From bbea1f63857e75082745c2b249857d35052acea2 Mon Sep 17 00:00:00 2001
From: Matteo Cargnelutti <mcargnelutti@law.harvard.edu>
Date: Fri, 22 Sep 2023 17:39:49 -0400
Subject: [PATCH] Add `Sccop.targetUrlResolved`

As discussed w/ @rebeccacremona: This new property helps keep track of redirects. It is updated during the non-web content detection phase of the capture process (`Scoop.#detectAndCaptureNonWebContent()`) and appears in `Scoop.summary()` / `--json-summary-output`.
---
 Scoop.js      | 10 ++++++++++
 Scoop.test.js |  1 +
 2 files changed, 11 insertions(+)

diff --git a/Scoop.js b/Scoop.js
index 5603626b..7fa20879 100644
--- a/Scoop.js
+++ b/Scoop.js
@@ -70,6 +70,13 @@ export class Scoop {
    */
   url = ''
 
+  /**
+   * URL to capture, resolved to account for redirects.
+   * Populated during non-web content detection step.
+   * @type {string}
+   */
+  targetUrlResolved = ''
+
   /**
    * Is the target url a web page?
    * Assumed `true` until detected otherwise.
@@ -186,6 +193,7 @@ export class Scoop {
     this.options = filterOptions(options)
     this.blocklist = this.options.blocklist.map(castBlocklistMatcher)
     this.url = this.filterUrl(url)
+    this.targetUrlResolved = this.url
 
     // Logging setup (level, output formatting)
     logPrefix.reg(this.log)
@@ -675,6 +683,7 @@ export class Scoop {
 
       headRequestTimeMs = after - before
 
+      this.targetUrlResolved = headRequest.url
       contentType = headRequest.headers.get('Content-Type')
       contentLength = headRequest.headers.get('Content-Length')
     } catch (err) {
@@ -1413,6 +1422,7 @@ export class Scoop {
       state: this.state,
       states: Object.keys(Scoop.states), // So summary.states[summary.state] = 'NAME-OF-STATE'
       targetUrl: this.url,
+      targetUrlResolved: this.targetUrlResolved,
       targetUrlIsWebPage: this.targetUrlIsWebPage,
       targetUrlContentType: this.targetUrlContentType,
       startedAt: this.startedAt,
diff --git a/Scoop.test.js b/Scoop.test.js
index 06b10893..b075112c 100644
--- a/Scoop.test.js
+++ b/Scoop.test.js
@@ -93,6 +93,7 @@ await test('Scoop - capture of a web page.', async (t) => {
     const summary = await capture.summary()
     assert(summary)
     assert.equal(summary.targetUrl, capture.url)
+    assert.equal(summary.targetUrlResolved, capture.targetUrlResolved)
     assert.equal(summary.targetUrlContentType, 'text/html; charset=UTF-8')
     assert.equal(summary.state, Scoop.states.COMPLETE)
     assert.equal(summary.exchangeUrls.length, capture.exchanges.length)