Fixed snake case name translation

khellang · Feb 8, 2019 · 7c8a37a · 7c8a37a
1 parent 064fa94
commit 7c8a37a
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 52 deletions.
diff --git a/src/Npgsql/NameTranslation/NpgsqlSnakeCaseNameTranslator.cs b/src/Npgsql/NameTranslation/NpgsqlSnakeCaseNameTranslator.cs
@@ -1,4 +1,5 @@
-using System.Linq;
+using System.Globalization;
+using System.Linq;
 using System.Text;
 
 namespace Npgsql.NameTranslation
@@ -38,74 +39,63 @@ public string TranslateMemberName(string clrName) => LegacyMode
         /// <summary>
         /// Converts a string to its snake_case equivalent.
         /// </summary>
-        /// <remarks>
-        /// Code borrowed from Newtonsoft.Json.
-        /// See https://github.com/JamesNK/Newtonsoft.Json/blob/f012ba857f36fe75b1294a210b9104130a4db4d5/Src/Newtonsoft.Json/Utilities/StringUtils.cs#L200-L276.
-        /// </remarks>
         /// <param name="value">The value to convert.</param>
         public static string ConvertToSnakeCase(string value)
         {
             if (string.IsNullOrEmpty(value))
                 return value;
 
-            var sb = new StringBuilder();
-            var state = SnakeCaseState.Start;
+            const char underscore = '_';
+            const UnicodeCategory noneCategory = UnicodeCategory.Control;
 
-            for (var i = 0; i < value.Length; i++)
+            var builder = new StringBuilder();
+            var previousCategory = noneCategory;
+
+            for (var currentIndex = 0; currentIndex< value.Length; currentIndex++)
             {
-                if (value[i] == ' ')
+                var currentChar = value[currentIndex];
+                if (currentChar == underscore)
                 {
-                    if (state != SnakeCaseState.Start)
-                        state = SnakeCaseState.NewWord;
+                    builder.Append(underscore);
+                    previousCategory = noneCategory;
+                    continue;
                 }
-                else if (char.IsUpper(value[i]))
+
+                var currentCategory = char.GetUnicodeCategory(currentChar);
+                switch (currentCategory)
                 {
-                    switch (state)
-                    {
-                    case SnakeCaseState.Upper:
-                        var hasNext = (i + 1 < value.Length);
-                        if (i > 0 && hasNext)
+                    case UnicodeCategory.UppercaseLetter:
+                    case UnicodeCategory.TitlecaseLetter:
+                        if (previousCategory == UnicodeCategory.SpaceSeparator ||
+                            previousCategory == UnicodeCategory.LowercaseLetter ||
+                            previousCategory != UnicodeCategory.DecimalDigitNumber &&
+                            currentIndex > 0 &&
+                            currentIndex + 1 < value.Length &&
+                            char.IsLower(value[currentIndex + 1]))
                         {
-                            var nextChar = value[i + 1];
-                            if (!char.IsUpper(nextChar) && nextChar != '_')
-                            {
-                                sb.Append('_');
-                            }
+                            builder.Append(underscore);
                         }
+
+                        currentChar = char.ToLower(currentChar);
                         break;
-                    case SnakeCaseState.Lower:
-                    case SnakeCaseState.NewWord:
-                        sb.Append('_');
+
+                    case UnicodeCategory.LowercaseLetter:
+                    case UnicodeCategory.DecimalDigitNumber:
+                        if (previousCategory == UnicodeCategory.SpaceSeparator)
+                            builder.Append(underscore);
                         break;
-                    }
 
-                    sb.Append(char.ToLowerInvariant(value[i]));
-                    state = SnakeCaseState.Upper;
-                }
-                else if (value[i] == '_')
-                {
-                    sb.Append('_');
-                    state = SnakeCaseState.Start;
+                    default:
+                        if (previousCategory != noneCategory)
+                            previousCategory = UnicodeCategory.SpaceSeparator;
+                        continue;
                 }
-                else
-                {
-                    if (state == SnakeCaseState.NewWord)
-                        sb.Append('_');
 
-                    sb.Append(value[i]);
-                    state = SnakeCaseState.Lower;
-                }
+                builder.Append(currentChar);
+                previousCategory = currentCategory;
             }
 
-            return sb.ToString();
-        }
-
-        enum SnakeCaseState
-        {
-            Start,
-            Lower,
-            Upper,
-            NewWord
+            return builder.ToString();
         }
     }
 }
diff --git a/test/Npgsql.Tests/SnakeCaseNameTranslatorTests.cs b/test/Npgsql.Tests/SnakeCaseNameTranslatorTests.cs
@@ -18,16 +18,16 @@ public string TranslateMemberName(string value, bool legacyMode)
 
         static IEnumerable<TestCaseData> TestCases => new (string value, string legacyResult, string result)[]
         {
-            ("Hi!! This is text. Time to test.", "hi!! _this is text. _time to test.", "hi!!_this_is_text._time_to_test."),
-            ("9999-12-31T23:59:59.9999999Z", "9999-12-31_t23:59:59.9999999_z", "9999-12-31_t23:59:59.9999999_z"),
+            ("Hi!! This is text. Time to test.", "hi!! _this is text. _time to test.", "hi_this_is_text_time_to_test"),
+            ("9999-12-31T23:59:59.9999999Z", "9999-12-31_t23:59:59.9999999_z", "9999_12_31t23_59_59_9999999z"),
             ("already_snake_case_ ", "already_snake_case_ ", "already_snake_case_"),
             ("SHOUTING_CASE", "s_h_o_u_t_i_n_g__c_a_s_e", "shouting_case"),
             ("IsJSONProperty", "is_j_s_o_n_property", "is_json_property"),
             ("SnA__ kEcAsE", "sn_a__ k_ec_as_e", "sn_a__k_ec_as_e"),
             ("SnA__kEcAsE", "sn_a__k_ec_as_e", "sn_a__k_ec_as_e"),
             ("SnAkEcAsE", "sn_ak_ec_as_e", "sn_ak_ec_as_e"),
             ("URLValue", "u_r_l_value", "url_value"),
-            ("Xml2Json", "xml2_json", "xml2_json"),
+            ("Xml2Json", "xml2_json", "xml2json"),
             (" IPhone ", " _i_phone ", "i_phone"),
             ("I  Phone", "i  _phone", "i_phone"),
             (" IPhone", " _i_phone", "i_phone"),