From 0abbc11a6bb5e52977fd8a81f12a86bcdf44d9f4 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <github@sweeting.me>
Date: Tue, 15 Oct 2024 22:32:36 -0700
Subject: [PATCH] add get_scope_config method to merge config down to scope

---
 archivebox/abx/archivebox/hookspec.py | 13 +++++++++++++
 archivebox/abx/archivebox/reads.py    | 27 +++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/archivebox/abx/archivebox/hookspec.py b/archivebox/abx/archivebox/hookspec.py
index b92dae24..54bf1113 100644
--- a/archivebox/abx/archivebox/hookspec.py
+++ b/archivebox/abx/archivebox/hookspec.py
@@ -4,15 +4,28 @@ from typing import Dict, Any
 
 from .. import hookspec
 
+from .base_binary import BaseBinary, BaseBinProvider
 from .base_configset import BaseConfigSet
 from .base_extractor import BaseExtractor
 from .base_searchbackend import BaseSearchBackend
 
 
+@hookspec
+def get_PLUGIN() -> Dict[str, Dict[str, Any]]:
+    return {}
+
 @hookspec
 def get_CONFIG() -> Dict[str, BaseConfigSet]:
     return {}
 
+@hookspec
+def get_BINARIES() -> Dict[str, BaseBinary]:
+    return {}
+
+@hookspec
+def get_BINPROVIDERS() -> Dict[str, BaseBinProvider]:
+    return {}
+
 @hookspec
 def get_EXTRACTORS() -> Dict[str, BaseExtractor]:
     return {}
diff --git a/archivebox/abx/archivebox/reads.py b/archivebox/abx/archivebox/reads.py
index f2479b5b..5653a7fd 100644
--- a/archivebox/abx/archivebox/reads.py
+++ b/archivebox/abx/archivebox/reads.py
@@ -128,3 +128,30 @@ def get_SEARCHBACKENDS() -> Dict[str, 'BaseSearchBackend']:
         for plugin_searchbackends in pm.hook.get_SEARCHBACKENDS()
             for searchbackend_id,searchbackend in plugin_searchbackends.items()
     })
+
+
+
+def get_scope_config(defaults=settings.CONFIG, persona=None, seed=None, crawl=None, snapshot=None, archiveresult=None, extra_config=None):
+    """Get all the relevant config for the given scope, in correct precedence order"""
+    
+    snapshot = snapshot or (archiveresult and archiveresult.snapshot)
+    crawl = crawl or (snapshot and snapshot.crawl)
+    seed = seed or (crawl and crawl.seed)
+    persona = persona or (crawl and crawl.persona)
+    
+    persona_config = persona.config if persona else {}
+    seed_config = seed.config if seed else {}
+    crawl_config = crawl.config if crawl else {}
+    snapshot_config = snapshot.config if snapshot else {}
+    archiveresult_config = archiveresult.config if archiveresult else {}
+    extra_config = extra_config or {}
+    
+    return {
+        **defaults,                     # defaults / config file / environment variables
+        **persona_config,               # lowest precedence
+        **seed_config,
+        **crawl_config,
+        **snapshot_config,
+        **archiveresult_config,
+        **extra_config,                 # highest precedence
+    }