src/PdfEncoder.Infrastructure/Pipeline/ExtractedFieldFormatValidator.cs - master - afanasevn/PdfEncoder

src/PdfEncoder.Infrastructure/Pipeline/ExtractedFieldFormatValidator.cs
85 строк3 KB
IBS\NAfanasev
Field validation, integration tests, UI polish
06 июл 2026, 17:18
06 июл 2026, 17:180e936b0
using System.Globalization;
using System.Text.RegularExpressions;

namespace PdfEncoder.Infrastructure.Pipeline;

/// <summary>
/// Проверка форматов значений полей после LLM-extract (даты ISO 8601, числа).
/// </summary>
internal static partial class ExtractedFieldFormatValidator
{
    private static readonly Regex DateKeyPattern = DateKeyRegex();
    private static readonly Regex NumericKeyPattern = NumericKeyRegex();

    /// <summary>
    /// Проверяет значение поля и возвращает предупреждения формата.
    /// </summary>
    /// <param name="key">Ключ поля.</param>
    /// <param name="value">Значение поля.</param>
    internal static IReadOnlyList<string> ValidateFieldFormat(
        string key,
        string? value)
    {
        if (string.IsNullOrWhiteSpace(value))
        {
            return [];
        }

        if (IsDateKey(key) && !IsIso8601Date(value))
        {
            return [$"field-format-invalid:{key}:expected-iso8601-date"];
        }

        if (IsNumericKey(key) && !IsDecimalNumber(value))
        {
            return [$"field-format-invalid:{key}:expected-decimal-number"];
        }

        return [];
    }

    private static bool IsDateKey(string key)
    {
        return DateKeyPattern.IsMatch(key);
    }

    private static bool IsNumericKey(string key)
    {
        return NumericKeyPattern.IsMatch(key);
    }

    private static bool IsIso8601Date(string value)
    {
        return DateOnly.TryParse(value, CultureInfo.InvariantCulture, DateTimeStyles.None, out _)
            || DateTime.TryParse(
                value,
                CultureInfo.InvariantCulture,
                DateTimeStyles.AssumeUniversal | DateTimeStyles.AdjustToUniversal,
                out _);
    }

    private static bool IsDecimalNumber(string value)
    {
        var normalized = value
            .Replace(" ", string.Empty, StringComparison.Ordinal)
            .Replace('\u00A0', ' ')
            .Trim();

        if (normalized.EndsWith("RUB", StringComparison.OrdinalIgnoreCase))
        {
            normalized = normalized[..^3].Trim();
        }

        return decimal.TryParse(
            normalized,
            NumberStyles.Number | NumberStyles.AllowCurrencySymbol,
            CultureInfo.InvariantCulture,
            out _);
    }

    [GeneratedRegex("date", RegexOptions.IgnoreCase | RegexOptions.CultureInvariant)]
    private static partial Regex DateKeyRegex();

    [GeneratedRegex("(amount|total|sum|balance|price|qty|quantity)", RegexOptions.IgnoreCase | RegexOptions.CultureInvariant)]
    private static partial Regex NumericKeyRegex();
}