Unverified Commit f556ac8b authored by Simo Lin's avatar Simo Lin Committed by GitHub
Browse files

[router] add json tool parser (#9516)

parent 110a6598
use async_trait::async_trait;
use regex::Regex;
use serde_json::Value;
use crate::tool_parser::{
errors::{ToolParserError, ToolParserResult},
partial_json::PartialJson,
state::ParseState,
traits::ToolParser,
types::{FunctionCall, StreamResult, ToolCall},
};
/// JSON format parser for tool calls
///
/// Handles various JSON formats for function calling:
/// - Single tool call: {"name": "fn", "arguments": {...}}
/// - Multiple tool calls: [{"name": "fn1", "arguments": {...}}, ...]
/// - With parameters instead of arguments: {"name": "fn", "parameters": {...}}
///
/// Supports configurable token markers for different models
pub struct JsonParser {
/// Token(s) that mark the start of tool calls
start_tokens: Vec<String>,
/// Token(s) that mark the end of tool calls
end_tokens: Vec<String>,
/// Separator between multiple tool calls (reserved for future use)
_separator: String,
/// Parser for handling incomplete JSON during streaming
partial_json: PartialJson,
/// Regex patterns for extracting content between tokens
extractors: Vec<Regex>,
}
impl JsonParser {
/// Create a new JSON parser with default configuration
pub fn new() -> Self {
Self::with_config(
vec![], // No wrapper tokens by default
vec![],
", ".to_string(),
)
}
/// Create a parser with custom token configuration
pub fn with_config(
start_tokens: Vec<String>,
end_tokens: Vec<String>,
separator: String,
) -> Self {
// Build extraction patterns for each token pair
let extractors = start_tokens
.iter()
.zip(end_tokens.iter())
.filter_map(|(start, end)| {
if !start.is_empty() && !end.is_empty() {
// Use (?s) flag to enable DOTALL mode so . matches newlines
let pattern =
format!(r"(?s){}(.*?){}", regex::escape(start), regex::escape(end));
Regex::new(&pattern).ok()
} else {
None
}
})
.collect();
Self {
start_tokens,
end_tokens,
_separator: separator,
partial_json: PartialJson::default(),
extractors,
}
}
/// Extract JSON content from text, handling wrapper tokens if configured
fn extract_json_content<'a>(&self, text: &'a str) -> &'a str {
let mut content = text.trim();
// Try each extractor pattern
for extractor in &self.extractors {
if let Some(captures) = extractor.captures(content) {
if let Some(matched) = captures.get(1) {
content = matched.as_str().trim();
break;
}
}
}
// Handle special case where there's a start token but no end token
for (start, end) in self.start_tokens.iter().zip(self.end_tokens.iter()) {
if !start.is_empty() && end.is_empty() {
content = content.strip_prefix(start).unwrap_or(content);
}
}
content
}
/// Parse a single JSON object into a ToolCall
fn parse_single_object(&self, obj: &Value) -> ToolParserResult<Option<ToolCall>> {
// Check if this looks like a tool call
let name = obj
.get("name")
.or_else(|| obj.get("function"))
.and_then(|v| v.as_str());
if let Some(name) = name {
// Get arguments - support both "arguments" and "parameters" keys
let empty_obj = Value::Object(serde_json::Map::new());
let args = obj
.get("arguments")
.or_else(|| obj.get("parameters"))
.unwrap_or(&empty_obj);
// Convert arguments to JSON string
let arguments = serde_json::to_string(args)
.map_err(|e| ToolParserError::ParsingFailed(e.to_string()))?;
// Generate a unique ID if not provided
let id = obj
.get("id")
.and_then(|v| v.as_str())
.map(String::from)
.unwrap_or_else(|| format!("call_{}", uuid::Uuid::new_v4()));
Ok(Some(ToolCall {
id,
r#type: "function".to_string(),
function: FunctionCall {
name: name.to_string(),
arguments,
},
}))
} else {
Ok(None)
}
}
/// Parse JSON value(s) into tool calls
fn parse_json_value(&self, value: &Value) -> ToolParserResult<Vec<ToolCall>> {
let mut tools = Vec::new();
match value {
Value::Array(arr) => {
// Parse each element in the array
for item in arr {
if let Some(tool) = self.parse_single_object(item)? {
tools.push(tool);
}
}
}
Value::Object(_) => {
// Single tool call
if let Some(tool) = self.parse_single_object(value)? {
tools.push(tool);
}
}
_ => {
// Not a valid tool call format
return Ok(vec![]);
}
}
Ok(tools)
}
/// Check if text contains potential tool call markers
fn has_tool_markers(&self, text: &str) -> bool {
// If no start tokens configured, check for JSON structure
if self.start_tokens.is_empty() {
// For JSON, we just need to see the start of an object or array
return text.contains('{') || text.contains('[');
}
// Check for any start token
self.start_tokens.iter().any(|token| text.contains(token))
}
}
impl Default for JsonParser {
fn default() -> Self {
Self::new()
}
}
#[async_trait]
impl ToolParser for JsonParser {
async fn parse_complete(&self, text: &str) -> ToolParserResult<Vec<ToolCall>> {
// Extract JSON content from wrapper tokens if present
let json_content = self.extract_json_content(text);
// Try to parse as JSON
match serde_json::from_str::<Value>(json_content) {
Ok(value) => self.parse_json_value(&value),
Err(_) => {
// Not valid JSON, return empty
Ok(vec![])
}
}
}
async fn parse_incremental(
&self,
chunk: &str,
state: &mut ParseState,
) -> ToolParserResult<StreamResult> {
state.buffer.push_str(chunk);
// Check if we have potential tool calls
if !self.has_tool_markers(&state.buffer) {
// No tool markers, return as incomplete
return Ok(StreamResult::Incomplete);
}
// Extract JSON content
let json_content = self.extract_json_content(&state.buffer);
// Try to parse with partial JSON parser
match self.partial_json.parse_value(json_content) {
Ok((value, consumed)) => {
// Check if we have a complete JSON structure
if consumed == json_content.len() {
// Complete JSON, parse tool calls
let tools = self.parse_json_value(&value)?;
if !tools.is_empty() {
// Clear buffer since we consumed everything
state.buffer.clear();
// Return the first tool as complete (simplified for Phase 2)
if let Some(tool) = tools.into_iter().next() {
return Ok(StreamResult::ToolComplete(tool));
}
}
} else {
// Partial JSON, try to extract tool name
if let Some(name) = value.get("name").and_then(|v| v.as_str()) {
// Simple implementation for Phase 2
// Just return the tool name once we see it
if !state.in_string {
state.in_string = true; // Use as a flag for "name sent"
return Ok(StreamResult::ToolName {
index: 0,
name: name.to_string(),
});
}
// Check for complete arguments
if let Some(args) =
value.get("arguments").or_else(|| value.get("parameters"))
{
if let Ok(args_str) = serde_json::to_string(args) {
// Return arguments as a single update
return Ok(StreamResult::ToolArguments {
index: 0,
arguments: args_str,
});
}
}
}
}
}
Err(_) => {
// Failed to parse even as partial JSON
// Keep buffering
}
}
Ok(StreamResult::Incomplete)
}
fn detect_format(&self, text: &str) -> bool {
// Check if text contains JSON-like structure
if self.has_tool_markers(text) {
// Try to extract and parse
let json_content = self.extract_json_content(text);
// Check if it looks like valid JSON for tool calls
if let Ok(value) = serde_json::from_str::<Value>(json_content) {
match value {
Value::Object(ref obj) => {
// Check for tool call structure
obj.contains_key("name") || obj.contains_key("function")
}
Value::Array(ref arr) => {
// Check if array contains tool-like objects
arr.iter().any(|v| {
v.as_object().is_some_and(|o| {
o.contains_key("name") || o.contains_key("function")
})
})
}
_ => false,
}
} else {
false
}
} else {
false
}
}
}
#[cfg(test)]
mod tests {
use super::*;
#[tokio::test]
async fn test_parse_single_tool_call() {
let parser = JsonParser::new();
let input = r#"{"name": "get_weather", "arguments": {"location": "San Francisco"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "get_weather");
}
#[tokio::test]
async fn test_parse_multiple_tool_calls() {
let parser = JsonParser::new();
let input = r#"[
{"name": "get_weather", "arguments": {"location": "SF"}},
{"name": "search", "arguments": {"query": "news"}}
]"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 2);
assert_eq!(result[0].function.name, "get_weather");
assert_eq!(result[1].function.name, "search");
}
#[tokio::test]
async fn test_parse_with_parameters_key() {
let parser = JsonParser::new();
let input = r#"{"name": "calculate", "parameters": {"x": 10, "y": 20}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "calculate");
assert!(result[0].function.arguments.contains("10"));
}
#[tokio::test]
async fn test_parse_with_wrapper_tokens() {
let parser = JsonParser::with_config(
vec!["<tool>".to_string()],
vec!["</tool>".to_string()],
", ".to_string(),
);
let input = r#"<tool>{"name": "test", "arguments": {}}</tool>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test");
}
#[test]
fn test_detect_format() {
let parser = JsonParser::new();
assert!(parser.detect_format(r#"{"name": "test", "arguments": {}}"#));
assert!(parser.detect_format(r#"[{"name": "test"}]"#));
assert!(!parser.detect_format("plain text"));
assert!(!parser.detect_format(r#"{"key": "value"}"#));
}
#[tokio::test]
async fn test_streaming_parse() {
// Phase 2 simplified streaming test
// Just verify that streaming eventually produces a complete tool call
let parser = JsonParser::new();
let mut state = ParseState::new();
// Send complete JSON in one go (simplified for Phase 2)
let full_json = r#"{"name": "get_weather", "arguments": {"location": "SF"}}"#;
let result = parser
.parse_incremental(full_json, &mut state)
.await
.unwrap();
// Should get a complete tool immediately with complete JSON
match result {
StreamResult::ToolComplete(tool) => {
assert_eq!(tool.function.name, "get_weather");
assert!(tool.function.arguments.contains("SF"));
}
_ => panic!("Expected ToolComplete for complete JSON input"),
}
}
}
/// Tool parser module for handling function/tool calls in model outputs
///
/// This module provides infrastructure for parsing tool calls from various model formats.
/// Phase 1 focuses on core infrastructure: types, traits, registry, and partial JSON parsing.
pub mod errors;
pub mod json_parser;
pub mod partial_json;
pub mod registry;
pub mod state;
......@@ -14,6 +14,7 @@ mod tests;
// Re-export commonly used types
pub use errors::{ToolParserError, ToolParserResult};
pub use json_parser::JsonParser;
pub use registry::ParserRegistry;
pub use state::{ParsePhase, ParseState};
pub use traits::{PartialJsonParser, ToolParser};
......
use crate::tool_parser::json_parser::JsonParser;
use crate::tool_parser::traits::ToolParser;
use std::collections::HashMap;
use std::sync::Arc;
......@@ -21,6 +22,9 @@ impl ParserRegistry {
default_parser: "json".to_string(),
};
// Register default parsers
registry.register_default_parsers();
// Register default model mappings
registry.register_default_mappings();
......@@ -75,6 +79,14 @@ impl ParserRegistry {
.collect()
}
/// Register default parsers
fn register_default_parsers(&mut self) {
// JSON parser - most common format
self.register_parser("json", Arc::new(JsonParser::new()));
// Note: Additional parsers (mistral, qwen, llama) will be added in later phases
}
/// Register default model mappings
fn register_default_mappings(&mut self) {
// OpenAI models
......@@ -85,16 +97,16 @@ impl ParserRegistry {
// Anthropic models
self.map_model("claude-*", "json");
// Mistral models
self.map_model("mistral-*", "mistral");
self.map_model("mixtral-*", "mistral");
// Mistral models (will use json until mistral parser is implemented)
self.map_model("mistral-*", "json");
self.map_model("mixtral-*", "json");
// Qwen models
self.map_model("qwen*", "qwen");
// Qwen models (will use json until qwen parser is implemented)
self.map_model("qwen*", "json");
// Llama models
self.map_model("llama-*", "llama");
self.map_model("meta-llama-*", "llama");
// Llama models (will use json until llama parser is implemented)
self.map_model("llama-*", "json");
self.map_model("meta-llama-*", "json");
// Other models default to JSON
self.map_model("gemini-*", "json");
......
use super::*;
use crate::tool_parser::json_parser::JsonParser;
use crate::tool_parser::partial_json::{
compute_diff, find_common_prefix, is_complete_json, PartialJson,
};
use crate::tool_parser::traits::ToolParser;
#[test]
fn test_parse_state_new() {
......@@ -247,3 +249,638 @@ fn test_partial_tool_call() {
assert!(partial.name_sent);
assert_eq!(partial.streamed_args, r#"{"key": "#);
}
#[tokio::test]
async fn test_json_parser_complete_single() {
let parser = JsonParser::new();
// Test single tool call with arguments
let input = r#"{"name": "get_weather", "arguments": {"location": "San Francisco", "units": "celsius"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "get_weather");
assert!(result[0].function.arguments.contains("San Francisco"));
assert!(result[0].function.arguments.contains("celsius"));
}
#[tokio::test]
async fn test_json_parser_complete_array() {
let parser = JsonParser::new();
// Test array of tool calls
let input = r#"[
{"name": "get_weather", "arguments": {"location": "SF"}},
{"name": "get_news", "arguments": {"query": "technology"}}
]"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 2);
assert_eq!(result[0].function.name, "get_weather");
assert_eq!(result[1].function.name, "get_news");
}
#[tokio::test]
async fn test_json_parser_with_parameters() {
let parser = JsonParser::new();
// Test with "parameters" instead of "arguments"
let input = r#"{"name": "calculate", "parameters": {"x": 10, "y": 20, "operation": "add"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "calculate");
assert!(result[0].function.arguments.contains("10"));
assert!(result[0].function.arguments.contains("20"));
assert!(result[0].function.arguments.contains("add"));
}
#[tokio::test]
async fn test_json_parser_with_tokens() {
// Test with custom wrapper tokens
let parser = JsonParser::with_config(
vec!["[TOOL_CALLS] [".to_string()],
vec!["]".to_string()],
", ".to_string(),
);
let input = r#"[TOOL_CALLS] [{"name": "search", "arguments": {"query": "rust programming"}}]"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "search");
}
#[tokio::test]
async fn test_multiline_json_with_tokens() {
// Test that regex with (?s) flag properly handles multi-line JSON
let parser = JsonParser::with_config(
vec!["<tool>".to_string()],
vec!["</tool>".to_string()],
", ".to_string(),
);
// Pretty-printed multi-line JSON
let input = r#"<tool>{
"name": "get_weather",
"arguments": {
"location": "San Francisco",
"units": "celsius",
"include_forecast": true
}
}</tool>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "get_weather");
assert!(result[0].function.arguments.contains("San Francisco"));
assert!(result[0].function.arguments.contains("celsius"));
assert!(result[0].function.arguments.contains("true"));
}
#[tokio::test]
async fn test_multiline_json_array() {
// Test multi-line JSON array without wrapper tokens
let parser = JsonParser::new();
let input = r#"[
{
"name": "function1",
"arguments": {
"param1": "value1",
"param2": 42
}
},
{
"name": "function2",
"parameters": {
"data": [1, 2, 3],
"flag": false
}
}
]"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 2);
assert_eq!(result[0].function.name, "function1");
assert_eq!(result[1].function.name, "function2");
assert!(result[0].function.arguments.contains("value1"));
assert!(result[1].function.arguments.contains("[1,2,3]"));
}
#[test]
fn test_json_parser_format_detection() {
let parser = JsonParser::new();
// Should detect valid tool call formats
assert!(parser.detect_format(r#"{"name": "test", "arguments": {}}"#));
assert!(parser.detect_format(r#"{"name": "test", "parameters": {"x": 1}}"#));
assert!(parser.detect_format(r#"[{"name": "test"}]"#));
// Should not detect non-tool formats
assert!(!parser.detect_format("plain text"));
assert!(!parser.detect_format(r#"{"key": "value"}"#));
assert!(!parser.detect_format(r#"{"data": {"nested": true}}"#));
}
#[tokio::test]
async fn test_json_parser_streaming() {
// Phase 2 simplified streaming test
let parser = JsonParser::new();
let mut state = ParseState::new();
// Test with complete JSON (simplified for Phase 2)
let full_json = r#"{"name": "get_weather", "arguments": {"location": "San Francisco"}}"#;
let result = parser
.parse_incremental(full_json, &mut state)
.await
.unwrap();
match result {
StreamResult::ToolComplete(tool) => {
assert_eq!(tool.function.name, "get_weather");
assert!(tool.function.arguments.contains("San Francisco"));
}
_ => panic!("Expected ToolComplete for complete JSON"),
}
}
#[tokio::test]
async fn test_registry_with_json_parser() {
let registry = ParserRegistry::new();
// JSON parser should be registered by default
assert!(registry.has_parser("json"));
// Should get JSON parser for OpenAI models
let parser = registry.get_parser("gpt-4-turbo").unwrap();
// Test that the parser works
let input = r#"{"name": "test", "arguments": {"x": 1}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test");
}
#[tokio::test]
async fn test_json_parser_invalid_input() {
let parser = JsonParser::new();
// Invalid JSON should return empty results
assert_eq!(parser.parse_complete("not json").await.unwrap().len(), 0);
assert_eq!(parser.parse_complete("{invalid}").await.unwrap().len(), 0);
assert_eq!(parser.parse_complete("").await.unwrap().len(), 0);
}
#[tokio::test]
async fn test_json_parser_empty_arguments() {
let parser = JsonParser::new();
// Tool call with no arguments
let input = r#"{"name": "get_time"}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "get_time");
assert_eq!(result[0].function.arguments, "{}");
}
#[cfg(test)]
mod failure_cases {
use super::*;
#[tokio::test]
async fn test_malformed_tool_missing_name() {
let parser = JsonParser::new();
// Missing name field
let input = r#"{"arguments": {"x": 1}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 0, "Should return empty for tool without name");
// Empty name
let input = r#"{"name": "", "arguments": {"x": 1}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1, "Should accept empty name string");
assert_eq!(result[0].function.name, "");
}
#[tokio::test]
async fn test_invalid_arguments_json() {
let parser = JsonParser::new();
// Arguments is a string instead of object
let input = r#"{"name": "test", "arguments": "not an object"}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
// Should serialize the string as JSON
assert!(result[0].function.arguments.contains("not an object"));
// Arguments is a number
let input = r#"{"name": "test", "arguments": 42}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.arguments, "42");
// Arguments is null
let input = r#"{"name": "test", "arguments": null}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.arguments, "null");
}
#[tokio::test]
async fn test_broken_wrapper_tokens() {
let parser = JsonParser::with_config(
vec!["<tool>".to_string()],
vec!["</tool>".to_string()],
", ".to_string(),
);
// Missing end token
let input = r#"<tool>{"name": "test", "arguments": {}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(
result.len(),
0,
"Should fail to parse without complete wrapper"
);
// Missing start token - parser looks for complete wrapper, so this won't parse
let input = r#"{"name": "test", "arguments": {}}</tool>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(
result.len(),
0,
"Should not parse JSON with incomplete wrapper"
);
// Mismatched tokens
let input = r#"<tool>{"name": "test", "arguments": {}}</wrong>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 0, "Should fail with mismatched tokens");
}
#[tokio::test]
async fn test_invalid_json_structures() {
let parser = JsonParser::new();
// Trailing comma
let input = r#"{"name": "test", "arguments": {"x": 1,}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 0, "Should reject JSON with trailing comma");
// Missing quotes on keys
let input = r#"{name: "test", arguments: {}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 0, "Should reject invalid JSON syntax");
// Unclosed object
let input = r#"{"name": "test", "arguments": {"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 0, "Should reject incomplete JSON");
}
}
#[cfg(test)]
mod edge_cases {
use super::*;
#[tokio::test]
async fn test_unicode_in_names_and_arguments() {
let parser = JsonParser::new();
// Unicode in function name
let input = r#"{"name": "获取天气", "arguments": {"location": "北京"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "获取天气");
assert!(result[0].function.arguments.contains("北京"));
// Emoji in arguments
let input = r#"{"name": "send_message", "arguments": {"text": "Hello 👋 World 🌍"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("👋"));
assert!(result[0].function.arguments.contains("🌍"));
}
#[tokio::test]
async fn test_escaped_characters() {
let parser = JsonParser::new();
// Escaped quotes in arguments
let input = r#"{"name": "echo", "arguments": {"text": "He said \"hello\""}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains(r#"\"hello\""#));
// Escaped backslashes
let input = r#"{"name": "path", "arguments": {"dir": "C:\\Users\\test"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("\\\\"));
// Newlines and tabs
let input = r#"{"name": "format", "arguments": {"text": "line1\nline2\ttabbed"}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("\\n"));
assert!(result[0].function.arguments.contains("\\t"));
}
#[tokio::test]
async fn test_very_large_payloads() {
let parser = JsonParser::new();
// Large arguments object
let mut large_args = r#"{"name": "process", "arguments": {"#.to_string();
for i in 0..1000 {
large_args.push_str(&format!(r#""field_{}": "value_{}","#, i, i));
}
large_args.push_str(r#""final": "value"}}"#);
let result = parser.parse_complete(&large_args).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "process");
assert!(result[0].function.arguments.contains("field_999"));
// Large array of tool calls
let mut large_array = "[".to_string();
for i in 0..100 {
if i > 0 {
large_array.push(',');
}
large_array.push_str(&format!(r#"{{"name": "func_{}", "arguments": {{}}}}"#, i));
}
large_array.push(']');
let result = parser.parse_complete(&large_array).await.unwrap();
assert_eq!(result.len(), 100);
assert_eq!(result[99].function.name, "func_99");
}
#[tokio::test]
async fn test_mixed_array_tools_and_non_tools() {
let parser = JsonParser::new();
// Array with both tool calls and non-tool objects
let input = r#"[
{"name": "tool1", "arguments": {}},
{"not_a_tool": "just_data"},
{"name": "tool2", "parameters": {"x": 1}},
{"key": "value", "another": "field"}
]"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 2, "Should only parse valid tool calls");
assert_eq!(result[0].function.name, "tool1");
assert_eq!(result[1].function.name, "tool2");
}
#[tokio::test]
async fn test_duplicate_keys_in_json() {
let parser = JsonParser::new();
// JSON with duplicate keys (last one wins in most parsers)
let input = r#"{"name": "first", "name": "second", "arguments": {"x": 1, "x": 2}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(
result[0].function.name, "second",
"Last duplicate key should win"
);
assert!(
result[0].function.arguments.contains("2"),
"Last duplicate value should win"
);
}
#[tokio::test]
async fn test_null_values_in_arguments() {
let parser = JsonParser::new();
// Null values in arguments
let input = r#"{"name": "test", "arguments": {"required": "value", "optional": null}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("null"));
// Array with null
let input = r#"{"name": "test", "arguments": {"items": [1, null, "three"]}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("null"));
}
#[tokio::test]
async fn test_multiple_token_pairs_with_conflicts() {
// Test with overlapping token patterns
let parser = JsonParser::with_config(
vec!["<<".to_string(), "<tool>".to_string()],
vec![">>".to_string(), "</tool>".to_string()],
", ".to_string(),
);
// First pattern
let input = r#"<<{"name": "test1", "arguments": {}}>>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test1");
// Second pattern
let input = r#"<tool>{"name": "test2", "arguments": {}}</tool>"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test2");
// Nested patterns (should use first match)
let input = r#"<<tool>{"name": "test3", "arguments": {}}</tool>>"#;
let result = parser.parse_complete(input).await.unwrap();
// This is tricky - depends on regex behavior
// The parser should handle this gracefully
assert!(result.len() <= 1, "Should not parse multiple times");
}
#[tokio::test]
async fn test_streaming_with_partial_chunks() {
let parser = JsonParser::new();
// Test 1: Very incomplete JSON (just opening brace) should return Incomplete
let mut state1 = ParseState::new();
let partial = r#"{"#;
let result = parser
.parse_incremental(partial, &mut state1)
.await
.unwrap();
assert!(
matches!(result, StreamResult::Incomplete),
"Should return Incomplete for just opening brace"
);
// Test 2: Complete JSON should return ToolComplete
let mut state2 = ParseState::new();
let complete = r#"{"name": "get_weather", "arguments": {"location": "SF"}}"#;
let result = parser
.parse_incremental(complete, &mut state2)
.await
.unwrap();
match result {
StreamResult::ToolComplete(tool) => {
assert_eq!(tool.function.name, "get_weather");
let args: serde_json::Value =
serde_json::from_str(&tool.function.arguments).unwrap();
assert_eq!(args["location"], "SF");
}
_ => panic!("Expected ToolComplete for complete JSON"),
}
// Test 3: Partial JSON with name - Phase 2 behavior
// The PartialJson parser can complete partial JSON by filling in missing values
let mut state3 = ParseState::new();
let partial_with_name = r#"{"name": "test", "argum"#;
let result = parser
.parse_incremental(partial_with_name, &mut state3)
.await
.unwrap();
match result {
StreamResult::ToolComplete(tool) => {
assert_eq!(tool.function.name, "test");
// Arguments will be empty object since "argum" is incomplete
assert_eq!(tool.function.arguments, "{}");
}
StreamResult::ToolName { name, .. } => {
assert_eq!(name, "test");
}
StreamResult::Incomplete => {
// Also acceptable if parser decides to wait
}
_ => panic!("Unexpected result for partial JSON with name"),
}
}
#[tokio::test]
async fn test_special_json_values() {
let parser = JsonParser::new();
// Boolean values
let input = r#"{"name": "toggle", "arguments": {"enabled": true, "disabled": false}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("true"));
assert!(result[0].function.arguments.contains("false"));
// Numbers (including float and negative)
let input = r#"{"name": "calc", "arguments": {"int": 42, "float": 3.14, "negative": -17}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("42"));
assert!(result[0].function.arguments.contains("3.14"));
assert!(result[0].function.arguments.contains("-17"));
// Empty arrays and objects
let input = r#"{"name": "test", "arguments": {"empty_arr": [], "empty_obj": {}}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("[]"));
assert!(result[0].function.arguments.contains("{}"));
}
#[tokio::test]
async fn test_function_field_alternative() {
let parser = JsonParser::new();
// Using "function" instead of "name"
let input = r#"{"function": "test_func", "arguments": {"x": 1}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test_func");
// Both "name" and "function" present (name should take precedence)
let input = r#"{"name": "primary", "function": "secondary", "arguments": {}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "primary");
}
#[tokio::test]
async fn test_whitespace_handling() {
let parser = JsonParser::new();
// Extra whitespace everywhere
let input = r#" {
"name" : "test" ,
"arguments" : {
"key" : "value"
}
} "#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "test");
// Minified JSON (no whitespace)
let input = r#"{"name":"compact","arguments":{"a":1,"b":2}}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, "compact");
}
}
#[cfg(test)]
mod stress_tests {
use super::*;
#[tokio::test]
async fn test_deeply_nested_arguments() {
let parser = JsonParser::new();
// Deeply nested structure
let input = r#"{
"name": "nested",
"arguments": {
"level1": {
"level2": {
"level3": {
"level4": {
"level5": {
"value": "deep"
}
}
}
}
}
}
}"#;
let result = parser.parse_complete(input).await.unwrap();
assert_eq!(result.len(), 1);
assert!(result[0].function.arguments.contains("deep"));
}
#[tokio::test]
async fn test_concurrent_parser_usage() {
// Test that parser can be used concurrently
let parser = std::sync::Arc::new(JsonParser::new());
let mut handles = vec![];
for i in 0..10 {
let parser_clone = parser.clone();
let handle = tokio::spawn(async move {
let input = format!(r#"{{"name": "func_{}", "arguments": {{}}}}"#, i);
let result = parser_clone.parse_complete(&input).await.unwrap();
assert_eq!(result.len(), 1);
assert_eq!(result[0].function.name, format!("func_{}", i));
});
handles.push(handle);
}
for handle in handles {
handle.await.unwrap();
}
}
}
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