taiqane/UmlautAdaptarr
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor TItleMatchingService

Browse Source
This commit is contained in:
pcjones
2024-02-14 21:00:24 +01:00
parent e739affb39
commit 96f8ff9332
1 changed files with 13 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
UmlautAdaptarr/Services/TitleMatchingService.cs
View File

@@ -20,7 +20,7 @@ namespace UmlautAdaptarr.Services
if (titleElement != null) if (titleElement != null)
{ {
var originalTitle = titleElement.Value; var originalTitle = titleElement.Value;
var normalizedOriginalTitle = NormalizeTitle(originalTitle); var cleanTitleSeperatedBySpace = ReplaceSeperatorsWithSpace(originalTitle.RemoveAccentButKeepGermanUmlauts());
var categoryElement = item.Element("category"); var categoryElement = item.Element("category");
var category = categoryElement?.Value; var category = categoryElement?.Value;
@@ -34,7 +34,7 @@ namespace UmlautAdaptarr.Services
if (useCacheService) if (useCacheService)
{ {
// Use CacheService to find a matching SearchItem by title // Use CacheService to find a matching SearchItem by title
searchItem = cacheService.SearchItemByTitle(mediaType, normalizedOriginalTitle); searchItem = cacheService.SearchItemByTitle(mediaType, cleanTitleSeperatedBySpace);
} }
if (searchItem == null) if (searchItem == null)
@@ -46,10 +46,10 @@ namespace UmlautAdaptarr.Services
switch (mediaType) switch (mediaType)
{ {
case "tv": case "tv":
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, normalizedOriginalTitle!); FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, cleanTitleSeperatedBySpace!);
break; break;
case "movie": case "movie":
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, normalizedOriginalTitle!); FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, cleanTitleSeperatedBySpace!);
break; break;
case "audio": case "audio":
FindAndReplaceForAudio(searchItem, titleElement, originalTitle!); FindAndReplaceForAudio(searchItem, titleElement, originalTitle!);
@@ -68,9 +68,9 @@ namespace UmlautAdaptarr.Services
var authorMatch = FindBestMatch(searchItem.AuthorMatchVariations, originalTitle.NormalizeForComparison(), originalTitle); var authorMatch = FindBestMatch(searchItem.AuthorMatchVariations, originalTitle.NormalizeForComparison(), originalTitle);
var titleMatch = FindBestMatch(searchItem.TitleMatchVariations, originalTitle.NormalizeForComparison(), originalTitle); var titleMatch = FindBestMatch(searchItem.TitleMatchVariations, originalTitle.NormalizeForComparison(), originalTitle);
if (authorMatch.Item1 && titleMatch.Item1) if (authorMatch.foundMatch && titleMatch.foundMatch)
{ {
int matchEndPositionInOriginal = Math.Max(authorMatch.Item3, titleMatch.Item3); int matchEndPositionInOriginal = Math.Max(authorMatch.bestEndInOriginal, titleMatch.bestEndInOriginal);
// Check and adjust for immediate following delimiter // Check and adjust for immediate following delimiter
char[] delimiters = [' ', '-', '_', '.']; char[] delimiters = [' ', '-', '_', '.'];
@@ -96,7 +96,7 @@ namespace UmlautAdaptarr.Services
} }
private Tuple<bool, int, int> FindBestMatch(string[] variations, string normalizedOriginal, string originalTitle) private (bool foundMatch, int bestStart, int bestEndInOriginal) FindBestMatch(string[] variations, string normalizedOriginal, string originalTitle)
{ {
bool found = false; bool found = false;
int bestStart = int.MaxValue; int bestStart = int.MaxValue;
@@ -119,8 +119,8 @@ namespace UmlautAdaptarr.Services
} }
} }
if (!found) return Tuple.Create(false, 0, 0); if (!found) return (false, 0, 0);
return Tuple.Create(found, bestStart, bestEndInOriginal); return (found, bestStart, bestEndInOriginal);
} }
// Maps an index from the normalized string back to a corresponding index in the original string // Maps an index from the normalized string back to a corresponding index in the original string
@@ -159,6 +159,7 @@ namespace UmlautAdaptarr.Services
var titleMatchVariations = searchItem.TitleMatchVariations; var titleMatchVariations = searchItem.TitleMatchVariations;
var expectedTitle = searchItem.ExpectedTitle; var expectedTitle = searchItem.ExpectedTitle;
var variationsOrderedByLength = titleMatchVariations!.OrderByDescending(variation => variation.Length); var variationsOrderedByLength = titleMatchVariations!.OrderByDescending(variation => variation.Length);
// Attempt to find a variation that matches the start of the original title // Attempt to find a variation that matches the start of the original title
foreach (var variation in variationsOrderedByLength) foreach (var variation in variationsOrderedByLength)
{ {
@@ -198,8 +199,8 @@ namespace UmlautAdaptarr.Services
} }
} }
// Clean up any leading separators from the suffix // Clean up any leading separator from the suffix
suffix = Regex.Replace(suffix, "^[._ ]+", ""); suffix = Regex.Replace(suffix, "^ +", "");
// TODO EVALUTE! definitely make this optional - this adds GERMAN to the title is the title is german to make sure it's recognized as german // TODO EVALUTE! definitely make this optional - this adds GERMAN to the title is the title is german to make sure it's recognized as german
// can lead to problems with shows such as "dark" that have international dubs // can lead to problems with shows such as "dark" that have international dubs
@@ -225,9 +226,8 @@ namespace UmlautAdaptarr.Services
} }
} }
private static string NormalizeTitle(string title) private static string ReplaceSeperatorsWithSpace(string title)
{ {
title = title.RemoveAccentButKeepGermanUmlauts();
// Replace all known separators with space for normalization // Replace all known separators with space for normalization
return WordSeperationCharRegex().Replace(title, " ".ToString()); return WordSeperationCharRegex().Replace(title, " ".ToString());
} }