chapter-organizer/Core/Parsers/EventOccurrence/LocationPatternMatcher.cs

using System.Text.RegularExpressions;

namespace Core.Parsers.EventOccurrence;

/// <summary>
/// Matches location strings against configurable patterns.
/// Supports exact matches and wildcard patterns (e.g., "Room *", "Exhibit Hall *").
/// </summary>
public static class LocationPatternMatcher
{
    /// <summary>
    /// Matches location text against configured patterns and returns the matched location.
    /// </summary>
    /// <param name="locationText">The location text to match.</param>
    /// <param name="patterns">The list of patterns to match against.</param>
    /// <returns>The matched location if a pattern matches, or empty string if no match is found.</returns>
    public static string Match(string locationText, IReadOnlyList<string> patterns)
    {
        // Normalize location text for matching (trim and handle variations)
        var normalizedLocation = locationText.Trim();
        
        // If location is empty after normalization, return empty
        if (string.IsNullOrWhiteSpace(normalizedLocation))
            return string.Empty;
        
        foreach (var pattern in patterns)
        {
            var normalizedPattern = pattern.Trim();
            
            // Skip empty patterns
            if (string.IsNullOrWhiteSpace(normalizedPattern))
                continue;
            
            // Handle exact matches (patterns without wildcards like "Online", "Virtual", "TBD")
            if (!normalizedPattern.Contains('*'))
            {
                if (string.Equals(normalizedPattern, normalizedLocation, StringComparison.OrdinalIgnoreCase))
                {
                    return normalizedLocation;
                }
                continue;
            }

            // Convert pattern to regex: escape special chars, replace * with .*
            // This handles patterns like "Exhibit Hall *", "Room *", "Mtg. Room *", etc.
            var escapedPattern = Regex.Escape(normalizedPattern);
            escapedPattern = escapedPattern.Replace(@"\*", ".*?");
            
            // Use case-insensitive matching
            // Note: For dynamic patterns, we compile on demand. This is acceptable since patterns
            // are configured and reused across parsing sessions.
            var regex = new Regex($"^{escapedPattern}$", RegexOptions.IgnoreCase | RegexOptions.Compiled);
            if (regex.IsMatch(normalizedLocation))
            {
                return normalizedLocation; // Return the full matched location
            }
        }

        return string.Empty;
    }
}