Merge pull request #47 from neilharvey/nonstandard-office-files

Nonstandard office files
neilharvey · May 15, 2022 · e14962f · e14962f
2 parents 33fef7c + 8fe0642
commit e14962f
Show file tree

Hide file tree

Showing 7 changed files with 42 additions and 8 deletions.
diff --git a/.gitignore b/.gitignore
@@ -24,6 +24,8 @@ bld/
 
 # Visual Studio 2015 cache/options directory
 .vs/
+# Visual Studio code
+.vscode/
 # Uncomment if you have tasks that create the project's static files in wwwroot
 #wwwroot/
 

diff --git a/src/FileSignatures/FileSignatures.csproj b/src/FileSignatures/FileSignatures.csproj
@@ -13,13 +13,13 @@
     <GenerateAssemblyConfigurationAttribute>false</GenerateAssemblyConfigurationAttribute>
     <GenerateAssemblyCompanyAttribute>false</GenerateAssemblyCompanyAttribute>
     <GenerateAssemblyProductAttribute>false</GenerateAssemblyProductAttribute>
-    <Version>4.3.0</Version>
+    <Version>4.3.1</Version>
     <GeneratePackageOnBuild>true</GeneratePackageOnBuild>
     <PackageLicenseExpression>MIT</PackageLicenseExpression>
-    <AssemblyVersion>4.3.0</AssemblyVersion>
-    <LangVersion>8.0</LangVersion>
+    <AssemblyVersion>4.3.1</AssemblyVersion>
+    <LangVersion>Latest</LangVersion>
     <Nullable>enable</Nullable>
-    <FileVersion>4.3.0</FileVersion>
+    <FileVersion>4.3.1</FileVersion>
     <SignAssembly>True</SignAssembly>
     <AssemblyOriginatorKeyFile>..\..\FileSignatures.snk</AssemblyOriginatorKeyFile>
   </PropertyGroup>

diff --git a/src/FileSignatures/Formats/OfficeOpenXml.cs b/src/FileSignatures/Formats/OfficeOpenXml.cs
@@ -34,9 +34,14 @@ protected OfficeOpenXml(string identifiableEntry, string mediaType, string exten
 
         public bool IsMatch(IDisposable? file)
         {
-            if(file is ZipArchive archive)
+            if (file is ZipArchive archive)
             {
-                return archive.Entries.Any(e => e.FullName.Equals(IdentifiableEntry, StringComparison.OrdinalIgnoreCase));
+                // Match archives which contain a non-standard version of the identifiable entry, e.g. document2.xml instead of document.xml.
+                var index = Math.Max(0, IdentifiableEntry.LastIndexOf('.'));     
+                var fileName = IdentifiableEntry.Substring(0, IdentifiableEntry.Length - index);
+                var extension = IdentifiableEntry.Substring(index); 
+                return archive.Entries.Any(e => e.FullName.StartsWith(fileName, StringComparison.OrdinalIgnoreCase)
+                        && e.FullName.EndsWith(extension, StringComparison.OrdinalIgnoreCase));
             }
             else
             {

diff --git a/test/FileSignatures.Tests/FileSignatures.Tests.csproj b/test/FileSignatures.Tests/FileSignatures.Tests.csproj
@@ -1,14 +1,14 @@
 <Project Sdk="Microsoft.NET.Sdk">
 
   <PropertyGroup>
-    <TargetFramework>net5.0</TargetFramework>
+    <TargetFramework>net6.0</TargetFramework>
     <AssemblyName>FileSignatures.Tests</AssemblyName>
     <PackageId>FileSignatures.Tests</PackageId>
     <GenerateRuntimeConfigurationFiles>true</GenerateRuntimeConfigurationFiles>
     <GenerateAssemblyConfigurationAttribute>false</GenerateAssemblyConfigurationAttribute>
     <GenerateAssemblyCompanyAttribute>false</GenerateAssemblyCompanyAttribute>
     <GenerateAssemblyProductAttribute>false</GenerateAssemblyProductAttribute>
-    <LangVersion>8.0</LangVersion>
+    <LangVersion>Latest</LangVersion>
     <StartupObject>FileSignatures.Tests.Benchmarks.Program</StartupObject>
   </PropertyGroup>
 

diff --git a/test/FileSignatures.Tests/Formats/OfficeOpenXmlTests.cs b/test/FileSignatures.Tests/Formats/OfficeOpenXmlTests.cs
@@ -1,5 +1,6 @@
 using FileSignatures.Formats;
 using System.IO;
+using System.IO.Compression;
 using Xunit;
 
 namespace FileSignatures.Tests.Formats
@@ -17,5 +18,30 @@ public void InvalidZipArchiveDoesNotThrow()
             Assert.NotNull(format);
             Assert.IsType<Zip>(format);
         }
+
+        [Fact]
+        public void IdentifierWithoutExtensionDoesNotThrow()
+        {
+            var format = new TestOfficeOpenXml("test", "example/test", "test");
+
+            using var stream = new MemoryStream();
+            using(var createArchive = new ZipArchive(stream, ZipArchiveMode.Create, true))
+            {
+                createArchive.CreateEntry("test");
+            }
+
+            using var archive = new ZipArchive(stream, ZipArchiveMode.Read);
+
+            var result = format.IsMatch(archive);
+
+            Assert.True(result);
+        }
+
+        private class TestOfficeOpenXml : OfficeOpenXml
+        {
+            public TestOfficeOpenXml(string identifiableEntry, string mediaType, string extension) : base(identifiableEntry, mediaType, extension)
+            {
+            }
+        }
     }
 }
diff --git a/test/FileSignatures.Tests/FunctionalTests.cs b/test/FileSignatures.Tests/FunctionalTests.cs
@@ -16,6 +16,7 @@ public class FunctionalTests
         [InlineData("test.exif", "image/jpeg")]
         [InlineData("saved.msg", "application/vnd.ms-outlook")]
         [InlineData("dragndrop.msg", "application/vnd.ms-outlook")]
+        [InlineData("nonstandard.docx","application/vnd.openxmlformats-officedocument.wordprocessingml.document")]
         [InlineData("test.pdf", "application/pdf")]
         [InlineData("test.rtf", "application/rtf")]
         [InlineData("test.png", "image/png")]

diff --git a/test/FileSignatures.Tests/Samples/nonstandard.docx b/test/FileSignatures.Tests/Samples/nonstandard.docx