chapter-organizer/Core/Services/EventOccurrenceParserService.cs

using System.Text;
using System.Text.RegularExpressions;
using Core.Entities;
using Core.Models;
using Core.Parsers;
using Microsoft.Extensions.Configuration;
using SchoolLevel = Core.Models.SchoolLevel;

namespace Core.Services;

/// <summary>
/// Service implementation for parsing event occurrence text data.
/// Wraps EventOccurrenceParser to support text input and error collection.
/// </summary>
public class EventOccurrenceParserService : IEventOccurrenceParserService
{
    private readonly IConfiguration? _configuration;

    public EventOccurrenceParserService(IConfiguration? configuration = null)
    {
        _configuration = configuration;
    }

    /// <inheritdoc/>
    public EventOccurrenceParseResult ParseFromText(string text, ICollection<EventDefinition> events)
    {
        var result = new EventOccurrenceParseResult();

        if (string.IsNullOrWhiteSpace(text))
        {
            result.Errors.Add("Input text is empty or whitespace.");
            return result;
        }

        try
        {
            // Create a temporary file from the text content
            var tempFile = Path.GetTempFileName();
            try
            {
                File.WriteAllText(tempFile, text, Encoding.UTF8);
                var fileInfo = new FileInfo(tempFile);

                // Read SchoolLevel from configuration
                SchoolLevel? schoolLevel = null;
                if (_configuration != null)
                {
                    var schoolLevelStr = _configuration.GetSection("ChapterSettings:SchoolLevel").Get<string>();
                    if (!string.IsNullOrWhiteSpace(schoolLevelStr))
                    {
                        if (Enum.TryParse<SchoolLevel>(schoolLevelStr, ignoreCase: true, out var parsed))
                        {
                            schoolLevel = parsed;
                        }
                    }
                }

                // Use the existing EventOccurrenceParser with school level
                var parser = new EventOccurrenceParser(fileInfo, events, schoolLevel);
                var parserResult = parser.Parse();
                
                // Copy occurrences from parser result
                var parsedOccurrences = parserResult.Occurrences;

                // Convert parsed occurrences to result format, handling special event types
                foreach (var kvp in parsedOccurrences)
                {
                    var eventDefinition = kvp.Key;
                    var occurrences = kvp.Value;

                    // Check if this is a special event type (not stored in database)
                    if (eventDefinition == EventDefinition.GeneralSchedule || 
                        eventDefinition == EventDefinition.MeetTheCandidates ||
                        eventDefinition == EventDefinition.ChapterOfficerMeeting ||
                        eventDefinition == EventDefinition.VotingDelegateMeeting ||
                        eventDefinition == EventDefinition.SocialGathering)
                    {
                        // For special events, set EventDefinitionId to null and set SpecialEventType
                        foreach (var occurrence in occurrences)
                        {
                            occurrence.EventDefinitionId = null;
                            occurrence.SpecialEventType = eventDefinition switch
                            {
                                var ed when ed == EventDefinition.GeneralSchedule => "GeneralSchedule",
                                var ed when ed == EventDefinition.MeetTheCandidates => "MeetTheCandidates",
                                var ed when ed == EventDefinition.ChapterOfficerMeeting => "ChapterOfficerMeeting",
                                var ed when ed == EventDefinition.VotingDelegateMeeting => "VotingDelegateMeeting",
                                var ed when ed == EventDefinition.SocialGathering => "SocialGathering",
                                _ => throw new InvalidOperationException($"Unknown special event type: {eventDefinition.Name}")
                            };
                        }

                        // Add to result with the special EventDefinition as key
                        result.Occurrences[eventDefinition] = occurrences;
                    }
                    else
                    {
                        // For regular events, set EventDefinitionId and ensure SpecialEventType is null
                        foreach (var occurrence in occurrences)
                        {
                            occurrence.EventDefinitionId = eventDefinition.Id;
                            occurrence.SpecialEventType = null;
                        }

                        result.Occurrences[eventDefinition] = occurrences;
                    }
                }

                // Copy parsing issues from parser result
                result.Issues.AddRange(parserResult.Issues);

                // Copy skipped section headers from parser result
                result.SkippedHSSectionHeaders.AddRange(parserResult.SkippedHSSectionHeaders);
                result.SkippedMSSectionHeaders.AddRange(parserResult.SkippedMSSectionHeaders);
                result.SkippedMSEventCount = parserResult.SkippedMSEventCount;
                result.SkippedHSEventCount = parserResult.SkippedHSEventCount;

                // Copy footnotes from parser result
                foreach (var kvp in parserResult.Footnotes)
                {
                    result.Footnotes[kvp.Key] = kvp.Value;
                }
                
                // Add informational messages about skipped events
                if (parserResult.SkippedMSEventCount > 0)
                {
                    result.Warnings.Add($"Skipped {parserResult.SkippedMSEventCount} Middle School (MS) event occurrence(s) based on school level setting");
                }
                if (parserResult.SkippedHSEventCount > 0)
                {
                    result.Warnings.Add($"Skipped {parserResult.SkippedHSEventCount} High School (HS) event occurrence(s) based on school level setting");
                }

                // Validate locations and add warnings for problematic ones
                ValidateLocations(result);
            }
            finally
            {
                // Clean up temporary file
                try
                {
                    if (File.Exists(tempFile))
                    {
                        File.Delete(tempFile);
                    }
                }
                catch
                {
                    // Ignore cleanup errors
                }
            }
        }
        catch (Exception ex)
        {
            result.Errors.Add($"Error parsing text: {ex.Message}");
            if (ex.InnerException != null)
            {
                result.Errors.Add($"Inner exception: {ex.InnerException.Message}");
            }
        }

        return result;
    }

    /// <summary>
    /// Validates locations from parsed occurrences and adds warnings for problematic locations.
    /// </summary>
    private static void ValidateLocations(EventOccurrenceParseResult result)
    {
        // Collect all unique locations
        var locations = result.Occurrences.Values
            .SelectMany(list => list)
            .Select(eo => eo.Location)
            .Where(loc => !string.IsNullOrWhiteSpace(loc))
            .Distinct()
            .ToList();

        if (!locations.Any())
            return;

        // Check for long locations (>50 chars)
        var longLocations = locations.Where(loc => loc != null && loc.Length > 50).ToList();
        foreach (var loc in longLocations)
        {
            if (loc != null)
            {
                result.Warnings.Add($"Location '{loc}' is unusually long ({loc.Length} characters) and may contain multiple lines or extra text");
            }
        }

        // Check for date/time patterns
        // Pattern matches: month names with day numbers, time patterns (HH:MM AM/PM), and NOON
        var dateTimePattern = new Regex(
            @"\b(January|February|March|April|May|June|July|August|September|October|November|December)\s+\d{1,2}\b|\b\d{1,2}:\d{2}\s*(a|p)\.?m\.?\b|\bNOON\b",
            RegexOptions.IgnoreCase | RegexOptions.Compiled);
        
        var locationsWithDateTime = locations.Where(loc => loc != null && dateTimePattern.IsMatch(loc)).ToList();
        foreach (var loc in locationsWithDateTime)
        {
            if (loc != null)
            {
                var match = dateTimePattern.Match(loc);
                result.Warnings.Add($"Location '{loc}' may contain date/time information: '{match.Value}'");
            }
        }

        // Check for section header patterns (missing line break detection)
        // Pattern matches: text ending with " - MS", " - HS"
        // This indicates a missing line break where the next section header was concatenated to the location
        // Note: Input is already sanitized (en-dash/em-dash -> regular hyphen), so we only need to match regular hyphens
        var sectionHeaderPattern = new Regex(
            @"-\s*(MS|HS)\s*$",
            RegexOptions.IgnoreCase | RegexOptions.Compiled);
        
        var locationsWithSectionHeader = locations.Where(loc => loc != null && sectionHeaderPattern.IsMatch(loc)).ToList();
        foreach (var loc in locationsWithSectionHeader)
        {
            if (loc != null)
            {
                var match = sectionHeaderPattern.Match(loc);
                // Extract the section header part for better warning message
                var sectionHeaderPart = match.Value.Trim();
                result.Warnings.Add($"Location '{loc}' appears to contain a section header (ends with '{sectionHeaderPart}') - likely missing line break. The location may be corrupted.");
            }
        }
    }
}