Create a histogram serialization mechanism in components/metrics

components/metrics/chromeos/serialization_utils.cc

Index: components/metrics/chromeos/serialization_utils.cc
diff --git a/components/metrics/chromeos/serialization_utils.cc b/components/metrics/chromeos/serialization_utils.cc
new file mode 100644
index 0000000000000000000000000000000000000000..7c8159e0d90f485f7b5d6084b7f2d820c201f9cf
--- /dev/null
+++ b/components/metrics/chromeos/serialization_utils.cc
@@ -0,0 +1,205 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "components/metrics/chromeos/serialization_utils.h"
+
+#include <sys/file.h>
+
+#include <string>
+#include <vector>
+
+#include "base/file_util.h"
+#include "base/files/file_path.h"
+#include "base/files/scoped_file.h"
+#include "base/logging.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/memory/scoped_vector.h"
+#include "base/strings/string_split.h"
+#include "base/strings/string_util.h"
+#include "components/metrics/chromeos/metric_sample.h"
+
+#define READ_WRITE_ALL_FILE_FLAGS \
+  (S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP | S_IROTH | S_IWOTH)
+
+namespace metrics {
+
+scoped_ptr<MetricSample> SerializationUtils::ParseSample(
+    const std::string& sample) {
+  if (sample.empty())
+    return scoped_ptr<MetricSample>();
+
+  std::vector<std::string> parts;
+  base::SplitString(sample, '\0', &parts);
+  // We should have two null terminated strings so split should produce
+  // three chunks.
+  if (parts.size() != 3) {
+    DLOG(ERROR) << "splitting message on \\0 produced " << parts.size()
+                << " parts (expected 3)";
+    return scoped_ptr<MetricSample>();
+  }
+  const std::string& name = parts[0];
+  const std::string& value = parts[1];
+
+  if (LowerCaseEqualsASCII(name, "crash")) {
+    return MetricSample::CrashSample(value);
+  } else if (LowerCaseEqualsASCII(name, "histogram")) {
+    return MetricSample::ParseHistogram(value);
+  } else if (LowerCaseEqualsASCII(name, "linearhistogram")) {
+    return MetricSample::ParseLinearHistogram(value);
+  } else if (LowerCaseEqualsASCII(name, "sparsehistogram")) {
+    return MetricSample::ParseSparseHistogram(value);
+  } else if (LowerCaseEqualsASCII(name, "useraction")) {
+    return MetricSample::UserActionSample(value);
+  } else {
+    DLOG(ERROR) << "invalid event type: " << name << ", value: " << value;
+  }
+  return scoped_ptr<MetricSample>();
+}
+
+void SerializationUtils::ReadAndTruncateMetricsFromFile(
+    const std::string& filename,
+    ScopedVector<MetricSample>* metrics) {
+  struct stat stat_buf;
+  int result;
+
+  result = stat(filename.c_str(), &stat_buf);
+  if (result < 0) {
+    if (errno != ENOENT)
+      DPLOG(ERROR) << filename << ": bad metrics file stat";
+
+    // Nothing to collect---try later.
+    return;
+  }
+  if (stat_buf.st_size == 0) {
+    // Also nothing to collect.
+    return;
+  }
+  base::ScopedFD fd(open(filename.c_str(), O_RDWR));
+  if (fd.get() < 0) {
+    DPLOG(ERROR) << filename << ": cannot open";
+    return;
+  }
+  result = flock(fd.get(), LOCK_EX);
+  if (result < 0) {
+    DPLOG(ERROR) << filename << ": cannot lock";
+    return;
+  }
+
+  // This processes all messages in the log. When all messages are
+  // read and processed, or an error occurs, truncate the file to zero size.
+  for (;;) {
+    std::string message;
+
+    if (!ReadMessage(fd.get(), &message))
+      break;
+
+    scoped_ptr<MetricSample> sample = ParseSample(message);
+    if (sample)
+      metrics->push_back(sample.release());
+  }
+
+  result = ftruncate(fd.get(), 0);
+  if (result < 0)
+    DPLOG(ERROR) << "truncate metrics log";
+
+  result = flock(fd.get(), LOCK_UN);
+  if (result < 0)
+    DPLOG(ERROR) << "unlock metrics log";
+}
+
+bool SerializationUtils::ReadMessage(int file_descriptor,

[Alexei Svitkine (slow), 2014/05/15 19:41:44]

Nit: Put this function into an anon namespace at the top of the file, since it
doesn't need to be publicly exposed in the header.

[bsimonnet, 2014/05/15 21:19:42]

Done.

+                                     std::string* message) {
+  int result;
+  int32 message_size;
+  // The file containing the metrics do not leave the device so the writer and
+  // the reader will always have the same endianness.
+  result =
+      HANDLE_EINTR(read(file_descriptor, &message_size, sizeof(message_size)));
+  if (result < 0) {
+    DPLOG(ERROR) << "reading metrics message header";
+    return false;
+  }
+  if (result == 0) {
+    // This indicates a normal EOF.
+    return false;
+  }
+  if (result < static_cast<int>(sizeof(message_size))) {
+    DLOG(ERROR) << "bad read size " << result << ", expecting "
+                << sizeof(message_size);
+    return false;
+  }
+
+  // kMessageMaxLength applies to the entire message: the 4-byte
+  // length field and the content.
+  if (message_size > kMessageMaxLength) {
+    DLOG(ERROR) << "message too long : " << message_size;
+    if (HANDLE_EINTR(lseek(file_descriptor, message_size - 4, SEEK_CUR)) ==

[Alexei Svitkine (slow), 2014/05/15 19:41:44]

Nit: I think changing |file_descriptor| -> |fd| will make wrapping a bit nicer
in this function.

[bsimonnet, 2014/05/15 21:19:42]

Done.

+        -1) {
+      DLOG(ERROR) << "error while skipping message. abort";
+      return false;
+    }
+    *message = "";

[Alexei Svitkine (slow), 2014/05/15 19:41:44]

Nit: This isn't needed, right?

[bsimonnet, 2014/05/15 20:50:21]

Not in the current format, this is meant to be a protection against refactoring
as the behaviour of ReadMessage is not very standard.

As we dont know what is in message in the beginning of this function if message
contains a valid message, the code will be wrong. I'll add
DCHECK(message.empty()) at the beginning of the function to make sure it is used
properly.

[bsimonnet, 2014/05/15 21:19:42]

Imposing constraint on the pointer used to store the output does not seem like a
good idea.
I feel like it makes more sense to consider that reading a badly formatted
sample is like reading an empty sample.
I added a comment to explain it.
Would that work?

[Alexei Svitkine (slow), 2014/05/15 22:40:50]

Sorry, my comment was actually incorrect - I had missed that this returned true
and the empty message was so that parsing isn't aborted, but rather than the
message skipped. Comment SGTM, thanks!

Nit: Change *message = ""; to message->clear(); to avoid constructing
temporaries.

+    return true;
+  }
+
+  message_size -= sizeof(message_size);  // The message size includes itself.
+  char buffer[kMessageMaxLength];
+  if (!base::ReadFromFD(file_descriptor, buffer, message_size)) {
+    DPLOG(ERROR) << "reading metrics message body";
+    return false;
+  }
+  *message = std::string(buffer, message_size);
+  return true;
+}
+
+bool SerializationUtils::WriteMetricToFile(const MetricSample& sample,
+                                           const std::string& filename) {
+  if (!sample.IsValid())
+    return false;
+
+  base::ScopedFD file_descriptor(open(filename.c_str(),
+                                      O_WRONLY | O_APPEND | O_CREAT,
+                                      READ_WRITE_ALL_FILE_FLAGS));
+
+  if (file_descriptor.get() < 0) {
+    DLOG(ERROR) << "error openning the file";
+    return false;
+  }
+
+  fchmod(file_descriptor.get(), READ_WRITE_ALL_FILE_FLAGS);
+  // Grab a lock to avoid chrome truncating the file
+  // underneath us. Keep the file locked as briefly as possible.
+  // Freeing file_descriptor will close the file and and remove the lock.
+  if (HANDLE_EINTR(flock(file_descriptor.get(), LOCK_EX)) < 0) {
+    DLOG(ERROR) << "error locking" << filename << " : " << errno;
+    return false;
+  }
+
+  std::string msg = sample.ToString();
+  int32 size = msg.length() + sizeof(int32);
+  if (size > kMessageMaxLength) {
+    DLOG(ERROR) << "cannot write message: too long";
+    return false;
+  }
+
+  // The file containing the metrics samples will only be read by programs on
+  // the same device so we do not check endianness.
+  if (base::WriteFileDescriptor(file_descriptor.get(),
+                                reinterpret_cast<char*>(&size),
+                                sizeof(size)) != sizeof(size)) {
+    DLOG(ERROR) << "error writing message length " << errno;
+    return false;
+  }
+
+  if (base::WriteFileDescriptor(
+          file_descriptor.get(), msg.c_str(), msg.length()) !=
+      static_cast<int>(msg.length())) {
+    DLOG(ERROR) << "error writing message" << errno;
+    return false;
+  }
+
+  return true;
+}
+
+}  // namespace metrics