From 139c419268d97beaa112f1c751733393772f82ac Mon Sep 17 00:00:00 2001
From: makeworld <makeworld@protonmail.com>
Date: Mon, 22 Jul 2024 17:33:49 +0200
Subject: [PATCH] browsertrix: store crawl workflow metadata

---
 docs/attributes.md     | 1 +
 webhook/browsertrix.go | 3 +++
 2 files changed, 4 insertions(+)

diff --git a/docs/attributes.md b/docs/attributes.md
index d86328c..220cd40 100644
--- a/docs/attributes.md
+++ b/docs/attributes.md
@@ -35,6 +35,7 @@ The majority of these attributes are set automatically upon ingestion, but all c
 - `project_id`: the name for the project this asset was ingested under
 - `project_path`: the path for the project within the sync folder
 - `asset_origin_sig_key_name`: may exist if the ingestion process involved verifiying a known, named public key
+- Browsertrix crawl info: `crawl_workflow_name`, `crawl_workflow_description`, `crawl_workflow_tags`
 
 Encrypted files have the `encryption_type` attribute, currently always set to `secretstream`. See [encryption.md](./encryption.md) for more info.
 
diff --git a/webhook/browsertrix.go b/webhook/browsertrix.go
index 5c9ecb4..e8e6894 100644
--- a/webhook/browsertrix.go
+++ b/webhook/browsertrix.go
@@ -260,6 +260,9 @@ func handleBrowsertrixEvent(w http.ResponseWriter, r *http.Request) {
 	metadataMap["asset_origin_type"] = []string{"wacz"}
 	metadataMap["project_id"] = projectId
 	metadataMap["file_name"] = e.Resources[0].Name
+	metadataMap["crawl_workflow_name"] = crawlInfo.Name
+	metadataMap["crawl_workflow_description"] = crawlInfo.Description
+	metadataMap["crawl_workflow_tags"] = crawlInfo.Tags
 
 	err = util.MoveFile(tempFilePath, filepath.Join(outputDirectory, cid))
 	if err != nil {