!7902 [MD]add default for sentencepiece col_names processing and vocab_size must be provided

Merge pull request !7902 from xulei/lite_test0905

mindspore/ccsrc/minddata/dataset/api/python/de_pipeline.cc

@@ -1959,6 +1959,8 @@ Status DEPipeline::ParseBuildSentencePieceVocabOp(const py::dict &args, std::sha
     if (!value.is_none()) {
       if (key == "vocab_size") {
         builder->SetVocabSize(ToInt(value));
+      } else if (key == "columns") {
+        (void)builder->SetColumnNames(ToStringVector(value));
       } else if (key == "character_coverage") {
         (void)builder->SetCharacterCoverage(ToFloat(value));
       } else if (key == "params") {

mindspore/dataset/text/validators.py

@@ -436,6 +436,9 @@ def check_from_dataset_sentencepiece(method):
 
         if vocab_size is not None:
             check_uint32(vocab_size, "vocab_size")
+        else:
+            raise TypeError("vocab_size must be provided")
+
 
         if character_coverage is not None:
             type_check(character_coverage, (float,), "character_coverage")