From 3672174dad81f078f54c931b85cf86830073b5ec Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Fri, 2 Jan 2026 00:24:44 -0800
Subject: [PATCH] fix transition mid transition

---
 archivebox/core/models.py                 |  6 ++++--
 archivebox/crawls/models.py               | 13 ++++++++-----
 archivebox/plugins/chrome/chrome_utils.js |  1 +
 3 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/archivebox/core/models.py b/archivebox/core/models.py
index 471a410d..3a21041a 100755
--- a/archivebox/core/models.py
+++ b/archivebox/core/models.py
@@ -2241,10 +2241,12 @@ class SnapshotMachine(BaseStateMachine, strict_states=True):
     # Tick Event (polled by workers)
     tick = (
         queued.to.itself(unless='can_start') |
-        queued.to(started, cond='can_start')
+        queued.to(started, cond='can_start') |
+        started.to.itself(unless='is_finished') |
+        started.to(sealed, cond='is_finished')
     )
 
-    # Manual event (triggered by last ArchiveResult finishing)
+    # Manual event (can also be triggered by last ArchiveResult finishing)
     seal = started.to(sealed)
 
     def can_start(self) -> bool:
diff --git a/archivebox/crawls/models.py b/archivebox/crawls/models.py
index 40bbb6c2..52ed6c81 100755
--- a/archivebox/crawls/models.py
+++ b/archivebox/crawls/models.py
@@ -548,17 +548,20 @@ class CrawlMachine(BaseStateMachine, strict_states=True):
             if root_snapshot:
                 print(f'[cyan]🔄 Created root snapshot: {root_snapshot.url}[/cyan]', file=sys.stderr)
                 # Update status to STARTED
-                # Set retry_at to far future so workers don't claim us (we're waiting for snapshots to finish)
+                # Set retry_at to None so workers don't claim us (we wait for snapshots to finish)
                 # Last snapshot will manually call self.seal() when done
                 self.crawl.update_and_requeue(
-                    retry_at=timezone.now() + timedelta(days=365),
+                    retry_at=None,
                     status=Crawl.StatusChoices.STARTED,
                 )
             else:
                 # No snapshots (system crawl like archivebox://install)
-                print(f'[cyan]🔄 No snapshots created, sealing crawl immediately[/cyan]', file=sys.stderr)
-                # Seal immediately since there's no work to do
-                self.seal()
+                print(f'[cyan]🔄 No snapshots created, allowing immediate seal[/cyan]', file=sys.stderr)
+                # Set retry_at=now so next tick() will transition to sealed
+                self.crawl.update_and_requeue(
+                    retry_at=timezone.now(),
+                    status=Crawl.StatusChoices.STARTED,
+                )
 
         except Exception as e:
             print(f'[red]⚠️ Crawl {self.crawl.id} failed to start: {e}[/red]')
diff --git a/archivebox/plugins/chrome/chrome_utils.js b/archivebox/plugins/chrome/chrome_utils.js
index b0293356..dd9ad47b 100755
--- a/archivebox/plugins/chrome/chrome_utils.js
+++ b/archivebox/plugins/chrome/chrome_utils.js
@@ -1435,6 +1435,7 @@ function getTestEnv() {
         MACHINE_TYPE: machineType,
         LIB_DIR: libDir,
         NODE_MODULES_DIR: nodeModulesDir,
+        NODE_PATH: nodeModulesDir,  // Node.js uses NODE_PATH for module resolution
         NPM_BIN_DIR: path.join(libDir, 'npm', '.bin'),
         CHROME_EXTENSIONS_DIR: getExtensionsDir(),
     };