mmd-pdf/Services/PdfRedactText.cs

using System;
using System.Collections.Generic;
using PdfSharpCore.Drawing;
using PdfSharpCore.Pdf;
using PdfSharpCore.Pdf.Content;
using PdfSharpCore.Pdf.Content.Objects;

namespace MmdPdf.Services
{
    // ============================================================
    // Removes original page text that sits under an opaque cover.
    //
    // When a user edits existing text, the editor lays an opaque CoverAnnotation
    // over the original run and draws the replacement on top. The page LOOKED
    // right but the original glyphs were still in the content stream, so text
    // extraction and copy/paste returned the old value alongside the new one -
    // an edited quantity on a delivery note stayed recoverable.
    //
    // This pass walks the page content stream and blanks every text-showing
    // operator whose text origin falls inside a cover rectangle, so the old
    // glyphs are gone from the file, not merely hidden. Positioning operators
    // are left untouched, so nothing else on the page moves.
    //
    // Deliberate limitation: a run is removed only when its ORIGIN is inside the
    // cover. A single show-text operator that starts outside the cover and runs
    // into it is left alone rather than risk deleting text the user did not edit.
    // The editor's own covers are generated to bound the run they replace, so
    // this is exact for edits made in this application.
    // ============================================================
    internal static class PdfRedactText
    {
        /// <summary>
        /// Blanks text drawn with its origin inside any of <paramref name="rectsPdf"/>.
        /// Rectangles are in PDF user space (origin bottom-left, y up), points.
        /// Returns the number of show-text operators blanked.
        /// </summary>
        internal static int RemoveTextUnder(PdfPage page, IReadOnlyList<XRect> rectsPdf)
        {
            if (page is null || rectsPdf is null || rectsPdf.Count == 0) return 0;

            CSequence content;
            try { content = ContentReader.ReadContent(page); }
            catch { return 0; }   // unparsable stream: leave the page exactly as it was

            var state = new TextState();
            int blanked = Walk(content, state, rectsPdf);
            if (blanked == 0) return 0;

            try { page.Contents.ReplaceContent(content); }
            catch { return 0; }
            return blanked;
        }

        // ── graphics + text state ──────────────────────────────────────────

        private sealed class TextState
        {
            internal XMatrix Ctm = XMatrix.Identity;
            internal readonly Stack<XMatrix> CtmStack = new Stack<XMatrix>();
            internal XMatrix Tm = XMatrix.Identity;    // text matrix
            internal XMatrix Tlm = XMatrix.Identity;   // text line matrix
            internal double Leading;                   // TL
        }

        private static int Walk(CSequence seq, TextState st, IReadOnlyList<XRect> rects)
        {
            int blanked = 0;
            foreach (var obj in seq)
            {
                if (obj is CSequence inner && obj is not CArray)
                {
                    blanked += Walk(inner, st, rects);
                    continue;
                }
                if (obj is not COperator op) continue;

                string name = op.OpCode?.Name ?? "";
                switch (name)
                {
                    case "q":
                        st.CtmStack.Push(st.Ctm);
                        break;
                    case "Q":
                        if (st.CtmStack.Count > 0) st.Ctm = st.CtmStack.Pop();
                        break;
                    case "cm":
                        if (TryMatrix(op.Operands, out var cm)) st.Ctm = cm * st.Ctm;
                        break;

                    case "BT":
                        st.Tm = st.Tlm = XMatrix.Identity;
                        break;
                    case "ET":
                        st.Tm = st.Tlm = XMatrix.Identity;
                        break;

                    case "Tm":
                        if (TryMatrix(op.Operands, out var tm)) st.Tm = st.Tlm = tm;
                        break;
                    case "TL":
                        st.Leading = Num(op.Operands, 0);
                        break;
                    case "Td":
                        NextLine(st, Num(op.Operands, 0), Num(op.Operands, 1));
                        break;
                    case "TD":
                        st.Leading = -Num(op.Operands, 1);
                        NextLine(st, Num(op.Operands, 0), Num(op.Operands, 1));
                        break;
                    case "T*":
                        NextLine(st, 0, -st.Leading);
                        break;

                    case "'":
                        NextLine(st, 0, -st.Leading);
                        blanked += BlankIfInside(op, st, rects);
                        break;
                    case "\"":
                        NextLine(st, 0, -st.Leading);
                        blanked += BlankIfInside(op, st, rects);
                        break;
                    case "Tj":
                    case "TJ":
                        blanked += BlankIfInside(op, st, rects);
                        break;
                }
            }
            return blanked;
        }

        private static void NextLine(TextState st, double tx, double ty)
        {
            var t = new XMatrix(1, 0, 0, 1, tx, ty);
            st.Tlm = t * st.Tlm;
            st.Tm = st.Tlm;
        }

        private static int BlankIfInside(COperator op, TextState st, IReadOnlyList<XRect> rects)
        {
            var m = st.Tm * st.Ctm;
            var origin = new XPoint(m.OffsetX, m.OffsetY);

            bool hit = false;
            for (int i = 0; i < rects.Count; i++)
            {
                var r = rects[i];
                // A baseline sits slightly above the visual bottom of the glyphs, so allow a
                // small pad below the cover; otherwise a run whose baseline is a point under
                // the cover edge survives.
                if (origin.X >= r.X - 1.0 && origin.X <= r.X + r.Width + 1.0 &&
                    origin.Y >= r.Y - 2.0 && origin.Y <= r.Y + r.Height + 2.0)
                {
                    hit = true;
                    break;
                }
            }
            if (!hit) return 0;

            return BlankStrings(op.Operands) ? 1 : 0;
        }

        private static bool BlankStrings(CSequence operands)
        {
            if (operands is null) return false;
            bool any = false;
            for (int i = 0; i < operands.Count; i++)
            {
                switch (operands[i])
                {
                    case CString s when s.Value.Length > 0:
                        s.Value = "";
                        any = true;
                        break;
                    case CArray arr:
                        for (int j = 0; j < arr.Count; j++)
                        {
                            if (arr[j] is CString cs && cs.Value.Length > 0)
                            {
                                cs.Value = "";
                                any = true;
                            }
                        }
                        break;
                }
            }
            return any;
        }

        // ── operand helpers ────────────────────────────────────────────────

        private static double Num(CSequence operands, int index)
        {
            if (operands is null || index >= operands.Count) return 0;
            return operands[index] switch
            {
                CInteger i => i.Value,
                CReal r => r.Value,
                _ => 0
            };
        }

        private static bool TryMatrix(CSequence operands, out XMatrix m)
        {
            m = XMatrix.Identity;
            if (operands is null || operands.Count < 6) return false;
            m = new XMatrix(Num(operands, 0), Num(operands, 1), Num(operands, 2),
                            Num(operands, 3), Num(operands, 4), Num(operands, 5));
            return true;
        }
    }
}