Function getCharacterCodeData

src/tools/edit.ts:57–114

* Extract character code data from diff * @param expected The string that was searched for * @param actual The string that was found * @returns Character code statistics

(expected: string, actual: string)

 * @returns Character code statistics
 */
function getCharacterCodeData(expected: string, actual: string): {
    report: string;
    uniqueCount: number;
    diffLength: number;
} {
    // Find common prefix and suffix
    let prefixLength = 0;
    const minLength = Math.min(expected.length, actual.length);

    // Determine common prefix length
    while (prefixLength < minLength &&
           expected[prefixLength] === actual[prefixLength]) {
        prefixLength++;
    }

    // Determine common suffix length
    let suffixLength = 0;
    while (suffixLength < minLength - prefixLength &&
           expected[expected.length - 1 - suffixLength] === actual[actual.length - 1 - suffixLength]) {
        suffixLength++;
    }
    
    // Extract the different parts
    const expectedDiff = expected.substring(prefixLength, expected.length - suffixLength);
    const actualDiff = actual.substring(prefixLength, actual.length - suffixLength);
    
    // Count unique character codes in the diff
    const characterCodes = new Map<number, number>();
    const fullDiff = expectedDiff + actualDiff;
    
    for (let i = 0; i < fullDiff.length; i++) {
        const charCode = fullDiff.charCodeAt(i);
        characterCodes.set(charCode, (characterCodes.get(charCode) || 0) + 1);
    }
    
    // Create character codes string report
    const charCodeReport: string[] = [];
    characterCodes.forEach((count, code) => {
        // Include character representation for better readability
        const char = String.fromCharCode(code);
        // Make special characters more readable
        const charDisplay = code < 32 || code > 126 ? `\\x${code.toString(16).padStart(2, '0')}` : char;
        charCodeReport.push(`${code}:${count}[${charDisplay}]`);
    });
    
    // Sort by character code for consistency
    charCodeReport.sort((a, b) => {
        const codeA = parseInt(a.split(':')[0]);
        const codeB = parseInt(b.split(':')[0]);
        return codeA - codeB;
    });
    
    return {
        report: charCodeReport.join(','),
        uniqueCount: characterCodes.size,
        diffLength: fullDiff.length
    };
}