Integration with ChatGLM (langchain4j#360)

support [chatglm](https://github.com/THUDM/ChatGLM-6B), which was
mentioned in langchain4j#267 .

[chatglm2](https://github.com/THUDM/ChatGLM2-6B) and
[chatglm3](https://github.com/THUDM/ChatGLM3) api are compatible with
openai, so It is enough to support chatglm.

because chatglm does not have official docker image, so I don't know how
to use `Testcontainers` to do test. (I'm not familiar with
`Testcontainers`, so for now I have to copy test from the other modules,
lol). The test will update using `Testcontainers` after I learn about it
in few days.

langchain4j-bom/pom.xml

@@ -87,6 +87,12 @@
                 <version>${project.version}</version>
             </dependency>
 
+            <dependency>
+                <groupId>dev.langchain4j</groupId>
+                <artifactId>langchain4j-chatglm</artifactId>
+                <version>${project.version}</version>
+            </dependency>
+
             <!-- embedding stores -->
 
             <dependency>

langchain4j-chatglm/pom.xml

@@ -0,0 +1,70 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project xmlns="http://maven.apache.org/POM/4.0.0"
+         xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
+    <modelVersion>4.0.0</modelVersion>
+    <parent>
+        <groupId>dev.langchain4j</groupId>
+        <artifactId>langchain4j-parent</artifactId>
+        <version>0.25.0-SNAPSHOT</version>
+        <relativePath>../langchain4j-parent/pom.xml</relativePath>
+    </parent>
+
+    <artifactId>langchain4j-chatglm</artifactId>
+    <packaging>jar</packaging>
+
+    <name>LangChain4j integration with ChatGLM</name>
+
+    <dependencies>
+        <dependency>
+            <groupId>dev.langchain4j</groupId>
+            <artifactId>langchain4j-core</artifactId>
+        </dependency>
+
+        <dependency>
+            <groupId>com.squareup.retrofit2</groupId>
+            <artifactId>retrofit</artifactId>
+        </dependency>
+
+        <dependency>
+            <groupId>com.squareup.retrofit2</groupId>
+            <artifactId>converter-gson</artifactId>
+        </dependency>
+
+        <dependency>
+            <groupId>com.squareup.okhttp3</groupId>
+            <artifactId>okhttp</artifactId>
+        </dependency>
+
+        <dependency>
+            <groupId>org.projectlombok</groupId>
+            <artifactId>lombok</artifactId>
+            <scope>provided</scope>
+        </dependency>
+
+        <dependency>
+            <groupId>org.junit.jupiter</groupId>
+            <artifactId>junit-jupiter-engine</artifactId>
+            <scope>test</scope>
+        </dependency>
+
+        <dependency>
+            <groupId>org.junit.jupiter</groupId>
+            <artifactId>junit-jupiter-params</artifactId>
+            <scope>test</scope>
+        </dependency>
+
+        <dependency>
+            <groupId>org.assertj</groupId>
+            <artifactId>assertj-core</artifactId>
+            <scope>test</scope>
+        </dependency>
+
+        <dependency>
+            <groupId>org.testcontainers</groupId>
+            <artifactId>testcontainers</artifactId>
+            <scope>test</scope>
+        </dependency>
+    </dependencies>
+
+</project>

langchain4j-chatglm/src/main/java/dev/langchain4j/model/chatglm/ChatCompletionRequest.java

@@ -0,0 +1,21 @@
+package dev.langchain4j.model.chatglm;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+import java.util.List;
+
+@Data
+@NoArgsConstructor
+@AllArgsConstructor
+@Builder
+class ChatCompletionRequest {
+
+    private String prompt;
+    private Double temperature;
+    private Double topP;
+    private Integer maxLength;
+    private List<List<String>> history;
+}

langchain4j-chatglm/src/main/java/dev/langchain4j/model/chatglm/ChatCompletionResponse.java

@@ -0,0 +1,20 @@
+package dev.langchain4j.model.chatglm;
+
+import lombok.AllArgsConstructor;
+import lombok.Builder;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+import java.util.List;
+
+@Data
+@NoArgsConstructor
+@AllArgsConstructor
+@Builder
+class ChatCompletionResponse {
+
+    private String response;
+    private List<List<String>> history;
+    private Integer status;
+    private String time;
+}

langchain4j-chatglm/src/main/java/dev/langchain4j/model/chatglm/ChatGlmApi.java

@@ -0,0 +1,15 @@
+package dev.langchain4j.model.chatglm;
+
+import retrofit2.Call;
+import retrofit2.http.Body;
+import retrofit2.http.Headers;
+import retrofit2.http.POST;
+
+interface ChatGlmApi {
+
+    int OK = 200;
+
+    @POST("/")
+    @Headers({"Content-Type: application/json"})
+    Call<ChatCompletionResponse> chatCompletion(@Body ChatCompletionRequest chatCompletionRequest);
+}

langchain4j-chatglm/src/main/java/dev/langchain4j/model/chatglm/ChatGlmChatModel.java

@@ -0,0 +1,84 @@
+package dev.langchain4j.model.chatglm;
+
+import dev.langchain4j.data.message.AiMessage;
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.ChatMessageType;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.output.Response;
+import lombok.Builder;
+
+import java.time.Duration;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.stream.Collectors;
+
+import static dev.langchain4j.internal.RetryUtils.withRetry;
+import static dev.langchain4j.internal.Utils.getOrDefault;
+
+/**
+ * Support <a href="https://github.com/THUDM/ChatGLM-6B">ChatGLM</a>,
+ * ChatGLM2 and ChatGLM3 api are compatible with OpenAI API
+ */
+public class ChatGlmChatModel implements ChatLanguageModel {
+
+    private final ChatGlmClient client;
+    private final Double temperature;
+    private final Double topP;
+    private final Integer maxLength;
+    private final Integer maxRetries;
+
+    @Builder
+    public ChatGlmChatModel(String baseUrl, Duration timeout,
+                            Double temperature, Integer maxRetries,
+                            Double topP, Integer maxLength) {
+        this.client = new ChatGlmClient(baseUrl, timeout);
+        this.temperature = getOrDefault(temperature, 0.7);
+        this.maxRetries = getOrDefault(maxRetries, 3);
+        this.topP = topP;
+        this.maxLength = maxLength;
+    }
+
+
+    @Override
+    public Response<AiMessage> generate(List<ChatMessage> messages) {
+        // get last user message
+        String prompt = messages.get(messages.size() - 1).text();
+        List<List<String>> history = toHistory(messages.subList(0, messages.size() - 1));
+        ChatCompletionRequest request = ChatCompletionRequest.builder()
+                .prompt(prompt)
+                .temperature(temperature)
+                .topP(topP)
+                .maxLength(maxLength)
+                .history(history)
+                .build();
+
+        ChatCompletionResponse response = withRetry(() -> client.chatCompletion(request), maxRetries);
+
+        return Response.from(AiMessage.from(response.getResponse()));
+    }
+
+    private List<List<String>> toHistory(List<ChatMessage> historyMessages) {
+        // Order: User - AI - User - AI ...
+        // so the length of historyMessages must be divisible by 2
+        if (containsSystemMessage(historyMessages)) {
+            throw new IllegalArgumentException("ChatGLM does not support system prompt");
+        }
+
+        if (historyMessages.size() % 2 != 0) {
+            throw new IllegalArgumentException("History must be divisible by 2 because it's order User - AI - User - AI ...");
+        }
+
+        List<List<String>> history = new ArrayList<>();
+        for (int i = 0; i < historyMessages.size() / 2; i++) {
+            history.add(historyMessages.subList(i * 2, i * 2 + 2).stream()
+                    .map(ChatMessage::text)
+                    .collect(Collectors.toList()));
+        }
+
+        return history;
+    }
+
+    private boolean containsSystemMessage(List<ChatMessage> messages) {
+        return messages.stream().anyMatch(message -> message.type() == ChatMessageType.SYSTEM);
+    }
+}

langchain4j-chatglm/src/main/java/dev/langchain4j/model/chatglm/ChatGlmClient.java

@@ -0,0 +1,70 @@
+package dev.langchain4j.model.chatglm;
+
+import com.google.gson.Gson;
+import com.google.gson.GsonBuilder;
+import lombok.Builder;
+import okhttp3.OkHttpClient;
+import retrofit2.Response;
+import retrofit2.Retrofit;
+import retrofit2.converter.gson.GsonConverterFactory;
+
+import java.io.IOException;
+import java.time.Duration;
+
+import static com.google.gson.FieldNamingPolicy.LOWER_CASE_WITH_UNDERSCORES;
+import static dev.langchain4j.internal.Utils.getOrDefault;
+import static java.time.Duration.ofSeconds;
+
+class ChatGlmClient {
+
+    private final ChatGlmApi chatGLMApi;
+    private static final Gson GSON = new GsonBuilder()
+            .setFieldNamingPolicy(LOWER_CASE_WITH_UNDERSCORES)
+            .create();
+
+
+    @Builder
+    public ChatGlmClient(String baseUrl, Duration timeout) {
+        timeout = getOrDefault(timeout, ofSeconds(60));
+
+        OkHttpClient okHttpClient = new OkHttpClient.Builder()
+                .callTimeout(timeout)
+                .connectTimeout(timeout)
+                .readTimeout(timeout)
+                .writeTimeout(timeout)
+                .build();
+
+        Retrofit retrofit = new Retrofit.Builder()
+                .baseUrl(baseUrl)
+                .client(okHttpClient)
+                .addConverterFactory(GsonConverterFactory.create(GSON))
+                .build();
+
+        chatGLMApi = retrofit.create(ChatGlmApi.class);
+    }
+
+    public ChatCompletionResponse chatCompletion(ChatCompletionRequest request) {
+        try {
+            Response<ChatCompletionResponse> retrofitResponse
+                    = chatGLMApi.chatCompletion(request).execute();
+
+            if (retrofitResponse.isSuccessful() && retrofitResponse.body() != null
+                    && retrofitResponse.body().getStatus() == ChatGlmApi.OK) {
+                return retrofitResponse.body();
+            } else {
+                throw toException(retrofitResponse);
+            }
+        } catch (IOException e) {
+            throw new RuntimeException(e);
+        }
+    }
+
+    private RuntimeException toException(Response<?> response) throws IOException {
+        int code = response.code();
+        String body = response.errorBody().string();
+
+        String errorMessage = String.format("status code: %s; body: %s", code, body);
+        return new RuntimeException(errorMessage);
+    }
+
+}

langchain4j-chatglm/src/test/java/dev/langchain4j/model/chatglm/ChatGlmChatModelIT.java

@@ -0,0 +1,51 @@
+package dev.langchain4j.model.chatglm;
+
+import dev.langchain4j.data.message.AiMessage;
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.UserMessage;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.output.Response;
+import org.junit.jupiter.api.Disabled;
+import org.junit.jupiter.api.Test;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import static dev.langchain4j.data.message.UserMessage.userMessage;
+import static org.assertj.core.api.Assertions.assertThat;
+
+@Disabled("need local deployment of ChatGLM, see https://github.com/THUDM/ChatGLM-6B")
+class ChatGlmChatModelIT {
+
+    ChatLanguageModel model = ChatGlmChatModel.builder()
+            .baseUrl("http://localhost:8000")
+            .build();
+
+    @Test
+    void should_generate_answer() {
+        UserMessage userMessage = userMessage("你好，请问一下德国的首都是哪里呢？");
+        Response<AiMessage> response = model.generate(userMessage);
+        assertThat(response.content().text()).contains("柏林");
+    }
+
+    @Test
+    void should_generate_answer_from_history() {
+        // init history
+        List<ChatMessage> messages = new ArrayList<>();
+
+        // given question first time
+        UserMessage userMessage = userMessage("你好，请问一下德国的首都是哪里呢？");
+        Response<AiMessage> response = model.generate(userMessage);
+        assertThat(response.content().text()).contains("柏林");
+
+        // given question with history
+        messages.add(userMessage);
+        messages.add(response.content());
+
+        UserMessage secondUserMessage = userMessage("你能告诉我上个问题我问了你什么呢？请把我的问题原封不动的告诉我");
+        messages.add(secondUserMessage);
+
+        Response<AiMessage> secondResponse = model.generate(messages);
+        assertThat(secondResponse.content().text()).contains("德国"); // the answer should contain Germany in the First Question
+    }
+}

langchain4j-parent/pom.xml

@@ -497,4 +497,4 @@
             </build>
         </profile>
     </profiles>
-</project>
+</project>

pom.xml

@@ -27,6 +27,7 @@
         <module>langchain4j-open-ai</module>
         <module>langchain4j-vertex-ai</module>
         <module>langchain4j-ollama</module>
+        <module>langchain4j-chatglm</module>
 
         <!-- embedding stores -->
         <module>langchain4j-cassandra</module>