From bfef81393883fd1cfec5d2f13865c2d5ca3d6431 Mon Sep 17 00:00:00 2001
From: Jaap de Haan <261428+jdehaan@users.noreply.github.com>
Date: Sat, 25 Apr 2026 09:56:32 +0000
Subject: [PATCH] feat: Dump

---
 PdfInfoTool/Command/Dump.cs | 134 +++++++++++++++++++++++++++++++++++-
 PdfInfoTool/Options.cs      |   8 +++
 2 files changed, 141 insertions(+), 1 deletion(-)

diff --git a/PdfInfoTool/Command/Dump.cs b/PdfInfoTool/Command/Dump.cs
index abd32f2..2c14511 100644
--- a/PdfInfoTool/Command/Dump.cs
+++ b/PdfInfoTool/Command/Dump.cs
@@ -1,10 +1,142 @@
 using System;
+using System.Globalization;
+using System.IO;
+using System.Text;
+using SafeRapidPdf;
+using SafeRapidPdf.Objects;
+using SafeRapidPdf.Parsing;
+
 namespace PdfInfoTool;
 
 internal static partial class Command
 {
     internal static int RunDumpAndReturnExitCode(DumpOptions opts)
     {
-        throw new NotImplementedException();
+        if (opts.Hex && opts.Binary)
+        {
+            Console.Error.WriteLine("--hex and --binary are mutually exclusive.");
+            return 1;
+        }
+
+        try
+        {
+            var file = PdfFile.Parse(opts.FileName);
+
+            if (opts.Binary)
+            {
+                DumpBinary(file);
+                return 0;
+            }
+
+            if (!opts.Quiet)
+            {
+                Console.WriteLine("PDF Version: {0}", file.Version);
+            }
+
+            if (opts.Hex)
+            {
+                DumpHex(file);
+            }
+            else
+            {
+                WriteObject(file, depth: 0, includeType: opts.Verbose);
+
+                if (!opts.Quiet)
+                {
+                    Console.WriteLine("Top-level objects: {0}", file.Items.Count);
+                }
+            }
+
+            return 0;
+        }
+        catch (Exception ex) when (
+            ex is ParsingException or IOException or UnauthorizedAccessException)
+        {
+            Console.Error.WriteLine(ex.Message);
+            return 1;
+        }
+    }
+
+    /// <summary>
+    /// Recursively print all stream data as hex strings, one stream per line,
+    /// prefixed with the owning indirect object number.
+    /// </summary>
+    private static void DumpHex(IPdfObject obj, string prefix = "")
+    {
+        if (obj is PdfIndirectObject indirect)
+        {
+            prefix = indirect.ObjectNumber.ToString(CultureInfo.InvariantCulture);
+        }
+
+        if (obj is PdfStream stream)
+        {
+            byte[] decoded = stream.Decode();
+            Console.WriteLine("{0}: {1}", prefix, ToHexString(decoded));
+            return;
+        }
+
+        if (obj.IsContainer && obj.Items is not null)
+        {
+            foreach (var item in obj.Items)
+            {
+                DumpHex(item, prefix);
+            }
+        }
+    }
+
+    /// <summary>
+    /// Write the raw decoded bytes of every stream in the file to stdout.
+    /// Intended for piping to external tools.
+    /// </summary>
+    private static void DumpBinary(IPdfObject obj)
+    {
+        if (obj is PdfStream stream)
+        {
+            byte[] decoded = stream.Decode();
+            using var stdout = new BinaryWriter(Console.OpenStandardOutput(), Encoding.UTF8, leaveOpen: true);
+            stdout.Write(decoded);
+            return;
+        }
+
+        if (obj.IsContainer && obj.Items is not null)
+        {
+            foreach (var item in obj.Items)
+            {
+                DumpBinary(item);
+            }
+        }
+    }
+
+    private static string ToHexString(byte[] data)
+    {
+        var sb = new StringBuilder(data.Length * 2);
+        foreach (byte b in data)
+        {
+            _ = sb.Append(b.ToString("x2", CultureInfo.InvariantCulture));
+        }
+        return sb.ToString();
+    }
+
+    private static void WriteObject(IPdfObject obj, int depth, bool includeType)
+    {
+        var indent = new string(' ', depth * 2);
+        if (includeType)
+        {
+            Console.WriteLine($"{indent}{obj.ObjectType}: {obj.Text}");
+        }
+        else
+        {
+            Console.WriteLine($"{indent}{obj.Text}");
+        }
+
+        if (!obj.IsContainer || obj.Items is null)
+        {
+            return;
+        }
+
+        foreach (var item in obj.Items)
+        {
+            WriteObject(item, depth + 1, includeType);
+        }
     }
 }
diff --git a/PdfInfoTool/Options.cs b/PdfInfoTool/Options.cs
index 61d9de5..e1cb6fb 100644
--- a/PdfInfoTool/Options.cs
+++ b/PdfInfoTool/Options.cs
@@ -27,6 +27,14 @@ internal sealed class DumpOptions : IOptions
     public bool Verbose { get; set; }
     public bool Quiet { get; set; }
     public string FileName { get; set; }
+
+    [Option('x', "hex",
+        HelpText = "Dump decoded stream data as a hex string (for use in unit tests).")]
+    public bool Hex { get; set; }
+
+    [Option('b', "binary",
+        HelpText = "Write decoded stream data as raw bytes to stdout (for piping to other tools).")]
+    public bool Binary { get; set; }
 }
 
 [Verb("show", HelpText = "Show object contents in a human readable way.")]