From 6db8a08cf82823052075e1e86abb138c1c3439cc Mon Sep 17 00:00:00 2001
From: Jane Sandberg <sandbej@linnbenton.edu>
Date: Thu, 27 Aug 2020 19:27:59 -0700
Subject: [PATCH] Add a traject config for oai-pmh, helps with #309, helps with
 #308

To use it, you can run:

traject -i xml -r Traject::OaiPmhNokogiriReader \
-s oai_pmh.start_url=https://libarchive.linnbenton.edu/catalog/oai.xml?verb=ListRecords&metadataPrefix=oai_dc&set=type:OpenEducationalResource \
-c lib/tasks/data/config/config.rb \
-s solr.url=[YOUR SOLR URL \
-s solrj_writer.commit_on_close=true \
-c lib/tasks/data/config/oai.rb
---
 lib/tasks/data/config/oai.rb | 40 ++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
 create mode 100644 lib/tasks/data/config/oai.rb

diff --git a/lib/tasks/data/config/oai.rb b/lib/tasks/data/config/oai.rb
new file mode 100644
index 0000000..c453148
--- /dev/null
+++ b/lib/tasks/data/config/oai.rb
@@ -0,0 +1,40 @@
+require 'traject'
+
+settings do
+  provide "solr_writer.max_skipped", -1
+  provide "nokogiri.namespaces", {
+    "oai" => "http://www.openarchives.org/OAI/2.0/",
+    "dc" => "http://purl.org/dc/elements/1.1/",
+    "oai_dc" => "http://www.openarchives.org/OAI/2.0/oai_dc/"
+  }
+  provide "nokogiri.each_record_xpath", "//oai:record"
+end
+
+
+
+to_field "id", extract_xpath("/oai:record/oai:header/oai:identifier", to_text: false) do |record, accumulator|
+  accumulator.map! do |xml_node|
+    Digest::MD5.hexdigest(xml_node)
+  end
+end
+
+
+to_field "abstract_display", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:description")
+to_field "abstract_t", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:description")
+
+to_field "author_display", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:creator[1]")
+to_field "author_t", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:creator")
+
+to_field "contributor_display", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:contributor")
+to_field "contributor_t", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:contributor")
+
+to_field "is_electronic_facet", literal("Online")
+
+to_field "subject_t", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:subject")
+to_field "subject_topic_facet", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:subject")
+
+to_field "title_display", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:title[1]")
+to_field "title_t", extract_xpath("/oai:record/oai:metadata/oai_dc:dc/dc:title")
+
+
+