fix little bug

core/predictor/framework/bsf-inl.h

@@ -31,32 +31,32 @@ namespace im {
 namespace bsf {
 
 template <typename InItemT, typename OutItemT>
-bool Task<InItemT, OutItemT>::task_fetch_init(BatchTasks<TaskT>& baskTask) {
+bool Task<InItemT, OutItemT>::task_fetch_init(BatchTasks<TaskT>& batchTask) {
   // 双检锁，减少加锁的粒度
   if (!fetch_init) {
     if (taskmeta_num > 1) {
       // 对于task被拆分为多个taskmeta,需要加锁。
       AutoMutex lock(task_mut);
-      task_fetch_create(baskTask);
+      task_fetch_create(batchTask);
     } else {
       // 对于task只有1个taskmeta,不需要加锁。
-      task_fetch_create(baskTask);
+      task_fetch_create(batchTask);
     }
   }
   return true;
 }
 
 template <typename InItemT, typename OutItemT>
-bool Task<InItemT, OutItemT>::task_fetch_create(BatchTasks<TaskT>& baskTask) {
+bool Task<InItemT, OutItemT>::task_fetch_create(BatchTasks<TaskT>& batchTask) {
   if (!fetch_init) {
-    vector_fetch_lod_index = baskTask.vector_fetch_lod_index;
-    set_fetch_nobatch_index = baskTask.set_fetch_nobatch_index;
+    vector_fetch_lod_index = batchTask.vector_fetch_lod_index;
+    set_fetch_nobatch_index = batchTask.set_fetch_nobatch_index;
     OutVectorT taskMetaOutLodTensor;
-    size_t fetchvar_num = baskTask._batch_out.size();
+    size_t fetchvar_num = batchTask._batch_out.size();
     for (size_t fetchvar_index = 0; fetchvar_index < fetchvar_num;
          ++fetchvar_index) {
       size_t fetchvar_bytesize_index =
-          baskTask.fetchvar_bytesize(fetchvar_index);
+          batchTask.fetchvar_bytesize(fetchvar_index);
       size_t fetchvar_batch = 0;
       // 1. nobatch fetchvar情况
       if (set_fetch_nobatch_index.size() > 0 &&
@@ -79,14 +79,14 @@ bool Task<InItemT, OutItemT>::task_fetch_create(BatchTasks<TaskT>& baskTask) {
         fetchvar_batch = batch_size();
       }
       paddle::PaddleTensor tensor_out;
-      tensor_out.name = baskTask._batch_out[fetchvar_index].name;
+      tensor_out.name = batchTask._batch_out[fetchvar_index].name;
       tensor_out.dtype =
-          paddle::PaddleDType(baskTask._batch_out[fetchvar_index].dtype);
-      tensor_out.shape = baskTask._batch_out[fetchvar_index].shape;
+          paddle::PaddleDType(batchTask._batch_out[fetchvar_index].dtype);
+      tensor_out.shape = batchTask._batch_out[fetchvar_index].shape;
       tensor_out.shape[0] = fetchvar_batch;
       if (fetchvar_batch != 0) {
         // 此时 lod 为空。
-        tensor_out.lod = baskTask._batch_out[fetchvar_index].lod;
+        tensor_out.lod = batchTask._batch_out[fetchvar_index].lod;
         // resize all batch memory at one time
         size_t databuf_size = fetchvar_batch * fetchvar_bytesize_index;
         tensor_out.data.Resize(databuf_size);

core/predictor/framework/bsf.h

@@ -319,6 +319,7 @@ struct Task {
       for (size_t index = 0; index < vector_fetch_lod_index.size(); ++index) {
         size_t data_length = 0;
         size_t lod_length = 0;
+        size_t total_shape0 = 0;
         size_t feedvar_index = vector_fetch_lod_index[index];
         // 由于PaddleTensor的resize实现，是每次都会清空，所以必须先统计总长度。
         for (size_t taskmeta_index = 0; taskmeta_index < taskmeta_num;
@@ -326,6 +327,7 @@ struct Task {
           data_length +=
               outLodTensorVector[taskmeta_index][index].data.length();
           lod_length += outLodTensorVector[taskmeta_index][index].lod[0].size();
+          total_shape0 += outLodTensorVector[taskmeta_index][index].shape[0];
         }
         // 一次性扩容PaddleTensor中的data和lod
         paddle::PaddleTensor& fetchVarTensor = (*outVectorT_ptr)[feedvar_index];
@@ -368,8 +370,8 @@ struct Task {
     return true;
   }
 
-  bool task_fetch_init(BatchTasks<TaskT>& baskTask);
-  bool task_fetch_create(BatchTasks<TaskT>& baskTask);
+  bool task_fetch_init(BatchTasks<TaskT>& batchTask);
+  bool task_fetch_create(BatchTasks<TaskT>& batchTask);
 };
 
 // `Several Task` or `part of batch in Task` can be a TaskMeta.
@@ -788,6 +790,7 @@ class BatchTasks {
             paddle::PaddleTensor& fetchVarTensor =
                 task->outLodTensorVector[taskmeta_index][fetch_lod_index];
             size_t length = fetchvar_bytesize_index * shape0_length;
+            fetchVarTensor.shape[0] = shape0_length;
             fetchVarTensor.data.Resize(length);
             void* dst_ptr = fetchVarTensor.data.data();
             void* source_ptr = _batch_out[fetchvar_index].data.data() +
@@ -813,6 +816,7 @@ class BatchTasks {
             paddle::PaddleTensor& fetchVarTensor =
                 (*task->outVectorT_ptr)[fetchvar_index];
             size_t length = fetchvar_bytesize_index * shape0_length;
+            fetchVarTensor.shape[0] = shape0_length;
             fetchVarTensor.data.Resize(length);
             void* dst_ptr = fetchVarTensor.data.data();
             void* source_ptr = _batch_out[fetchvar_index].data.data() +