An open API service indexing awesome lists of open source software.

https://github.com/hehonghui/mdict-analysis


https://github.com/hehonghui/mdict-analysis

Last synced: 5 months ago
JSON representation

Awesome Lists containing this project

README

          

An Analysis of MDX/MDD File Format
==================================

MDict is a multi-platform open dictionary

which are both questionable. It is not available for every platform, e.g. OS X, Linux.
Its dictionary file format is not open. But this has not hindered its popularity,
and many dictionaries have been created for it.

This is an attempt to reveal MDX/MDD file format, so that my favorite dictionaries,
created by MDict users, could be used elsewhere.

MDict Files
===========
MDict stores the dictionary definitions, i.e. (key word, explanation) in MDX file and
the dictionary reference data, e.g. images, pronunciations, stylesheets in MDD file.
Although holding different contents, these two file formats share the same structure.

MDict v1 and v2 Format
======================
MdxBuilder 2.x creates v1 format and MdxBuilder 3.x creates v2 format.

.. image:: MDX.svg

.. image:: MDD.svg

MDict v3 Format
===============
MdxBuilder 4.x creates v3 format. The major changes are,

* The file is divided into 4 blocks, i.e. key index, key data, record index and record data.
And all blocks have the same structure. The decoded block holds type dependent data.
* Partial block data (first 16 bytes as now) is always encrypted (after possible compression), via _fast_encrypt or _salsa_encrypt.
The encryption key derives globally from the dictionary header's UUID value (since MdxBuilder 4.0 RC4), ::

key = xxh64_digest(uuid[:18]) + xxh64_digest(uuid[18:])

or individually from each block data's Adler-32 checksum (until MdxBuilder 4.0 RC2), ::

key = ripemd128(alder)

.. image:: MDict3.svg

Example Programs
================

readmdict.py
------------
readmdict.py is an example implementation in Python. This program can read/extract mdx/mdd files.

.. note:: python-lzo is required to read mdx files created with engine 1.2.
xxhash is required to read mdx files created with engine 3.0.
Get Windows version from http://www.lfd.uci.edu/~gohlke/pythonlibs

It can be used as a command line tool. Suppose one has *oald8.mdx* and *oald8.mdd*::

$ python readmdict.py -x oald8.mdx

This will create a dictionary file *oald8.txt* and a folder *data* for images, pronunciation audio files.

On Windows, one can also double click it and select the file in the popup dialog.

Or as a module::

In [1]: from readmdict import MDX, MDD

Read MDX file and print the first entry::

In [2]: mdx = MDX('oald8.mdx')

In [3]: items = mdx.items()

In [4]: next(items)
Out[4]:
('A',
'.........')

``mdx`` is an object having all info from a MDX file. ``items`` is an iterator producing 2-item tuples.
Of each tuple, the first element is the entry text and the second is the explanation. Both are UTF-8 encoded strings.

Read MDD file and print the first entry::

In [5]: mdd = MDD('oald8.mdd')

In [6]: items = mdd.items()

In [7]: next(items)
Out[7]:
(u'\\pic\\accordion_concertina.jpg',
'\xff\xd8\xff\xe0\x00\x10JFIF...........')

``mdd`` is an object having all info from a MDD file. ``items`` is an iterator producing 2-item tuples.
Of each tuple, the first element is the file name and the second element is the corresponding file content.
The file name is encoded in UTF-8. The file content is a plain bytes array.

Acknowledge
===========
The file format gets fully disclosed by https://github.com/zhansliu/writemdict.
The encryption part is taken into this project.