Compare commits

..

2 Commits

Author SHA1 Message Date
Gatefixer 88d8a69a99 fix(python): scope Instructor compatibility shim 2026-08-06 01:30:35 +00:00
Gatefixer bd779bb7d5 fix(python): support legacy InstructorEmbedding downloads 2026-08-06 01:12:01 +00:00
9 changed files with 112 additions and 569 deletions
-21
View File
@@ -1,21 +0,0 @@
# Keep Node's Linux x64 addons compatible with pre-Haswell CPUs.
# lance-linalg dispatches hot vector kernels to newer SIMD tiers at runtime.
[env]
LANCEDB_NODE_ENFORCE_X86_64_V2 = "1"
[target.x86_64-unknown-linux-gnu]
rustflags = [
"-C",
"target-cpu=x86-64-v2",
"-C",
"target-feature=-avx,-avx2,-fma,-f16c",
]
# Preserve the workspace's dynamic C runtime configuration for musl.
[target.x86_64-unknown-linux-musl]
rustflags = [
"-C",
"target-cpu=x86-64-v2",
"-C",
"target-feature=-crt-static,-avx,-avx2,-fma,-f16c",
]
-4
View File
@@ -12,10 +12,6 @@ categories.workspace = true
[lib]
crate-type = ["cdylib"]
[[test]]
name = "x86_64_v2_build_flags"
path = "build_support/x86_64_v2.rs"
[dependencies]
async-trait.workspace = true
arrow-ipc.workspace = true
-50
View File
@@ -1,50 +0,0 @@
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright The LanceDB Authors
import * as tmp from "tmp";
import { connect } from "../lancedb";
import {
Field,
FixedSizeList,
Float32,
Int32,
Schema,
makeArrowTable,
} from "../lancedb/arrow";
test("cosine vector search runs on the pre-Haswell build baseline", async () => {
const tmpDir = tmp.dirSync({ unsafeCleanup: true });
try {
const db = await connect(tmpDir.name);
const schema = new Schema([
new Field("id", new Int32(), false),
new Field(
"vector",
new FixedSizeList(3, new Field("item", new Float32(), false)),
false,
),
]);
const data = makeArrowTable(
[
{ id: 1, vector: [1, 0, 0] },
{ id: 2, vector: [0, 1, 0] },
],
{ schema },
);
const table = await db.createTable("vectors", data);
const results = await table
.vectorSearch([1, 0, 0])
.distanceType("cosine")
.limit(1)
.toArray();
expect(results).toHaveLength(1);
expect(results[0].id).toBe(1);
expect(results[0]._distance).toBeCloseTo(0);
} finally {
tmpDir.removeCallback();
}
}, 60_000);
-23
View File
@@ -3,29 +3,6 @@
extern crate napi_build;
use std::env;
#[path = "build_support/x86_64_v2.rs"]
mod x86_64_v2;
const ENFORCE_BASELINE: &str = "LANCEDB_NODE_ENFORCE_X86_64_V2";
fn main() {
napi_build::setup();
println!("cargo:rerun-if-env-changed={ENFORCE_BASELINE}");
println!("cargo:rerun-if-env-changed=CARGO_ENCODED_RUSTFLAGS");
let is_linux_x64 = env::var("CARGO_CFG_TARGET_ARCH").as_deref() == Ok("x86_64")
&& env::var("CARGO_CFG_TARGET_OS").as_deref() == Ok("linux");
let is_release = env::var("PROFILE").as_deref() == Ok("release");
let is_node_build = env::var(ENFORCE_BASELINE).as_deref() == Ok("1");
if !is_linux_x64 || (!is_release && !is_node_build) {
return;
}
let encoded_rustflags = env::var("CARGO_ENCODED_RUSTFLAGS").unwrap_or_default();
x86_64_v2::validate_encoded_rustflags(&encoded_rustflags).unwrap_or_else(|error| {
panic!("Linux x64 Node addons must use the x86-64-v2 baseline; {error}")
});
}
-430
View File
@@ -1,430 +0,0 @@
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright The LanceDB Authors
use std::collections::{BTreeMap, BTreeSet};
const SAFE_TARGET_CPUS: [&str; 2] = ["x86-64", "x86-64-v2"];
const BASELINE_FEATURES: [&str; 9] = [
"cmpxchg16b",
"fxsr",
"popcnt",
"sse",
"sse2",
"sse3",
"sse4.1",
"sse4.2",
"ssse3",
];
pub(crate) fn validate_encoded_rustflags(encoded: &str) -> Result<(), String> {
let mut target_cpu = None;
let mut feature_states = BTreeMap::new();
let mut required_disables = BTreeSet::new();
let mut unsupported_features = BTreeSet::new();
for option in codegen_options(encoded)? {
let Some((name, value)) = option.split_once('=') else {
continue;
};
let name = name.replace('_', "-");
match name.as_str() {
"target-cpu" => target_cpu = Some(value),
"target-feature" => {
for toggle in value.split(',').filter(|toggle| !toggle.is_empty()) {
let (enabled, feature) = match toggle.as_bytes()[0] {
b'+' => (true, &toggle[1..]),
b'-' => (false, &toggle[1..]),
_ => return Err(format!("invalid target feature flag: {toggle}")),
};
feature_states.insert(feature, enabled);
if enabled && !BASELINE_FEATURES.contains(&feature) {
match feature {
// These are inherited from the workspace configuration and
// explicitly canceled by the Node configuration. Account for
// their implied AVX prerequisite as well as the named feature.
"avx" => {
required_disables.insert("avx");
}
"avx2" => {
required_disables.extend(["avx", "avx2"]);
}
"f16c" => {
required_disables.extend(["avx", "f16c"]);
}
"fma" => {
required_disables.extend(["avx", "fma"]);
}
_ => {
unsupported_features.insert(feature);
}
}
}
}
}
// LLVM arguments can independently alter the target feature set and
// cannot be proven safe by inspecting rustc's target options.
"llvm-args" => return Err("LLVM arguments can override the CPU baseline".to_owned()),
_ => {}
}
}
if let Some(cpu) = target_cpu.filter(|cpu| !SAFE_TARGET_CPUS.contains(cpu)) {
return Err(format!(
"effective target CPU is {}, expected x86-64-v2 or lower",
cpu
));
}
if !unsupported_features.is_empty() {
return Err(format!(
"features above v2: {}",
unsupported_features
.into_iter()
.collect::<Vec<_>>()
.join(", ")
));
}
let not_disabled = required_disables
.into_iter()
.filter(|feature| feature_states.get(feature) != Some(&false))
.collect::<Vec<_>>();
if !not_disabled.is_empty() {
return Err(format!(
"inherited features not fully disabled: {}",
not_disabled.join(", ")
));
}
Ok(())
}
fn codegen_options(encoded: &str) -> Result<Vec<&str>, String> {
let arguments = encoded.split('\u{1f}').collect::<Vec<_>>();
if arguments.iter().any(|argument| argument.starts_with('@')) {
return Err("rustc response-file arguments cannot be validated".to_owned());
}
let mut options = Vec::new();
let mut index = 0;
while index < arguments.len() {
let argument = arguments[index];
if argument == "--cfg" {
index += 1;
let cfg = arguments
.get(index)
.copied()
.ok_or_else(|| "missing value after --cfg".to_owned())?;
reject_builtin_target_feature_cfg(cfg)?;
} else if let Some(cfg) = argument.strip_prefix("--cfg=") {
reject_builtin_target_feature_cfg(cfg)?;
} else if argument == "-C" || argument == "--codegen" {
index += 1;
let option = arguments
.get(index)
.copied()
.ok_or_else(|| format!("missing value after {argument}"))?;
options.push(option.trim_start_matches('='));
} else if let Some(option) = argument.strip_prefix("-C") {
if !option.is_empty() {
options.push(option.trim_start_matches('='));
}
} else if let Some(option) = argument.strip_prefix("--codegen=") {
options.push(option);
}
index += 1;
}
Ok(options)
}
fn reject_builtin_target_feature_cfg(cfg: &str) -> Result<(), String> {
let key = cfg.split_once('=').map_or(cfg, |(key, _)| key);
if key.contains("/*") || key.contains("//") {
return Err("comment-bearing cfgs cannot be validated".to_owned());
}
if !key.is_ascii() {
return Err("non-ASCII cfg keys cannot be validated".to_owned());
}
let name = key.trim();
let name = name.strip_prefix("r#").unwrap_or(name);
if name == "target_feature" {
return Err("built-in target_feature cfgs can override runtime CPU detection".to_owned());
}
Ok(())
}
#[cfg(test)]
mod tests {
use super::*;
fn encoded(arguments: &[&str]) -> String {
arguments.join("\u{1f}")
}
#[test]
fn accepts_merged_workspace_and_node_flags() {
let flags = encoded(&[
"-C",
"target-cpu=haswell",
"-C",
"target-feature=+avx2,+fma,+f16c",
"-C",
"target-cpu=x86-64-v2",
"-C",
"target-feature=-avx,-avx2,-fma,-f16c",
]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_non_merging_v2_boundary() {
let flags = encoded(&["-Ctarget-cpu=x86-64-v2"]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_default_cpu_with_non_codegen_flags() {
let flags = encoded(&["-D", "warnings"]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_unrelated_custom_cfg() {
let flags = encoded(&["--cfg=tokio_unstable"]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_comment_like_syntax_in_cfg_value() {
let flags = encoded(&[r#"--cfg=endpoint="https://example.com/*""#]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_explicit_x86_64_v1_cpu() {
let flags = encoded(&["-Ctarget-cpu=x86-64"]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn accepts_musl_dynamic_crt_configuration() {
let flags = encoded(&[
"-C",
"target-cpu=haswell",
"-C",
"target-feature=-crt-static,+avx2,+fma,+f16c",
"-C",
"target-cpu=x86-64-v2",
"-C",
"target-feature=-crt-static,-avx,-avx2,-fma,-f16c",
]);
assert_eq!(validate_encoded_rustflags(&flags), Ok(()));
}
#[test]
fn rejects_feature_omitted_from_target_cfg() {
let flags = encoded(&["-Ctarget-cpu=x86-64-v2", "-Ctarget-feature=+apxf"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("features above v2: apxf".to_owned())
);
}
#[test]
fn rejects_underscore_spelling_above_baseline_feature() {
let flags = encoded(&["-Ctarget_cpu=x86-64-v2", "-Ctarget_feature=+apxf"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("features above v2: apxf".to_owned())
);
}
#[test]
fn rejects_unexpected_above_baseline_feature() {
let flags = encoded(&[
"--codegen=target-cpu=x86-64-v2",
"--codegen",
"target-feature=+bmi2",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("features above v2: bmi2".to_owned())
);
}
#[test]
fn rejects_cpu_above_v2() {
let flags = encoded(&["-Ctarget-cpu=haswell"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("effective target CPU is haswell, expected x86-64-v2 or lower".to_owned())
);
}
#[test]
fn rejects_incompletely_disabled_feature_implications() {
let flags = encoded(&["-Ctarget-cpu=x86-64-v2", "-Ctarget-feature=+avx2,-avx2"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("inherited features not fully disabled: avx".to_owned())
);
}
#[test]
fn rejects_llvm_feature_overrides() {
let flags = encoded(&["-Ctarget-cpu=x86-64-v2", "-Cllvm-args=-mattr=+apxf"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("LLVM arguments can override the CPU baseline".to_owned())
);
}
#[test]
fn rejects_underscore_spelling_llvm_feature_overrides() {
let flags = encoded(&["-Ctarget_cpu=x86-64-v2", "-Cllvm_args=-mattr=+apxf"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("LLVM arguments can override the CPU baseline".to_owned())
);
}
#[test]
fn rejects_response_file_arguments() {
let flags = encoded(&["-Ctarget-cpu=x86-64-v2", "@flags.rsp"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("rustc response-file arguments cannot be validated".to_owned())
);
}
#[test]
fn rejects_split_builtin_target_feature_cfg() {
let flags = encoded(&[
"-Ctarget-cpu=x86-64-v2",
"--cfg",
r#"target_feature="avx2""#,
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("built-in target_feature cfgs can override runtime CPU detection".to_owned())
);
}
#[test]
fn rejects_equals_builtin_target_feature_cfg() {
let flags = encoded(&[
"-Ctarget-cpu=x86-64-v2",
r#"--cfg=target_feature="avx2""#,
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("built-in target_feature cfgs can override runtime CPU detection".to_owned())
);
}
#[test]
fn rejects_raw_identifier_builtin_target_feature_cfg() {
let flags = encoded(&[
"-Ctarget-cpu=x86-64-v2",
r#"--cfg=r#target_feature="avx2""#,
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("built-in target_feature cfgs can override runtime CPU detection".to_owned())
);
}
#[test]
fn rejects_block_comment_after_cfg_name() {
let flags = encoded(&[
"--cfg",
r#"target_feature/*gate*/="avx2""#,
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("comment-bearing cfgs cannot be validated".to_owned())
);
}
#[test]
fn rejects_block_comment_before_cfg_name() {
let flags = encoded(&[
r#"--cfg=/*gate*/target_feature="avx2""#,
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("comment-bearing cfgs cannot be validated".to_owned())
);
}
#[test]
fn rejects_line_comment_cfg_trivia() {
let flags = encoded(&[
"--cfg",
"target_feature// gate\n=\"avx2\"",
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("comment-bearing cfgs cannot be validated".to_owned())
);
}
#[test]
fn rejects_leading_bom_in_cfg_key() {
let flags = encoded(&[
"--cfg",
"\u{feff}target_feature=\"avx2\"",
"-Aexplicit_builtin_cfgs_in_flags",
]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("non-ASCII cfg keys cannot be validated".to_owned())
);
}
#[test]
fn rejects_non_ascii_pattern_whitespace_in_cfg_key() {
for whitespace in ['\u{200e}', '\u{200f}'] {
let cfg = format!("{whitespace}target_feature=\"avx2\"");
let flags = encoded(&["--cfg", &cfg, "-Aexplicit_builtin_cfgs_in_flags"]);
assert_eq!(
validate_encoded_rustflags(&flags),
Err("non-ASCII cfg keys cannot be validated".to_owned())
);
}
}
}
-37
View File
@@ -1,37 +0,0 @@
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright The LanceDB Authors
const assert = require("node:assert/strict");
const fs = require("node:fs");
const os = require("node:os");
const path = require("node:path");
const { connect } = require("../dist");
async function main() {
const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "lancedb-cpu-"));
try {
const db = await connect(tmpDir);
const table = await db.createTable("vectors", [
{ id: 1, vector: [1, 0, 0] },
{ id: 2, vector: [0, 1, 0] },
]);
const results = await table
.vectorSearch([1, 0, 0])
.distanceType("cosine")
.limit(1)
.toArray();
assert.equal(results.length, 1);
assert.equal(results[0].id, 1);
assert.ok(Math.abs(results[0]._distance) < 1e-6);
} finally {
fs.rmSync(tmpDir, { force: true, recursive: true });
}
}
main().catch((error) => {
console.error(error);
process.exitCode = 1;
});
-1
View File
@@ -88,7 +88,6 @@
"lint-fix": "biome check --write . && biome format --write .",
"prepublishOnly": "napi prepublish -t npm",
"test": "jest --verbose",
"test:pre-haswell": "node ci/pre_haswell_smoke.js",
"integration": "S3_TEST=1 pnpm test",
"universal": "napi universalize",
"version": "napi version"
+56 -3
View File
@@ -3,6 +3,7 @@
from typing import List
from urllib.parse import unquote, urlparse
import numpy as np
@@ -125,9 +126,20 @@ class InstructorEmbeddingFunction(TextEmbeddingFunction):
@weak_lru(maxsize=1)
def get_model(self):
instructor_embedding = attempt_import_or_raise(
"InstructorEmbedding", "InstructorEmbedding"
)
huggingface_hub = attempt_import_or_raise("huggingface_hub", "huggingface-hub")
missing = object()
original_cached_download = getattr(huggingface_hub, "cached_download", missing)
if original_cached_download is missing:
huggingface_hub.cached_download = _cached_download(huggingface_hub)
try:
instructor_embedding = attempt_import_or_raise(
"InstructorEmbedding", "InstructorEmbedding"
)
finally:
if original_cached_download is missing:
del huggingface_hub.cached_download
torch = attempt_import_or_raise("torch", "torch")
model = instructor_embedding.INSTRUCTOR(self.name)
@@ -140,3 +152,44 @@ class InstructorEmbeddingFunction(TextEmbeddingFunction):
model, {torch.nn.Linear}, dtype=torch.qint8
)
return model
def _cached_download(huggingface_hub):
"""Provide the legacy download API used by sentence-transformers 2.2.x."""
def cached_download(
*,
url,
cache_dir=None,
force_filename=None,
library_name=None,
library_version=None,
user_agent=None,
use_auth_token=None,
**_,
):
path = urlparse(url).path.lstrip("/")
try:
repo_id, resolved_path = path.split("/resolve/", maxsplit=1)
revision, filename = resolved_path.split("/", maxsplit=1)
except ValueError as err:
raise ValueError(f"Unsupported Hugging Face Hub URL: {url}") from err
repo_id = unquote(repo_id)
revision = unquote(revision)
filename = unquote(filename)
# sentence-transformers derives force_filename from this Hub path with
# os.path.join. Using the URL path beneath local_dir produces the same
# local destination without sending Windows separators to the Hub.
return huggingface_hub.hf_hub_download(
repo_id=repo_id,
filename=filename,
revision=revision,
local_dir=cache_dir,
library_name=library_name,
library_version=library_version,
user_agent=user_agent,
token=use_auth_token,
)
return cached_download
+56
View File
@@ -1,8 +1,11 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright The LanceDB Authors
import ntpath
import os
import pickle
import sys
from types import ModuleType
from typing import List, Optional, Union
from unittest.mock import MagicMock, patch
@@ -522,6 +525,59 @@ def test_embedding_function_safe_model_dump(embedding_type):
)
def test_instructor_embedding_supports_huggingface_hub_without_cached_download(
tmp_path, monkeypatch
):
from lancedb.embeddings.instructor import InstructorEmbeddingFunction
hub_download = MagicMock(return_value="/cache/1_Pooling/config.json")
huggingface_hub = ModuleType("huggingface_hub")
huggingface_hub.hf_hub_download = hub_download
torch = ModuleType("torch")
monkeypatch.setitem(sys.modules, "huggingface_hub", huggingface_hub)
monkeypatch.setitem(sys.modules, "torch", torch)
monkeypatch.delitem(sys.modules, "InstructorEmbedding", raising=False)
monkeypatch.syspath_prepend(str(tmp_path))
(tmp_path / "InstructorEmbedding.py").write_text(
"from huggingface_hub import cached_download\n\n"
"class INSTRUCTOR:\n"
" def __init__(self, name):\n"
" self.name = name\n"
)
embedding = InstructorEmbeddingFunction.create(show_progress_bar=False)
instructor_model = embedding.get_model()
assert instructor_model.name == "hkunlp/instructor-base"
assert not hasattr(huggingface_hub, "cached_download")
instructor_embedding = sys.modules["InstructorEmbedding"]
path = instructor_embedding.cached_download(
url=(
"https://huggingface.co/hkunlp/instructor-base/resolve/abc123/"
"1_Pooling/config.json"
),
cache_dir="/cache",
force_filename=ntpath.join("1_Pooling", "config.json"),
library_name="sentence-transformers",
library_version="2.2.2",
use_auth_token="token",
)
assert path == "/cache/1_Pooling/config.json"
hub_download.assert_called_once_with(
repo_id="hkunlp/instructor-base",
filename="1_Pooling/config.json",
revision="abc123",
local_dir="/cache",
library_name="sentence-transformers",
library_version="2.2.2",
user_agent=None,
token="token",
)
@patch("time.sleep")
def test_retry(mock_sleep):
test_function = MagicMock(side_effect=[Exception] * 9 + ["result"])