Compare commits

..

2 Commits

Author SHA1 Message Date
EliotJones
3131dae49e passing -r will repeat parsing the set n times, -f will run a single file 2025-07-26 17:30:43 -05:00
EliotJones
52c0635273 support performance profiling information in console runner 2025-07-26 15:04:03 -05:00
14 changed files with 342 additions and 70 deletions

View File

@@ -1,4 +1,5 @@
using System.Buffers;
using System;
using System.Buffers;
namespace UglyToad.PdfPig.Core;

View File

@@ -71,7 +71,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return obj is IndirectReference other && Equals(other);
}

View File

@@ -16,7 +16,7 @@
/// <summary>
/// Convert the string to bytes using the ISO 8859-1 encoding.
/// </summary>
public static byte[]? StringAsLatin1Bytes(string? s)
public static byte[] StringAsLatin1Bytes(string s)
{
if (s == null)
{

View File

@@ -264,7 +264,7 @@
/// Try to convert raw bytes to a PdfDocEncoding encoded string. If unsupported characters are encountered
/// meaning we cannot safely round-trip the value to bytes this will instead return false.
/// </summary>
public static bool TryConvertBytesToString(ReadOnlySpan<byte> bytes, out string? result)
public static bool TryConvertBytesToString(ReadOnlySpan<byte> bytes, out string result)
{
result = null;
if (bytes.Length == 0)

View File

@@ -70,7 +70,7 @@
/// <summary>
/// Returns a value indicating whether this <see cref="PdfLine"/> is equal to a specified <see cref="PdfLine"/> .
/// </summary>
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return obj is PdfLine other && Equals(other);
}

View File

@@ -83,7 +83,7 @@
/// <summary>
/// Returns a value indicating whether this <see cref="PdfPoint"/> is equal to a specified <see cref="PdfPoint"/> .
/// </summary>
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return obj is PdfPoint other && Equals(other);
}

View File

@@ -177,7 +177,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return obj is PdfRectangle other && Equals(other);
}

View File

@@ -241,8 +241,8 @@
public bool IsClosed()
{
var filteredCount = 0;
IPathCommand? last = null;
IPathCommand? first = null;
IPathCommand last = null;
IPathCommand first = null;
for (int i = Commands.Count - 1; i >= 0; i--)
{
var cmd = Commands[i];
@@ -376,14 +376,14 @@
/// Gets a <see cref="PdfRectangle"/> which entirely contains the geometry of the defined path.
/// </summary>
/// <returns>For paths which don't define any geometry this returns <see langword="null"/>.</returns>
public static PdfRectangle? GetBoundingRectangle(IReadOnlyList<PdfSubpath>? path)
public static PdfRectangle? GetBoundingRectangle(IReadOnlyList<PdfSubpath> path)
{
if (path == null || path.Count == 0)
{
return null;
}
var bboxes = path.Select(x => x.GetBoundingRectangle()).Where(x => x.HasValue).Select(x => x!.Value).ToList();
var bboxes = path.Select(x => x.GetBoundingRectangle()).Where(x => x.HasValue).Select(x => x.Value).ToList();
if (bboxes.Count == 0)
{
return null;
@@ -433,7 +433,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return (obj is Close);
}
@@ -479,7 +479,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
if (obj is Move move)
{
@@ -545,7 +545,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
if (obj is Line line)
{
@@ -651,7 +651,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
if (obj is QuadraticBezierCurve curve)
{
@@ -809,7 +809,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
if (obj is CubicBezierCurve curve)
{
@@ -944,7 +944,7 @@
/// <summary>
/// Compares two <see cref="PdfSubpath"/>s for equality. Paths will only be considered equal if the commands which construct the paths are in the same order.
/// </summary>
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
if (!(obj is PdfSubpath path) || Commands.Count != path.Commands.Count)
{

View File

@@ -463,7 +463,7 @@
}
/// <inheritdoc />
public override bool Equals(object? obj)
public override bool Equals(object obj)
{
return obj is TransformationMatrix other && Equals(other);
}

View File

@@ -7,8 +7,6 @@
<GenerateDocumentationFile>true</GenerateDocumentationFile>
<SignAssembly>true</SignAssembly>
<AssemblyOriginatorKeyFile>..\pdfpig.snk</AssemblyOriginatorKeyFile>
<Nullable>enable</Nullable>
<WarningsAsErrors>nullable</WarningsAsErrors>
</PropertyGroup>
<ItemGroup Condition="'$(TargetFramework)'=='net462'">
<PackageReference Include="System.ValueTuple" Version="4.5.0" />

View File

@@ -89,7 +89,7 @@ namespace UglyToad.PdfPig.Core
/// <inheritdoc />
public override bool TryGetSecond(out B b)
{
b = default!;
b = default(B);
return false;
}
@@ -135,7 +135,7 @@ namespace UglyToad.PdfPig.Core
/// <inheritdoc />
public override bool TryGetFirst(out A a)
{
a = default!;
a = default(A);
return false;
}

View File

@@ -1,5 +1,10 @@
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Diagnostics.CodeAnalysis;
using System.Globalization;
using System.IO;
using System.Linq;
using System.Text;
using Console = System.Console;
@@ -7,6 +12,128 @@ namespace UglyToad.PdfPig.ConsoleRunner
{
public static class Program
{
private class OptionalArg
{
public required string ShortSymbol { get; init; }
public required string Symbol { get; init; }
public required bool SupportsValue { get; init; }
public string? Value { get; set; }
}
private class ParsedArgs
{
public required IReadOnlyList<OptionalArg> SuppliedArgs { get; init; }
public required string SuppliedDirectoryPath { get; init; }
}
private const string FileSymbol = "f";
private const string RepeatSymbol = "r";
private static IReadOnlyList<OptionalArg> GetSupportedArgs() =>
[
new OptionalArg
{
SupportsValue = false,
ShortSymbol = "nr",
Symbol = "no-recursion"
},
new OptionalArg
{
SupportsValue = true,
ShortSymbol = "o",
Symbol = "output"
},
new OptionalArg
{
SupportsValue = true,
ShortSymbol = "l",
Symbol = "limit"
},
new OptionalArg
{
SupportsValue = true,
ShortSymbol = "f",
Symbol = "file"
},
new OptionalArg
{
SupportsValue = true,
ShortSymbol = "r",
Symbol = "repeats"
}
];
private static bool TryParseArgs(
string[] args,
[NotNullWhen(true)] out ParsedArgs? parsed)
{
parsed = null;
string? path = null;
var suppliedOpts = new List<OptionalArg>();
var opts = GetSupportedArgs();
for (var i = 0; i < args.Length; i++)
{
var str = args[i];
var isOptFlag = str.StartsWith('-');
if (!isOptFlag)
{
if (path == null)
{
path = str;
}
else
{
return false;
}
}
else
{
var item = opts.SingleOrDefault(x =>
string.Equals("-" + x.ShortSymbol, str, StringComparison.OrdinalIgnoreCase)
|| string.Equals("--" + x.Symbol, str, StringComparison.OrdinalIgnoreCase));
if (item == null)
{
return false;
}
if (item.SupportsValue)
{
if (i == args.Length - 1)
{
return false;
}
i++;
item.Value = args[i];
}
suppliedOpts.Add(item);
}
}
if (path == null)
{
return false;
}
parsed = new ParsedArgs
{
SuppliedArgs = suppliedOpts,
SuppliedDirectoryPath = path
};
return true;
}
public static int Main(string[] args)
{
if (args.Length == 0)
@@ -15,71 +142,140 @@ namespace UglyToad.PdfPig.ConsoleRunner
return 7;
}
var path = args[0];
if (!Directory.Exists(path))
if (!TryParseArgs(args, out var parsed))
{
Console.WriteLine($"The provided path is not a valid directory: {path}.");
var strJoined = string.Join(" ", args);
Console.WriteLine($"Unrecognized arguments passed: {strJoined}");
return 7;
}
var maxCount = default(int?);
if (args.Length > 1 && int.TryParse(args[1], out var countIn))
if (!Directory.Exists(parsed.SuppliedDirectoryPath))
{
maxCount = countIn;
Console.WriteLine($"The provided path is not a valid directory: {parsed.SuppliedDirectoryPath}.");
return 7;
}
int? maxCount = null;
var limit = parsed.SuppliedArgs.SingleOrDefault(x => x.ShortSymbol == "l");
if (limit?.Value != null && int.TryParse(limit.Value, CultureInfo.InvariantCulture, out var maxCountArg))
{
Console.WriteLine($"Limiting input files to first: {maxCountArg}");
maxCount = maxCountArg;
}
var noRecursionMode = parsed.SuppliedArgs.Any(x => x.ShortSymbol == "nr");
var outputOpt = parsed.SuppliedArgs.SingleOrDefault(x => x.ShortSymbol == "o" && x.Value != null);
var fileOpt = parsed.SuppliedArgs.SingleOrDefault(x => x.ShortSymbol == FileSymbol && x.Value != null);
var hasError = false;
var errorBuilder = new StringBuilder();
var fileList = Directory.GetFiles(path, "*.pdf", SearchOption.AllDirectories);
var fileList = Directory.GetFiles(
parsed.SuppliedDirectoryPath,
"*.pdf",
noRecursionMode ? SearchOption.TopDirectoryOnly : SearchOption.AllDirectories)
.OrderBy(x => x).ToList();
var runningCount = 0;
Console.WriteLine($"Found {fileList.Length} files.");
Console.WriteLine($"{GetCleanFilename("File")}| Size\t| Words\t| Pages");
foreach (var file in fileList)
if (fileOpt?.Value != null)
{
if (maxCount.HasValue && runningCount >= maxCount)
{
break;
}
fileList = fileList.Where(x => x.EndsWith(fileOpt.Value, StringComparison.OrdinalIgnoreCase)).ToList();
}
try
var repeatOpt = parsed.SuppliedArgs.SingleOrDefault(x => x.ShortSymbol == RepeatSymbol);
var repeats = 1;
if (repeatOpt?.Value != null && int.TryParse(repeatOpt.Value, CultureInfo.InvariantCulture, out repeats))
{
}
Console.WriteLine($"Found {fileList.Count} files.");
Console.WriteLine();
PrintTableColumns("File", "Size", "Words", "Pages", "Open cost (μs)", "Total cost (μs)", "Page cost (μs)");
var dataList = new List<DataRecord>();
var sw = new Stopwatch();
for (int i = 0; i < repeats; i++)
{
foreach (var file in fileList)
{
var numWords = 0;
var numPages = 0;
using (var pdfDocument = PdfDocument.Open(file))
if (maxCount.HasValue && runningCount >= maxCount)
{
foreach (var page in pdfDocument.GetPages())
{
numPages++;
foreach (var word in page.GetWords())
{
if (word != null)
{
numWords++;
}
}
}
break;
}
var filename = Path.GetFileName(file);
try
{
var numWords = 0;
var numPages = 0;
long openMicros;
long totalPageMicros;
var size = new FileInfo(file);
sw.Reset();
sw.Start();
Console.WriteLine($"{GetCleanFilename(filename)}| {size.Length}\t| {numWords}\t| {numPages}");
using (var pdfDocument = PdfDocument.Open(file))
{
sw.Stop();
openMicros = sw.Elapsed.Microseconds;
sw.Start();
foreach (var page in pdfDocument.GetPages())
{
numPages++;
foreach (var word in page.GetWords())
{
if (word != null)
{
numWords++;
}
}
}
sw.Stop();
totalPageMicros = sw.Elapsed.Microseconds;
}
var filename = Path.GetFileName(file);
var size = new FileInfo(file);
var item = new DataRecord
{
FileName = filename,
OpenCostMicros = openMicros,
Pages = numPages,
Size = size.Length,
Words = numWords,
TotalCostMicros = totalPageMicros + openMicros,
PerPageMicros = Math.Round(totalPageMicros / (double)Math.Max(numPages, 1), 2)
};
dataList.Add(item);
PrintTableColumns(
item.FileName,
item.Size,
item.Words,
item.Pages,
item.OpenCostMicros,
item.TotalCostMicros,
item.PerPageMicros);
}
catch (Exception ex)
{
hasError = true;
errorBuilder.AppendLine($"Parsing document {file} failed due to an error.")
.Append(ex)
.AppendLine();
}
runningCount++;
}
catch (Exception ex)
{
hasError = true;
errorBuilder.AppendLine($"Parsing document {file} failed due to an error.")
.Append(ex)
.AppendLine();
}
runningCount++;
}
if (hasError)
@@ -88,12 +284,71 @@ namespace UglyToad.PdfPig.ConsoleRunner
return 5;
}
if (outputOpt != null && outputOpt.Value != null)
{
WriteOutput(outputOpt.Value, dataList);
}
Console.WriteLine("Complete! :)");
return 0;
}
private static string GetCleanFilename(string name, int maxLength = 30)
private static void WriteOutput(string outPath, IReadOnlyList<DataRecord> records)
{
using var fs = File.OpenWrite(outPath);
using var sw = new StreamWriter(fs);
sw.WriteLine("File,Size,Words,Pages,Open Cost,Total Cost,Per Page");
foreach (var record in records)
{
var sizeStr = record.Size.ToString("D", CultureInfo.InvariantCulture);
var wordsStr = record.Words.ToString("D", CultureInfo.InvariantCulture);
var pagesStr = record.Pages.ToString("D", CultureInfo.InvariantCulture);
var openCostStr = record.OpenCostMicros.ToString("D", CultureInfo.InvariantCulture);
var totalCostStr = record.TotalCostMicros.ToString("D", CultureInfo.InvariantCulture);
var ppcStr = record.PerPageMicros.ToString("F2", CultureInfo.InvariantCulture);
var numericPartsStr = string.Join(",",
[
sizeStr,
wordsStr,
pagesStr,
openCostStr,
totalCostStr,
ppcStr
]);
sw.WriteLine($"\"{record.FileName}\",{numericPartsStr}");
}
sw.Flush();
}
private static void PrintTableColumns(params object[] values)
{
for (var i = 0; i < values.Length; i++)
{
var value = values[i];
var valueStr = value.ToString();
var cleaned = GetCleanStr(valueStr ?? string.Empty);
var padChars = 16 - cleaned.Length;
var padding = padChars > 0 ? new string(' ', padChars) : string.Empty;
var padded = cleaned + padding;
Console.Write("| ");
Console.Write(padded);
}
Console.WriteLine();
}
private static string GetCleanStr(string name, int maxLength = 16)
{
if (name.Length <= maxLength)
{
@@ -105,4 +360,21 @@ namespace UglyToad.PdfPig.ConsoleRunner
return name.Substring(0, maxLength);
}
}
internal class DataRecord
{
public required string FileName { get; init; }
public required long Size { get; init; }
public required int Words { get; init; }
public required int Pages { get; init; }
public required long OpenCostMicros { get; init; }
public required long TotalCostMicros { get; init; }
public required double PerPageMicros { get; init; }
}
}

View File

@@ -1,8 +1,8 @@
{
"profiles": {
"UglyToad.PdfPig.ConsoleRunner": {
"commandName": "Project",
"commandLineArgs": "\"C:\\temp\\pdfs\\archive\""
"commandName": "Project",
"commandLineArgs": "\"C:\\temp\\pdfs\\archive\""
}
}
}

View File

@@ -4,6 +4,7 @@
<LangVersion>latest</LangVersion>
<OutputType>Exe</OutputType>
<TargetFramework>net8</TargetFramework>
<Nullable>enable</Nullable>
</PropertyGroup>
<ItemGroup>