QWERTYkez.OpenXmlProcessors/QWERTYkez.ExcelProcessor/PlaceholderFinder.cs

namespace QWERTYkez.ExcelProcessor;

internal static class PlaceholderFinder
{
    /// <summary>Находит плейсхолдеры $...$ в указанных листах.</summary>
    public static ISet<string> FindInDocument(SpreadsheetDocument doc, params WorksheetPart[] worksheets)
    {
        var result = new NormalizedSet();
        if (doc.WorkbookPart is null) return result;

        var sharedStringTable = doc.WorkbookPart.SharedStringTablePart?.SharedStringTable;
        foreach (var ws in worksheets)
            ProcessWorksheet(ws, sharedStringTable, result);

        return result;
    }

    /// <summary>Находит плейсхолдеры во всех листах документа.</summary>
    public static ISet<string> FindInDocument(SpreadsheetDocument doc)
    {
        if (doc.WorkbookPart is null) return new NormalizedSet();
        return FindInDocument(doc, [.. doc.WorkbookPart.WorksheetParts]);
    }

    static void ProcessWorksheet(WorksheetPart wsPart, SharedStringTable? sharedStrings, ISet<string> result)
    {
        var worksheet = wsPart.Worksheet;
        if (worksheet is null) return;

        ProcessCells(worksheet, sharedStrings, result);
        ProcessHeaderFooter(worksheet, result);
    }

    static void ProcessCells(Worksheet worksheet, SharedStringTable? sharedStrings, ISet<string> result)
    {
        var sheetData = worksheet.GetFirstChild<SheetData>();
        if (sheetData is null) return;

        List<string>? tempList = null; // буфер для плейсхолдеров из одного текста

        foreach (var row in sheetData.Elements<Row>())
        {
            foreach (var cell in row.Elements<Cell>())
            {
                string text = GetCellValue(cell, sharedStrings);
                if (string.IsNullOrEmpty(text)) continue;

#if DEBUG
                if (text.Contains('$'))
                    Debug.WriteLine($"[PlaceholderFinder] Cell {cell.CellReference}: '{text}'");
#endif
                if (FindPlaceholdersInText(text, ref tempList))
                {
                    foreach (var ph in tempList!)
                        result.Add(ph);
                    tempList.Clear();
                }
            }
        }
    }

    static void ProcessHeaderFooter(Worksheet worksheet, ISet<string> result)
    {
        var hf = worksheet.Descendants<HeaderFooter>().FirstOrDefault();
        if (hf is null) return;

        var texts = new[]
        {
            hf.OddHeader?.Text, hf.OddFooter?.Text,
            hf.EvenHeader?.Text, hf.EvenFooter?.Text,
            hf.FirstHeader?.Text, hf.FirstFooter?.Text
        };

        List<string>? tempList = null;
        foreach (var text in texts)
        {
            if (string.IsNullOrEmpty(text)) continue;
            if (FindPlaceholdersInText(text!, ref tempList))
            {
                foreach (var ph in tempList!)
                    result.Add(ph);
                tempList.Clear();
            }
        }
    }

    static string GetCellValue(Cell cell, SharedStringTable? sharedStrings)
    {
        if (cell?.CellValue is null && cell?.InlineString is null)
            return string.Empty;

        // Встроенная строка
        if (cell.InlineString is not null)
            return string.Concat(cell.InlineString.Descendants<Text>().Select(t => t.Text));

        // Обычное значение или общая строка
        string raw = cell.CellValue!.InnerText;
        if (cell.DataType?.Value == CellValues.SharedString)
        {
            if (sharedStrings is not null && int.TryParse(raw, out int idx) && idx >= 0 && idx < sharedStrings.Count!)
            {
                var si = sharedStrings.ElementAt(idx);
                return string.Concat(si.Descendants<Text>().Select(t => t.Text));
            }
            return string.Empty;
        }
        return raw;
    }

    static unsafe bool FindPlaceholdersInText(string text, ref List<string>? output)
    {
        fixed (char* pText = text)
        {
            char* start = pText;
            char* end = pText + text.Length;
            bool found = false;

            while (start < end)
            {
                // Ищем '$'
                while (start < end && *start != '$') start++;
                if (start >= end) break;
                char* open = start++;
                if (start >= end) break;

                // Ищем закрывающий '$'
                char* contentStart = start;
                while (start < end && *start != '$') start++;
                if (start >= end) break;
                char* close = start++;

                int len = (int)(close - contentStart);
                if (len > 0)
                {
                    found = true;
                    output ??= [];
                    output.Add(new string(contentStart, 0, len));
                }
            }
            return found;
        }
    }
}