feat: support doc url

src/Cnblogs.DashScope.AI/DashScopeChatClient.cs

@@ -248,7 +248,7 @@ public async IAsyncEnumerable<ChatResponseUpdate> GetStreamingResponseAsync(
                         Role = streamedRole
                     };
 
-                    if (response.Output.Choices?.FirstOrDefault()?.Message.Content is { Length: > 0 })
+                    if (response.Output.Choices?.FirstOrDefault()?.Message.Content.ToString() is { Length: > 0 })
                     {
                         update.Contents.Add(new TextContent(response.Output.Choices[0].Message.Content));
                     }

src/Cnblogs.DashScope.Core/Internals/TextChatMessageContentConvertor.cs

@@ -0,0 +1,62 @@
+using System.Text.Json;
+using System.Text.Json.Serialization;
+
+namespace Cnblogs.DashScope.Core.Internals;
+
+internal class TextChatMessageContentConvertor : JsonConverter<TextChatMessageContent>
+{
+    /// <inheritdoc />
+    public override TextChatMessageContent? Read(
+        ref Utf8JsonReader reader,
+        Type typeToConvert,
+        JsonSerializerOptions options)
+    {
+        if (reader.TokenType == JsonTokenType.String)
+        {
+            var s = reader.GetString();
+            return s == null ? null : new TextChatMessageContent(s);
+        }
+
+        if (reader.TokenType == JsonTokenType.StartArray)
+        {
+            var contents = JsonSerializer.Deserialize<List<TextChatMessageContentInternal>>(ref reader, options);
+            if (contents == null)
+            {
+                // impossible
+                return null;
+            }
+
+            var text = contents.FirstOrDefault(x => string.IsNullOrEmpty(x.Text) == false)?.Text
+                       ?? throw new JsonException("No text found in content array");
+            var docUrls = contents.FirstOrDefault(x => x.DocUrl != null)?.DocUrl;
+            return new TextChatMessageContent(text, docUrls);
+        }
+
+        throw new JsonException("Unknown type for TextChatMessageContent");
+    }
+
+    /// <inheritdoc />
+    public override void Write(Utf8JsonWriter writer, TextChatMessageContent value, JsonSerializerOptions options)
+    {
+        if (value.DocUrls != null)
+        {
+            JsonSerializer.Serialize(
+                writer,
+                new List<TextChatMessageContentInternal>()
+                {
+                    new() { Type = "text", Text = value.Text },
+                    new()
+                    {
+                        Type = "doc_url",
+                        DocUrl = value.DocUrls.ToList(),
+                        FileParsingStrategy = "auto"
+                    }
+                },
+                options);
+        }
+        else
+        {
+            JsonSerializer.Serialize(writer, value.Text, options);
+        }
+    }
+}

src/Cnblogs.DashScope.Core/Internals/TextChatMessageContentInternal.cs

@@ -0,0 +1,9 @@
+namespace Cnblogs.DashScope.Core.Internals;
+
+internal class TextChatMessageContentInternal
+{
+    public string Type { get; set; } = "text";
+    public string? Text { get; set; }
+    public List<string>? DocUrl { get; set; }
+    public string? FileParsingStrategy { get; set; }
+}

src/Cnblogs.DashScope.Core/TextChatMessage.cs

@@ -6,7 +6,7 @@ namespace Cnblogs.DashScope.Core;
 /// <summary>
 /// Represents a chat message between the user and the model.
 /// </summary>
-public record TextChatMessage : IMessage<string>
+public record TextChatMessage : IMessage<TextChatMessageContent>
 {
     /// <summary>
     /// Create chat message from an uploaded DashScope file.
@@ -38,7 +38,7 @@ public TextChatMessage(IEnumerable<DashScopeFileId> fileIds)
     [JsonConstructor]
     public TextChatMessage(
         string role,
-        string content,
+        TextChatMessageContent content,
         string? toolCallId = null,
         bool? partial = null,
         string? reasoningContent = null,
@@ -56,7 +56,7 @@ public TextChatMessage(
     public string Role { get; init; }
 
     /// <summary>The content of this message.</summary>
-    public string Content { get; init; }
+    public TextChatMessageContent Content { get; init; }
 
     /// <summary>Used when role is tool, represents the function name of this message generated by.</summary>
     public string? ToolCallId { get; init; }
@@ -105,6 +105,17 @@ public static TextChatMessage File(IEnumerable<DashScopeFileId> fileIds)
         return new TextChatMessage(fileIds);
     }
 
+    /// <summary>
+    /// Create a docUrls message.
+    /// </summary>
+    /// <param name="prompt">Text input.</param>
+    /// <param name="docUrls">The doc urls.</param>
+    /// <returns></returns>
+    public static TextChatMessage DocUrl(string prompt, IEnumerable<string> docUrls)
+    {
+        return new TextChatMessage(DashScopeRoleNames.User, new TextChatMessageContent(prompt, docUrls));
+    }
+
     /// <summary>
     /// Create a user message.
     /// </summary>

src/Cnblogs.DashScope.Core/TextChatMessageContent.cs

@@ -0,0 +1,62 @@
+using System.Text.Json.Serialization;
+using Cnblogs.DashScope.Core.Internals;
+
+namespace Cnblogs.DashScope.Core;
+
+/// <summary>
+/// Content of the <see cref="TextChatMessage"/>.
+/// </summary>
+[JsonConverter(typeof(TextChatMessageContentConvertor))]
+public class TextChatMessageContent
+{
+    /// <summary>
+    /// The text part of the content.
+    /// </summary>
+    public string Text { get; }
+
+    /// <summary>
+    /// Optional doc urls.
+    /// </summary>
+    public IEnumerable<string>? DocUrls { get; }
+
+    /// <summary>
+    /// Creates a <see cref="TextChatMessageContent"/> with text content.
+    /// </summary>
+    /// <param name="text">The text content.</param>
+    public TextChatMessageContent(string text)
+    {
+        Text = text;
+        DocUrls = null;
+    }
+
+    /// <summary>
+    /// Creates a <see cref="TextChatMessageContent"/> with text content and doc urls.
+    /// </summary>
+    /// <param name="text">The text content.</param>
+    /// <param name="docUrls">The doc urls.</param>
+    public TextChatMessageContent(string text, IEnumerable<string>? docUrls)
+    {
+        Text = text;
+        DocUrls = docUrls;
+    }
+
+    /// <summary>
+    /// Convert string to TextChatMessageContent implicitly.
+    /// </summary>
+    /// <param name="value">The string value to convert.</param>
+    /// <returns></returns>
+    public static implicit operator TextChatMessageContent(string value) => new(value);
+
+    /// <summary>
+    /// Convert to string implicitly.
+    /// </summary>
+    /// <param name="value">The value to convert.</param>
+    /// <returns></returns>
+    public static implicit operator string(TextChatMessageContent value) => value.Text;
+
+    /// <inheritdoc />
+    public override string ToString()
+    {
+        return Text;
+    }
+}

src/Cnblogs.DashScope.Core/TextGenerationTokenUsage.cs

@@ -35,4 +35,9 @@ public class TextGenerationTokenUsage
     /// The total number of token.
     /// </summary>
     public int TotalTokens { get; set; }
+
+    /// <summary>
+    /// Cached token count.
+    /// </summary>
+    public int? CachedTokens { get; set; }
 }

test/Cnblogs.DashScope.Sdk.UnitTests/FileSerializationTests.cs

@@ -1,5 +1,4 @@
 using Cnblogs.DashScope.Core;
-using Cnblogs.DashScope.Core.Internals;
 using Cnblogs.DashScope.Tests.Shared.Utils;
 using NSubstitute;
 

test/Cnblogs.DashScope.Sdk.UnitTests/TextChatMessageContentJsonConvertorTests.cs

@@ -0,0 +1,97 @@
+using System.Text.Json;
+using Cnblogs.DashScope.Core;
+using Cnblogs.DashScope.Core.Internals;
+
+namespace Cnblogs.DashScope.Sdk.UnitTests;
+
+public class TextChatMessageContentJsonConvertorTests
+{
+    private const string TextJson = "\"some text\"";
+
+    private const string DocUrlJson =
+        "[{\"type\":\"text\",\"text\":\"some content\"},{\"type\":\"doc_url\",\"doc_url\":[\"url1\"],\"file_parsing_strategy\":\"auto\"}]";
+
+    [Fact]
+    public void Serialize_Text_StringAsync()
+    {
+        // Arrange
+        var content = new TextChatMessageContent("some text");
+
+        // Act
+        var json = JsonSerializer.Serialize(content, DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.Equal(TextJson, json);
+    }
+
+    [Fact]
+    public void Serialize_DocUrl_ObjectAsync()
+    {
+        // Arrange
+        var content = new TextChatMessageContent("some content", new[] { "url1" });
+
+        // Act
+        var json = JsonSerializer.Serialize(content, DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.Equal(DocUrlJson, json);
+    }
+
+    [Fact]
+    public void Deserialize_InvalidType_ThrowAsync()
+    {
+        // Arrange
+        const string errJson = "{}";
+
+        // Act
+        var act = () => JsonSerializer.Deserialize<TextChatMessageContent>(
+            errJson,
+            DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.Throws<JsonException>(act);
+    }
+
+    [Fact]
+    public void Deserialize_InvalidArray_ThrowAsync()
+    {
+        // Arrange
+        const string errJson = "[{\"type\":\"doc_url\", \"doc_url\":[]}]";
+
+        // Act
+        var act = () => JsonSerializer.Deserialize<TextChatMessageContent>(
+            errJson,
+            DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.Throws<JsonException>(act);
+    }
+
+    [Fact]
+    public void Deserialize_Text_SetTextOnlyAsync()
+    {
+        // Act
+        var content = JsonSerializer.Deserialize<TextChatMessageContent>(
+            TextJson,
+            DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.NotNull(content);
+        Assert.Equal("some text", content.Text);
+        Assert.Null(content.DocUrls);
+    }
+
+    [Fact]
+    public void Deserialize_DocUrl_SetUrlsAsync()
+    {
+        // Act
+        var content = JsonSerializer.Deserialize<TextChatMessageContent>(
+            DocUrlJson,
+            DashScopeDefaults.SerializationOptions);
+
+        // Assert
+        Assert.NotNull(content);
+        Assert.Equal("some content", content.Text);
+        Assert.Equivalent(new[] { "url1" }, content.DocUrls);
+    }
+}

test/Cnblogs.DashScope.Sdk.UnitTests/TextGenerationSerializationTests.cs

@@ -225,7 +225,8 @@ public async Task ConversationCompletion_DeepResearchSse_SuccessAsync(
     public static readonly TheoryData<RequestSnapshot<ModelRequest<TextGenerationInput, ITextGenerationParameters>,
         ModelResponse<TextGenerationOutput, TextGenerationTokenUsage>>> ConversationMessageFormatSseData = new(
         Snapshots.TextGeneration.MessageFormat.ConversationMessageIncremental,
-        Snapshots.TextGeneration.MessageFormat.ConversationMessageWithFilesIncremental);
+        Snapshots.TextGeneration.MessageFormat.ConversationMessageWithFilesIncremental,
+        Snapshots.TextGeneration.MessageFormat.ConversationMessageWithDocUrlsIncremental);
 
     public static readonly TheoryData<RequestSnapshot<ModelRequest<TextGenerationInput, ITextGenerationParameters>,
         ModelResponse<TextGenerationOutput, TextGenerationTokenUsage>>> ConversationMessageFormatNoSseData = new(

test/Cnblogs.DashScope.Tests.Shared/RawHttpData/conversation-generation-message-with-doc-url-sse.request.body.json

@@ -0,0 +1,32 @@
+{
+  "model": "qwen-doc-turbo",
+  "input": {
+    "messages": [
+      {
+        "role": "system",
+        "content": "You are a helpful assistant."
+      },
+      {
+        "role": "user",
+        "content": [
+          {
+            "type": "text",
+            "text": "从这两份产品手册中，提取所有产品信息，并整理成一个标准的JSON数组。每个对象需要包含：model(产品的型号)、name(产品的名称)、price(价格（去除货币符号和逗号）)"
+          },
+          {
+            "type": "doc_url",
+            "doc_url": [
+              "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251107/jockge/%E7%A4%BA%E4%BE%8B%E4%BA%A7%E5%93%81%E6%89%8B%E5%86%8CA.docx",
+              "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251107/ztwxzr/%E7%A4%BA%E4%BE%8B%E4%BA%A7%E5%93%81%E6%89%8B%E5%86%8CB.docx"
+            ],
+            "file_parsing_strategy": "auto"
+          }
+        ]
+      }
+    ]
+  },
+  "parameters": {
+    "result_format": "message",
+    "incremental_output": true
+  }
+}

test/Cnblogs.DashScope.Tests.Shared/RawHttpData/conversation-generation-message-with-doc-url-sse.request.header.txt

@@ -0,0 +1,8 @@
+POST /api/v1/services/aigc/text-generation/generation HTTP/1.1
+Accept: text/event-stream
+Content-Type: application/json
+Cache-Control: no-cache
+Host: dashscope.aliyuncs.com
+Accept-Encoding: gzip, deflate, br
+Connection: keep-alive
+Content-Length: 1341