3 Commits

Author SHA1 Message Date
pcjones
96f8ff9332 Refactor TItleMatchingService 2024-02-14 21:00:24 +01:00
pcjones
e739affb39 Use TitleMatchVariations instead of TitleSearchVariations in SearchItemByTitle 2024-02-14 20:55:13 +01:00
pcjones
92bdf14618 Remove test variable 2024-02-14 20:40:13 +01:00
2 changed files with 15 additions and 16 deletions

View File

@@ -84,7 +84,7 @@ namespace UmlautAdaptarr.Services
continue;
}
// After finding a potential item, compare normalizedTitle with each German title variation
foreach (var variation in item?.TitleSearchVariations ?? [])
foreach (var variation in item?.TitleMatchVariations ?? [])
{
var normalizedVariation = variation.RemoveAccentButKeepGermanUmlauts().ToLower();
if (normalizedTitle.StartsWith(variation, StringComparison.OrdinalIgnoreCase))

View File

@@ -20,7 +20,7 @@ namespace UmlautAdaptarr.Services
if (titleElement != null)
{
var originalTitle = titleElement.Value;
var normalizedOriginalTitle = NormalizeTitle(originalTitle);
var cleanTitleSeperatedBySpace = ReplaceSeperatorsWithSpace(originalTitle.RemoveAccentButKeepGermanUmlauts());
var categoryElement = item.Element("category");
var category = categoryElement?.Value;
@@ -34,7 +34,7 @@ namespace UmlautAdaptarr.Services
if (useCacheService)
{
// Use CacheService to find a matching SearchItem by title
searchItem = cacheService.SearchItemByTitle(mediaType, normalizedOriginalTitle);
searchItem = cacheService.SearchItemByTitle(mediaType, cleanTitleSeperatedBySpace);
}
if (searchItem == null)
@@ -46,10 +46,10 @@ namespace UmlautAdaptarr.Services
switch (mediaType)
{
case "tv":
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, normalizedOriginalTitle!);
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, cleanTitleSeperatedBySpace!);
break;
case "movie":
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, normalizedOriginalTitle!);
FindAndReplaceForMoviesAndTV(logger, searchItem, titleElement, originalTitle, cleanTitleSeperatedBySpace!);
break;
case "audio":
FindAndReplaceForAudio(searchItem, titleElement, originalTitle!);
@@ -68,13 +68,12 @@ namespace UmlautAdaptarr.Services
var authorMatch = FindBestMatch(searchItem.AuthorMatchVariations, 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);
var test = originalTitle[matchEndPositionInOriginal];
// Check and adjust for immediate following delimiter
char[] delimiters = new char[] { ' ', '-', '_', '.' };
char[] delimiters = [' ', '-', '_', '.'];
if (matchEndPositionInOriginal < originalTitle.Length && delimiters.Contains(originalTitle[matchEndPositionInOriginal]))
{
matchEndPositionInOriginal++; // Skip the delimiter if it's immediately after the match
@@ -97,7 +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;
int bestStart = int.MaxValue;
@@ -120,8 +119,8 @@ namespace UmlautAdaptarr.Services
}
}
if (!found) return Tuple.Create(false, 0, 0);
return Tuple.Create(found, bestStart, bestEndInOriginal);
if (!found) return (false, 0, 0);
return (found, bestStart, bestEndInOriginal);
}
// Maps an index from the normalized string back to a corresponding index in the original string
@@ -160,6 +159,7 @@ namespace UmlautAdaptarr.Services
var titleMatchVariations = searchItem.TitleMatchVariations;
var expectedTitle = searchItem.ExpectedTitle;
var variationsOrderedByLength = titleMatchVariations!.OrderByDescending(variation => variation.Length);
// Attempt to find a variation that matches the start of the original title
foreach (var variation in variationsOrderedByLength)
{
@@ -199,8 +199,8 @@ namespace UmlautAdaptarr.Services
}
}
// Clean up any leading separators from the suffix
suffix = Regex.Replace(suffix, "^[._ ]+", "");
// Clean up any leading separator from the 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
// can lead to problems with shows such as "dark" that have international dubs
@@ -226,9 +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
return WordSeperationCharRegex().Replace(title, " ".ToString());
}