/usr/share/doc/opendict/Plugin-HOWTO.html

<html>
<head>
   <meta http-equiv="Content-Type" content="text/html;
   charset=UTF-8"/>
   <title>OpenDict Add-Ons Development HOWTO</title>
</head>
<style>
body
{
    font-family: Verdana, Arial, Sans, sans-serif;
    font-size: 12px;
    color: #555;
}

h1
{
    font-size: 26px;
    text-align: center;
    font-weight: normal;
    color: #2b4d85;
}

h2
{
    color: #2b4d85;
    border-bottom: 1px dashed #2b4d85;
    margin: 0px;
    margin-top: 20px;
    margin-bottom: 10px;
}

div.footer
{
    text-align: center;
    font-size: 10px;
    background-color: #EEE;
    border: 1px solid #D9D9D9;
    color: #AAA;
}

a
{
    color: #2b4d85;
}

a:active, a:hover
{
    background-color: #DDD;
}

pre.file-contents
{
    border: 1px solid #AAA;
    margin-left: 30px;
    width: 600px;
    background-color: #EEE;
}

table
{
    font: inherit;
    border: 1px solid #AAA;
    border-collapse: collapse;
    margin-left: 30px;
    width: 600px;
}

table td, table, th
{
    border: 1px solid #DDD;
    padding: 5px;
}
</style>
<body>
<h1>OpenDict Add-Ons Development HOWTO</h1>

<h2>Introduction</h2>
<p>
OpenDict is free multiplatform dictionary. More information about it
can be found on <a href="http://opendict.idiles.com">OpenDict website</a>.
</p>
<p>
This document is written for developers who want to make
easy-installable OpenDict dictionaries.
</p>
<p>
<i><b>Warning:</b> It is not necessary to make OpenDict dictionary if you
have a dictionary file in Slowo, Mova or DICT format. Such
dictionaries can be used right now by selecting <u>Dictionaries ->
Install Dictionary from File</u> from OpenDict menu.</i>
</p>

<h2>Types of OpenDict Dictionaries</h2>
<p>
OpenDict has two types of dictionaries: <i>plain</i> (simple) dictionaries and 
<i>plugin</i> (complex)
dictionaries. <i>Plain</i> dictionaries consists of dictionary file (type of
that dictionary must be supported by OpenDict) and description file in
XML format. <i>Plugin</i> dictionaries consists of Python module with code
that handles search process and description file in XML format; it
may also have dictionary file which is processed by Python module
mentioned above.
</p>
<p>
Plain dictionaries are simple and handy. It is very easy to install
them using OpenDict itself. If you have dictionary file (format of
that dictionary must be supported by OpenDict) and want to attach some
information to it, like the name, the author, the description, you may
want to make <i>plain</i> dictionary.
</p>
<p>
If you have more complex task, like search the web, etc, you may want
to make a <i>plugin</i> for OpenDict. The plugin is a chunk of code
written in <a href="http://www.python.org">Python</a> programming
language that OpenDict attaches to itself at runtime.
</p>
<p>
The next sections of this document will describe how to make
<i>plain</i> dictionaries and <i>plugins</i>.
</p>

<h2>Plain Dictionary Example</h2>
<p>
Assume we have a small dictionary file named <i>mydict.dwa</i> in Slowo
format:
</p>

<pre class="file-contents">
above = ant ; virš ;
abroad = visur ; užsienyje ;
acoustic = akustinis ;
acquaint = pranešti ;
</pre>

<p>
Say we want to call this dictionary "My Personal Dictionary" and
attacht a description "This is my personal dictionary. It is very
small, but it is not the size that matters :)" to it.
<i>Each</i> plain dictionary must have XML description file named
<i>config.xml</i>, otherwise it will be treated as invalid dictionary. So we
write an XML file called config.xml with our favorite text or XML editor: 
</p>

<pre class="file-contents">
&lt;?xml version='1.0' encoding='UTF-8'?&gt;
&lt;plain-dictionary&gt;
  &lt;format&gt;slowo&lt;/format&gt;
  &lt;name&gt;My Personal Dictionarys&lt;/name&gt;
  &lt;version&gt;0.1&lt;/version&gt;
  &lt;authors&gt;
     &lt;author name="Your Name" email="your@email.tdl"/&gt;
  &lt;/authors&gt;
  &lt;path&gt;mydict.dwa&lt;/path&gt;
  &lt;md5&gt;9c62810c32ca20fe018b79987789daef&lt;/md5&gt;
  &lt;encoding&gt;UTF-8&lt;/encoding&gt;
  &lt;description&gt;&lt;![CDATA[
     This is my personal dictionary. It is very small, but it is not the
     size that matters :)
  ]]&gt;
  &lt;/description&gt;
&lt;/plain-dictionary&gt;
</pre>

<p>
As you can see, a little more information must be added to description
(i.e. configuration) file. Here is a short description of each section:
</p>

<table>
<tr>
    <th>XML tag</th>
    <th>Description</th>
</tr>
<tr>
    <td>format</td>
    <td>Dictionary format. OpenDict supports the following formats:
        <ul>
	    <li>slowo -- Slowo format</li>
	    <li>mova -- Mova format</li>
	    <li>dict -- DICT format</li>
	</ul>
	<i>TODO: describe all supported format somewhere</i>
    </td>
</tr>
<tr>
    <td>name</td>
    <td>Name of the dictionary. It will be shown in the
    <i>Dictionaries</i> menu in OpenDict window.
</td>
</tr>
<tr>
    <td>version</td>
    <td>Version value is very important. It shows the freshness of
    your dictionary. The more recent the dictionary, the greater
    version value must be set. For example: 0.1, 0.2, 0.3, etc; or 1,
    2, 3, etc.</td>
</tr>
<tr>
    <td>author</td>
    <td>
    <p>The author of the dictionary. Notice that <i>author</i> tag is
    inside <i>authors</i> tag, because there may be several
    authors of the dictionary. So add as may <i>author</i> tags as you
    want, but make them the childs of the <i>authors</i> tag.</p>

    <p>This tag is a bit complicated, because you may not be the
    author of the dictionary file you are using. If so, you may treat
    the <i>author</i> tag as maintainance tag and write your <i>as
    maintainer</i> name inside it. In addition to this, you should
    write the name of the author in the <i>description</i> tag
    mentioned below.
</td>
</tr>
<tr>
    <td>path</td>
    <td>Path to the dictionary file. It may be an absolute (i.e. full)
    path to the dictionary file or relative path. If you write full
    path, dictionary would be taken from there. Buf if you write only
    the name of the dictionary (i.e. mydict.dwa), the dictionary will
    be treated to be located at <i>$DICTDIR/file/mydict.dwa</i>, where
    $DICTDIR is directory where you dictionary is located, for example
    <i>/home/mjoc/.opendict/dictionaries/plain/mydict.dwa/file/mydict.dwa</i>.
    If you are going to distribute dictionary file <i>with OpenDict
    plain dictionary</i>, you want to write only the file name inside
    <i>path</i> tag.
</td>
</tr>
<tr>
    <td>md5</td>
    <td>This is an MD5 checksum value of the dictionary file
    (i.e. mydict.dwa). This MD5 code is used to determine changes 
    of the dictionary file. That means user is able to modify
    dictionary file and OpenDict will recreate <i>index table</i>
    before using that dictionary next time. If you want to get MD5
    checksum value of you dictionary file, execute the following
    command on your system (Linux, BSD) shell:
    <pre>
        $ md5sum myfile.dwa
	9c62810c32ca20fe018b79987789daef  myfile.dwa
    </pre>
    You can see the command and the output above. Just copy that
    32-chars checksum to your XML file. <i>This is quite unfriendly,
    so I will think about easer ways some day. Nevertheless, we are 
    developers today.</i>
    </td>
</tr>
<tr>
    <td>encoding</td>
    <td>Character encoding of the dictionary file. Examples: UTF-8,
    UTF-16, ISO-8859-1, ISO-8859-13, etc.</td>
</tr>
<tr>
    <td>description</td>
    <td>Description of your dictionary. You should not remove that 
    <i>&lt;![CDATA[ ... ]]&gt;</i> tag, because it lets you to write
    any text inside using even XML (HTML) tags like <i>1 + 1 &lt;
    1</i> or <i>&lt;myemail@abcdef.tdl&gt;</i>
</td>
</tr>
</table>

<p>
The last file you need is an <i>index</i> file. This file contains
index table of the dictionary describing in what position what letter
begins. Index makes search <i>a lot faster</i>.
</p>

<p>
I do not recommend writing index file using your fingers and the
keyboard :), so we will use OpenDict to make one for ourselves. Lets
now make a directory tree containing these files and directories:
</p>

<pre>
mydict.dwa/
    |
    +-- conf/
    |     |
    |     +-- config.xml
    |
    +-- data/
    |     
    |
    +-- file/
          |
          +-- mydict.dwa
</pre>

<p>
As you see, the root direcory is called <i>mydict.dwa</i>, the same as
dictionary file is called. This is <i>the rule</i>. Now move that
directory to the directory where all the plain dictionaries are
located. On my machine I would do:
</p>

<pre>
    mjoc@kumo:~/tmp/dict-factory$ <i>mv mydict.dwa/ ~/.opendict/dictionaries/plain/</i>
</pre>

<p>
After that you should start OpenDict and load the dictionary you've
just made. If everything goes well, you will be informed about
reindexing the dictionary. Press <i>OK</i> and, if everything goes
well, you will have index.xml file located at data/ directory. On my
machine the full path would be
<i>/home/mjoc/.opendict/dictionaries/plain/mydict.dwa/data/index.xml</i>. 
Now the directory tree look in a way like that:
</p>

<pre>
mydict.dwa/
    |
    +-- conf/
    |     |
    |     +-- config.xml
    |
    +-- data/
    |     |
    |     +-- index.xml
    |
    +-- file/
          |
          +-- mydict.dwa
</pre>

<p>
Now we have all the files needed. They are located at $DICTDIR. The
last step is to make a ZIP archive of that directory. Notice that that
ZIP file must contain the dictionary directory (i.e. mydict.dwa/)
itself, not only files inside it. To zip the directory, I would do: 
</p>

<pre>
    mjoc@kumo:~/tmp/dict-factory$ cd ~/.opendict/dictionaries/plain/
    mjoc@kumo:~/.opendict/dictionaries/plain$ zip -r MyDictionary-0.1.zip mydict.dwa/
    mjoc@kumo:~/.opendict/dictionaries/plain$ mv MyDictionary-0.1.zip ~
    mjoc@kumo:~/.opendict/dictionaries/plain$ cd
</pre>

<p>
Now file <i>MyDictionary-0.1.zip</i> can be found at my home direcory 
(/home/mjoc/). That's all. Now you have made a dictionary that can be
installed by everyone using OpenDict by selecting <i>Dictionaries ->
Install Dictionary from File</i> from the menu.
</p>

<h2>Simple Plugin Example</h2>
<p>
In this section I am going to give an example of simple OpenDict
plugin dictionary. Good example is a network dictionary plugin that
fetches translations from the web and processed them locally.
</p>

<p>
At first we have to create file called <i>plugin.xml</i>. The contents
of this file may be similar to this one:
</p>

<pre class="file-contents">
&lt;?xml version="1.0" encoding="UTF-8"?&gt;
&lt;plugin type="dictionary"&gt;
   &lt;name&gt;Network Test&lt;/name&gt;
   &lt;version&gt;0.1&lt;/version&gt;
   &lt;authors&gt;
      &lt;author name="Martynas Jocius" email="mjoc@akl.lt" /&gt;
   &lt;/authors&gt;
   &lt;module lang="Python"&gt;mydict.py&lt;/module&gt;
   &lt;encoding&gt;UTF-8&lt;/encoding&gt;
   &lt;uses-word-list&gt;False&lt;/uses-word-list&gt;
   &lt;opendict-version&gt;0.5.7&lt;/opendict-version&gt;
   &lt;python-version&gt;2.3&lt;/python-version&gt;
   &lt;platforms&gt;
      &lt;platform name="Linux" /&gt;
      &lt;platform name="BSD" /&gt;
      &lt;platform name="Windows" /&gt;
   &lt;/platforms&gt;
   &lt;description&gt;
   Example plugin dictioary. Does nothing.
   &lt;/description&gt;
&lt;/plugin&gt;
</pre>

<p>
As you may already noticed, some XML tags are the same as in the
previous <i>config.xml</i> example. There is the description of new
tags, like <i>encoding</i>, <i>module</i>, etc.
</p>

<table>
<tr>
    <th>XML tag</th>
    <th>Description</th>
</tr>
<tr>
    <td>module</td>
    <td>Main <a href="http://www.python.org">Python</a> module file name.</td>
</tr>
<tr>
    <td>encoding</td>
    <td>Character encoding of the result string that plugin module
    returns (examples include UTF-8, ISO-8859-15, etc)</td>
</tr>
<tr>
    <td>uses-word-list</td>
    <td>This should be set to <i>True</i> if dictionary needs word
    list to be shown, <i>False</i> otherwise. If dictionary uses word
    list, it must return a list of alternative words in addition with
    translation string (<i>see below for more information</i>)</td>
</tr>
<tr>
    <td>opendict-version</td>
    <td>The lowest OpenDict version plugin requires, for example
    0.5.1, 0.5.7, etc. This is important when plugin structure is
    different from one that old OpenDict versions provide (for example 0.5.1
    and 0.5.7 plugin structure differs a lot)</td>
</tr>
<tr>
    <td>python-version</td>
    <td>The lowest Python version that plugin requires.</td>
</tr>
<tr>
    <td>python-version</td>
    <td>The lowest Python version that plugin requires.</td>
</tr>
<tr>
    <td>platforms</td>
    <td>Parent node for <i>platform</i> tag.</td>
</tr>
<tr>
    <td>platform</td>
    <td>This tag with attribute <i>name</i> value X means that this
    plugin is tested and works on platform X. For example, Linux, BSD,
    MacOS. <i>Currently not used</i></td>
</tr>
</table>

<p>
Now we should write the main module in Python programming
language. Module file name is <i>mydict.py</i> as we called it in
<i>plugin.xml</i>. The following code might be correct OpenDict plugin:
</p>

<pre class="file-contents">
#!/usr/bin/env python

#
# MyDict 0.1
# Copyright (c) 2005 Matynas Jocius &lt;mjoc@akl.lt&gt;
#
# Simple plugin dictionary for OpenDict.
#
# This code is licensed under the GNU GPL v2.
#

"""
Simple OpenDict plugin module
"""

import sys
import httplib
import urllib


def init(libraryPath):
    """Return dictionary instance"""

    sys.path.insert(0, libraryPath)

    return MyDict()



class MyDict:
    """MyDict plugin class"""
    
    def __init__(self):
        """Initialize variables"""

        # This trick is needed to have accessible modules from
        # OpenDict library
        from lib import errortype, meta

        self.errorModule = errortype
        self.metaModule = meta

        
    def search(self, word):
        """Search and return HTML code."""

        print type(word), len(word)

        result = self.metaModule.SearchResult()

        try:
            self.conn = httplib.HTTPConnection("mjoc.sig.lt")
            self.conn.request("GET", "/index.html")
            response = self.conn.getresponse()
            data = response.read()

            trans = "&lt;html&gt;&lt;body&gt;&lt;h3&gt;"

            if word in data:
                trans += "Word &lt;i&gt;%s&lt;/i&gt; was found on " \
                         "http://mjoc.sig.lt/index.html" % word
            else:
                trans += "Word &lt;i&gt;%s&lt;/i&gt; was not found on " \
                         "http://mjoc.sig.lt/index.html" % word

            trans += "&lt;/h3&gt;&lt;/body&gt;&lt;/html&gt;"

            result.setTranslation(trans)

        except Exception, e:
            import traceback
            traceback.print_exc()

            result.setError(self.errorModule.INTERNAL_ERROR)

            return result

        return result

</pre>

<p>
<i>TODO: describe how to write the class and other details.</i>
</p>

<p>
To make an installable plugin file in ZIP archive format, move
<i>plugin.xml</i> and <i>mydict.py</i> into some directory
(i.e. <i>mydict-0.1</i>) and zip that directory. After performing
these actions your new OpenDict plugin will be ready to be installed.
</p>


<h2>Conclusion</h2>
<p>
<i>TODO: This document is very short and fuzzy. It must be improved in
the future.</i>
</p>

<div class="footer">Copyright &copy; 2005 
<a href="mailto:mjoc@akl.lt">Martynas Jocius</a>
 -- Last modified: 2005-09-06 22:57
</div>

</body>
</html>
opendict 0.6.3-4 / usr / share / doc / opendict / Plugin-HOWTO.html
