fix: keep docling imports lazy

Add a subprocess regression test for import crewai so future changes do not pull docling, docling_core, torch, or transformers into the base import path.

Also pin torch and torchvision through the workspace overrides so Python 3.13 resolves torch 2.12 and the matching torchvision line. Torch 2.11.0 fails to import on CPython 3.13.8, while older Python versions keep looser upper bounds and are not forced below the 2.12 line.

Author: devin-lai

lib/crewai/tests/test_imports.py

@@ -1,5 +1,9 @@
 """Test that all public API classes are properly importable."""
 
+import os
+import subprocess
+import sys
+
 
 def test_task_output_import():
     """Test that TaskOutput can be imported from crewai."""
@@ -13,3 +17,34 @@ def test_crew_output_import():
     from crewai import CrewOutput
 
     assert CrewOutput is not None
+
+
+def test_import_crewai_does_not_import_heavy_optional_deps():
+    """`import crewai` must not eagerly import heavy optional dependencies.
+
+    ``docling`` and its transitive dependencies (``torch``, ``transformers``) are
+    optional and only needed when ``CrewDoclingSource`` is instantiated. Importing
+    them at module load made ``import crewai`` slow and could fail outright on
+    environments where the optional stack does not import cleanly (e.g. ``torch``
+    on some CPython 3.13 builds). They must be imported lazily; this guards against
+    a regression. A subprocess is used so the check is unaffected by modules other
+    tests may have already imported into this process.
+    """
+    code = (
+        "import sys\n"
+        "import crewai  # noqa: F401\n"
+        "heavy = [m for m in ('torch', 'docling', 'docling_core', 'transformers') if m in sys.modules]\n"
+        "print(','.join(sorted(heavy)))\n"
+    )
+    result = subprocess.run(
+        [sys.executable, "-c", code],
+        capture_output=True,
+        text=True,
+        check=True,
+        env={**os.environ, "CREWAI_DISABLE_TELEMETRY": "true", "OTEL_SDK_DISABLED": "true"},
+    )
+    leaked = result.stdout.strip()
+    assert not leaked, (
+        f"`import crewai` eagerly imported heavy optional dependencies: {leaked}. "
+        "Import them lazily (only when actually used)."
+    )

pyproject.toml

@@ -192,6 +192,8 @@ exclude-newer = "3 days"
 # starlette <1.0.1 has PYSEC-2026-161 (missing Host header validation poisons request.url.path, bypassing path-based auth). Transitive via fastapi.
 # litellm 1.83.8+ hard-pins openai==2.24.0, missing openai.types.responses used by crewai;
 # override to >=2.30.0 (the version litellm 1.83.7 used) until upstream relaxes the pin.
+# torch 2.11.0 fails to import on CPython 3.13.8; torchvision 0.27.0 is the matching
+# release for torch 2.12.0.
 override-dependencies = [
     "openai>=2.30.0,<3",
     "rich>=13.7.1",
@@ -201,6 +203,10 @@ override-dependencies = [
     "langchain-text-splitters>=1.1.2,<2",
     "urllib3>=2.7.0",
     "transformers>=5.4.0; python_version >= '3.10'",
+    "torch<2.13.0; python_version < '3.13'",
+    "torch>=2.12.0,<2.13.0; python_version >= '3.13'",
+    "torchvision<0.28.0; python_version < '3.13'",
+    "torchvision>=0.27.0,<0.28.0; python_version >= '3.13'",
     "cryptography>=46.0.7",
     "pypdf>=6.10.2,<7",
     "uv>=0.11.15,<1",