[tf.data] Add metric to record file names read by tf.data Datasets.

PiperOrigin-RevId: 355887516
Change-Id: Ic32e6cc9cd07e0ec24f77401cb3d2c6908a0a2cd

tensorflow/core/framework/metrics.cc

@@ -115,6 +115,10 @@ auto* tf_data_iterator_lifetime_counter = monitoring::Counter<0>::New(
 auto* tf_data_optimization_counter = monitoring::Counter<1>::New(
     "/tensorflow/data/optimization", "tf.data optimization", "name");
 
+auto* tf_data_filename_counter = monitoring::Counter<2>::New(
+    "/tensorflow/data/filename", "The file name read by a tf.data Dataset.",
+    "name", "filename");
+
 auto* parse_dense_feature_counter = monitoring::Counter<0>::New(
     "/tensorflow/data/dense_feature",
     "The number of dense features parsed by ops for parsing tf.Example.");
@@ -218,6 +222,10 @@ void RecordTFDataOptimization(const string& name, int64 num_changes) {
   tf_data_optimization_counter->GetCell(name)->IncrementBy(num_changes);
 }
 
+void RecordTFDataFilename(const string& name, const string& filename) {
+  tf_data_filename_counter->GetCell(name, filename)->IncrementBy(1);
+}
+
 void RecordParseDenseFeature(int64 num_features) {
   static auto* parse_dense_feature_counter_cell =
       parse_dense_feature_counter->GetCell();

tensorflow/core/framework/metrics.h

@@ -84,6 +84,11 @@ void RecordTFDataIteratorLifetime(uint64 duration_us);
 // The `name` argument identifies the optimization (e.g. "noop_elimination").
 void RecordTFDataOptimization(const string& name, int64 num_changes);
 
+// Records the file name read by a tf.data Dataset.
+//
+// The `name` argument identifies the Dataset type (e.g. "TFRecordDataset").
+void RecordTFDataFilename(const string& name, const string& filename);
+
 // Records parsing of dense tensor features.
 void RecordParseDenseFeature(int64 num_features);
 

tensorflow/core/kernels/data/fixed_length_record_dataset_op.cc

@@ -471,6 +471,7 @@ void FixedLengthRecordDatasetOp::MakeDataset(OpKernelContext* ctx,
   filenames.reserve(filenames_tensor->NumElements());
   for (int i = 0; i < filenames_tensor->NumElements(); ++i) {
     filenames.push_back(filenames_tensor->flat<tstring>()(i));
+    metrics::RecordTFDataFilename(kDatasetType, filenames[i]);
   }
 
   int64 header_bytes = -1;

tensorflow/core/kernels/data/text_line_dataset_op.cc

@@ -275,6 +275,7 @@ void TextLineDatasetOp::MakeDataset(OpKernelContext* ctx,
   filenames.reserve(filenames_tensor->NumElements());
   for (int i = 0; i < filenames_tensor->NumElements(); ++i) {
     filenames.push_back(filenames_tensor->flat<tstring>()(i));
+    metrics::RecordTFDataFilename(kDatasetType, filenames[i]);
   }
 
   *output = new Dataset(ctx, std::move(filenames), compression_type,

tensorflow/core/kernels/data/tf_record_dataset_op.cc

@@ -292,6 +292,7 @@ void TFRecordDatasetOp::MakeDataset(OpKernelContext* ctx,
     filenames.push_back(filenames_tensor->flat<tstring>()(i));
     is_gcs_fs &= absl::StartsWith(filenames[i], kGcsFsPrefix);
     is_s3_fs &= absl::StartsWith(filenames[i], kS3FsPrefix);
+    metrics::RecordTFDataFilename(kDatasetType, filenames[i]);
   }
 
   tstring compression_type;