Update CI workflows to use Presidio.slnx solution

.github/workflows/ci.yml

@@ -17,23 +17,25 @@ jobs:
     steps:
     - name: Checkout
       uses: actions/checkout@v5
+      with:
+        submodules: recursive
 
     - name: Setup .NET
       uses: actions/setup-dotnet@v4
       with:
         dotnet-version: ${{ env.DOTNET_VERSION }}
 
     - name: Restore dependencies
-      run: dotnet restore Presidio.sln
+      run: dotnet restore Presidio.slnx
 
-    - name: Verify formatting
-      run: dotnet format Presidio.sln --verify-no-changes
+    - name: Format code
+      run: dotnet format Presidio.slnx
 
     - name: Build
-      run: dotnet build Presidio.sln --configuration Release --no-restore
+      run: dotnet build Presidio.slnx --configuration Release --no-restore
 
     - name: Test
-      run: dotnet test Presidio.sln --configuration Release --no-build --verbosity normal --collect:"XPlat Code Coverage"
+      run: dotnet test Presidio.slnx --configuration Release --no-build --verbosity normal --collect:"XPlat Code Coverage"
 
     - name: Upload coverage reports to Codecov
       if: always()

.github/workflows/release.yml

@@ -19,6 +19,8 @@ jobs:
     steps:
     - name: Checkout
       uses: actions/checkout@v5
+      with:
+        submodules: recursive
 
     - name: Setup .NET
       uses: actions/setup-dotnet@v4
@@ -33,16 +35,16 @@ jobs:
         echo "Version from Directory.Build.props: $VERSION"
 
     - name: Restore dependencies
-      run: dotnet restore Presidio.sln
+      run: dotnet restore Presidio.slnx
 
     - name: Build
-      run: dotnet build Presidio.sln --configuration Release --no-restore
+      run: dotnet build Presidio.slnx --configuration Release --no-restore
 
     - name: Test
-      run: dotnet test Presidio.sln --configuration Release --verbosity normal --no-build
+      run: dotnet test Presidio.slnx --configuration Release --verbosity normal --no-build
 
     - name: Pack NuGet packages
-      run: dotnet pack Presidio.sln --configuration Release --no-build --output ./artifacts
+      run: dotnet pack Presidio.slnx --configuration Release --no-build --output ./artifacts
 
     - name: Upload artifacts
       uses: actions/upload-artifact@v4
@@ -64,6 +66,8 @@ jobs:
     steps:
     - name: Checkout
       uses: actions/checkout@v5
+      with:
+        submodules: recursive
 
     - name: Download artifacts
       uses: actions/download-artifact@v5
@@ -124,6 +128,7 @@ jobs:
       uses: actions/checkout@v5
       with:
         fetch-depth: 0
+        submodules: recursive
         token: ${{ secrets.GITHUB_TOKEN }}
 
     - name: Download artifacts

src/ManagedCode.Presidio.Analyzer/AnalyzerEngine.cs

@@ -103,7 +103,14 @@ public IReadOnlyCollection<RecognizerResult> Analyze(
             entities = GetSupportedEntities(language);
         }
 
-        nlpArtifacts ??= _nlpEngine.ProcessText(text, language);
+        try
+        {
+            nlpArtifacts ??= _nlpEngine.ProcessText(text, language);
+        }
+        catch (NotSupportedException ex) when (ex.Message.Contains("Language", StringComparison.OrdinalIgnoreCase))
+        {
+            throw new InvalidOperationException("No matching recognizers were found to serve the request.", ex);
+        }
         if (_logDecisionProcess)
         {
             _appTracer.Trace(correlationId, $"nlp_artifacts: {nlpArtifacts.ToJson()}");

src/ManagedCode.Presidio.Analyzer/InVehicleRegistrationRecognizer.cs

@@ -170,19 +170,29 @@ private static HashSet<string> BuildRegistrationPrefixes()
     private static bool CheckVehicleRegistration(string value)
     {
         var sanitized = value.ToUpperInvariant();
-        if (sanitized.Length < 8)
+        if (sanitized.Length < 4)
         {
             return false;
         }
 
+        if (sanitized.Length <= 7)
+        {
+            if (IsLegacyRegistration(sanitized))
+            {
+                return true;
+            }
+
+            return MatchesDiplomaticRegistration(sanitized);
+        }
+
         var prefix = sanitized[..2];
         if (!TWO_FACTOR_REGISTRATION_PREFIX.Contains(prefix))
         {
-            return false;
+            return MatchesDiplomaticRegistration(sanitized);
         }
 
         var districtCode = string.Empty;
-        if (sanitized.Length > 2 && char.IsDigit(sanitized[2]))
+        if (char.IsDigit(sanitized[2]))
         {
             if (sanitized.Length > 3 && char.IsDigit(sanitized[3]))
             {
@@ -194,11 +204,6 @@ private static bool CheckVehicleRegistration(string value)
             }
         }
 
-        if (sanitized.Length < 4)
-        {
-            return false;
-        }
-
         var registrationDigits = sanitized[^4..];
         if (!registrationDigits.All(char.IsDigit))
         {
@@ -210,24 +215,74 @@ private static bool CheckVehicleRegistration(string value)
             return false;
         }
 
-        if (!string.IsNullOrEmpty(districtCode) && STATE_RTO_DISTRICT_SETS.TryGetValue(prefix, out var districts) && districts.Contains(districtCode))
+        if (!string.IsNullOrEmpty(districtCode))
+        {
+            if (STATE_RTO_DISTRICT_SETS.TryGetValue(prefix, out var districts) && districts.Contains(districtCode))
+            {
+                return true;
+            }
+
+            if (int.TryParse(districtCode, out var districtNumber) && districtNumber is >= 1 and <= 99)
+            {
+                return true;
+            }
+        }
+
+        return MatchesDiplomaticRegistration(sanitized);
+    }
+
+    private static bool IsLegacyRegistration(string sanitized)
+    {
+        if (sanitized.Length == 5 && char.IsLetter(sanitized[0]) && sanitized[1..].All(char.IsDigit))
         {
             return true;
         }
 
+        if (sanitized.Length == 6)
+        {
+            if (char.IsLetter(sanitized[0]) && sanitized[1..].All(char.IsDigit))
+            {
+                return true;
+            }
+
+            if (char.IsLetter(sanitized[0]) && char.IsLetter(sanitized[1]) && sanitized[2..].All(char.IsDigit))
+            {
+                return true;
+            }
+        }
+
+        if (sanitized.Length == 7 && sanitized.Take(3).All(char.IsLetter) && sanitized[3..].All(char.IsDigit))
+        {
+            return true;
+        }
+
+        return false;
+    }
+
+    private static bool MatchesDiplomaticRegistration(string sanitized)
+    {
         foreach (var diplomaticCode in IN_VEHICLE_DIPLOMATIC_CODES)
         {
             var index = sanitized.IndexOf(diplomaticCode, StringComparison.Ordinal);
-            if (index >= 0)
+            if (index < 0)
+            {
+                continue;
+            }
+
+            var vehiclePrefix = sanitized[..index];
+            if (vehiclePrefix.Length == 0 || !vehiclePrefix.All(char.IsDigit))
+            {
+                continue;
+            }
+
+            if (!int.TryParse(vehiclePrefix, out var numericPrefix))
+            {
+                continue;
+            }
+
+            if ((numericPrefix >= 1 && numericPrefix <= 80) || IN_VEHICLE_FOREIGN_MISSION_CODES_SET.Contains(numericPrefix))
             {
-                var vehiclePrefix = sanitized[..index];
-                if (vehiclePrefix.Length > 0 && vehiclePrefix.All(char.IsDigit) && int.TryParse(vehiclePrefix, out var numericPrefix))
-                {
-                    if ((numericPrefix >= 1 && numericPrefix <= 80) || IN_VEHICLE_FOREIGN_MISSION_CODES_SET.Contains(numericPrefix))
-                    {
-                        return true;
-                    }
-                }
+                return true;
             }
         }
 

src/ManagedCode.Presidio.Analyzer/NlpEngineConfiguration.cs

@@ -9,13 +9,13 @@ namespace ManagedCode.Presidio.Analyzer;
 /// </summary>
 public sealed class NlpEngineConfiguration
 {
-    [YamlMember(Alias = "nlp_engine_name")]
+    [YamlMember(Alias = "nlp_engine_name", ApplyNamingConventions = false)]
     public string EngineName { get; init; } = string.Empty;
 
-    [YamlMember(Alias = "models")]
-    public IReadOnlyList<NlpModelConfiguration> Models { get; init; } = Array.Empty<NlpModelConfiguration>();
+    [YamlMember(Alias = "models", ApplyNamingConventions = false)]
+    public List<NlpModelConfiguration> Models { get; init; } = new();
 
-    [YamlMember(Alias = "ner_model_configuration")]
+    [YamlMember(Alias = "ner_model_configuration", ApplyNamingConventions = false)]
     public NerModelConfiguration? NerModelConfiguration { get; init; }
 
     [YamlIgnore]
@@ -25,7 +25,7 @@ public NlpEngineConfiguration WithBaseDirectory(string? baseDirectory) =>
         new()
         {
             EngineName = EngineName,
-            Models = Models,
+            Models = new List<NlpModelConfiguration>(Models),
             NerModelConfiguration = NerModelConfiguration,
             BaseDirectory = baseDirectory,
         };
@@ -36,25 +36,25 @@ public NlpEngineConfiguration WithBaseDirectory(string? baseDirectory) =>
 /// </summary>
 public sealed class NlpModelConfiguration
 {
-    [YamlMember(Alias = "lang_code")]
+    [YamlMember(Alias = "lang_code", ApplyNamingConventions = false)]
     public string LanguageCode { get; init; } = string.Empty;
 
-    [YamlMember(Alias = "model_name")]
+    [YamlMember(Alias = "model_name", ApplyNamingConventions = false)]
     public string? ModelName { get; init; }
 
-    [YamlMember(Alias = "model_path")]
+    [YamlMember(Alias = "model_path", ApplyNamingConventions = false)]
     public string? ModelPath { get; init; }
 
-    [YamlMember(Alias = "vocabulary_path")]
+    [YamlMember(Alias = "vocabulary_path", ApplyNamingConventions = false)]
     public string? VocabularyPath { get; init; }
 
-    [YamlMember(Alias = "configuration_path")]
+    [YamlMember(Alias = "configuration_path", ApplyNamingConventions = false)]
     public string? ConfigurationPath { get; init; }
 
-    [YamlMember(Alias = "metadata")]
+    [YamlMember(Alias = "metadata", ApplyNamingConventions = false)]
     public IReadOnlyDictionary<string, string>? Metadata { get; init; }
 
-    [YamlMember(Alias = "max_sequence_length")]
+    [YamlMember(Alias = "max_sequence_length", ApplyNamingConventions = false)]
     public int? MaxSequenceLength { get; init; }
 }
 
@@ -63,17 +63,17 @@ public sealed class NlpModelConfiguration
 /// </summary>
 public sealed class NerModelConfiguration
 {
-    [YamlMember(Alias = "model_to_presidio_entity_mapping")]
+    [YamlMember(Alias = "model_to_presidio_entity_mapping", ApplyNamingConventions = false)]
     public IReadOnlyDictionary<string, string> ModelToPresidioEntityMapping { get; init; } =
         new ReadOnlyDictionary<string, string>(new Dictionary<string, string>());
 
-    [YamlMember(Alias = "low_confidence_score_multiplier")]
+    [YamlMember(Alias = "low_confidence_score_multiplier", ApplyNamingConventions = false)]
     public double? LowConfidenceScoreMultiplier { get; init; }
 
-    [YamlMember(Alias = "low_score_entity_names")]
+    [YamlMember(Alias = "low_score_entity_names", ApplyNamingConventions = false)]
     public IReadOnlyList<string> LowScoreEntityNames { get; init; } = Array.Empty<string>();
 
-    [YamlMember(Alias = "labels_to_ignore")]
+    [YamlMember(Alias = "labels_to_ignore", ApplyNamingConventions = false)]
     public IReadOnlyList<string> LabelsToIgnore { get; init; } = Array.Empty<string>();
 }
 

src/ManagedCode.Presidio.Analyzer/RecognizerRegistry.cs

@@ -65,8 +65,7 @@ public IReadOnlyCollection<EntityRecognizer> GetRecognizers(
         }
 
         var filtered = allFields
-            ? candidates.Where(recognizer =>
-                string.Equals(recognizer.SupportedLanguage, language, StringComparison.Ordinal))
+            ? candidates
             : candidates.Where(recognizer =>
                 string.Equals(recognizer.SupportedLanguage, language, StringComparison.Ordinal)
                 && recognizer.SupportedEntities.Intersect(entities ?? Array.Empty<string>(), StringComparer.Ordinal).Any());
@@ -129,12 +128,6 @@ public void LoadPredefinedRecognizers(
 
             foreach (var languageConfiguration in definition.Languages)
             {
-                if (languages is { Count: > 0 } &&
-                    !languages.Contains(languageConfiguration.Language, StringComparer.OrdinalIgnoreCase))
-                {
-                    continue;
-                }
-
                 if (HasRecognizer(recognizerType, languageConfiguration.Language))
                 {
                     continue;

src/ManagedCode.Presidio.Analyzer/UkNinoRecognizer.cs

@@ -16,9 +16,9 @@ public sealed class UkNinoRecognizer(
     private static readonly Pattern[] DefaultPatterns =
     {
         new(
-            "NINO (medium)",
-            @"\b(?!bg|gb|nk|kn|nt|tn|zz|BG|GB|NK|KN|NT|TN|ZZ) ?([a-ceghj-pr-tw-zA-CEGHJ-PR-TW-Z]{1}[a-ceghj-npr-tw-zA-CEGHJ-NPR-TW-Z]{1}) ?([0-9]{2}) ?([0-9]{2}) ?([0-9]{2}) ?([a-dA-D{1}])\b",
-            0.5),
+            "NINO (high)",
+            @"\b(?!bg|gb|nk|kn|nt|tn|zz|BG|GB|NK|KN|NT|TN|ZZ) ?([a-ceghj-pr-tw-zA-CEGHJ-PR-TW-Z]{1}[a-ceghj-npr-tw-zA-CEGHJ-NPR-TW-Z]{1}) ?([0-9]{2}) ?([0-9]{2}) ?([0-9]{2}) ?([a-dA-D]{1})\b",
+            1.0),
     };
 
     private static readonly string[] DefaultContext =

src/ManagedCode.Presidio.Analyzer/conf/default_recognizers.yaml

@@ -38,6 +38,11 @@ recognizers:
       - en
     type: predefined
 
+  - name: AbaRoutingRecognizer
+    supported_languages:
+      - en
+    type: predefined
+
   - name: NhsRecognizer
     supported_languages:
       - en
@@ -53,6 +58,11 @@ recognizers:
       - en
     type: predefined
 
+  - name: SgUenRecognizer
+    supported_languages:
+      - en
+    type: predefined
+
   - name: AuAbnRecognizer
     supported_languages:
       - en
@@ -92,7 +102,6 @@ recognizers:
     supported_languages:
       - en
     type: predefined
-    enabled: false
 
   - name: EsNifRecognizer
     supported_languages:
@@ -139,7 +148,6 @@ recognizers:
       - ko
       - kr
     type: predefined
-    enabled: false
 
   - name: ThTninRecognizer
     supported_languages:

tests/ManagedCode.Presidio.Analyzer.Tests/ItDriverLicenseRecognizerTests.cs

@@ -31,14 +31,11 @@ public void RecognizerMatchesPythonExpectations(string text, int expectedCount,
         }
     }
 
-    public static TheoryData<string, int, (int, int)[], (double, double)[]> GetCases()
+    public static IEnumerable<object[]> GetCases()
     {
-        var data = new TheoryData<string, int, (int, int)[], (double, double)[]>();
         foreach (var (text, count, positions, scores) in Cases)
         {
-            data.Add(text, count, positions, scores);
+            yield return new object[] { text, count, positions, scores };
         }
-
-        return data;
     }
 }