New dictionary format of firmware 2.14

[gouni]

Tshering

From this moment I'm in the fourth dimension.

I'll slowly slowly

[tshering]

Originally Posted by gouni

From this moment I'm in the fourth dimension.

Does this mean you are now enlightened? Confused? Happy? Motivated?

Don't forget to make back ups of your files after every step. In case something goes wrong (and this will certainly happen at one point or another) you can always return to your last back up, rather than start the whole work anew.

Which text editor do you use? For replacing some text by some other text almost any editor will do (for instance the simple "Editor" of Microsoft); for replacing line breaks by something you need another tool. If you have MS word you can use it. I guess notepad+ can also do it.

[gouni]

I am all at once. I have never done that.

I use my windows notepad.

I'll install notepad +

I'll go little by little, I will put a year but I'll.

My cat tells me I'll go faster by learning the dictionary upside down.

[ShellShock]

Originally Posted by tshering

The description of which words go into the 11.html file might easily be misunderstood. Actually, a word/expression goes there if the first or/and second character is not a letter. Examples for the second character being not a letter are <a name="I'd">, <a name="o'clock">, <a name="T4 cell">.

Maybe it would be good to shift the whole explanation about CR and LF to point 3. I do not think that CRs in the html prevent correct processing or displaying, and they have no connection with the index file. On a side note, in the original dicthtml, CR and LF are both omitted, presumably because of performance concerns.

Thanks very much for this; especially the point about first or second characters not being a letter means using 11.html, which I did not realise. I need to go back and update my dictionaries to take that into account.

[ShellShock]

Actually, in my dictionary "o'clock" is defined in o'.html, and the lookup works fine. So I am not convinced about this second character business and having to use 11.html. Maybe it works either way?

[Koboyashi]

Originally Posted by gouni

... I use my windows notepad. I'll install notepad + ...

Notepad++ is a fine tool. Especially to edit files from a linux device under windows easily.
Here you go: http://notepad-plus-plus.org/

[tshering]

Originally Posted by ShellShock

Actually, in my dictionary "o'clock" is defined in o'.html, and the lookup works fine. So I am not convinced about this second character business and having to use 11.html. Maybe it works either way?

Could well be. I only reported what I found in the 11.htmls of the Kobo dictionaries. Maybe only those words with the first or second letter being forbidden in filenames must go to the 11.html, and some other cases are optional.

[gouni]

Kobojashi

[gouni]

Originally Posted by tshering

gouni, You can try this:

To get minimal formatting replace line breaks by <br/>

I do not know to do this

I use your notepad +++ which is wonderful.

At the end of the line I have a LF in black, it is the function to replace it with <br/> ?

Thanks for your help

[AlPe]

I guess tshering's suggestion is to use:

<p>First phrase.<br/>Second phrase.</p>

instead of

<p>First phrase.</p><p>Second phrase.</p>

to gain some typographical space on the device screen.

[gouni]

I don't understand

I have this on the screen with notepad +++

http://www.mediafire.com/?5g5y2pam055olvg

[Koboyashi]

This is from your example pic:



The black'n white "LF" comes from Notepad++, its a config thingy to show you the normally invisible line feeds in
the editor window. The file content itself is not affected. For a Linux file it is only "LF", for Windows it's "CRLF" and
for Mac it's "CR".

Notepad++ recognizes the line feeds of a file automatically and will keep them while saving. Please do never use
the small Notepad delivered with Windows to edit Linux text files.

You may toggle the visibility of the line feeds as follows:
French: View -> Show Symbol -> Afficher les symboles de fin de ligne
English: View -> Show Symbol -> Show End of Line

[gouni]

Thank Koboyashi.


I chose the end of line feed LF as shown above.



I think I'll now move to the next step

Create a dictionary with a few words and see if it works.


It is thrilling.

[tshering]

Congratulations! Your dicitonary file looks good. Does your sample dictionary already work?
You will see that the whole definition is in one block without any formatting and difficult to read. I thought it might be nice to add a minimal formatting. One easy way would be by replacing all LF by <b/>. In the replace box of notepad++ select "Search mode: extended." You can search LF by "\n".

Find what: \n
Replace with: <b/>

This will give you, for instance

Synonymes :<b/> - aveuli<b/> - mou<b/> - ramolli<b/> - se relâcher<b/>

But it will give you also </w><b/>. This however we do not want, therefore we replace </w><b/> by </w>.

The drawback of this formatting is that it uses a lot of space on the screen, and you will probably have to go through several pages in order to read the definition of an entry. Therefore, it maybe better to replace the LF by something else that gives you a hint about the the structure of the information, maybe "||" or any good visible symbol.

Another thing: When producing the index file choose for encoding "encode in UTF-8 without BOM". If you choose "encode in UTF-8" the first line will not be taken into account by the dictionary engine.

[gouni]

I have just read to you. tshering

I have a long time before you understand how to run a command by windows for marisa.

I'll redo with utf8-bom I had a - 1 to my first and last index entry.

At the moment I am testing marisa it build and marisa lookup.

My eyes are like those of my cats at night.

I learn while on the job, I have the head that heats.


I start again with the right encoding


I had on my five-word index with marisa lookup

-1 1 2 3 -1

with utf8-bom i have

1 2 0 3 -1

I have a bug I can do