PdfMarker_Web/PdfMarker/Services/PdfQuotaExtractor.cs

using System.Globalization;
using PdfMarker.Models;
using UglyToad.PdfPig;
using UglyToad.PdfPig.Content;

namespace PdfMarker.Services;

public class PdfQuotaExtractor
{
    // percentuale inferiore della pagina da escludere (cartiglio)
    private const double CartiglioCutoffY = 0.15; // basso
    private const double CartiglioMarginX = 0.05; // bordi laterali 5%

    
    public List<QuotaCandidate> Extract(string filePath)
    {
        var result = new List<QuotaCandidate>();

        using var document = PdfDocument.Open(filePath);

        foreach (var page in document.GetPages())
        {
            var pageWidth = page.Width;
            var pageHeight = page.Height;

            foreach (var word in page.GetWords())
            {
                if (!IsQuotaLikeText(word.Text))
                    continue;

                var type = ClassifyQuota(word.Text);
                if (type == QuotaType.Unknown)
                    continue;

                var box = word.BoundingBox;

                var x = box.Left / pageWidth;
                var y = 1 - (box.Bottom / pageHeight);
                var w = box.Width / pageWidth;
                var h = box.Height / pageHeight;

                // --- FILTRI CARTIGLIO ---

                // parte bassa del foglio
                if (y < CartiglioCutoffY)
                    continue;

                // bordi laterali
                if (x < CartiglioMarginX || x > 1 - CartiglioMarginX)
                    continue;

                // testi troppo piccoli (tipici del cartiglio)
                /*if (h < 0.003)
                    continue;*/

                // testi troppo corti e non significativi
                var t = word.Text.Trim();
                if (t.Length <= 1 && !t.Contains("Ø") && !t.Contains("R") && !t.Contains("0"))
                    continue;

                // --- AGGIUNTA QUOTA ---

                result.Add(new QuotaCandidate
                {
                    RawText = word.Text,
                    Type = type,

                    X = x,
                    Y = y,
                    Width = w,
                    Height = h,

                    IsHorizontal = box.Width >= box.Height,
                    IsVertical = box.Height > box.Width,

                    Confidence = ComputeConfidence(type, word)
                });
            }
        }

        return MergeSplitTexts(result);
    }

    private bool IsQuotaLikeText(string text)
    {
        if (string.IsNullOrWhiteSpace(text))
            return false;

        text = text.Replace(" ", "");

        // deve contenere almeno un numero
        if (!text.Any(char.IsDigit))
            return false;

        return
            text.Contains("Ø") ||
            text.StartsWith("R") ||
            text.Contains("°") ||
            text.Contains("x") ||
            text.Contains("×") ||
            IsNumeric(text);
    }


    private QuotaType ClassifyQuota(string text)
    {
        text = text.Replace(" ", "");

        if (text.Contains("Ø"))
            return QuotaType.Diameter;

        if (text.StartsWith("R"))
            return QuotaType.Radius;

        if (text.Contains("°"))
            return QuotaType.Angle;

        if (text.Contains("x") || text.Contains("×"))
            return QuotaType.Chamfer;

        if (IsNumeric(text))
            return QuotaType.Linear;

        return QuotaType.Unknown;
    }

    private bool IsNumeric(string text)
    {
        return double.TryParse(
            text.Replace(",", "."),
            NumberStyles.Any,
            CultureInfo.InvariantCulture,
            out _);
    }

    private double ComputeConfidence(QuotaType type, Word word)
    {
        var confidence = type switch
        {
            QuotaType.Diameter => 0.9,
            QuotaType.Radius   => 0.85,
            QuotaType.Angle    => 0.85,
            QuotaType.Chamfer  => 0.8,
            QuotaType.Linear   => 0.6,
            _                  => 0.3
        };

        var box = word.BoundingBox;
        var text = word.Text.Replace(" ", "");

        // penalizza testi verticali
        if (box.Height > box.Width)
            confidence *= 0.7;

        // penalizza numeri interi corti (1, 2, 3…)
        if (IsNumeric(text) && !text.Contains(",") && !text.Contains("."))
            confidence *= 0.7;

        return confidence;
    }
    
    private bool IsDecimalNumber(string text)
    {
        // accetta 37,50 – 12.7 – 1,5
        return double.TryParse(
                   text.Replace(",", "."),
                   System.Globalization.NumberStyles.Any,
                   System.Globalization.CultureInfo.InvariantCulture,
                   out _)
               && (text.Contains(",") || text.Contains("."));
    }
    
    private List<QuotaCandidate> MergeSplitTexts(List<QuotaCandidate> input)
    {
        var merged = new List<QuotaCandidate>();
        var used = new HashSet<QuotaCandidate>();

        for (int i = 0; i < input.Count; i++)
        {
            var current = input[i];
            if (used.Contains(current))
                continue;

            var cluster = new List<QuotaCandidate> { current };
            used.Add(current);

            foreach (var other in input)
            {
                if (used.Contains(other))
                    continue;

                // stesso tipo
                if (other.Type != current.Type)
                    continue;

                // molto vicini
                var dx = Math.Abs(current.X - other.X);
                var dy = Math.Abs(current.Y - other.Y);

                if (dx < 0.02 && dy < 0.02)
                {
                    cluster.Add(other);
                    used.Add(other);
                }
            }

            if (cluster.Count == 1)
            {
                merged.Add(current);
                continue;
            }

            // merge testi
            var mergedText = string.Join(" ",
                cluster
                    .Select(q => q.RawText.Trim())
                    .OrderBy(t => t.Length));

            merged.Add(new QuotaCandidate
            {
                RawText = mergedText,
                Type = current.Type,

                X = cluster.Average(q => q.X),
                Y = cluster.Average(q => q.Y),
                Width = cluster.Max(q => q.Width),
                Height = cluster.Max(q => q.Height),

                IsHorizontal = cluster.Any(q => q.IsHorizontal),
                IsVertical = cluster.Any(q => q.IsVertical),

                Confidence = cluster.Average(q => q.Confidence)
            });
        }

        return merged;
    }
}