public static IEnumerable <MYDRecord> Records(MYDBinaryReader br)
{
foreach (byte[] record_data in MYDBlockReader.Blocks(br))
{
yield return(ProcessRecord(record_data));
}
}
private static MYDRecord ProcessRecord(byte[] record_data)
{
//DumpRecord(record_data);
//return;
using (MemoryStream ms = new MemoryStream(record_data))
{
MYDBinaryReader br = new MYDBinaryReader(ms);
{
int NULLIES = 7;
// Read the nullyinfo bits
byte[] nulls_bytes = br.ReadBytes(NULLIES);
string nulls = "";
for (int i = 0; i < NULLIES; ++i)
{
nulls = Convert.ToString(nulls_bytes[i], 2).PadLeft(8, '0') + nulls;
}
nulls = StringTools.Reverse(nulls);
// Read the id
int id = -1;
if ((nulls_bytes[0] & 0x01) == 0x00)
{
byte[] id_bytes = br.ReadBytes(4);
id = 1 * id_bytes[0] + 256 * id_bytes[1] + 256 * 256 * id_bytes[2] + 256 * 256 * 256 * id_bytes[3];
}
// Read the reference type
int reference_type = -1;
if ((nulls_bytes[0] & 0x02) == 0x00)
{
byte[] reference_type_bytes = br.ReadBytes(2);
reference_type = 1 * reference_type_bytes[0] + 256 * reference_type_bytes[1];
}
// Read the varfields
Dictionary <string, string> datas = new Dictionary <string, string>();
for (int field = 2; field < 54; ++field)
{
if (nulls[field] == '0')
{
byte[] data_length_bytes = br.ReadBytes(3);
int data_length = 1 * data_length_bytes[0] + 256 * data_length_bytes[1] + 256 * 256 * data_length_bytes[2];
byte[] data_bytes = br.ReadBytes(data_length);
string data = Encoding.ASCII.GetString(data_bytes);
datas[EndnoteConstants.FIELD_NAMES[field]] = data;
}
}
MYDRecord myd_record = new MYDRecord();
myd_record.id = id;
myd_record.reference_type = reference_type;
myd_record.fields = datas;
return(myd_record);
}
}
}
public static IEnumerable <byte[]> Blocks(MYDBinaryReader br)
{
while (br.Position < br.Length)
{
MYDBlock root_block = MYDBlock.Read(br);
//root_block.DumpStats();
long post_block_processing_return_offset = br.Position;
// Only follow blocks that are a 'starting block'
if (!root_block.is_block_first)
{
continue;
}
// Concatenate all blocks
byte[] record_data = new byte[root_block.rec_len];
{
MYDBlock block = root_block;
int record_data_i = 0;
while (true)
{
Array.Copy(block.data, 0, record_data, record_data_i, block.data_len);
record_data_i += block.data_len;
if (block.is_block_last)
{
break;
}
if (0 == block.next_filepos)
{
throw new GenericException("Expecting a file offset if we have more blocks!");
}
br.Seek(block.next_filepos);
//Logging.Info("Appending additional data from block at {0}", block.next_filepos);
block = MYDBlock.Read(br);
//block.DumpStats("\t");
}
if (record_data_i != root_block.rec_len)
{
throw new GenericException("Block building length {0} is different to expected length of {1}!", record_data_i, root_block.rec_len);
}
}
br.Seek(post_block_processing_return_offset);
yield return(record_data);
}
}
private static void DumpRecord(byte[] record_data)
{
using (MemoryStream ms = new MemoryStream(record_data))
{
MYDBinaryReader br = new MYDBinaryReader(ms);
{
int NULLIES = 20;
// Read the nullyinfo bits
byte[] nulls_bytes = br.ReadBytes(NULLIES);
string binary = "";
string hex = "";
string ascii = "";
for (int i = 0; i < NULLIES; ++i)
{
binary = binary + Convert.ToString(nulls_bytes[i], 2).PadLeft(8, '0') + " ";
hex = hex + String.Format("{0:X2}", nulls_bytes[i]);
ascii = ascii + ((nulls_bytes[i] < 32) ? "." : ASCIIEncoding.ASCII.GetString(nulls_bytes, i, 1));
}
Logging.Info("{0} {1} {2}", ascii, hex, binary);
}
}
}
internal static MYDBlock Read(MYDBinaryReader br)
{
MYDBlock block = new MYDBlock();
// Debugging purposes
block.start_position = br.Position;
// Start of header
block.header = br.ReadByte();
block.rec_len = 0;
block.data_len = 0;
block.block_len = 0;
block.next_filepos = 0;
// See http://dev.mysql.com/doc/internals/en/layout-record-storage-frame.html
// Or bottom of this code
switch (block.header)
{
case 0:
block.block_len = br.Read3();
block.block_len -= 4; // Deleted blocks include the header in their total length...sigh.
block.is_block_deleted = true;
break;
case 1:
block.rec_len = block.data_len = block.block_len = br.Read2();
block.is_block_first = block.is_block_last = true;
break;
case 2:
block.rec_len = block.data_len = block.block_len = br.Read3();
block.is_block_first = block.is_block_last = true;
break;
case 13:
block.rec_len = br.Read4();
block.data_len = block.block_len = br.Read3();
block.next_filepos = br.Read8();
block.is_block_first = true;
break;
case 3:
block.rec_len = block.data_len = br.Read2();
block.block_len = block.rec_len + br.Read1();
block.is_block_first = block.is_block_last = true;
break;
case 4:
block.rec_len = block.data_len = br.Read3();
block.block_len = block.rec_len + br.Read1();
block.is_block_first = block.is_block_last = true;
break;
case 5:
block.rec_len = br.Read2();
block.data_len = block.block_len = br.Read2();
block.next_filepos = br.Read8();
block.is_block_first = true;
break;
case 6:
block.rec_len = br.Read3();
block.data_len = block.block_len = br.Read3();
block.next_filepos = br.Read8();
block.is_block_first = true;
break;
case 7:
block.data_len = block.block_len = br.Read2();
block.is_block_last = true;
break;
case 8:
block.data_len = block.block_len = br.Read3();
block.is_block_last = true;
break;
case 9:
block.data_len = br.Read2();
block.block_len = block.data_len + br.Read1();
block.is_block_last = true;
break;
case 10:
block.data_len = br.Read3();
block.block_len = block.rec_len + br.Read1();
block.is_block_last = true;
break;
case 11:
block.data_len = block.block_len = br.Read2();
block.next_filepos = br.Read8();
break;
case 12:
block.data_len = block.block_len = br.Read3();
block.next_filepos = br.Read8();
break;
default:
break;
}
block.data = br.ReadBytes(block.block_len);
return(block);
}