json \ "creator_email").as[String],
       (json \ "created").as[Date],
       (json \ "created_as").as[String],
       (json \ "parent_of").as[List[String]],

app/services/ElasticsearchPlugin.scala

@@ -301,6 +301,10 @@ class ElasticsearchPlugin(application: Application) extends Plugin {
             .startObject("default")
               .field("type", "standard")
             .endObject()
+            .startObject("email_analyzer")
+              .field("type", "custom")
+              .field("tokenizer", "uax_url_email")
+            .endObject()
           .endObject()
         .endObject()
         .startObject("index")
@@ -476,6 +480,8 @@ class ElasticsearchPlugin(application: Application) extends Plugin {
               .startObject()
               // BASIC INFO
               .field("creator", eso.creator)
+              .field("creator_name", eso.creator_name)
+              .field("creator_email", eso.creator_email)
               .field("created", eso.created)
               .field("created_as", eso.created_as)
               .field("resource_type", eso.resource.resourceType.name)
@@ -719,6 +725,8 @@ class ElasticsearchPlugin(application: Application) extends Plugin {
             |"child_of": {"type": "string", "include_in_all": false},
             |"parent_of": {"type": "string", "include_in_all": false},
             |"creator": {"type": "string", "include_in_all": false},
+            |"creator_name": {"type": "string"},
+            |"creator_email": {"type": "string", "search_analyzer": "email_analyzer", "analyzer": "email_analyzer"},
             |"created_as": {"type": "string"},
             |"created": {"type": "date", "format": "dateOptionalTime", "include_in_all": false},
             |"metadata": {"type": "object"},
@@ -852,7 +860,7 @@ class ElasticsearchPlugin(application: Application) extends Plugin {
     }
 
     // If a term is specified that isn't in this list, it's assumed to be a metadata field
-    val official_terms = List("name", "creator", "resource_type", "in", "contains", "tag")
+    val official_terms = List("name", "creator", "email", "resource_type", "in", "contains", "tag")
 
     // Create list of "key:value" terms for parsing by builder
     val terms = ListBuffer[String]()
@@ -867,6 +875,10 @@ class ElasticsearchPlugin(application: Application) extends Plugin {
           currterm += "child_of:"
         else if (mt == "contains")
           currterm += "parent_of:"
+        else if (mt == "creator")
+          currterm += "creator_name:"
+        else if (mt == "email")
+          currterm += "creator_email:"
         else if (!official_terms.contains(mt))
           currterm += "metadata."+mt+":"
         else

app/util/SearchUtils.scala

@@ -97,6 +97,8 @@ object SearchUtils {
       ResourceRef(ResourceRef.file, id),
       f.filename,
       f.author.id.toString,
+      f.author.fullName.toString,
+      f.author.email.getOrElse("").toString,
       f.uploadDate,
       f.originalname,
       List.empty,
@@ -168,6 +170,8 @@ object SearchUtils {
       ResourceRef(ResourceRef.dataset, id),
       ds.name,
       ds.author.id.toString,
+      ds.author.fullName.toString,
+      ds.author.email.getOrElse("").toString,
       ds.created,
       "",
       parent_of_distinct,
@@ -196,6 +200,8 @@ object SearchUtils {
       ResourceRef(ResourceRef.collection, c.id),
       c.name,
       c.author.id.toString,
+      c.author.fullName.toString,
+      c.author.email.getOrElse("").toString,
       c.created,
       "",
       parent_of,
@@ -252,6 +258,8 @@ object SearchUtils {
       ResourceRef(ResourceRef.section, id),
       "section-"+id.toString,
       "",
+      "",
+      "",
       new Date,
       "",
       List.empty,
@@ -269,6 +277,8 @@ object SearchUtils {
       ResourceRef(ResourceRef.file, file.id),
       file.filename,
       "",
+      "",
+      "",
       file.uploadDate,
       "",
       List.empty,

app/views/searchResults.scala.html

@@ -50,8 +50,11 @@ <h1>Search</h1>
                             <td>name:VIS_SV_180_z1_1207272.png</td>
                             <td>@Messages("help.search.name", Messages("dataset.title").toLowerCase, Messages("collection.title").toLowerCase)</td></tr>
                         <tr><td>creator</td>
-                            <td>creator:5a8c4bd574d559ca9b46ef58</td>
+                            <td>creator:Smith</td>
                             <td>@Messages("help.search.creator")</td></tr>
+                        <tr><td>email</td>
+                            <td>email:janesmith@@email.com</td>
+                            <td>@Messages("help.search.email")</td></tr>
                         <tr><td>resource_type</td>
                             <td>resource_type:collection</td>
                             <td>@Messages("help.search.resource_type", Messages("dataset.title").toLowerCase, Messages("collection.title").toLowerCase)</td></tr>

conf/messages

@@ -70,7 +70,8 @@ help.search.col2=example
 help.search.allfields=name, description, creator name, and tag
 help.search.regex1=get anything with "trees" or "streetcars" in basic fields
 help.search.name=searches file, {0} or {1} names
-help.search.creator=creator ID can be found in their profile URL
+help.search.creator=creator name
+help.search.email=creator email address
 help.search.resource_type=can be file, {0} or {1}
 help.search.tag=filter search results by specific tags
 help.search.in=a {0} or {1} ID can be specified