llm-json-fix/lib/esm/utils/stringUtils.js

Fix malformed JSON outputs from Large Language Models (LLMs)

/**
 * Checks if a character is a whitespace character
 */
export function isWhitespace(char) {
    return /\s/.test(char);
}
/**
 * Checks if a character is a line terminator
 */
export function isLineTerminator(char) {
    return char === '\n' || char === '\r';
}
/**
 * Checks if a character is a digit
 */
export function isDigit(char) {
    return /[0-9]/.test(char);
}
/**
 * Checks if a character is a letter
 */
export function isAlpha(char) {
    return /[a-zA-Z]/.test(char);
}
/**
 * Checks if a character can be part of a valid identifier
 */
export function isIdentifierChar(char) {
    return isAlpha(char) || isDigit(char) || char === '_' || char === '$';
}
/**
 * Get the position in the text as line:column
 */
export function getPositionDetails(text, index) {
    const lines = text.slice(0, index).split('\n');
    const line = lines.length;
    const column = lines[lines.length - 1].length + 1;
    return { line, column };
}
/**
 * Format a position as a string
 */
export function formatPosition(text, index) {
    const { line, column } = getPositionDetails(text, index);
    return `line ${line}, column ${column}`;
}
/**
 * Extracts text that is likely markdown-formatted code blocks
 */
export function extractMarkdownCodeBlocks(text) {
    const codeBlockRegex = /```(?:json)?([^`]+)```/g;
    const matches = text.match(codeBlockRegex);
    return matches ? Array.from(matches) : null;
}
/**
 * Normalizes different quote styles to standard double quotes
 */
export function normalizeQuotes(text) {
    // Replace fancy quotes with standard double quotes
    return text
        .replace(/[""]/g, '"')
        .replace(/['']/g, "'");
}
/**
 * Checks if text contains common markdown indicators
 */
export function containsMarkdown(text) {
    const markdownIndicators = [
        /```/, // Code blocks
        /^#+\s+/m, // Headers
        /\*\*.+\*\*/, // Bold
        /\*.+\*/, // Italic
        /\[.+\]\(.+\)/ // Links
    ];
    return markdownIndicators.some(pattern => pattern.test(text));
}
/**
 * Checks if the text appears to be LLM formatted output with natural language
 */
export function isLLMStyleOutput(text) {
    // Look for patterns common in LLM outputs
    const llmPatterns = [
        /here(?:'|')?s\s+(?:the|an?|your)\s+(?:json|output|response)/i,
        /I(?:'|')?(?:ve|ll|m)\s+(?:generated|created|provided)/i,
        /```json/i,
        /^\s*[\w\s]+:\s*$/m // Explanatory labels followed by a colon
    ];
    return llmPatterns.some(pattern => pattern.test(text));
}