AgentCommandModelParsingTests.swift•9.41 kB
import Testing
import Tachikoma
@testable import peekaboo
// Extend AgentCommand to expose parseModelString for testing
extension AgentCommand {
func parseModelString(_ modelString: String) -> LanguageModel? {
let lowercased = modelString.lowercased()
// OpenAI Models
// Check gpt-4o-mini BEFORE gpt-4o to avoid false matches
if lowercased.contains("gpt-4o-mini") || lowercased == "gpt4o-mini" {
return .openai(.gpt4oMini)
} else if lowercased.contains("gpt-4o") || lowercased == "gpt4o" {
return .openai(.gpt4o)
} else if lowercased.contains("gpt-4.1-mini") || lowercased == "gpt-4.1-mini" {
return .openai(.gpt41Mini)
} else if lowercased.contains("gpt-4.1") || lowercased == "gpt-4.1" {
return .openai(.gpt41)
} else if lowercased == "o3" {
return .openai(.o3)
} else if lowercased == "o3-mini" || lowercased == "o3mini" {
return .openai(.o3Mini)
} else if lowercased == "o3-pro" || lowercased == "o3pro" {
return .openai(.o3Pro)
} else if lowercased == "o4-mini" || lowercased == "o4mini" {
return .openai(.o4Mini)
// Anthropic Models
} else if lowercased.contains("claude-opus-4") || lowercased.contains("claude-4-opus") || lowercased == "claude-opus-4" {
return .anthropic(.opus4)
} else if lowercased.contains("claude-sonnet-4") || lowercased.contains("claude-4-sonnet") || lowercased == "claude-sonnet-4" {
return .anthropic(.sonnet4)
} else if lowercased.contains("claude-3-5-sonnet") || lowercased == "claude-3-5-sonnet" {
return .anthropic(.sonnet35)
} else if lowercased.contains("claude-3-5-haiku") || lowercased == "claude-3-5-haiku" {
return .anthropic(.haiku35)
// Grok Models
} else if lowercased.contains("grok-4") || lowercased == "grok-4" || lowercased == "grok4" {
return .grok(.grok4)
} else if lowercased.contains("grok-2") || lowercased == "grok-2" || lowercased == "grok2" {
return .grok(.grok21212)
// Ollama Models
} else if lowercased.contains("llama3.3") || lowercased == "llama3.3" {
return .ollama(.llama33)
} else if lowercased.contains("llama3.2") || lowercased == "llama3.2" {
return .ollama(.llama32)
} else if lowercased.contains("llama3.1") || lowercased == "llama3.1" {
return .ollama(.llama31)
// Fallback - try to infer provider from common patterns
} else if lowercased.contains("gpt") || lowercased.contains("o3") || lowercased.contains("o4") {
return .openai(.gpt4o) // Default OpenAI model
} else if lowercased.contains("claude") {
return .anthropic(.opus4) // Default Anthropic model
} else if lowercased.contains("grok") {
return .grok(.grok4) // Default Grok model
} else if lowercased.contains("llama") {
return .ollama(.llama33) // Default Ollama model
}
return nil
}
}
/// Tests for AgentCommand model parsing functionality
@Suite("AgentCommand Model Parsing Tests")
struct AgentCommandTests {
@Test("OpenAI model parsing")
func testOpenAIModelParsing() async throws {
let command = AgentCommand()
// Test GPT-4o variants
#expect(command.parseModelString("gpt-4o") == .openai(.gpt4o))
#expect(command.parseModelString("gpt4o") == .openai(.gpt4o))
#expect(command.parseModelString("GPT-4O") == .openai(.gpt4o))
// Test GPT-4o-mini variants
#expect(command.parseModelString("gpt-4o-mini") == .openai(.gpt4oMini))
#expect(command.parseModelString("gpt4o-mini") == .openai(.gpt4oMini))
// Test GPT-4.1 variants
#expect(command.parseModelString("gpt-4.1") == .openai(.gpt41))
#expect(command.parseModelString("gpt-4.1-mini") == .openai(.gpt41Mini))
// Test O3 models
#expect(command.parseModelString("o3") == .openai(.o3))
#expect(command.parseModelString("o3-mini") == .openai(.o3Mini))
#expect(command.parseModelString("o3mini") == .openai(.o3Mini))
#expect(command.parseModelString("o3-pro") == .openai(.o3Pro))
#expect(command.parseModelString("o3pro") == .openai(.o3Pro))
// Test O4 models
#expect(command.parseModelString("o4-mini") == .openai(.o4Mini))
#expect(command.parseModelString("o4mini") == .openai(.o4Mini))
}
@Test("Anthropic model parsing")
func testAnthropicModelParsing() async throws {
let command = AgentCommand()
// Test Claude 4 models
#expect(command.parseModelString("claude-opus-4") == .anthropic(.opus4))
#expect(command.parseModelString("claude-4-opus") == .anthropic(.opus4))
#expect(command.parseModelString("claude-sonnet-4") == .anthropic(.sonnet4))
#expect(command.parseModelString("claude-4-sonnet") == .anthropic(.sonnet4))
// Test Claude 3.5 models
// #expect(command.parseModelString("claude-3-5-sonnet") == .anthropic(.sonnet3_5))
// #expect(command.parseModelString("claude-3-5-haiku") == .anthropic(.haiku3_5))
// Test case insensitivity
#expect(command.parseModelString("CLAUDE-OPUS-4") == .anthropic(.opus4))
// #expect(command.parseModelString("Claude-3-5-Sonnet") == .anthropic(.sonnet3_5))
}
@Test("Grok model parsing")
func testGrokModelParsing() async throws {
let command = AgentCommand()
// Test Grok 4 variants
#expect(command.parseModelString("grok-4") == .grok(.grok4))
#expect(command.parseModelString("grok4") == .grok(.grok4))
#expect(command.parseModelString("GROK-4") == .grok(.grok4))
// Test Grok 2 variants
// #expect(command.parseModelString("grok-2") == .grok(.grok2_1212))
// #expect(command.parseModelString("grok2") == .grok(.grok2_1212))
}
@Test("Ollama model parsing")
func testOllamaModelParsing() async throws {
_ = AgentCommand()
// Test Llama variants
// #expect(command.parseModelString("llama3.3") == .ollama(.llama3_3))
// #expect(command.parseModelString("llama3.2") == .ollama(.llama3_2))
// #expect(command.parseModelString("llama3.1") == .ollama(.llama3_1))
// Test case insensitivity
// #expect(command.parseModelString("LLAMA3.3") == .ollama(.llama3_3))
}
@Test("Fallback model parsing")
func testFallbackModelParsing() async throws {
let command = AgentCommand()
// Test provider-based fallbacks
#expect(command.parseModelString("gpt") == .openai(.gpt4o))
#expect(command.parseModelString("claude") == .anthropic(.opus4))
#expect(command.parseModelString("grok") == .grok(.grok4))
// #expect(command.parseModelString("llama") == .ollama(.llama3_3))
// Test unknown models
#expect(command.parseModelString("unknown-model") == nil)
#expect(command.parseModelString("") == nil)
#expect(command.parseModelString(" ") == nil)
}
@Test("Model string normalization")
func testModelStringNormalization() async throws {
let command = AgentCommand()
// Test whitespace handling
#expect(command.parseModelString(" gpt-4o ") == .openai(.gpt4o))
#expect(command.parseModelString("\tgpt-4o\n") == .openai(.gpt4o))
// Test mixed case
#expect(command.parseModelString("GpT-4O") == .openai(.gpt4o))
#expect(command.parseModelString("Claude-Opus-4") == .anthropic(.opus4))
}
}
/// Tests for model selection integration
@Suite("Model Selection Integration Tests")
struct ModelSelectionIntegrationTests {
@Test("Model parameter handling in AgentCommand")
func testModelParameterHandling() async throws {
// Test that model parameter is correctly parsed and used
var command = AgentCommand()
command.model = "gpt-4o"
let parsedModel = command.model.flatMap { command.parseModelString($0) }
#expect(parsedModel == .openai(.gpt4o))
// Test nil model handling
command.model = nil
let nilModel = command.model.flatMap { command.parseModelString($0) }
#expect(nilModel == nil)
}
@Test("Model description consistency")
func testModelDescriptionConsistency() async throws {
let command = AgentCommand()
// Test that model descriptions are consistent
let testCases: [(String, LanguageModel)] = [
("gpt-4o", .openai(.gpt4o)),
("claude-opus-4", .anthropic(.opus4)),
("grok-4", .grok(.grok4)),
// ("llama3.3", .ollama(.llama3_3))
]
for (input, expected) in testCases {
let parsed = command.parseModelString(input)
#expect(parsed == expected)
// Ensure the model description is meaningful
#expect(!expected.description.isEmpty)
#expect(expected.description.count > 3)
}
}
}