Simple PMP reader to parse Picasa's metadata

webkit/fileapi/media/picasa/pmp_column_reader.h

Index: webkit/fileapi/media/picasa/pmp_column_reader.h
diff --git a/webkit/fileapi/media/picasa/pmp_column_reader.h b/webkit/fileapi/media/picasa/pmp_column_reader.h
new file mode 100644
index 0000000000000000000000000000000000000000..a8c27d6d09aca996624f67d3a72c91512fe16ad7
--- /dev/null
+++ b/webkit/fileapi/media/picasa/pmp_column_reader.h
@@ -0,0 +1,68 @@
+// Copyright 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.
+
+#ifndef WEBKIT_FILEAPI_MEDIA_PICASA_PMP_COLUMN_READER_H_
+#define WEBKIT_FILEAPI_MEDIA_PICASA_PMP_COLUMN_READER_H_
+
+#include <string>
+#include <vector>
+
+#include "base/basictypes.h"
+#include "base/memory/scoped_ptr.h"
+#include "webkit/storage/webkit_storage_export.h"
+
+namespace base {
+class FilePath;
+}
+
+namespace fileapi {
+
+// Parses a PMP bytestream from either a file or directly from memory.

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

nit: update or remove comment.

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

Done.

+class WEBKIT_STORAGE_EXPORT_PRIVATE PmpColumnReader {
+ public:
+  explicit PmpColumnReader(std::string column_name);
+  virtual ~PmpColumnReader();
+
+  // Returns true if read successfully. |rows_read| undefined if returns false.
+  bool ReadFromMemory(uint32* rows_read, const uint8* data, const size_t len);

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

output paramaters should go at the end of the list, so:
bool ReadFromMemory(const uint8* data, const size_t len, uint32* rows_read);

Does the "spec" have a limit on the number of rows?  If so, you could return a
signed 32 bit int and use negative for errors.

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

Number of rows is defined as a uint32. We will run into the 50MB file limit way
before that, of course. I'll keep it as a bool for now to make it consistent
will all the other methods.

+
+  // Family of functions to read fields.

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

A bit more description wouldn't hurt here.  Only one type per column right?
So... "Read a row from this column. These return false if the column isn't of
the requested type or the requested row is out of range."

A better name for the second arg might be result instead of target.

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

Done.

+  bool ReadString(const uint32 row, std::string* target) const;
+  bool ReadUInt32(const uint32 row, uint32* target) const;
+  bool ReadDouble64(const uint32 row, double* target) const;
+  bool ReadUInt8(const uint32 row, uint8* target) const;
+  bool ReadUInt64(const uint32 row, uint64* target) const;
+
+  const std::string column_name() const {
+    return column_name_;
+  }
+
+ private:
+  // Private template method to support all the numeric Read*() methods.
+  template<class T> bool ReadPrimitive(const uint32 row, T* target) const;
+
+  // Only called from ReadFromMemory().
+  bool ParseData(uint32* rows_read);
+  bool IndexStrings();
+  template<class T> bool SetAndValidateFieldsize();
+
+  const std::string column_name_;
+
+  scoped_array<uint8> data_;
+  size_t length_;
+  uint32 rows_;
+
+  // Index of string start locations if fields are strings. Empty otherwise.
+  std::vector<const char*> strings_;
+
+  // Size in bytes of each field. This is for validation of Read requests.
+  // |fieldsize_| is -1 for uninitialized cases or string field types.
+  int fieldsize_;
+
+  DISALLOW_COPY_AND_ASSIGN(PmpColumnReader);
+};
+
+}  // namespace fileapi
+
+#endif  // WEBKIT_FILEAPI_MEDIA_PICASA_PMP_COLUMN_READER_H_