Slightly better structured content support

rampaa · Oct 19, 2023 · edd174b · edd174b
1 parent 6baebc8
commit edd174b
Show file tree

Hide file tree

Showing 2 changed files with 105 additions and 31 deletions.
diff --git a/JL.Core/Dicts/EPWING/EpwingYomichan/EpwingYomichanRecord.cs b/JL.Core/Dicts/EPWING/EpwingYomichan/EpwingYomichanRecord.cs
@@ -67,8 +67,7 @@ public EpwingYomichanRecord(List<JsonElement> jsonElement)
         //jsonElement[4].TryGetInt32(out int score);
         //Score = score;
 
-        List<string> definitionList = GetDefinitionsFromJsonArray(jsonElement[5]);
-        Definitions = definitionList.TrimStringListToStringArray() ?? Array.Empty<string>();
+        Definitions = GetDefinitions(jsonElement[5]) ?? Array.Empty<string>();
         Definitions.DeduplicateStringsInArray();
 
         //jsonElement[6].TryGetInt32(out int sequence);
@@ -142,50 +141,113 @@ public int GetFrequency(Freq freq)
         return frequency;
     }
 
-    private static List<string>? GetDefinitionsFromJsonElement(JsonElement jsonElement)
+    private static string[]? GetDefinitions(JsonElement jsonElement)
     {
-        return jsonElement.ValueKind switch
+        List<string> definitions = new();
+        foreach (JsonElement definitionElement in jsonElement.EnumerateArray())
         {
-            JsonValueKind.Array => GetDefinitionsFromJsonArray(jsonElement),
-            JsonValueKind.Object => GetDefinitionsFromJsonObject(jsonElement),
-            JsonValueKind.String => GetDefinitionsFromJsonString(jsonElement),
-            JsonValueKind.Number => null,
-            JsonValueKind.Undefined => null,
-            JsonValueKind.True => null,
-            JsonValueKind.False => null,
-            JsonValueKind.Null => null,
-            _ => null
-        };
+            string? definition = null;
+            if (definitionElement.ValueKind is JsonValueKind.String)
+            {
+                definition = definitionElement.GetString();
+            }
+
+            else if (definitionElement.ValueKind is JsonValueKind.Array)
+            {
+                definition = GetDefinitionsFromJsonArray(definitionElement);
+            }
+
+            else if (definitionElement.ValueKind is JsonValueKind.Object)
+            {
+                definition = GetDefinitionsFromJsonObject(definitionElement).Content;
+            }
+
+            if (definition is not null)
+            {
+                definitions.Add(definition.GetPooledString());
+            }
+        }
+
+        return definitions.TrimStringListToStringArray();
     }
 
-    private static List<string> GetDefinitionsFromJsonArray(JsonElement jsonElement)
+    private static string? GetDefinitionsFromJsonArray(JsonElement jsonElement, string? parentTag = null)
     {
-        List<string> definitions = new();
+        StringBuilder stringBuilder = new();
 
+        bool first = true;
         foreach (JsonElement definitionElement in jsonElement.EnumerateArray())
         {
-            List<string>? defs = GetDefinitionsFromJsonElement(definitionElement);
-            if (defs is not null)
+            if (definitionElement.ValueKind is JsonValueKind.String)
             {
-                definitions.AddRange(defs);
+                _ = stringBuilder.Append(definitionElement.GetString());
             }
-        }
 
-        return definitions;
-    }
+            else if (definitionElement.ValueKind is JsonValueKind.Array)
+            {
+                _ = stringBuilder.Append(GetDefinitionsFromJsonArray(definitionElement));
+            }
 
-    private static List<string>? GetDefinitionsFromJsonObject(JsonElement jsonElement)
-    {
-        return jsonElement.TryGetProperty("content", out JsonElement contentElement)
-            ? GetDefinitionsFromJsonElement(contentElement)
+            else if (definitionElement.ValueKind is JsonValueKind.Object)
+            {
+                if (first)
+                {
+                    first = false;
+                    parentTag = null;
+                }
+
+                YomichanContent contentResult = GetDefinitionsFromJsonObject(definitionElement, parentTag);
+                if (contentResult.Content is not null)
+                {
+                    if (contentResult.Tag is null or "span")
+                    {
+                        _ = stringBuilder.Append(contentResult.Content);
+                    }
+
+                    else if (contentResult.Tag is "ruby" or "rt")
+                    {
+                        _ = stringBuilder.Append(CultureInfo.InvariantCulture, $" ({contentResult.Content}) ");
+                    }
+
+                    else //if (contentResult.Tag is "div" or "a" or "li" or "ul" or "ol" or "p" or "h1" or "h2" or "h3" or "h4" or "h5" or "h6")
+                    {
+                        _ = stringBuilder.Append(CultureInfo.InvariantCulture, $"\n{contentResult.Content.TrimStart('\n')}");
+                    }
+                }
+            }
+        }
+
+        return stringBuilder.Length > 0
+            ? stringBuilder.ToString()
             : null;
     }
 
-    private static List<string> GetDefinitionsFromJsonString(JsonElement jsonElement)
+    private static YomichanContent GetDefinitionsFromJsonObject(JsonElement jsonElement, string? parentTag = null)
     {
-        return jsonElement.ToString()
-            .Split('\n', StringSplitOptions.TrimEntries | StringSplitOptions.RemoveEmptyEntries)
-            .Select(static s => s.Replace("\\\"", "\"", StringComparison.Ordinal))
-            .ToList();
+        if (jsonElement.TryGetProperty("content", out JsonElement contentElement))
+        {
+            string? tag = null;
+            if (jsonElement.TryGetProperty("tag", out JsonElement tagElement))
+            {
+                tag = tagElement.GetString();
+            }
+
+            if (contentElement.ValueKind is JsonValueKind.String)
+            {
+                return new YomichanContent(parentTag ?? tag, contentElement.GetString()?.Trim());
+            }
+
+            else if (contentElement.ValueKind is JsonValueKind.Array)
+            {
+                return new YomichanContent(parentTag ?? tag, GetDefinitionsFromJsonArray(contentElement, tag));
+            }
+
+            else if (contentElement.ValueKind is JsonValueKind.Object)
+            {
+                return GetDefinitionsFromJsonObject(contentElement, parentTag ?? tag);
+            }
+        }
+
+        return default;
     }
 }
diff --git a/JL.Core/Dicts/EPWING/EpwingYomichan/YomichanContent.cs b/JL.Core/Dicts/EPWING/EpwingYomichan/YomichanContent.cs
@@ -0,0 +1,12 @@
+namespace JL.Core.Dicts.EPWING.EpwingYomichan;
+internal readonly struct YomichanContent
+{
+    public string? Tag { get; }
+    public string? Content { get; }
+
+    public YomichanContent(string? tag, string? content)
+    {
+        Tag = tag;
+        Content = content;
+    }
+}