wehub-resource-sync
c3749daf48
Tests / test-linux (3.13) (push) Failing after 0s
Tests / test-linux (3.11) (push) Failing after 1s
Tests / lint (push) Failing after 0s
Tests / test-linux (3.9) (push) Failing after 1s
Docker / build (push) Failing after 1s
Docker / build-gpu (push) Failing after 2s
Tests / test-windows (push) Has been cancelled
Tests / test-macos (push) Has been cancelled
61 lines
2.7 KiB
Python
61 lines
2.7 KiB
Python
"""MemPalace — Give your AI a memory. No API key required."""
|
|
|
|
import logging
|
|
import os
|
|
import sys
|
|
|
|
|
|
def _strip_leaked_pythonpath_from_sys_path() -> None:
|
|
# Venvs inherit PYTHONPATH; on multi-Python systems it can cause
|
|
# transitive imports to load compiled extensions (pydantic_core,
|
|
# chromadb_rust_bindings) from the wrong ABI. Remove sys.path entries
|
|
# the interpreter populated from PYTHONPATH so this process imports
|
|
# only the venv's own packages. Comparison normalizes case + separators
|
|
# so Windows paths and trailing-separator quirks do not slip through
|
|
# string equality. The empty-string CWD marker on sys.path is preserved
|
|
# regardless, so PYTHONPATH=. does not collapse the implicit current
|
|
# directory.
|
|
#
|
|
# os.environ is intentionally NOT modified here. CLI entry points
|
|
# (mempalace.cli:main, mempalace.mcp_server:main) drop PYTHONPATH from
|
|
# the env themselves so any subprocess they spawn starts clean. Host
|
|
# applications that embed mempalace as a library (e.g. import
|
|
# mempalace.searcher) keep their PYTHONPATH intact for their own
|
|
# unrelated subprocesses.
|
|
leaked = os.environ.get("PYTHONPATH", None)
|
|
if not leaked:
|
|
return
|
|
|
|
def _norm(path: str) -> str:
|
|
return os.path.normcase(os.path.normpath(path))
|
|
|
|
leaked_entries = {_norm(p) for p in leaked.split(os.pathsep) if p}
|
|
sys.path[:] = [p for p in sys.path if not p or _norm(p) not in leaked_entries]
|
|
|
|
|
|
_strip_leaked_pythonpath_from_sys_path()
|
|
|
|
from .version import __version__ # noqa: E402
|
|
|
|
# chromadb telemetry: posthog capture() was broken in 0.6.x causing noisy stderr
|
|
# warnings ("capture() takes 1 positional argument but 3 were given"). In 1.x the
|
|
# posthog client is a no-op stub, so this is now harmless — kept as a guard in
|
|
# case future chromadb versions re-introduce real telemetry calls.
|
|
logging.getLogger("chromadb.telemetry.product.posthog").setLevel(logging.CRITICAL)
|
|
|
|
# NOTE: the previous block set ``ORT_DISABLE_COREML=1`` on macOS arm64 as a
|
|
# supposed workaround for the #74 ARM64 segfault. Two problems:
|
|
#
|
|
# 1. ONNX Runtime does not read that env var -- it has no global way to
|
|
# disable a single execution provider, so the setdefault was a no-op.
|
|
# 2. #74 is a null-pointer crash in ``chromadb_rust_bindings.abi3.so``, not
|
|
# an ONNX issue, so disabling CoreML would not have fixed it anyway.
|
|
#
|
|
# #521 has since traced the actual macOS arm64 crashes (both in mine and
|
|
# search paths) to the 0.x chromadb hnswlib binding. Filtering
|
|
# CoreMLExecutionProvider at the ONNX layer leaves the hnswlib C++ crash
|
|
# intact, so the real fix is upgrading chromadb to 1.5.4+, which #581
|
|
# proposes. See #397 for the history of this line.
|
|
|
|
__all__ = ["__version__"]
|