mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-22 00:31:44 +00:00
63d994ade4
Extracted from #41733 without the router loop, the cache machine layer, streaming, or the error, timeout and route-pruning work that moved to #41745 litellm-callbacks holds the contract a native call and its host share: Machine, HostOp, CallEvent, the in-process run loop, and Passthrough, which is built only by comparing the caller's inputs with the body the route sends, so a route can never mark a key it rewrote. litellm-host-python (formerly python-interop) owns the CPython driver and the Execution handle, and litellm-callbacks-legacy is the @client wrapper as the native call sees it: function_setup, the deployment hooks, pre_call and post_call, the success and failure fan-out and the deferred proxy release. OCR is the one route on it, and the old core and bridge lifecycles are gone The passthrough rule is the structural fix for the bug #41719 patched in core and #41716 reworks: an inlined remote document no longer counts as the caller's value, so the legacy adapter never hands the caller's URL back into the body. core/tests/ocr/passthrough.rs pins it for every route and document source, including that unchanged values stay passthrough, and callbacks-legacy/tests/payload.rs pins the adapter side with a real pre_call callback Python OCR integration tests that only exercised core behavior now live as Rust tests, so tests/test_litellm_rust keeps the cases that need the full Python stack
102 lines
3.6 KiB
Rust
102 lines
3.6 KiB
Rust
use std::hint::black_box;
|
|
use std::time::Duration;
|
|
|
|
use criterion::{BenchmarkId, Criterion, criterion_group, criterion_main};
|
|
use litellm_host_python::{from_py, to_py};
|
|
use pyo3::prelude::*;
|
|
use pyo3::types::PyDict;
|
|
use serde_json::{Value, json};
|
|
|
|
const PAYLOAD_SIZES: &[(&str, usize)] = &[
|
|
("1_KiB", 1024),
|
|
("64_KiB", 64 * 1024),
|
|
("1_MiB", 1024 * 1024),
|
|
("4_MiB", 4 * 1024 * 1024),
|
|
("16_MiB", 16 * 1024 * 1024),
|
|
];
|
|
|
|
fn former_json_roundtrip_from_py(py: Python<'_>, value: &Bound<'_, PyAny>) -> Value {
|
|
let json = py.import("json").expect("Python json module should import");
|
|
let encoded: String = json
|
|
.call_method1("dumps", (value,))
|
|
.expect("payload should serialize")
|
|
.extract()
|
|
.expect("json.dumps should return a string");
|
|
serde_json::from_str(&encoded).expect("serialized JSON should parse")
|
|
}
|
|
|
|
fn pythonize_from_py(value: &Bound<'_, PyAny>) -> Value {
|
|
from_py(value).expect("payload should depythonize")
|
|
}
|
|
|
|
fn former_json_roundtrip_to_py(py: Python<'_>, value: &Value) -> Py<PyAny> {
|
|
let json = py.import("json").expect("Python json module should import");
|
|
let encoded = serde_json::to_string(value).expect("response should serialize");
|
|
json.call_method1("loads", (encoded,))
|
|
.expect("serialized response should parse in Python")
|
|
.unbind()
|
|
}
|
|
|
|
fn pythonize_to_py(py: Python<'_>, value: &Value) -> Py<PyAny> {
|
|
to_py(py, value).expect("response should pythonize")
|
|
}
|
|
|
|
fn bridge_serialization(c: &mut Criterion) {
|
|
Python::initialize();
|
|
Python::attach(|py| {
|
|
for &(label, payload_bytes) in PAYLOAD_SIZES {
|
|
let data_uri = format!("data:image/png;base64,{}", "A".repeat(payload_bytes));
|
|
let document = PyDict::new(py);
|
|
document
|
|
.set_item("type", "image_url")
|
|
.expect("document type should be set");
|
|
document
|
|
.set_item("image_url", &data_uri)
|
|
.expect("document URL should be set");
|
|
let response = json!({
|
|
"pages": [{
|
|
"index": 0,
|
|
"markdown": "OCR text",
|
|
"images": [{"image_base64": data_uri}],
|
|
}],
|
|
"model": "mistral-ocr-latest",
|
|
"document_annotation": null,
|
|
"usage_info": {"pages_processed": 1},
|
|
"object": "ocr",
|
|
});
|
|
|
|
c.bench_with_input(
|
|
BenchmarkId::new("python_to_rust_json", label),
|
|
&document,
|
|
|b, document| {
|
|
b.iter(|| former_json_roundtrip_from_py(py, black_box(document.as_any())))
|
|
},
|
|
);
|
|
c.bench_with_input(
|
|
BenchmarkId::new("python_to_rust_pythonize", label),
|
|
&document,
|
|
|b, document| b.iter(|| pythonize_from_py(black_box(document.as_any()))),
|
|
);
|
|
c.bench_with_input(
|
|
BenchmarkId::new("rust_to_python_json", label),
|
|
&response,
|
|
|b, response| b.iter(|| former_json_roundtrip_to_py(py, black_box(response))),
|
|
);
|
|
c.bench_with_input(
|
|
BenchmarkId::new("rust_to_python_pythonize", label),
|
|
&response,
|
|
|b, response| b.iter(|| pythonize_to_py(py, black_box(response))),
|
|
);
|
|
}
|
|
});
|
|
}
|
|
|
|
criterion_group! {
|
|
name = benches;
|
|
config = Criterion::default()
|
|
.sample_size(20)
|
|
.warm_up_time(Duration::from_secs(1))
|
|
.measurement_time(Duration::from_secs(4));
|
|
targets = bridge_serialization
|
|
}
|
|
criterion_main!(benches);
|