pwiz: Fixed four vendor readers failing from an installer-based install

pwiz-sharp/Pwiz.sln

@@ -93,6 +93,10 @@ Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Installer.Tests", "pwiz\tes
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "MsBenchmark", "Tools\Commandline\MsBenchmark\src\MsBenchmark.csproj", "{20ABA6CB-AA2D-40B2-99B0-DD7A85BC94DF}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "MsDiff", "Tools\Commandline\MsDiff\src\MsDiff.csproj", "{5D1FF001-3D1F-4A55-9E01-000000A5D1FF}"
+EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "MsDiff.Tests", "Tools\Commandline\MsDiff\test\MsDiff.Tests.csproj", "{5D1FF002-3D1F-4A55-9E01-000000B5D1FF}"
+EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "TraData", "pwiz\src\TraData\TraData.csproj", "{8F972942-1D04-4F46-B541-6CBC521FF292}"
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "TraData.Tests", "pwiz\test\TraData.Tests\TraData.Tests.csproj", "{DC3E3099-0E07-44E1-8F04-055DD3FD3572}"
@@ -308,6 +312,14 @@ Global
 		{20ABA6CB-AA2D-40B2-99B0-DD7A85BC94DF}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{20ABA6CB-AA2D-40B2-99B0-DD7A85BC94DF}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{20ABA6CB-AA2D-40B2-99B0-DD7A85BC94DF}.Release|Any CPU.Build.0 = Release|Any CPU
+		{5D1FF001-3D1F-4A55-9E01-000000A5D1FF}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{5D1FF001-3D1F-4A55-9E01-000000A5D1FF}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{5D1FF001-3D1F-4A55-9E01-000000A5D1FF}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{5D1FF001-3D1F-4A55-9E01-000000A5D1FF}.Release|Any CPU.Build.0 = Release|Any CPU
+		{5D1FF002-3D1F-4A55-9E01-000000B5D1FF}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{5D1FF002-3D1F-4A55-9E01-000000B5D1FF}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{5D1FF002-3D1F-4A55-9E01-000000B5D1FF}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{5D1FF002-3D1F-4A55-9E01-000000B5D1FF}.Release|Any CPU.Build.0 = Release|Any CPU
 		{8F972942-1D04-4F46-B541-6CBC521FF292}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
 		{8F972942-1D04-4F46-B541-6CBC521FF292}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{8F972942-1D04-4F46-B541-6CBC521FF292}.Release|Any CPU.ActiveCfg = Release|Any CPU
@@ -409,6 +421,8 @@ Global
 		{FF96077C-6894-4611-8B73-211F356B88A5} = {050D548F-5BA2-4CC0-839A-A485500BF642}
 		{2A9B3D7E-4C81-4F12-9E3A-58D1F2B7E406} = {9B31505B-3D7C-4F5F-88CC-2C018DCC041A}
 		{20ABA6CB-AA2D-40B2-99B0-DD7A85BC94DF} = {1044EA12-4417-4A1F-802D-3B7E0AE3324E}
+		{5D1FF001-3D1F-4A55-9E01-000000A5D1FF} = {1044EA12-4417-4A1F-802D-3B7E0AE3324E}
+		{5D1FF002-3D1F-4A55-9E01-000000B5D1FF} = {9B31505B-3D7C-4F5F-88CC-2C018DCC041A}
 		{8F972942-1D04-4F46-B541-6CBC521FF292} = {1044EA12-4417-4A1F-802D-3B7E0AE3324E}
 		{DC3E3099-0E07-44E1-8F04-055DD3FD3572} = {9B31505B-3D7C-4F5F-88CC-2C018DCC041A}
 		{F6D07FD2-A9F1-4A86-8F3C-9370E1EBF5C7} = {050D548F-5BA2-4CC0-839A-A485500BF642}

pwiz-sharp/Tools/Commandline/MsDiff/src/ArgParser.cs

@@ -0,0 +1,96 @@
+using Pwiz.Data.Common.Diff;
+
+namespace Pwiz.Tools.MsDiff;
+
+/// <summary>Thrown when the user asked for help rather than a comparison.</summary>
+internal sealed class ArgParseHelpRequested : Exception;
+
+/// <summary>Parsed msdiff-sharp command line.</summary>
+internal sealed class Config
+{
+    public string FileA { get; set; } = string.Empty;
+    public string FileB { get; set; } = string.Empty;
+    public DiffConfig DiffConfig { get; } = new();
+}
+
+/// <summary>
+/// Command-line parsing for msdiff-sharp, matching cpp <c>msdiff</c>'s boost::program_options
+/// surface: two positional filenames plus <c>-p</c> and <c>-i</c>.
+/// </summary>
+internal static class ArgParser
+{
+    public static string Usage() =>
+        """
+        Usage: msdiff [options] filename1 filename2
+        Compare two mass spec data files.
+
+        Options:
+          -p [ --precision ] arg (=1e-06)  : set floating point precision for comparing binary data
+          -i [ --ignore ]                  : ignore metadata (compare scan binary data and
+                                             important scan metadata only)
+
+        Questions, comments, and bug reports:
+        https://github.com/ProteoWizard
+        support@proteowizard.org
+        """;
+
+    public static Config Parse(string[] args)
+    {
+        ArgumentNullException.ThrowIfNull(args);
+        var config = new Config();
+        var positional = new List<string>();
+
+        for (int i = 0; i < args.Length; i++)
+        {
+            string a = args[i];
+            switch (a)
+            {
+                case "-h" or "--help" or "-?":
+                    throw new ArgParseHelpRequested();
+
+                case "-i" or "--ignore":
+                    // cpp declares this zero_tokens, so it never consumes a value.
+                    config.DiffConfig.IgnoreMetadata = true;
+                    break;
+
+                case "-p" or "--precision":
+                    config.DiffConfig.Precision = ParsePrecision(Next(args, ref i, a));
+                    break;
+
+                default:
+                    if (a.StartsWith("--precision=", StringComparison.Ordinal))
+                        config.DiffConfig.Precision = ParsePrecision(a["--precision=".Length..]);
+                    else if (a.StartsWith('-') && a.Length > 1)
+                        throw new ArgumentException($"unrecognized option '{a}'");
+                    else
+                        positional.Add(a);
+                    break;
+            }
+        }
+
+        // cpp throws the usage string when it does not get exactly two filenames.
+        if (positional.Count != 2)
+            throw new ArgumentException(
+                $"expected exactly 2 filenames, got {positional.Count}");
+
+        config.FileA = positional[0];
+        config.FileB = positional[1];
+        return config;
+    }
+
+    private static string Next(string[] args, ref int i, string option)
+    {
+        if (i + 1 >= args.Length) throw new ArgumentException($"option '{option}' requires a value");
+        return args[++i];
+    }
+
+    private static double ParsePrecision(string s)
+    {
+        // Invariant culture on purpose: "1e-5" must parse the same wherever this runs, and the
+        // container sets no particular culture.
+        if (!double.TryParse(s, System.Globalization.NumberStyles.Float,
+                             System.Globalization.CultureInfo.InvariantCulture, out double v))
+            throw new ArgumentException($"could not parse precision '{s}'");
+        return v;
+    }
+}

pwiz-sharp/Tools/Commandline/MsDiff/src/Differ.cs

@@ -0,0 +1,89 @@
+using Pwiz.Data.MsData;
+using Pwiz.Data.MsData.Diff;
+using Pwiz.Data.MsData.Readers;
+using Pwiz.Vendor.Bruker;
+using Pwiz.Vendor.Thermo;
+using Pwiz.Vendor.Waters;
+
+namespace Pwiz.Tools.MsDiff;
+
+/// <summary>Reads two data files and reports how they differ.</summary>
+internal sealed class Differ
+{
+    private readonly ReaderList _readers;
+
+    public Differ()
+    {
+        // The same list msconvert builds, so msdiff can read anything msconvert can write from.
+        _readers = ThermoReaderRegistration.CreateDefaultWithThermo();
+        _readers.Add(new Reader_Bruker());
+        _readers.Add(new Reader_Waters());
+        _readers.Add(new Pwiz.Vendor.Agilent.Reader_Agilent());
+        _readers.Add(new Pwiz.Vendor.Sciex.Reader_Sciex());
+        _readers.Add(new Pwiz.Vendor.Shimadzu.Reader_Shimadzu());
+        _readers.Add(new Pwiz.Vendor.UNIFI.Reader_UNIFI());
+        _readers.Add(new Pwiz.Vendor.UIMF.Reader_UIMF());
+        _readers.Add(new Pwiz.Vendor.Mobilion.Reader_Mobilion());
+    }
+
+    /// <summary>
+    /// Compares the two files named in <paramref name="config"/>, writing the report to
+    /// <paramref name="output"/>. Returns 1 when they differ and 0 when they do not, matching
+    /// cpp msdiff's <c>return diff;</c>.
+    /// </summary>
+    public int Run(Config config, TextWriter output)
+    {
+        // Disposed: a vendor- or mzML-backed MSData holds its source file open, and a caller
+        // that diffs then replaces one of the inputs would otherwise hit a sharing violation.
+        using var a = Read(config.FileA);
+        using var b = Read(config.FileB);
+
+        var result = MSDataDiff.Compare(a, b, config.DiffConfig);
+
+        // The two summary lines come first and are always written, even when nothing differs.
+        //
+        // cpp reaches the same wording by printing the diff object through TextWriter, which
+        // renders each list as "<name> (N <units>)" holding only the entries that differ. It
+        // prints nothing at all when the files match, because the whole report is guarded by
+        // `if (diff)`. Callers that grep for "0 spectra" therefore see a FAILURE for two
+        // identical files - the one case where matching cpp byte for byte would be worse than
+        // being correct, so these lines are unconditional here.
+        output.WriteLine(Summary("spectrumList", result.SpectraDiffering, "spectra"));
+        output.WriteLine(Summary("chromatogramList", result.ChromatogramsDiffering, "chromatograms"));
+
+        if (!result.Differs) return 0;
+
+        output.Write(result.Report);
+        return 1;
+    }
+
+    /// <summary>
+    /// Renders one list's summary line: <c>spectrumList (0 spectra)</c> when nothing differs,
+    /// <c>spectrumList (12 differing spectra)</c> when something does.
+    /// </summary>
+    /// <remarks>
+    /// The extra word in the non-zero form is deliberate, and it is the one place this port
+    /// does not simply reproduce cpp's wording. Consumers test for equality by grepping for
+    /// the unanchored string "0 spectra" - the container's vendor sweep does exactly that -
+    /// and cpp's uniform "(N spectra)" makes that grep match the tail of any count ending in
+    /// zero. A Mobilion fixture differing in all 19570 spectra prints "(19570 spectra)",
+    /// which contains "0 spectra", so the sweep would report a PASS for a file that matched
+    /// nothing. Putting a word between the digits and the unit means only a real zero can
+    /// match. The right long-term fix is to anchor the grep on "(0 spectra)"; until every
+    /// consumer does, emitting a string that cannot be misread is the safer half of the deal.
+    /// </remarks>
+    private static string Summary(string listName, int differing, string unit) =>
+        differing == 0
+            ? $"{listName} (0 {unit})"
+            : $"{listName} ({differing} differing {unit})";
+
+    private MSData Read(string filename)
+    {
+        if (!File.Exists(filename) && !Directory.Exists(filename))
+            throw new FileNotFoundException($"no such file or directory: {filename}", filename);
+
+        var msd = new MSData();
+        _readers.Read(filename, msd, new ReaderConfig());
+        return msd;
+    }
+}

pwiz-sharp/Tools/Commandline/MsDiff/src/MsDiff.csproj

@@ -0,0 +1,75 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <RootNamespace>Pwiz.Tools.MsDiff</RootNamespace>
+    <AssemblyName>msdiff</AssemblyName>
+    <Description>Command-line comparison of two mass spec data files. Port of pwiz_tools/commandline/msdiff.</Description>
+
+    <!-- Inherits from Pwiz.Vendor.Thermo - the Thermo SDK needs en-US culture. -->
+    <InvariantGlobalization>false</InvariantGlobalization>
+
+    <!-- See the same switch in MsConvert.csproj: the wiff2 SDK deserializes its bundled
+         assembly resources through BinaryFormatter. -->
+    <EnableUnsafeBinaryFormatterSerialization>true</EnableUnsafeBinaryFormatterSerialization>
+    <UseWindowsForms>False</UseWindowsForms>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Util\Util.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Common\Common.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\MsData\MsData.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Analysis\Analysis.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Common\Vendor.Common.csproj" />
+  </ItemGroup>
+
+  <!-- The same vendor set msconvert carries. cpp msdiff builds an ExtendedReaderList, so it can
+       diff a vendor file directly against an mzML; keeping that means a reader missing here
+       would silently degrade to "unknown format" rather than compare. -->
+  <ItemGroup>
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Thermo\Thermo.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Bruker\Bruker.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Waters\Waters.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Agilent\Agilent.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Sciex\Sciex.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Shimadzu\Shimadzu.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\UNIFI\UNIFI.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\UIMF\UIMF.csproj" />
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Mobilion\Mobilion.csproj" />
+    <!-- Runtime-only plugin; see the identical block in MsConvert.csproj. -->
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Sciex\Wiff2\Sciex.Wiff2.csproj">
+      <ReferenceOutputAssembly>false</ReferenceOutputAssembly>
+      <OutputItemType>Content</OutputItemType>
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </ProjectReference>
+    <ProjectReference Include="$(PwizSharpRoot)pwiz\src\Vendor\Sciex\OfxLoggingStub\OfxLoggingStub.csproj"
+                      Condition="'$(NativeVendorsAvailable)' == 'true'">
+      <ReferenceOutputAssembly>false</ReferenceOutputAssembly>
+      <OutputItemType>Content</OutputItemType>
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </ProjectReference>
+  </ItemGroup>
+
+  <ItemGroup>
+    <InternalsVisibleTo Include="MsDiff.Tests" />
+  </ItemGroup>
+
+  <!-- Side-by-side wiff2 dlls. Same rationale and the same trailing-slash trap documented in
+       MsConvert.csproj's copy of this target; without it every .wiff2 fails to open. -->
+  <Target Name="StageWiff2NativeAssemblies" AfterTargets="Build" Condition="'$(NativeVendorsAvailable)' == 'true'">
+    <PropertyGroup>
+      <_SciexWiff2DepsDir>$(PwizVendorArchivesPath)\Sciex</_SciexWiff2DepsDir>
+      <_SciexAssembliesDir>$(PwizVendorAssembliesPath)\Sciex</_SciexAssembliesDir>
+      <_Wiff2OutDir>$([MSBuild]::EnsureTrailingSlash('$(OutputPath)'))wiff2</_Wiff2OutDir>
+    </PropertyGroup>
+    <MakeDir Directories="$(_Wiff2OutDir)" />
+    <Copy SourceFiles="$(_SciexWiff2DepsDir)\System.Data.SQLite.dll" DestinationFiles="$(_Wiff2OutDir)\System.Data.SQLite.dll" OverwriteReadOnlyFiles="true" />
+    <Copy SourceFiles="$(_SciexAssembliesDir)\SQLite.Interop.dll" DestinationFiles="$(_Wiff2OutDir)\SQLite.Interop.dll" OverwriteReadOnlyFiles="true" />
+    <Copy SourceFiles="$(_SciexWiff2DepsDir)\Unity.Abstractions.dll" DestinationFiles="$(_Wiff2OutDir)\Unity.Abstractions.dll" OverwriteReadOnlyFiles="true" />
+    <Copy SourceFiles="@(VendorNativeCrtFile)" DestinationFolder="$(_Wiff2OutDir)" SkipUnchangedFiles="true" OverwriteReadOnlyFiles="true" />
+  </Target>
+
+  <!-- VC++ redistributable beside the vendor DLLs; see the targets file. -->
+  <Import Project="$(PwizRepoRoot)\pwiz_tools\Shared\Lib\VendorNativeCrt.targets" />
+
+</Project>

pwiz-sharp/Tools/Commandline/MsDiff/src/Program.cs

@@ -0,0 +1,39 @@
+namespace Pwiz.Tools.MsDiff;
+
+/// <summary>Entry point for the msdiff-sharp CLI.</summary>
+public static class Program
+{
+    /// <summary>
+    /// Runs msdiff-sharp. Returns 0 when the two files match, 1 when they differ or the run
+    /// failed, and 2 for a bad command line - the first two matching cpp msdiff, whose
+    /// <c>main</c> returns the diff itself and 1 from its catch blocks.
+    /// </summary>
+    public static int Main(string[] args)
+    {
+        // Hook the vendor SDK on-demand resolver before any Reader_* is touched, exactly as
+        // msconvert does; without it a vendor input fails to load its SDK.
+        Pwiz.Vendor.Common.VendorSdkLoader.RegisterAssemblyResolver();
+        try
+        {
+            var config = ArgParser.Parse(args);
+            return new Differ().Run(config, Console.Out);
+        }
+        catch (ArgParseHelpRequested)
+        {
+            Console.Out.WriteLine(ArgParser.Usage());
+            return 0;
+        }
+        catch (ArgumentException ex)
+        {
+            Console.Error.WriteLine("Error: " + ex.Message);
+            Console.Error.WriteLine();
+            Console.Error.WriteLine(ArgParser.Usage());
+            return 2;
+        }
+        catch (Exception ex)
+        {
+            Console.Error.WriteLine(ex.Message);
+            return 1;
+        }
+    }
+}