Rework architecture

DiscordChatExporter.Domain/Markdown/Matching/AggregateMatcher.cs

@@ -0,0 +1,46 @@
+using System.Collections.Generic;
+
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal class AggregateMatcher<T> : IMatcher<T>
+    {
+        private readonly IReadOnlyList<IMatcher<T>> _matchers;
+
+        public AggregateMatcher(IReadOnlyList<IMatcher<T>> matchers)
+        {
+            _matchers = matchers;
+        }
+
+        public AggregateMatcher(params IMatcher<T>[] matchers)
+            : this((IReadOnlyList<IMatcher<T>>) matchers)
+        {
+        }
+
+        public ParsedMatch<T>? TryMatch(StringPart stringPart)
+        {
+            ParsedMatch<T>? earliestMatch = null;
+
+            // Try to match the input with each matcher and get the match with the lowest start index
+            foreach (var matcher in _matchers)
+            {
+                // Try to match
+                var match = matcher.TryMatch(stringPart);
+
+                // If there's no match - continue
+                if (match == null)
+                    continue;
+
+                // If this match is earlier than previous earliest - replace
+                if (earliestMatch == null || match.StringPart.StartIndex < earliestMatch.StringPart.StartIndex)
+                    earliestMatch = match;
+
+                // If the earliest match starts at the very beginning - break,
+                // because it's impossible to find a match earlier than that
+                if (earliestMatch.StringPart.StartIndex == stringPart.StartIndex)
+                    break;
+            }
+
+            return earliestMatch;
+        }
+    }
+}

DiscordChatExporter.Domain/Markdown/Matching/IMatcher.cs

@@ -0,0 +1,49 @@
+using System;
+using System.Collections.Generic;
+
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal interface IMatcher<T>
+    {
+        ParsedMatch<T>? TryMatch(StringPart stringPart);
+    }
+
+    internal static class MatcherExtensions
+    {
+        public static IEnumerable<ParsedMatch<T>> MatchAll<T>(this IMatcher<T> matcher,
+            StringPart stringPart, Func<StringPart, T> transformFallback)
+        {
+            // Loop through segments divided by individual matches
+            var currentIndex = stringPart.StartIndex;
+            while (currentIndex < stringPart.EndIndex)
+            {
+                // Find a match within this segment
+                var match = matcher.TryMatch(stringPart.Slice(currentIndex, stringPart.EndIndex - currentIndex));
+
+                // If there's no match - break
+                if (match == null)
+                    break;
+
+                // If this match doesn't start immediately at current index - transform and yield fallback first
+                if (match.StringPart.StartIndex > currentIndex)
+                {
+                    var fallbackPart = stringPart.Slice(currentIndex, match.StringPart.StartIndex - currentIndex);
+                    yield return new ParsedMatch<T>(fallbackPart, transformFallback(fallbackPart));
+                }
+
+                // Yield match
+                yield return match;
+
+                // Shift current index to the end of the match
+                currentIndex = match.StringPart.StartIndex + match.StringPart.Length;
+            }
+
+            // If EOL wasn't reached - transform and yield remaining part as fallback
+            if (currentIndex < stringPart.EndIndex)
+            {
+                var fallbackPart = stringPart.Slice(currentIndex);
+                yield return new ParsedMatch<T>(fallbackPart, transformFallback(fallbackPart));
+            }
+        }
+    }
+}

DiscordChatExporter.Domain/Markdown/Matching/ParsedMatch.cs

@@ -0,0 +1,15 @@
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal class ParsedMatch<T>
+    {
+        public StringPart StringPart { get; }
+
+        public T Value { get; }
+
+        public ParsedMatch(StringPart stringPart, T value)
+        {
+            StringPart = stringPart;
+            Value = value;
+        }
+    }
+}

DiscordChatExporter.Domain/Markdown/Matching/RegexMatcher.cs

@@ -0,0 +1,40 @@
+using System;
+using System.Text.RegularExpressions;
+
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal class RegexMatcher<T> : IMatcher<T>
+    {
+        private readonly Regex _regex;
+        private readonly Func<StringPart, Match, T> _transform;
+
+        public RegexMatcher(Regex regex, Func<StringPart, Match, T> transform)
+        {
+            _regex = regex;
+            _transform = transform;
+        }
+
+        public RegexMatcher(Regex regex, Func<Match, T> transform)
+            : this(regex, (p, m) => transform(m))
+        {
+        }
+
+        public ParsedMatch<T>? TryMatch(StringPart stringPart)
+        {
+            var match = _regex.Match(stringPart.Target, stringPart.StartIndex, stringPart.Length);
+            if (!match.Success)
+                return null;
+
+            // Overload regex.Match(string, int, int) doesn't take the whole string into account,
+            // it effectively functions as a match check on a substring.
+            // Which is super weird because regex.Match(string, int) takes the whole input in context.
+            // So in order to properly account for ^/$ regex tokens, we need to make sure that
+            // the expression also matches on the bigger part of the input.
+            if (!_regex.IsMatch(stringPart.Target.Substring(0, stringPart.EndIndex), stringPart.StartIndex))
+                return null;
+
+            var stringPartMatch = stringPart.Slice(match.Index, match.Length);
+            return new ParsedMatch<T>(stringPartMatch, _transform(stringPartMatch, match));
+        }
+    }
+}

DiscordChatExporter.Domain/Markdown/Matching/StringMatcher.cs

@@ -0,0 +1,36 @@
+using System;
+
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal class StringMatcher<T> : IMatcher<T>
+    {
+        private readonly string _needle;
+        private readonly StringComparison _comparison;
+        private readonly Func<StringPart, T> _transform;
+
+        public StringMatcher(string needle, StringComparison comparison, Func<StringPart, T> transform)
+        {
+            _needle = needle;
+            _comparison = comparison;
+            _transform = transform;
+        }
+
+        public StringMatcher(string needle, Func<StringPart, T> transform)
+            : this(needle, StringComparison.Ordinal, transform)
+        {
+        }
+
+        public ParsedMatch<T>? TryMatch(StringPart stringPart)
+        {
+            var index = stringPart.Target.IndexOf(_needle, stringPart.StartIndex, stringPart.Length, _comparison);
+
+            if (index >= 0)
+            {
+                var stringPartMatch = stringPart.Slice(index, _needle.Length);
+                return new ParsedMatch<T>(stringPartMatch, _transform(stringPartMatch));
+            }
+
+            return null;
+        }
+    }
+}

DiscordChatExporter.Domain/Markdown/Matching/StringPart.cs

@@ -0,0 +1,36 @@
+using System.Text.RegularExpressions;
+
+namespace DiscordChatExporter.Domain.Markdown.Matching
+{
+    internal readonly struct StringPart
+    {
+        public string Target { get; }
+
+        public int StartIndex { get; }
+
+        public int Length { get; }
+
+        public int EndIndex { get; }
+
+        public StringPart(string target, int startIndex, int length)
+        {
+            Target = target;
+            StartIndex = startIndex;
+            Length = length;
+            EndIndex = startIndex + length;
+        }
+
+        public StringPart(string target)
+            : this(target, 0, target.Length)
+        {
+        }
+
+        public StringPart Slice(int newStartIndex, int newLength) => new StringPart(Target, newStartIndex, newLength);
+
+        public StringPart Slice(int newStartIndex) => Slice(newStartIndex, EndIndex - newStartIndex);
+
+        public StringPart Slice(Capture capture) => Slice(capture.Index, capture.Length);
+
+        public override string ToString() => Target.Substring(StartIndex, Length);
+    }
+}