PlotDirector/PlotLine/Services/ManuscriptTextStatisticsService.cs

using PlotLine.Data;
using PlotLine.Models;

namespace PlotLine.Services;

public interface IManuscriptTextStatisticsService
{
    ManuscriptTextStatistics Calculate(string? sourceText);
    ManuscriptTextStatistics Aggregate(IEnumerable<ManuscriptTextStatistics?> statistics);
    Task RecalculateSceneAsync(int sceneId);
    Task RecalculateChapterAsync(int chapterId);
    Task RecalculateBookAsync(int bookId);
}

public sealed class ManuscriptTextStatisticsService(
    IManuscriptTextStatisticsRepository repository,
    IOnboardingBuildRepository manuscriptSources) : IManuscriptTextStatisticsService
{
    public ManuscriptTextStatistics Calculate(string? sourceText)
    {
        var paragraphs = StoryIntelligenceParagraphs.Split(sourceText);
        var totalWords = 0;
        var dialogueWords = 0;
        var proseWords = 0;
        var dialogueParagraphs = 0;
        var proseParagraphs = 0;
        var inDialogue = false;

        foreach (var paragraph in paragraphs)
        {
            var paragraphStartInDialogue = inDialogue;
            var paragraphHasDialogue = paragraphStartInDialogue;
            var paragraphHasProse = false;
            var firstContentIndex = FirstNonWhitespaceIndex(paragraph);
            var segmentStart = 0;

            for (var index = 0; index < paragraph.Length; index++)
            {
                var quote = QuoteKind(paragraph[index]);
                if (quote == ManuscriptQuoteKind.None)
                {
                    continue;
                }

                if (inDialogue && IsContinuationOpeningQuote(quote, index, firstContentIndex))
                {
                    AddSegment(paragraph, segmentStart, index, inDialogue, ref totalWords, ref dialogueWords, ref proseWords, ref paragraphHasDialogue, ref paragraphHasProse);
                    segmentStart = index + 1;
                    continue;
                }

                AddSegment(paragraph, segmentStart, index, inDialogue, ref totalWords, ref dialogueWords, ref proseWords, ref paragraphHasDialogue, ref paragraphHasProse);

                if (quote == ManuscriptQuoteKind.SmartOpen)
                {
                    inDialogue = true;
                    paragraphHasDialogue = true;
                }
                else if (quote == ManuscriptQuoteKind.SmartClose)
                {
                    inDialogue = false;
                }
                else
                {
                    inDialogue = !inDialogue;
                    if (inDialogue)
                    {
                        paragraphHasDialogue = true;
                    }
                }

                segmentStart = index + 1;
            }

            AddSegment(paragraph, segmentStart, paragraph.Length, inDialogue, ref totalWords, ref dialogueWords, ref proseWords, ref paragraphHasDialogue, ref paragraphHasProse);

            if (paragraphHasDialogue)
            {
                dialogueParagraphs++;
            }

            if (paragraphHasProse)
            {
                proseParagraphs++;
            }
        }

        return new ManuscriptTextStatistics
        {
            TotalWordCount = totalWords,
            DialogueWordCount = dialogueWords,
            ProseWordCount = proseWords,
            TotalParagraphCount = paragraphs.Count,
            DialogueParagraphCount = dialogueParagraphs,
            ProseParagraphCount = proseParagraphs
        };
    }

    public ManuscriptTextStatistics Aggregate(IEnumerable<ManuscriptTextStatistics?> statistics)
    {
        var rows = statistics.Where(row => row is not null).Select(row => row!).ToList();
        return new ManuscriptTextStatistics
        {
            TotalWordCount = rows.Sum(row => row.TotalWordCount),
            DialogueWordCount = rows.Sum(row => row.DialogueWordCount),
            ProseWordCount = rows.Sum(row => row.ProseWordCount),
            TotalParagraphCount = rows.Sum(row => row.TotalParagraphCount),
            DialogueParagraphCount = rows.Sum(row => row.DialogueParagraphCount),
            ProseParagraphCount = rows.Sum(row => row.ProseParagraphCount)
        };
    }

    public async Task RecalculateSceneAsync(int sceneId)
    {
        var source = await repository.GetSourceBySceneAsync(sceneId);
        if (source is null)
        {
            return;
        }

        await repository.SaveAsync(source.SceneID, Calculate(source.SourceText));
    }

    public async Task RecalculateChapterAsync(int chapterId)
    {
        var sources = await manuscriptSources.ListSceneManuscriptSourcesByChapterAsync(chapterId);
        foreach (var source in sources)
        {
            await repository.SaveAsync(source.SceneID, Calculate(source.SourceText));
        }
    }

    public async Task RecalculateBookAsync(int bookId)
    {
        var sources = await manuscriptSources.ListSceneManuscriptSourcesByBookAsync(bookId);
        foreach (var source in sources)
        {
            await repository.SaveAsync(source.SceneID, Calculate(source.SourceText));
        }
    }

    private static void AddSegment(
        string paragraph,
        int start,
        int end,
        bool isDialogue,
        ref int totalWords,
        ref int dialogueWords,
        ref int proseWords,
        ref bool paragraphHasDialogue,
        ref bool paragraphHasProse)
    {
        if (end <= start)
        {
            return;
        }

        var words = CountWords(paragraph[start..end]);
        if (words == 0)
        {
            return;
        }

        totalWords += words;
        if (isDialogue)
        {
            dialogueWords += words;
            paragraphHasDialogue = true;
        }
        else
        {
            proseWords += words;
            paragraphHasProse = true;
        }
    }

    private static int CountWords(string value)
        => string.IsNullOrWhiteSpace(value)
            ? 0
            : value.Split((char[]?)null, StringSplitOptions.RemoveEmptyEntries).Length;

    private static int FirstNonWhitespaceIndex(string paragraph)
    {
        for (var index = 0; index < paragraph.Length; index++)
        {
            if (!char.IsWhiteSpace(paragraph[index]))
            {
                return index;
            }
        }

        return -1;
    }

    private static bool IsContinuationOpeningQuote(ManuscriptQuoteKind quote, int index, int firstContentIndex)
        => index == firstContentIndex
            && (quote == ManuscriptQuoteKind.SmartOpen || quote == ManuscriptQuoteKind.Straight);

    private static ManuscriptQuoteKind QuoteKind(char value)
        => value switch
        {
            '"' => ManuscriptQuoteKind.Straight,
            '\u201c' => ManuscriptQuoteKind.SmartOpen,
            '\u201d' => ManuscriptQuoteKind.SmartClose,
            _ => ManuscriptQuoteKind.None
        };

    private enum ManuscriptQuoteKind
    {
        None,
        Straight,
        SmartOpen,
        SmartClose
    }
}