From e1eb5693c93ae7841349dbdc1501027dd39047dd Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Mon, 23 Mar 2026 13:16:47 -0700
Subject: [PATCH] split CrawlSetup into Install phase with new Binary +
 BinaryRequest events

---
 archivebox/services/runner.py       |  6 +++++-
 archivebox/tests/test_cli_piping.py | 11 ++---------
 archivebox/tests/test_runner.py     |  4 +++-
 3 files changed, 10 insertions(+), 11 deletions(-)

diff --git a/archivebox/services/runner.py b/archivebox/services/runner.py
index 6ab96cf4..3f86a214 100644
--- a/archivebox/services/runner.py
+++ b/archivebox/services/runner.py
@@ -110,7 +110,11 @@ def _installed_binary_config_overrides(plugins: dict[str, Plugin]) -> dict[str,
             node_modules_dir = node_modules_dir or resolved_path.parent.parent
             npm_home = npm_home or resolved_path.parent.parent.parent
             shared_lib_dir = shared_lib_dir or resolved_path.parent.parent.parent.parent
-        elif resolved_path.parent.name == "bin" and resolved_path.parent.parent.name == "venv" and resolved_path.parent.parent.parent.name == "pip":
+        elif (
+            resolved_path.parent.name == "bin"
+            and resolved_path.parent.parent.name == "venv"
+            and resolved_path.parent.parent.parent.name == "pip"
+        ):
             pip_bin_dir = pip_bin_dir or resolved_path.parent
             pip_home = pip_home or resolved_path.parent.parent.parent
             shared_lib_dir = shared_lib_dir or resolved_path.parent.parent.parent.parent
diff --git a/archivebox/tests/test_cli_piping.py b/archivebox/tests/test_cli_piping.py
index 9a9eb02a..721f82b3 100644
--- a/archivebox/tests/test_cli_piping.py
+++ b/archivebox/tests/test_cli_piping.py
@@ -337,11 +337,7 @@ def test_binary_create_stdout_pipes_into_run(initialized_archive):
     assert create_code == 0, create_stderr
     _assert_stdout_is_jsonl_only(create_stdout)
 
-    binary = next(
-        record
-        for record in parse_jsonl_output(create_stdout)
-        if record.get("type") in {"BinaryRequest", "Binary"}
-    )
+    binary = next(record for record in parse_jsonl_output(create_stdout) if record.get("type") in {"BinaryRequest", "Binary"})
 
     run_stdout, run_stderr, run_code = run_archivebox_cmd(
         ["run"],
@@ -353,10 +349,7 @@ def test_binary_create_stdout_pipes_into_run(initialized_archive):
     _assert_stdout_is_jsonl_only(run_stdout)
 
     run_records = parse_jsonl_output(run_stdout)
-    assert any(
-        record.get("type") in {"BinaryRequest", "Binary"} and record.get("id") == binary["id"]
-        for record in run_records
-    )
+    assert any(record.get("type") in {"BinaryRequest", "Binary"} and record.get("id") == binary["id"] for record in run_records)
 
     status = _db_value(
         initialized_archive,
diff --git a/archivebox/tests/test_runner.py b/archivebox/tests/test_runner.py
index 5e84dd5d..43b4dcd3 100644
--- a/archivebox/tests/test_runner.py
+++ b/archivebox/tests/test_runner.py
@@ -318,7 +318,9 @@ def test_installed_binary_config_overrides_include_valid_installed_binaries(monk
     monkeypatch.setattr(
         Path,
         "is_file",
-        lambda self: str(self) in {sys.executable, mercury_binary.abspath, wget_binary.abspath, puppeteer_binary.abspath, ytdlp_binary.abspath},
+        lambda self: (
+            str(self) in {sys.executable, mercury_binary.abspath, wget_binary.abspath, puppeteer_binary.abspath, ytdlp_binary.abspath}
+        ),
     )
     monkeypatch.setattr(
         runner_module.os,