From cc29ecd5accd87f40f09e190a0cc5196666e55b4 Mon Sep 17 00:00:00 2001
From: "Bruce W. Herr II" <bherr@indiana.edu>
Date: Mon, 10 Jun 2024 09:47:35 -0400
Subject: [PATCH] Added code to generate a compiled JSON of all raw data from
 ASCT+B tables for a collection (needed for CxG collab)

---
 .prettierrc                           |  3 ++-
 src/cli.js                            | 10 ++++++++++
 src/gen-asctb-collection-json.js      | 24 ++++++++++++++++++++++++
 src/normalization/normalize-asct-b.js |  6 +++---
 4 files changed, 39 insertions(+), 4 deletions(-)
 create mode 100644 src/gen-asctb-collection-json.js

diff --git a/.prettierrc b/.prettierrc
index 544138b..0981b7c 100644
--- a/.prettierrc
+++ b/.prettierrc
@@ -1,3 +1,4 @@
 {
-  "singleQuote": true
+  "singleQuote": true,
+  "printWidth": 120
 }
diff --git a/src/cli.js b/src/cli.js
index 0ad1ebf..520722b 100755
--- a/src/cli.js
+++ b/src/cli.js
@@ -7,6 +7,7 @@ import { bumpDraft } from './drafting/bump-draft.js';
 import { newDraft } from './drafting/new-draft.js';
 import { enrich } from './enrichment/enrich.js';
 import { finalize } from './finalizing/finalize.js';
+import { genAsctbCollectionJson } from './gen-asctb-collection-json.js';
 import { list } from './list.js';
 import { migrateCcfLandmarks } from './migration/ccf-landmarks/migrate.js';
 import { migrateCcfOwl } from './migration/ccf-owl/migrate.js';
@@ -82,6 +83,15 @@ program
     updateRefOrganCrosswalk(getContext(program, command, str));
   });
 
+program
+  .command('gen-asct-b-collection-json')
+  .description('Generate a json file of all raw ASCT+B data in json format from a given collection.')
+  .argument('<collection-path>', 'Path to the collection digital object relative to DO_HOME')
+  .option('-o, --output <json-path>', 'Path to the output JSON file to generate')
+  .action((str, _options, command) => {
+    genAsctbCollectionJson(getContext(program, command, str));
+  });
+
 program
   .command('enrich')
   .description(
diff --git a/src/gen-asctb-collection-json.js b/src/gen-asctb-collection-json.js
new file mode 100644
index 0000000..98944a6
--- /dev/null
+++ b/src/gen-asctb-collection-json.js
@@ -0,0 +1,24 @@
+import { readFileSync, writeFileSync } from 'fs';
+import { load } from 'js-yaml';
+import { resolve } from 'path';
+import { getRawData } from './normalization/normalize-asct-b.js';
+import { getDigitalObjectInformation } from './utils/digital-object.js';
+
+export async function genAsctbCollectionJson(context) {
+  const obj = context.selectedDigitalObject;
+  const doListing = resolve(obj.path, 'raw/digital-objects.yaml');
+  const childObjects = load(readFileSync(doListing))['digital-objects'];
+  const asctbObjects = childObjects.filter((obj) => obj.startsWith('asct-b/') && !obj.includes('-crosswalk/'));
+
+  const result = {};
+  for (const doName of asctbObjects) {
+    const obj = getDigitalObjectInformation(resolve(context.doHome, doName), context.purlIri);
+    const rawData = await getRawData({
+      ...context,
+      selectedDigitalObject: obj,
+    });
+    result[obj.name] = rawData;
+  }
+
+  writeFileSync(context.output, JSON.stringify(result, null, 2));
+}
diff --git a/src/normalization/normalize-asct-b.js b/src/normalization/normalize-asct-b.js
index bd3cf00..233433d 100644
--- a/src/normalization/normalize-asct-b.js
+++ b/src/normalization/normalize-asct-b.js
@@ -25,11 +25,11 @@ export function normalizeAsctbMetadata(context) {
 
 export async function normalizeAsctbData(context) {
   const rawData = await getRawData(context);
-  const normalizedData = normalizeData(context, rawData);
+  const normalizedData = normalizeData(context, rawData).data;
   writeNormalizedData(context, normalizedData);
 }
 
-async function getRawData(context) {
+export async function getRawData(context) {
   const { path } = context.selectedDigitalObject;
 
   const metadata = readMetadata(context);
@@ -53,7 +53,7 @@ async function getRawData(context) {
     more(`Please review the warnings at ${warningsFile}`);
   }
 
-  return data.data;
+  return data;
 }
 
 function normalizeData(context, data) {