可用，后面需要优化性能

Author: anyshu

tools/server/server-diffusion.cpp

@@ -913,7 +913,6 @@ struct swa_checkpoint {
     std::vector<uint8_t> data;
 };
 
-//last output chunk, sent when the generation is finished
 struct server_task_result_cmpl_final : server_task_result {
     int index = 0;
 
@@ -4101,7 +4100,7 @@ struct server_context {
 
                             SRV_INF("Diffusion generation completed: n_generated=%d, generated_size=%zu, filtered_size=%zu, output_text_length=%zu\n", 
                                     n_generated, generated_tokens.size(), filtered_tokens.size(), output_text.size());
-                            SRV_INF("Generated text preview (first 500 chars): %.500s\n", output_text.c_str());
+                            SRV_INF("Generated text preview (first 1000 chars): %.1000s\n", output_text.c_str());
 
                             slot.generated_text = output_text;
                             slot.generated_tokens = filtered_tokens;
@@ -5534,15 +5533,12 @@ int main(int argc, char ** argv) {
             OAICOMPAT_TYPE_NONE);
     };
 
-    const auto handle_completions_oai = [&handle_completions_impl](const httplib::Request & req, httplib::Response & res) {
+    const auto handle_completions_oai = [&ctx_server,&handle_completions_impl](const httplib::Request & req, httplib::Response & res) {
         json data = oaicompat_completion_params_parse(json::parse(req.body));
         std::vector<raw_buffer> files; // dummy
 
         // Check if this is a diffusion request by looking for diffusion-specific parameters
-        bool is_diffusion = data.contains("diffusion_steps") || 
-                           data.contains("diffusion_algorithm") || 
-                           data.contains("cfg_scale") ||
-                           data.contains("visual_mode");
+        bool is_diffusion = llama_model_is_diffusion(ctx_server.model);
 
         server_task_type task_type = is_diffusion ? SERVER_TASK_TYPE_DIFFUSION : SERVER_TASK_TYPE_COMPLETION;
 
@@ -5645,11 +5641,7 @@ int main(int argc, char ** argv) {
             files);
 
         // Check if this is a diffusion request by looking for diffusion-specific parameters
-        bool is_diffusion = data.contains("diffusion_steps") || 
-                           data.contains("diffusion_algorithm") || 
-                           data.contains("cfg_scale") ||
-                           data.contains("visual_mode") ||
-                           data.contains("max_length");
+        bool is_diffusion = llama_model_is_diffusion(ctx_server.model)
 
         server_task_type task_type = is_diffusion ? SERVER_TASK_TYPE_DIFFUSION : SERVER_TASK_TYPE_COMPLETION;