Run ArchiveBox tests one file per CI job

.github/scripts/discover_test_matrix.py

@@ -1,28 +1,9 @@
 #!/usr/bin/env python3
-"""Discover every ArchiveBox test file and pack it into one CI matrix."""
+"""Discover every ArchiveBox test file and expose it as one CI job."""
 
 import json
 from pathlib import Path
 
-TARGET_SHARDS = 18
-
-SLOW_TEST_WEIGHTS = {
-    "test_cli_archiveresult.py": 200,
-    "test_opencode_agent.py": 180,
-    "test_cli_piping.py": 135,
-    "test_api_v1_crawls_crawl_crawl_id.py": 130,
-    "test_binary_service.py": 80,
-    "test_api_v1_workflow_core_token_auth_side_effects.py": 55,
-    "test_api_v1_cli_schedule.py": 50,
-    "test_api_v1_cli_remove.py": 40,
-    "test_api_v1_workflow_frozen_crawl_config_sources.py": 35,
-    "test_cli_help.py": 35,
-}
-
-
-def test_weight(path: Path) -> int:
-    return SLOW_TEST_WEIGHTS.get(path.name, 25)
-
 
 def main() -> None:
     root = Path.cwd().resolve()
@@ -31,26 +12,16 @@ def main() -> None:
     if not archivebox_tests:
         raise SystemExit("No ArchiveBox tests discovered")
 
-    shard_count = min(TARGET_SHARDS, len(archivebox_tests))
-    shards: list[list[Path]] = [[] for _ in range(shard_count)]
-    shard_weights = [0 for _ in range(shard_count)]
-
-    for path in sorted(archivebox_tests, key=lambda item: (-test_weight(item), item.as_posix())):
-        shard_index = min(range(shard_count), key=lambda index: (shard_weights[index], index))
-        shards[shard_index].append(path)
-        shard_weights[shard_index] += test_weight(path)
-
     matrix: list[dict[str, object]] = []
-    for index, shard in enumerate(shards, start=1):
-        shard_paths = sorted(path.relative_to(root).as_posix() for path in shard)
-        paths_arg = " ".join(shard_paths)
+    for path in archivebox_tests:
+        test_path = path.relative_to(root).as_posix()
         matrix.append(
             {
-                "name": f"main/shard-{index:02d}",
-                "paths": shard_paths,
-                "paths_arg": paths_arg,
-                "extra": "ldap" if any(path.endswith("/test_auth_ldap.py") for path in shard_paths) else "",
-                "count": len(shard_paths),
+                "name": f"main/{path.stem}",
+                "paths": [test_path],
+                "paths_arg": test_path,
+                "extra": "ldap" if test_path.endswith("/test_auth_ldap.py") else "",
+                "count": 1,
             },
         )
 
@@ -60,7 +31,7 @@ def main() -> None:
     if sorted(discovered_paths) != [path.relative_to(root).as_posix() for path in archivebox_tests]:
         raise SystemExit("Discovered test shard coverage does not match test files")
 
-    print(f"Discovered {len(discovered_paths)} test files exactly once across {len(matrix)} balanced shards")
+    print(f"Discovered {len(discovered_paths)} test files exactly once across {len(matrix)} jobs")
     print(json.dumps(matrix, separators=(",", ":")))
 
 

.github/workflows/test-parallel.yml

@@ -64,7 +64,7 @@ jobs:
           test "${#output[@]}" -eq 2
           echo "tests=${output[1]}" >> "$GITHUB_OUTPUT"
           echo "${output[0]}"
-          "$JQ_BINARY" -e 'length > 0 and length <= 32 and ([.[].paths[]] | length == (unique | length))' <<< "${output[1]}"
+          "$JQ_BINARY" -e 'length > 0 and length <= 256 and ([.[].paths[]] | length == (unique | length))' <<< "${output[1]}"
 
   tests:
     name: ${{ matrix.test.name }}

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "archivebox"
-version = "0.9.35rc141"
+version = "0.9.35rc142"
 requires-python = ">=3.13"
 description = "Self-hosted internet archiving solution."
 authors = [{name = "Nick Sweeting", email = "pyproject.toml@archivebox.io"}]
@@ -348,7 +348,7 @@ Donate = "https://github.com/ArchiveBox/ArchiveBox/wiki/Donations"
 
 
 [tool.bumpver]
-current_version = "v0.9.35rc141"
+current_version = "v0.9.35rc142"
 version_pattern = "vMAJOR.MINOR.PATCH[PYTAGNUM]"
 commit_message = "bump version {old_version} -> {new_version}"
 tag_message = "{new_version}"

uv.lock

@@ -123,7 +123,7 @@ wheels = [
 
 [[package]]
 name = "archivebox"
-version = "0.9.35rc141"
+version = "0.9.35rc142"
 source = { editable = "." }
 dependencies = [
     { name = "abx-dl" },