From fb0755eb89b4471c7a5b9897546d14aef1e5df9c Mon Sep 17 00:00:00 2001
From: Glax <johanna@bernloehr.eu>
Date: Sat, 27 Apr 2024 19:09:12 +0200
Subject: [PATCH] Use NeedlemanWunsch for string comparison on Mangasee.cs
 Resolves #132 #167

---
 README.md                          |  1 +
 Tranga/MangaConnectors/Mangasee.cs | 54 +++++++-----------------------
 Tranga/Tranga.csproj               |  1 +
 3 files changed, 14 insertions(+), 42 deletions(-)

diff --git a/README.md b/README.md
index 05d6aee..6137396 100644
--- a/README.md
+++ b/README.md
@@ -85,6 +85,7 @@ That is why I wanted to create my own project, in a language I understand, and t
 - Newtonsoft.JSON
 - [PuppeteerSharp](https://www.puppeteersharp.com/)
 - [Html Agility Pack (HAP)](https://html-agility-pack.net/)
+- [Soenneker.Utils.String.NeedlemanWunsch](https://github.com/soenneker/soenneker.utils.string.needlemanwunsch)
 - 💙 Blåhaj 🦈
 
 <p align="right">(<a href="#readme-top">back to top</a>)</p>
diff --git a/Tranga/MangaConnectors/Mangasee.cs b/Tranga/MangaConnectors/Mangasee.cs
index e1ce0c2..067f840 100644
--- a/Tranga/MangaConnectors/Mangasee.cs
+++ b/Tranga/MangaConnectors/Mangasee.cs
@@ -4,6 +4,7 @@ using System.Text.RegularExpressions;
 using System.Xml.Linq;
 using HtmlAgilityPack;
 using Newtonsoft.Json;
+using Soenneker.Utils.String.NeedlemanWunsch;
 using Tranga.Jobs;
 
 namespace Tranga.MangaConnectors;
@@ -41,14 +42,6 @@ public class Mangasee : MangaConnector
             SearchResult[] filteredResults = FilteredResults(publicationTitle, searchResults);
             Log($"Total available manga: {searchResults.Length} Filtered down to: {filteredResults.Length}");
             
-            /*
-            Dictionary<SearchResult, int> levenshteinRelation = filteredResults.ToDictionary(result => result,
-                result =>
-                {
-                    Log($"Levenshtein {result.s}");
-                    return LevenshteinDistance(publicationTitle.Replace(" ", "").ToLower(), result.s.Replace(" ", "").ToLower());
-                });
-            Log($"After levenshtein: {levenshteinRelation.Count}");*/
 
             string[] urls = filteredResults.Select(result => $"https://mangasee123.com/manga/{result.i}").ToArray();
             List<Manga> searchResultManga = new();
@@ -70,42 +63,19 @@ public class Mangasee : MangaConnector
 
     private SearchResult[] FilteredResults(string publicationTitle, SearchResult[] unfilteredSearchResults)
     {
-        string[] bannedStrings = {"a", "the", "of", "as", "to", "no", "for", "on", "with", "be", "and", "in", "wa", "at"};
-        string[] cleanSplitPublicationTitle = publicationTitle.Split(' ')
-            .Where(part => part.Length > 0 && !bannedStrings.Contains(part.ToLower())).ToArray();
-        
-        return unfilteredSearchResults.Where(usr =>
+        Dictionary<SearchResult, int> similarity = new();
+        foreach (SearchResult sr in unfilteredSearchResults)
         {
-            string cleanSearchResultString = string.Join(' ', usr.s.Split(' ').Where(part => part.Length > 0 && !bannedStrings.Contains(part.ToLower())));
-            foreach(string splitPublicationTitlePart in cleanSplitPublicationTitle)
-                if (cleanSearchResultString.Contains(splitPublicationTitlePart, StringComparison.InvariantCultureIgnoreCase) ||
-                    cleanSearchResultString.Contains(splitPublicationTitlePart, StringComparison.InvariantCultureIgnoreCase))
-                    return true;
-            return false;
-        }).ToArray();
-    }
-
-    private int LevenshteinDistance(string a, string b)
-    {
-        if (b.Length == 0)
-            return a.Length;
-        if (a.Length == 0)
-            return b.Length;
-        if (a[0] == b[0])
-            return LevenshteinDistance(a[1..], b[1..]);
-
-        int case1 = LevenshteinDistance(a, b[1..]);
-        int case2 = LevenshteinDistance(a[1..], b[1..]);
-        int case3 = LevenshteinDistance(a[1..], b);
-
-        if (case1 < case2)
-        {
-            return 1 + (case1 < case3 ? case1 : case3);
-        }
-        else
-        {
-            return 1 + (case2 < case3 ? case2 : case3);
+            List<int> scores = new();
+            foreach (string se in sr.a)
+                scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(se.ToLower(), publicationTitle.ToLower()));
+            scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(sr.s.ToLower(), publicationTitle.ToLower()));
+            similarity.Add(sr, scores.Sum() / scores.Count);
         }
+
+        SearchResult[] similarity90 = similarity.Where(s => s.Value < 10).Select(s => s.Key).ToArray();
+
+        return similarity90;
     }
 
     public override Manga? GetMangaFromId(string publicationId)
diff --git a/Tranga/Tranga.csproj b/Tranga/Tranga.csproj
index 2e4242b..47b765f 100644
--- a/Tranga/Tranga.csproj
+++ b/Tranga/Tranga.csproj
@@ -11,6 +11,7 @@
       <PackageReference Include="HtmlAgilityPack" Version="1.11.46" />
       <PackageReference Include="Newtonsoft.Json" Version="13.0.3" />
       <PackageReference Include="PuppeteerSharp" Version="10.0.0" />
+      <PackageReference Include="Soenneker.Utils.String.NeedlemanWunsch" Version="2.1.301" />
     </ItemGroup>
 
     <ItemGroup>