judgeval
Version:
Judgment SDK for TypeScript/JavaScript
208 lines ā¢ 8.9 kB
JavaScript
/**
* Logger utilities and result printing for the JudgEval TypeScript SDK
*/
// Remove fs and path imports if no longer needed
// import * as fs from 'fs';
// import * as path from 'path';
import logger from './logger-instance.js'; // Import the configured winston logger
// Track current example info (Keep for potential context integration)
let currentExampleId = null;
let currentTimestamp = null;
/**
* Log a debug message
*/
export function debug(message, meta) {
// Prioritize meta.exampleId over currentExampleId
const exampleId = (meta === null || meta === void 0 ? void 0 : meta.exampleId) !== undefined ? meta.exampleId : currentExampleId;
logger.debug(message, Object.assign(Object.assign({}, meta), { exampleId, timestamp: currentTimestamp }));
}
/**
* Log an info message (alias for info)
*/
export function log(message, meta) {
// Prioritize meta.exampleId over currentExampleId
const exampleId = (meta === null || meta === void 0 ? void 0 : meta.exampleId) !== undefined ? meta.exampleId : currentExampleId;
logger.info(message, Object.assign(Object.assign({}, meta), { exampleId, timestamp: currentTimestamp }));
}
/**
* Log an info message
*/
export function info(message, meta) {
// Prioritize meta.exampleId over currentExampleId
const exampleId = (meta === null || meta === void 0 ? void 0 : meta.exampleId) !== undefined ? meta.exampleId : currentExampleId;
logger.info(message, Object.assign(Object.assign({}, meta), { exampleId, timestamp: currentTimestamp }));
}
/**
* Log a warning message
*/
export function warning(message, meta) {
// Prioritize meta.exampleId over currentExampleId
const exampleId = (meta === null || meta === void 0 ? void 0 : meta.exampleId) !== undefined ? meta.exampleId : currentExampleId;
logger.warn(message, Object.assign(Object.assign({}, meta), { exampleId, timestamp: currentTimestamp }));
}
/**
* Alias for warning
*/
export function warn(message, meta) {
warning(message, meta);
}
/**
* Log an error message
*/
export function error(message, meta) {
// Prioritize meta.exampleId over currentExampleId
const exampleId = (meta === null || meta === void 0 ? void 0 : meta.exampleId) !== undefined ? meta.exampleId : currentExampleId;
logger.error(message, Object.assign(Object.assign({}, meta), { exampleId, timestamp: currentTimestamp }));
}
/**
* Set the current example context for logging (Keep for potential context integration)
*/
export function setExampleContext(exampleId, timestamp) {
currentExampleId = exampleId;
currentTimestamp = timestamp;
}
/**
* Clear the current example context (Keep for potential context integration)
*/
export function clearExampleContext() {
currentExampleId = null;
currentTimestamp = null;
}
/**
* Create a context for example-specific logging (Keep for potential context integration)
*/
export function withExampleContext(exampleId, timestamp, fn) {
setExampleContext(exampleId, timestamp);
try {
return fn();
}
finally {
clearExampleContext();
}
}
/**
* Format evaluation results for display
* This matches the Python SDK's output format with clickable links
*/
export function formatEvaluationResults(results, projectName, evalName) {
var _a;
let output = '';
// Print summary information
if (results.length > 0) {
output += `\n=== Evaluation Results (${results.length} examples) ===\n\n`;
// Calculate success rate
const successfulExamples = results.filter(r => { var _a, _b; return r.success || ((_b = (_a = r.scorersData) === null || _a === void 0 ? void 0 : _a.every((s) => s.success)) !== null && _b !== void 0 ? _b : false); }).length;
const successRate = (successfulExamples / results.length) * 100;
output += `Success Rate: ${successRate.toFixed(2)}% (${successfulExamples}/${results.length})\n\n`;
// Print failures if any
const failures = results.filter(r => { var _a, _b; return !r.success || ((_b = (_a = r.scorersData) === null || _a === void 0 ? void 0 : _a.some((s) => !s.success)) !== null && _b !== void 0 ? _b : false); });
if (failures.length > 0) {
output += `Failures (${failures.length}):\n`;
for (const [index, failure] of failures.entries()) {
output += `\nExample ${index + 1}:\n`;
output += `Input: ${((_a = failure.example) === null || _a === void 0 ? void 0 : _a.input) || 'N/A'}\n`;
if (failure.scorersData) {
output += 'Scorer Failures:\n';
for (const scorer of failure.scorersData) {
if (!scorer.success) {
output += ` - ${scorer.name}: ${scorer.error || 'Unknown error'}\n`;
}
}
}
else if (!failure.success) {
output += `Error: ${failure.error || 'Unknown error'}\n`;
}
}
}
}
return output;
}
/**
* Print evaluation results to the console
* This matches the Python SDK's output format exactly
*/
export function printResults(results, projectName, evalName) {
// Always print a URL if projectName and evalName are provided
if (projectName && evalName) {
const baseUrl = 'https://app.judgmentlabs.ai/app/experiment';
const urlParams = `?project_name=${projectName}&eval_run_name=${evalName}`;
const resultsUrl = `${baseUrl}${urlParams}`;
// Print the URL
console.log(`\nš View results: ${resultsUrl}\n`);
}
// Format the results - only includes failure details
const formattedResults = formatEvaluationResults(results, projectName, evalName);
// Print the results to the console directly
if (formattedResults) {
console.log(formattedResults);
}
// Print raw results in the same format as Python SDK
console.log(JSON.stringify(results, null, 2));
}
/**
* Simplified print function for results - matches Python SDK's print(results) behavior
* This is the preferred way to print results
*/
export function print(data) {
if (Array.isArray(data)) {
// Handle array of results (evaluation results)
let projectName, evalName;
// Try to extract project name and eval name from the first result
if (data.length > 0 && data[0].metadata) {
projectName = data[0].metadata.project_name;
evalName = data[0].metadata.eval_name;
}
printResults(data, projectName, evalName);
}
else if (data && typeof data === 'object' && data.traceId) {
// Handle trace object
console.log(`\n--- Trace: ${data.name || 'Unnamed'} (ID: ${data.traceId}) ---`);
if (data.projectName) {
const traceUrl = `https://app.judgmentlabs.ai/app/monitor?project_name=${data.projectName}&trace_id=${data.traceId}&trace_name=${data.name || 'trace'}&show_trace=true`;
console.log(`\nš View trace: ${traceUrl}\n`);
}
console.log(JSON.stringify(data, null, 2));
}
else if (data && typeof data === 'object' && data.title === "Workflow Analysis Results") {
// Handle workflow analysis results
console.log(`\n=== ${data.title} ===\n`);
// Print scorer performance
console.log('Scorer Performance Summary:');
console.log('----------------------------');
if (Array.isArray(data.scorerPerformance)) {
data.scorerPerformance.forEach((scorer) => {
console.log(`${scorer.name.padEnd(30)} ${scorer.score.toFixed(2)} (${scorer.rating})`);
});
}
// Print areas for improvement
if (Array.isArray(data.areasForImprovement) && data.areasForImprovement.length > 0) {
console.log('\nAreas for Improvement:');
console.log('----------------------');
data.areasForImprovement.forEach((area, index) => {
console.log(`${index + 1}. ${area}`);
});
}
// Print strengths
if (Array.isArray(data.strengths) && data.strengths.length > 0) {
console.log('\nStrengths:');
console.log('----------');
data.strengths.forEach((strength, index) => {
console.log(`${index + 1}. ${strength}`);
});
}
}
else if (data && typeof data === 'object' && data.title && data.recommendations) {
// Handle recommendations
console.log(`\n=== ${data.title} ===`);
if (Array.isArray(data.recommendations)) {
data.recommendations.forEach((rec, index) => {
console.log(`${index + 1}. ${rec}`);
});
}
}
else {
// Handle any other object
console.log(JSON.stringify(data, null, 2));
}
}
//# sourceMappingURL=logger.js.map