Conversational speech tool: timing model with data access.

webrtc/modules/audio_processing/test/conversational_speech/timing.cc

Index: webrtc/modules/audio_processing/test/conversational_speech/timing.cc
diff --git a/webrtc/modules/audio_processing/test/conversational_speech/timing.cc b/webrtc/modules/audio_processing/test/conversational_speech/timing.cc
new file mode 100644
index 0000000000000000000000000000000000000000..83c99047ac1c726c5931808fa1d9bee5f1350f57
--- /dev/null
+++ b/webrtc/modules/audio_processing/test/conversational_speech/timing.cc
@@ -0,0 +1,89 @@
+/*
+ *  Copyright (c) 2017 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "webrtc/modules/audio_processing/test/conversational_speech/timing.h"
+
+#include <fstream>
+#include <iostream>
+#include <utility>
+
+#include "webrtc/base/stringencode.h"
+
+namespace webrtc {
+namespace test {
+namespace conversational_speech {
+
+bool Timing::Turn::operator ==(const Turn &b) const {
+  return b.speaker_name == speaker_name &&
+         b.audiotrack_file_name == audiotrack_file_name &&
+         b.offset == offset;
+}
+
+Timing::Timing() {}
+
+Timing::~Timing() {
+  Clear();

[kwiberg-webrtc, 2017/03/16 14:10:08]

Not necessary, right?

[AleBzk, 2017/03/20 14:25:28]

Acknowledged.

+}
+
+void Timing::Clear() {
+  // TODO(alessiob): check if pointers to Turn instances must be explicitly
+  // deleted (I assume not).

[kwiberg-webrtc, 2017/03/16 14:10:09]

No, unique_ptr will delete the thing it owns---that's pretty much its only
job...

[AleBzk, 2017/03/20 14:25:28]

Acknowledged.

+  turns_.clear();
+}
+
+void Timing::AppendTurn(std::unique_ptr<Timing::Turn> turn) {
+  turns_.push_back(std::move(turn));
+}
+
+void Timing::Load(const std::string& timing_filepath) {
+  std::string line;
+
+  // Line parser.
+  auto parse_line_ = [&line]() {
+    std::vector<std::string> fields;
+    rtc::split(line, ' ', &fields);
+    RTC_CHECK_EQ(fields.size(), 3);
+    return std::unique_ptr<Timing::Turn>(new Turn(
+        fields[0], fields[1], std::atol(fields[2].c_str())));
+  };
+
+  // Parse lines.
+  std::ifstream infile(timing_filepath);
+  while (std::getline(infile, line)) {
+    if (line.empty())
+      continue;
+    AppendTurn(parse_line_());

[kwiberg-webrtc, 2017/03/16 14:10:09]

Pass line as an argument here, instead of capturing a reference to it in the
lambda! As it is, it's very easy to miss this implicit argument.

[AleBzk, 2017/03/20 14:25:28]

Done.

+  }
+  infile.close();
+}
+
+void Timing::Save(const std::string& timing_filepath) const {
+  std::ofstream outfile(timing_filepath);
+  // TODO(alessio): check if file open for writing.
+  for (const auto& turn : turns_) {
+    outfile << turn->speaker_name << " " << turn->audiotrack_file_name
+        << " " << turn->offset << std::endl;
+  }
+  outfile.close();
+}
+
+const std::vector<std::unique_ptr<Timing::Turn>>& Timing::turns() const {
+  // TODO(alessiob): not sure if this is the best way to return, maybe we can
+  // return an iterator.

[kwiberg-webrtc, 2017/03/16 14:10:09]

This is fine. You could also return an ArrayView<std::unique_ptr<Timing::Turn>>
(ArrayView<Timing::Turn> if you follow my other advice).

[AleBzk, 2017/03/20 14:25:28]

I'm new with ArrayView, I guess you refer to the one defined in
//webrtc/base/array_view.h

[kwiberg-webrtc, 2017/03/21 22:35:14]

Yes. (I wrote it, so if you have questions or complaints, direct them my way...)

+  return turns_;
+}
+
+std::size_t Timing::Size() const {
+  return turns_.size();
+}

[kwiberg-webrtc, 2017/03/16 14:10:08]

Is this necessary? Callers can do .turns().size().

[AleBzk, 2017/03/20 14:25:28]

Removed

+
+}  // namespace conversational_speech
+}  // namespace test
+}  // namespace webrtc