stardict.js

JavaScript library for handling dictionaries in StarDict format.

The code depends on two further javascript libraries, namely:

• https://github.com/tuxor1337/dictzip.js
• https://github.com/nodeca/pako

referred to in the code as "DictZipFile" and "pako" respectively.

Example Code

Note that the code in the "demo" subdirectory depends on the files "pako_inflate.js", "dictzip.js" and "dictzip_sync.js" from the pako project and the dictzip.js project respectively. If you want to run the demo code, copy versions of those files into the demo directory.

Documentation: Synchronous interface

Use importScripts to import stardict_sync.js into your worker's scope. This will define the global variable StarDict. You create a new instance of StarDict as follows:

var dict = new StarDict(files);

files is an array of File objects (as in the File API http://www.w3.org/TR/FileAPI/). It's supposed to contain exactly one file of type *.dict(.dz), *.idx(.dz), *.ifo. Each instance of StarDict provides the following methods:

var value = dict.keyword(key);

key is a string and value is the value of the respective key in the dictionary's *.ifo file or null if key is not defined in the *.ifo file. Format specific keys include version, bookname, wordcount, synwordcount, idxfilesize and sametypesequence. You will typically be intrested in the value of bookname.

A dictionary in the StarDict format comes with a word index and optionally synonyms with references to this index. You access the index as follows:

var aIdx = dict.index(options);

Here aIdx is an array of objects, even if it contains only one or no entry. The form of the objects contained in aIdx is determined by the optional options parameter:

options = {
    start_offset: unsigned int, // default: 0
    count: unsigned int,        // default: 1 if start_offset != 0 else `wordcount`
    include_term: boolean,      // default: true
    include_dictpos: boolean,   // default: true
    include_offset: boolean     // default: false
};

All properties are optional. start_offset is the offset in the dictionary's *.idx file from which to start reading and count the (maximal) number of objects aIdx will contain. The include_\* properties determine the format of the objects in aIdx:

aIdx[n] = {
    term: string,                // if include_term == true
    dictpos: [startbyte, size],  // if include_dictpos == true
    offset: unsigned int         // if include_offset == true
};

Here dictpos is of the form which the entry method expects (see below) and offset corresponds to the start_offset property of options, thus represents the offset of this term in *.idx.

In a perfectly analogous way you can access the dictionary's synonyms:

var aSyns = dict.synonyms(options);

Here aSyns is an array of objects, even if it contains only one or no entry. The form of the objects contained in aSyns is determined by the optional options parameter:

options = {
    start_offset: unsigned int, // default: 0
    count: unsigned int,        // default: 1 if start_offset != 0 else `wordcount`
    include_term: boolean,      // default: true
    include_wid: boolean,       // default: true
    include_offset: boolean     // default: false
};

All properties are optional. start_offset is the offset in the dictionary's *.syn file from which to start reading and count the (maximal) number of objects aSyns will contain. The include_\* properties determine the format of the objects in aSyns:

aSyns[n] = {
    term: string,          // if include_term == true
    wid: unsigned int,     // if include_wid == true
    offset: unsigned int   // if include_offset == true
};

offset corresponds to the start_offset property of options, thus represents the offset of this term in *.syn. Note that wid is not the value corresponding to the start_offset option of the index method (cf. above). Instead it's the position of the corresponding object in the array dict.index() of all index terms.

You access the articles/entries of the dictionary (contained in *.dict) via the method

var aData = dict.entry(dictpos);

dictpos is an array [startbyte, size] as provided by the index method. Because each entry in *.dict is composed of several data blocks, aData is an array of data objects of the following form:

aData[n] = {
    type: string,                   // one of m,l,g,t,x,y,k,w,h,r,W,P,X
    content: ArrayBuffer or string
};

The type of a data block is defined in the StarDict format documentation (see link below), e.g. m is text/plain, h is text/html etc. content is a string for the data types m,g,t,x,y,k,w,h,r and an ArrayBuffer else. It's up to you to interpret the other types according to the StarDict format documentation.

Finally, you can access the resources provided with the dictionary (and usually referenced in the entries) using

var blob = dict.resource(name);

blob is a Blob object (or null if the resource doesn't exist) and name is a string.

Documentation: Asynchronous interface

The asynchronous version of the StarDict object provided in stardict.js needs to be initialized using the load method:

dict.load().then(function () {
    // start using the other methods
},
function (err) {
    // in case of errors; `err` is an instance of Error
});

After that, all definitions from the synchronous case also apply to this case. But the methods return corresponding Promise object, e.g.

dict.synonym(options).then(function (aSyns) {
    // do something with aSyns
});

Further reading

Format documentation (original documentation is lost): http://code.google.com/p/babiloo/wiki/StarDict_format

Python lib: http://code.google.com/p/pytoolkits/source/browse/trunk/utils/stardict/StarDict.py Java lib: http://code.google.com/p/toolkits/source/browse/trunk/android/YAStarDict/src/com/googlecode/toolkits/stardict/StarDict.java