feat: add parsing from files

Endpoint/Sync/ScheduleSynchronizer.cs

@@ -6,9 +6,11 @@ using Mirea.Api.DataAccess.Persistence;
 using Mirea.Api.Endpoint.Common.Interfaces;
 using Mirea.Api.Endpoint.Configuration.Model;
 using Mirea.Api.Endpoint.Sync.Common;
+using Mirea.Tools.Schedule.Parser.Domain;
 using Mirea.Tools.Schedule.WebParser;
 using Mirea.Tools.Schedule.WebParser.Common.Domain;
 using System;
+using System.Collections.Concurrent;
 using System.Collections.Generic;
 using System.Diagnostics;
 using System.Linq;
@@ -220,35 +222,15 @@ internal partial class ScheduleSynchronizer(UberDbContext dbContext, IOptionsSna
         await dbContext.LessonAssociations.BulkSynchronizeAsync(_lessonAssociation.GetAll(), bulkOperation => bulkOperation.BatchSize = 1000, cancellationToken);
     }
 
-    public async Task StartSync(CancellationToken cancellationToken)
+    private async Task Sync(Func<CancellationToken, Task<List<GroupResult>>> parseDataAsync, CancellationToken cancellationToken)
     {
-        var pairPeriods = config.Value.ScheduleSettings?.PairPeriod;
-        var startTerm = config.Value.ScheduleSettings?.StartTerm;
-
-        if (pairPeriods == null || startTerm == null)
-        {
-            logger.LogWarning("It is not possible to synchronize the schedule due to the fact that the {Arg1} or {Arg2} variable is not initialized.",
-                nameof(pairPeriods),
-                nameof(startTerm));
-
-            return;
-        }
-
         Stopwatch watch = new();
         watch.Start();
 
-        var parser = new Parser
-        {
-            Pairs = pairPeriods
-                .ToDictionary(x => x.Key,
-                    x => (x.Value.Start, x.Value.End)),
-            TermStart = startTerm.Value.ToDateTime(new TimeOnly(0, 0, 0))
-        };
-
         try
         {
             logger.LogDebug("Start parsing schedule");
-            var data = await parser.ParseAsync(cancellationToken);
+            var data = await parseDataAsync(cancellationToken);
 
             watch.Stop();
             var parsingTime = watch.ElapsedMilliseconds;
@@ -282,11 +264,255 @@ internal partial class ScheduleSynchronizer(UberDbContext dbContext, IOptionsSna
         catch (Exception ex)
         {
             logger.LogError(ex, "An error occurred during synchronization.");
-            maintenanceMode.DisableMaintenanceMode();
             throw;
         }
+        finally
+        {
+            maintenanceMode.DisableMaintenanceMode();
+        }
     }
 
+    public async Task StartSync(CancellationToken cancellationToken)
+    {
+        var pairPeriods = config.Value.ScheduleSettings?.PairPeriod
+            .ToDictionary(x => x.Key, x => (x.Value.Start, x.Value.End));
+
+        var startTerm = config.Value.ScheduleSettings?.StartTerm;
+
+        if (pairPeriods == null || startTerm == null)
+        {
+            logger.LogWarning("It is not possible to synchronize the schedule due to the fact that the {Arg1} or {Arg2} variable is not initialized.",
+                nameof(pairPeriods),
+                nameof(startTerm));
+
+            return;
+        }
+
+        var parser = new Parser
+        {
+            Pairs = pairPeriods
+                .ToDictionary(x => x.Key,
+                    x => (x.Value.Start, x.Value.End)),
+            TermStart = startTerm.Value.ToDateTime(new TimeOnly(0, 0, 0))
+        };
+
+        try
+        {
+            await Sync(parser.ParseAsync, cancellationToken);
+        }
+        catch (Exception ex)
+        {
+            logger.LogError(ex, "An error occurred during synchronization.");
+            throw;
+        }
+        finally
+        {
+            maintenanceMode.DisableMaintenanceMode();
+        }
+    }
+
+    public async Task StartSync(List<(string File, string Campus)> files, CancellationToken cancellationToken)
+    {
+        await Task.Yield();
+        var pairPeriods = config.Value.ScheduleSettings?.PairPeriod
+            .ToDictionary(x => x.Key, x => (x.Value.Start, x.Value.End));
+
+        if (pairPeriods == null)
+        {
+            logger.LogWarning("It is not possible to synchronize the schedule due to the fact that the {Arg1} variable is not initialized.",
+                nameof(pairPeriods));
+
+            return;
+        }
+
+        try
+        {
+            Task<List<GroupResult>> ParseTask(CancellationToken ctx)
+            {
+                var mappedData = new ConcurrentBag<GroupResult>();
+
+                ParallelOptions options = new() { CancellationToken = ctx, MaxDegreeOfParallelism = Environment.ProcessorCount };
+                Parallel.ForEach(files, options, (file) =>
+                {
+                    var parser = new Tools.Schedule.Parser.Parser();
+                    var result = ConvertToGroupResults(parser.Parse(file.File, pairPeriods), file.Campus);
+
+                    foreach (var item in result) mappedData.Add(item);
+                });
+
+                return Task.FromResult(mappedData.ToList());
+            }
+
+            await Sync(ParseTask, cancellationToken);
+        }
+        catch (Exception ex)
+        {
+            logger.LogError(ex, "An error occurred during synchronization.");
+            throw;
+        }
+        finally
+        {
+            maintenanceMode.DisableMaintenanceMode();
+        }
+    }
+
+    private static List<GroupResult> ConvertToGroupResults(IEnumerable<GroupInfo> groups, string campusDefault, CancellationToken cancellationToken = default)
+    {
+        var result = new List<GroupResult>();
+
+        foreach (var group in groups)
+        {
+            cancellationToken.ThrowIfCancellationRequested();
+
+            foreach (var day in group.Days)
+            {
+                foreach (var pair in day.Lessons)
+                {
+                    foreach (var lesson in pair.Value)
+                    {
+                        if (string.IsNullOrWhiteSpace(lesson.TypeOfOccupation))
+                            continue;
+
+                        var (weeks, isExclude) = ParseWeeks(lesson.Discipline);
+
+                        var (lectureHalls, campuses) = ParseLectureHalls(lesson.LectureHall, campusDefault);
+
+                        var groupResult = new GroupResult
+                        {
+                            Day = day.DayOfWeek,
+                            Pair = pair.Key,
+                            IsEven = lesson.IsEven,
+                            Group = group.GroupName,
+                            Discipline = NormalizeDiscipline(lesson.Discipline),
+                            Professor = ParseProfessors(lesson.Professor),
+                            TypeOfOccupation = lesson.TypeOfOccupation,
+                            LectureHalls = lectureHalls,
+                            Campuses = campuses,
+                            SpecialWeek = weeks,
+                            IsExclude = isExclude
+                        };
+
+                        result.Add(groupResult);
+                    }
+                }
+            }
+        }
+
+        return result;
+    }
+
+    private static string[]? ParseProfessors(string? input)
+    {
+        if (string.IsNullOrWhiteSpace(input)) return null;
+
+        var normalized = Regex.Replace(input
+                .Replace("\n", " ")
+                .Replace(",", " "),
+            @"\s+", " ").Trim();
+
+        return ProfessorFullName().Matches(normalized)
+            .Select(m => $"{m.Groups["surname"].Value} {m.Groups["initials"].Value}".Trim())
+            .Where(x => !string.IsNullOrEmpty(x))
+            .ToArray();
+    }
+
+    private static (int[]? weeks, bool? isExclude) ParseWeeks(string discipline)
+    {
+        var match = ParseSpecificWeeks().Match(discipline);
+
+        if (!match.Success) return (null, null);
+
+        var numbers = new List<int>();
+        var ranges = match.Groups[2].Value.Split(',');
+
+        foreach (var range in ranges)
+        {
+            if (range.Contains('-'))
+            {
+                var parts = range.Split('-');
+                if (int.TryParse(parts[0], out var start) &&
+                    int.TryParse(parts[1], out var end))
+                {
+                    numbers.AddRange(Enumerable.Range(start, end - start + 1));
+                }
+            }
+            else
+                if (int.TryParse(range, out var num)) numbers.Add(num);
+        }
+
+        return (
+            weeks: numbers.Distinct().OrderBy(x => x).ToArray(),
+            isExclude: match.Groups[1].Success
+        );
+    }
+
+    private static string NormalizeDiscipline(string input)
+    {
+        var normalized = Regex.Replace(input
+                .Replace("\n", " ")
+                .Replace("\r", " "),
+            @"\s{2,}", " ");
+
+        normalized = Regex.Replace(normalized,
+            @"(\S+)\s(\S{3,})",
+            "$1 $2");
+
+        normalized = ParseSpecificWeeks().Replace(normalized, "");
+
+        return normalized.Trim();
+    }
+
+    private static (string[]? lectureHalls, string[]? campuses) ParseLectureHalls(string? input, string defaultCampus)
+    {
+        if (string.IsNullOrWhiteSpace(input))
+            return (null, null);
+
+        var matches = ParseLectureCampus().Matches(input);
+        var lectureHalls = new List<string>();
+        var campuses = new List<string>();
+
+        foreach (Match match in matches)
+        {
+            if (match.Groups["lectureWithCampus"].Success)
+            {
+                var raw = match.Value.Split('(');
+                var campus = raw.LastOrDefault()?.Trim(')').Trim();
+                var lecture = raw.FirstOrDefault()?.Trim();
+
+                if (string.IsNullOrEmpty(campus) || string.IsNullOrEmpty(lecture))
+                    continue;
+
+                campuses.Add(campus);
+                lectureHalls.Add(lecture);
+            }
+            else if (match.Groups["lecture"].Success)
+            {
+                var lecture = match.Value.Trim();
+                if (string.IsNullOrEmpty(lecture))
+                    continue;
+
+                campuses.Add(defaultCampus);
+                lectureHalls.Add(lecture);
+            }
+        }
+
+        return (
+            lectureHalls: lectureHalls.ToArray(),
+            campuses: campuses.ToArray()
+        );
+    }
+
+
+
     [GeneratedRegex(@"\w{4}-\d{2}-\d{2}(?=\s?\d?\s?[Пп]/?[Гг]\s?\d?)?")]
     private static partial Regex OnlyGroupName();
+
+    [GeneratedRegex(@"(?<surname>[А-ЯЁ][а-яё]+(-[А-ЯЁ][а-яё]+)?)\s*(?<initials>[А-ЯЁ]\.[А-ЯЁ]?\.?)?", RegexOptions.IgnorePatternWhitespace)]
+    private static partial Regex ProfessorFullName();
+
+    [GeneratedRegex(@"([Кк]р\.?)?\s*((\d+-\d+|\d+)(,\s*\d+(-\d+)?)*)\s*[Нн]\.?", RegexOptions.IgnoreCase, "ru-RU")]
+    private static partial Regex ParseSpecificWeeks();
+
+    [GeneratedRegex(@"(?<lectureWithCampus>[^,.\n]+\s?\([А-Яа-яA-Za-z]+-?\d+\))|(?<lecture>[^,.\n]+)")]
+    private static partial Regex ParseLectureCampus();
 }