Simple PMP reader to parse Picasa's metadata

webkit/fileapi/media/picasa/pmp_table_reader.cc

Index: webkit/fileapi/media/picasa/pmp_table_reader.cc
diff --git a/webkit/fileapi/media/picasa/pmp_table_reader.cc b/webkit/fileapi/media/picasa/pmp_table_reader.cc
new file mode 100644
index 0000000000000000000000000000000000000000..bd008acd6edbe5849b372585fb399d65c632f263
--- /dev/null
+++ b/webkit/fileapi/media/picasa/pmp_table_reader.cc
@@ -0,0 +1,113 @@
+// Copyright (c) 2013 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 "webkit/fileapi/media/picasa/pmp_table_reader.h"
+
+#include <algorithm>
+
+#include "base/file_util.h"
+#include "base/files/file_path.h"
+#include "base/files/memory_mapped_file.h"
+#include "base/logging.h"
+#include "webkit/fileapi/media/picasa/pmp_column_reader.h"
+#include "webkit/fileapi/media/picasa/pmp_constants.h"
+
+namespace fileapi {
+
+namespace {
+  COMPILE_ASSERT(sizeof(double) == 8, double_must_be_8_bytes_long);

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

no indent on namespace, blank lines before and after.

[tommycli, 2013/03/27 19:34:30]

Done.

+  const size_t kPmpMaxFilesize = 50*1024*1024; // Maximum of 50 MB.
+}
+
+using std::string;

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

we don't "use" anything from std.

[tommycli, 2013/03/27 19:34:30]

Done.

+
+PmpTableReader::PmpTableReader() : column_readers_(), max_row_count_(0) {
+

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

nit: extra line.

[tommycli, 2013/03/27 19:34:30]

Done.

+}
+
+PmpTableReader::~PmpTableReader() {
+  PmpTableReader::FreeAndClearContents(column_readers_);
+}
+
+bool PmpTableReader::ReadFromDisk(const string& table_name,
+                                  const base::FilePath& directory_path,
+                                  const std::vector<std::string>& columns) {
+  // Expect to read some columns, otherwise, a programming error.
+  DCHECK(!columns.empty());
+
+  // Directory must exist.
+  if(!file_util::DirectoryExists(directory_path)) {

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

nit: no {}s

[tommycli, 2013/03/27 19:34:30]

Done.

+    return false;
+  }
+
+  string table_prefix = table_name + "_";
+  string pmp_suffix = ".pmp";

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

Maybe this should go in the constants file?

[tommycli, 2013/03/27 19:34:30]

Done.

+
+  // Look for the "%s_0".format(table_name_) file, indicating table existence.
+  base::FilePath indicator_file = directory_path.Append(table_prefix + "0");
+
+  // Expect the indicator file to exist but not be a directory.
+  if(!file_util::PathExists(indicator_file) ||
+     file_util::DirectoryExists(indicator_file)) {
+    return false;
+  }
+
+  std::vector<PmpColumnReader*> current_col_readers;
+  uint32 current_max_rows_parsed = 0;
+
+  // Construct the column readers.
+  for (std::vector<std::string>::const_iterator it = columns.begin();
+       it != columns.end(); it++) {
+    base::FilePath colfile_path = directory_path.Append(

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

nit: column_file_path

[tommycli, 2013/03/27 19:34:30]

Done.

+        table_prefix + *it + pmp_suffix);
+
+    // Constructed on the heap due to PmpColumnReader disallowing

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

Not sure this comment is necessary.

[tommycli, 2013/03/27 19:34:30]

Done.

+    // copies and assignment.
+    PmpColumnReader* column_reader = new PmpColumnReader(*it);
+    current_col_readers.push_back(column_reader);
+
+    base::MemoryMappedFile file;
+    uint32 row_cnt; // row count

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

unnecessary comment.

[tommycli, 2013/03/27 19:34:30]

Done.

+
+    // If file is not initialized, too small, or too large (50MB), fail out.
+    if (!file.Initialize(colfile_path) ||
+        file.length() < kPmpHeaderSize ||
+        file.length() > kPmpMaxFilesize ||
+        !column_reader->ReadFromMemory(&row_cnt, file.data(), file.length())) {

[vandebo (ex-Chrome), 2013/03/27 00:06:19]

This configuration forces a copy.  If you just pass the file name, the column
reader probably doesn't have to make a copy of the bytes, it can just use the
mmap'd file.

[tommycli, 2013/03/27 19:34:30]

OBE.

+      PmpTableReader::FreeAndClearContents(current_col_readers);
+      return false;
+    }
+
+    current_max_rows_parsed = std::max(current_max_rows_parsed, row_cnt);
+  }
+
+  // Actually succeeded. Set member data.
+  PmpTableReader::FreeAndClearContents(column_readers_);
+  column_readers_ = current_col_readers;
+  max_row_count_ = current_max_rows_parsed;
+
+  return true;
+}
+
+uint32 PmpTableReader::Count() const {
+  return max_row_count_;
+}
+
+std::vector<const PmpColumnReader*> PmpTableReader::GetColumns() const {
+  std::vector<const PmpColumnReader*> readers;
+  std::copy(column_readers_.begin(), column_readers_.end(),
+            std::back_inserter(readers));
+  return readers;
+}
+
+void PmpTableReader::FreeAndClearContents(
+    std::vector<PmpColumnReader*> readers) {
+  for (std::vector<PmpColumnReader*>::iterator it = readers.begin();
+       it != readers.end(); it++) {
+    delete *it;
+  }
+  readers.clear();
+}
+
+}  // namespace fileapi