Skip to content

GitLab

Unverified Commit d4b09631 authored by Ayush Agarwal's avatar Ayush Agarwal Committed by GitHub
Browse files

chore: added vllm tool + reasoning data (#3416) 


Signed-off-by: default avatarayushag <ayushag@nvidia.com>
parent 2160b913
Expand all Show whitespace changes
Inline Side-by-side
Showing with 1395 additions and 0 deletions
lib/llm/tests/data/vllm/gpt-oss-20b/chat_completion_stream_f0c86d72-tool.json 0 → 100644
View file @ d4b09631
{
"request_id": "f0c86d72-cdc7-4d76-b59f-79af124643c1",
"normal_content": "",
"reasoning_content": "User asks for weather in San Francisco in Celsius. Use function.",
"tool_calls": [{"index": 0, "id": "call-1", "type": "function", "function": {"name": "get_weather", "arguments": "{\"location\":\"San Francisco, CA\",\"unit\":\"celsius\"}"}}],
"data": [
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|channel|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"analysis","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|message|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"User","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" asks","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" for","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" weather","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" in","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" San","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" Francisco","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" in","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" Celsius","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":".","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" Use","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" function","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":".","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|end|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|start|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"assistant","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|channel|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"comment","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"ary","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" to","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"=","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"functions","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":".get","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"_weather","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" ","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|constrain|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"json","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|message|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"{\"","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"location","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"\":\"","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"San","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" Francisco","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":",","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":" CA","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"\",\"","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"unit","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"\":\"","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"c","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"elsius","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"\"}","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null}}]}},
{"data":{"id":"chatcmpl-f0c86d72-cdc7-4d76-b59f-79af124643c1","choices":[{"index":0,"delta":{"content":"<|call|>","function_call":null,"tool_calls":null,"role":"assistant","refusal":null,"reasoning_content":null},"finish_reason":"stop"}]}}
]
}
lib/llm/tests/test_parsers_e2e.rs 0 → 100644
View file @ d4b09631
// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// SPDX-License-Identifier: Apache-2.0
use dynamo_async_openai::types::ChatChoiceStream;
use dynamo_llm::preprocessor::OpenAIPreprocessor;
use dynamo_llm::protocols::openai::chat_completions::NvCreateChatCompletionStreamResponse;
use dynamo_runtime::protocols::annotated::Annotated;
use futures::{Stream, StreamExt, stream};
use std::pin::Pin;
const DATA_ROOT_PATH: &str = "tests/data/";
/// Test data structure containing expected results and stream data
struct TestData {
expected_normal_content: String,
expected_reasoning_content: String,
expected_tool_calls: Vec<serde_json::Value>,
stream_chunks: Vec<Annotated<NvCreateChatCompletionStreamResponse>>,
}
/// Helper function to load test data from a test data file
fn load_test_data(file_path: &str) -> TestData {
// Read the data from file
let data = std::fs::read_to_string(file_path).unwrap();
// Parse the file as JSON
let parsed_json: serde_json::Value = serde_json::from_str(&data).unwrap();
// Extract expected values
let expected_normal_content = parsed_json
.get("normal_content")
.and_then(|v| v.as_str())
.unwrap_or("")
.to_string();
let expected_reasoning_content = parsed_json
.get("reasoning_content")
.and_then(|v| v.as_str())
.unwrap_or("")
.to_string();
let expected_tool_calls = parsed_json
.get("tool_calls")
.and_then(|v| v.as_array())
.cloned()
.unwrap_or_default();
// Extract the data chunks with choices
let data_chunks = parsed_json
.get("data")
.and_then(|v| v.as_array())
.expect("No 'data' array found in JSON");
let stream_chunks = data_chunks
.iter()
.map(|chunk| {
let inner_data = chunk.get("data").expect("No 'data' field in chunk");
let id = inner_data
.get("id")
.and_then(|v| v.as_str())
.expect("No 'id' field")
.to_string();
let choices: Vec<ChatChoiceStream> = serde_json::from_value(
inner_data
.get("choices")
.cloned()
.expect("No 'choices' field"),
)
.expect("Failed to parse choices");
let response = NvCreateChatCompletionStreamResponse {
id: id.clone(),
choices,
created: 1234567890,
model: "test-model".to_string(),
system_fingerprint: None,
object: "chat.completion.chunk".to_string(),
usage: None,
service_tier: None,
};
Annotated {
id: Some(id),
data: Some(response),
event: None,
comment: None,
}
})
.collect();
TestData {
expected_normal_content,
expected_reasoning_content,
expected_tool_calls,
stream_chunks,
}
}
/// Helper function to parse response stream with optional reasoning and tool parsing
async fn parse_response_stream(
stream: impl Stream<Item = Annotated<NvCreateChatCompletionStreamResponse>> + Send + 'static,
tool_parse_enable: bool,
reasoning_enable: bool,
tool_parser_str: Option<String>,
reasoning_parser_str: Option<String>,
) -> Vec<Annotated<NvCreateChatCompletionStreamResponse>> {
// Apply reasoning parser if enabled
let stream: Pin<
Box<dyn Stream<Item = Annotated<NvCreateChatCompletionStreamResponse>> + Send>,
> = if reasoning_enable {
if let Some(reasoning_parser) = reasoning_parser_str {
Box::pin(OpenAIPreprocessor::parse_reasoning_content_from_stream(
stream,
reasoning_parser,
))
} else {
Box::pin(stream)
}
} else {
Box::pin(stream)
};
// Apply tool calling parser if enabled
let stream: Pin<
Box<dyn Stream<Item = Annotated<NvCreateChatCompletionStreamResponse>> + Send>,
> = if tool_parse_enable {
if let Some(tool_parser) = tool_parser_str {
Box::pin(OpenAIPreprocessor::apply_tool_calling_jail(
tool_parser,
stream,
))
} else {
Box::pin(stream)
}
} else {
Box::pin(stream)
};
// Collect all output chunks
let mut stream = std::pin::pin!(stream);
let mut output_chunks = Vec::new();
while let Some(chunk) = stream.next().await {
output_chunks.push(chunk);
}
output_chunks
}
/// Structure to hold aggregated results from chunks
struct AggregatedContent {
reasoning_content: String,
normal_content: String,
has_tool_calls: bool,
tool_calls: Vec<serde_json::Value>,
}
/// Helper function to assert tool calls match expected (ignoring random IDs)
fn assert_tool_calls(
actual_tool_calls: &[serde_json::Value],
expected_tool_calls: &[serde_json::Value],
) {
assert_eq!(actual_tool_calls.len(), expected_tool_calls.len());
if !expected_tool_calls.is_empty() {
let actual_fn = &actual_tool_calls[0]["function"];
let expected_fn = &expected_tool_calls[0]["function"];
let actual_name = actual_fn["name"].as_str().unwrap();
let expected_name = expected_fn["name"].as_str().unwrap();
assert_eq!(actual_name, expected_name);
let actual_args: serde_json::Value =
serde_json::from_str(actual_fn["arguments"].as_str().unwrap()).unwrap();
let expected_args: serde_json::Value =
serde_json::from_str(expected_fn["arguments"].as_str().unwrap()).unwrap();
assert_eq!(actual_args, expected_args);
}
}
/// Helper function to aggregate all content types from chunks
fn aggregate_content_from_chunks(
chunks: &[Annotated<NvCreateChatCompletionStreamResponse>],
) -> AggregatedContent {
let mut reasoning_content = String::new();
let mut normal_content = String::new();
let mut has_tool_calls = false;
let mut tool_calls = Vec::new();
for chunk in chunks.iter() {
if let Some(ref response_data) = chunk.data {
for choice in &response_data.choices {
// Collect reasoning content
if let Some(ref reasoning) = choice.delta.reasoning_content {
reasoning_content.push_str(reasoning);
}
// Collect normal content
if let Some(ref content) = choice.delta.content {
normal_content.push_str(content);
}
// Collect tool calls
if let Some(ref chunk_tool_calls) = choice.delta.tool_calls {
has_tool_calls = true;
if let Ok(json_array) = serde_json::to_value(chunk_tool_calls)
&& let Some(array) = json_array.as_array()
{
tool_calls.extend(array.iter().cloned());
}
}
}
}
}
AggregatedContent {
reasoning_content,
normal_content,
has_tool_calls,
tool_calls,
}
}
#[cfg(test)]
mod tests {
use super::*;
#[tokio::test]
async fn test_gpt_oss_e2e_with_no_tool_calls_vllm() {
// E2E Parsing test for GPT-OSS. The input stream does not contain tool calls.
// Just content and reasoning content.
// Test will call both reasoning parsing logic and tool calling parsing logic and verify the output
// Load test data from file
let file_path = format!(
"{}/vllm/gpt-oss-20b/chat_completion_stream_49f581c1-no-tool.json",
DATA_ROOT_PATH
);
let test_data = load_test_data(&file_path);
// Create a stream from the mock chunks
let input_stream = stream::iter(test_data.stream_chunks);
// Parse the response stream with reasoning and tool parsing enabled
let output_chunks = parse_response_stream(
input_stream,
true,
true,
Some("harmony".to_string()),
Some("gpt_oss".to_string()),
)
.await;
// Verify we got output chunks
assert!(!output_chunks.is_empty(), "Should have output chunks");
// Aggregate content from all chunks
let aggregated = aggregate_content_from_chunks(&output_chunks);
// Verify against expected content from test file
assert_eq!(
aggregated.reasoning_content, test_data.expected_reasoning_content,
"Reasoning content should match expected value"
);
assert_eq!(
aggregated.normal_content, test_data.expected_normal_content,
"Normal content should match expected value"
);
// Verify tool calls match expectations
let expected_has_tool_calls = !test_data.expected_tool_calls.is_empty();
assert_eq!(
aggregated.has_tool_calls, expected_has_tool_calls,
"Tool calls presence should match expected value"
);
}
#[tokio::test]
async fn test_gpt_oss_e2e_with_tool_calls_vllm() {
// E2E Parsing test for GPT-OSS. The input stream contains tool calls.
// Test will call both reasoning parsing logic and tool calling parsing logic and verify the output
// Load test data from file
let file_path = format!(
"{}/vllm/gpt-oss-20b/chat_completion_stream_f0c86d72-tool.json",
DATA_ROOT_PATH
);
let test_data = load_test_data(&file_path);
// Create a stream from the mock chunks
let input_stream = stream::iter(test_data.stream_chunks);
// Parse the response stream with reasoning and tool parsing enabled
let output_chunks = parse_response_stream(
input_stream,
true,
true,
Some("harmony".to_string()),
Some("gpt_oss".to_string()),
)
.await;
// Verify we got output chunks
assert!(!output_chunks.is_empty(), "Should have output chunks");
// Aggregate content from all chunks
let aggregated = aggregate_content_from_chunks(&output_chunks);
// Assert reasoning content was parsed
assert!(
!aggregated.reasoning_content.is_empty(),
"Should have extracted reasoning content from analysis channel. Got: '{}'",
aggregated.reasoning_content
);
// Assert normal content was parsed
assert!(
aggregated.normal_content.is_empty(),
"Normal content should be empty. Got: '{}'",
aggregated.normal_content
);
// Verify tool calls match expectations
let expected_has_tool_calls = !test_data.expected_tool_calls.is_empty();
assert_eq!(
aggregated.has_tool_calls, expected_has_tool_calls,
"Tool calls presence should match expected value"
);
// Verify tool calls
assert_tool_calls(&aggregated.tool_calls, &test_data.expected_tool_calls);
}
#[tokio::test]
async fn test_qwen_e2e_with_no_tools_vllm() {
// E2E Parsing test for Qwen with no tools.
let file_path = format!(
"{}/vllm/qwen3-0.6B/chat_completion_stream_5627a4c6-no-tool.json",
DATA_ROOT_PATH
);
let test_data = load_test_data(&file_path);
// Create a stream from the mock chunks
let input_stream = stream::iter(test_data.stream_chunks);
// Parse the response stream with reasoning and tool parsing disabled
let output_chunks = parse_response_stream(
input_stream,
true,
true,
Some("hermes".to_string()),
Some("qwen".to_string()),
)
.await;
// Verify we got output chunks
assert!(!output_chunks.is_empty(), "Should have output chunks");
// Aggregate content from output chunks
let aggregated = aggregate_content_from_chunks(&output_chunks);
// Assert that output content matches input content exactly (no parsing applied)
assert_eq!(
aggregated.normal_content, test_data.expected_normal_content,
"When parsing is disabled, output should match input exactly"
);
// Verify tool calls match expectations
let expected_has_tool_calls = !test_data.expected_tool_calls.is_empty();
assert_eq!(
aggregated.has_tool_calls, expected_has_tool_calls,
"Tool calls presence should match expected value"
);
}
#[tokio::test]
async fn test_qwen_e2e_with_tools_vllm() {
// E2E Parsing test for Qwen with tools.
// Test will call both reasoning parsing logic and tool calling parsing logic and verify the output
let file_path = format!(
"{}/vllm/qwen3-0.6B/chat_completion_stream_8f33c28b-tool.json",
DATA_ROOT_PATH
);
let test_data = load_test_data(&file_path);
// Create a stream from the mock chunks
let input_stream = stream::iter(test_data.stream_chunks);
// Parse the response stream with reasoning and tool parsing enabled
let output_chunks = parse_response_stream(
input_stream,
true,
true,
Some("hermes".to_string()),
Some("qwen".to_string()),
)
.await;
// Verify we got output chunks
assert!(!output_chunks.is_empty(), "Should have output chunks");
// Aggregate content from output chunks
let aggregated = aggregate_content_from_chunks(&output_chunks);
// Assert reasoning content was parsed
assert_eq!(
aggregated.reasoning_content, test_data.expected_reasoning_content,
"Should have extracted reasoning content.",
);
assert_eq!(
aggregated.normal_content, test_data.expected_normal_content,
"Normal content should match expected value.",
);
// Verify tool calls match expectations
let expected_has_tool_calls = !test_data.expected_tool_calls.is_empty();
assert_eq!(
aggregated.has_tool_calls, expected_has_tool_calls,
"Tool calls presence should match expected value"
);
// Verify tool calls
assert_tool_calls(&aggregated.tool_calls, &test_data.expected_tool_calls);
}
#[tokio::test]
async fn test_nemotron_e2e_with_tools_vllm() {
// E2E Parsing test for Nemotron with tools.
// Test will call both reasoning parsing logic and tool calling parsing logic and verify the output
let file_path = format!(
"{}/vllm/nemotron-49b/chat_completion_stream_3d40f925-tool.json",
DATA_ROOT_PATH
);
let test_data = load_test_data(&file_path);
// Create a stream from the mock chunks
let input_stream = stream::iter(test_data.stream_chunks);
// Parse the response stream with reasoning and tool parsing enabled
let output_chunks = parse_response_stream(
input_stream,
true,
true,
Some("nemotron_deci".to_string()),
Some("nemotron_deci".to_string()),
)
.await;
// Verify we got output chunks
assert!(!output_chunks.is_empty(), "Should have output chunks");
// Aggregate content from output chunks
let aggregated = aggregate_content_from_chunks(&output_chunks);
// Assert reasoning content was parsed
assert_eq!(
aggregated.reasoning_content, test_data.expected_reasoning_content,
"Should have extracted reasoning content.",
);
assert_eq!(
aggregated.normal_content, test_data.expected_normal_content,
"Normal content should match expected value.",
);
// Verify tool calls match expectations
let expected_has_tool_calls = !test_data.expected_tool_calls.is_empty();
assert_eq!(
aggregated.has_tool_calls, expected_has_tool_calls,
"Tool calls presence should match expected value"
);
// Verify tool calls
assert_tool_calls(&aggregated.tool_calls, &test_data.expected_tool_calls);
}
}
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment