Updates for webarchive-discovery 3.0.0 release.

Author: ruebot

solr/conf/schema.xml

@@ -28,7 +28,7 @@
 
         <!-- BL UKWA: additional -->
         <field name="access_terms" type="string" indexed="true" docValues="true" multiValued="true"/>
-        <field name="author" type="string" indexed="true" docValues="true"/>
+        <field name="author" type="string" indexed="true" docValues="true" multiValued="true"/>
         <field name="category" type="text_general" indexed="true" stored="true"/>
         <field name="collection" type="string" indexed="true" docValues="true" multiValued="true"/>
         <field name="collections" type="string" indexed="true" docValues="true" multiValued="true"/>

solr/warclight_warc-indexer.conf

@@ -4,21 +4,19 @@
 		"http_proxy": {},
 		"solr": {
 			"batch_size": 500,
-			"num_shards": 1,
 			"dummy_run": false,
-			"num_threads": 1,
 			"disablecommit": true,
 			"use_hash_url_id": true,
-			"check_solr_for_duplicates": true,
+			"check_solr_for_duplicates": false,
 
                         "field_setup": {
                             "fields" : {
-                                "content" : { "max_length" : 512K },
-                                "generator": {"max_length" : 2048 },
-                                "links" : { "max_length" : 2048 },
-                                "links_hosts" : { "max_length" : 2048 },
                                 "url" : { "max_length" : 2048 },
                                 "url_norm" : { "max_length" : 2048 },
+                                "links" : { "max_length" : 2048 },
+                                "links_hosts" : { "max_length" : 2048 },
+                                "content" : { "max_length" : 512K },
+                                "generator": {"max_length" : 2048 },
                             },
                         },
 		},
@@ -53,7 +51,7 @@
 				"file": "/path/to/exclude.txt"
 			},
 			"extract": {
-				"onDiskThreshold": "500M",
+				"onDiskThreshold": "1024M",
 				"response_include": [
 					"1",
 					"2",
@@ -74,7 +72,7 @@
 					"response",
 					"revisit"
 				],
-				"inMemoryThreshold": "500M",
+				"inMemoryThreshold": "1024M",
 				"content": {
 					"text": true,
 					"text_stored": true,

spec/test_app_templates/Gemfile.extra

@@ -1,2 +1 @@
-gem 'blacklight', github: 'projectblacklight/blacklight', branch: 'master'
-gem 'blacklight_range_limit', github: 'projectblacklight/blacklight_range_limit', branch: 'blacklight-7'
+gem 'blacklight_range_limit', '7.0.0.rc2'

tasks/warclight.rake

@@ -53,7 +53,7 @@ namespace :warclight do
   task :seed do
     puts 'Seeding index with data from spec/fixtures/warcs/...'
     # rubocop:disable Metrics/LineLength
-    system('curl -o ".internal_test_gem/tmp/warc-indexer.jar" "http://alpha.library.yorku.ca/warc-indexer-3.0.0-SNAPSHOT-jar-with-dependencies.jar"')
+    system('curl -o ".internal_test_gem/tmp/warc-indexer.jar" "http://alpha.library.yorku.ca/warc-indexer.jar"')
     system('java -Djava.io.tmpdir=.internal_test_gem/tmp -jar .internal_test_gem/tmp/warc-indexer.jar -c .internal_test_gem/solr/warclight_warc-indexer.conf -i "York University Libraries" -n "Test Collection" -u "12345" -s http://localhost:8983/solr/blacklight-core spec/fixtures/warcs/*.gz')
     system('curl "http://localhost:8983/solr/blacklight-core/update?commit=true&openSearcher=true"')
   end

warclight.gemspec

@@ -21,7 +21,7 @@ Gem::Specification.new do |s|
   s.executables     = s.files.grep(%r{^exe/}) { |f| File.basename(f) }
   s.require_paths   = ['lib']
 
-  s.add_dependency 'blacklight'
+  s.add_dependency 'blacklight', '7.0.0.rc1'
   s.add_dependency 'rails', '~> 5.0'
 
   s.add_development_dependency 'bundler', '~> 1.14'