From dcd7e2555e094b9f932257668491d2d89e6125e9 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <github@sweeting.me>
Date: Tue, 3 Dec 2024 02:14:56 -0800
Subject: [PATCH] add new archivebox_extract cli command

---
 archivebox/cli/archivebox_extract.py | 49 ++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100644 archivebox/cli/archivebox_extract.py

diff --git a/archivebox/cli/archivebox_extract.py b/archivebox/cli/archivebox_extract.py
new file mode 100644
index 00000000..fac9219f
--- /dev/null
+++ b/archivebox/cli/archivebox_extract.py
@@ -0,0 +1,49 @@
+#!/usr/bin/env python3
+
+__package__ = 'archivebox.cli'
+__command__ = 'archivebox extract'
+
+
+import sys
+from typing import TYPE_CHECKING, Generator
+
+import rich_click as click
+
+from django.db.models import Q
+
+from archivebox.misc.util import enforce_types, docstring
+
+
+if TYPE_CHECKING:
+    from core.models import ArchiveResult
+
+
+ORCHESTRATOR = None
+
+@enforce_types
+def extract(archiveresult_id: str) -> Generator['ArchiveResult', None, None]:
+    archiveresult = ArchiveResult.objects.get(Q(id=archiveresult_id) | Q(abid=archiveresult_id))
+    if not archiveresult:
+        raise Exception(f'ArchiveResult {archiveresult_id} not found')
+    
+    return archiveresult.EXTRACTOR.extract()
+
+# <user>@<machine_id>#<datetime>/absolute/path/to/binary
+# 2014.24.01
+
+@click.command()
+
+@click.argument('archiveresult_ids', nargs=-1, type=str)
+@docstring(extract.__doc__)
+def main(archiveresult_ids: list[str]):
+    """Add a new URL or list of URLs to your archive"""
+    
+    for archiveresult_id in (archiveresult_ids or sys.stdin):
+        print(f'Extracting {archiveresult_id}...')
+        archiveresult = extract(str(archiveresult_id))
+        print(archiveresult.as_json())
+
+
+if __name__ == '__main__':
+    main()
+