browse Concise Edict2 - kanji expressions only (Part 1)

[rshiplett]

Here is my blog post with a screenshot: http://kanjirecog.blogspot.ca/2012/09/concise-edict2.html

This first "chapter" covers about 350 kanji with about 13,500 entries. The idea is that you can browse the dictionary.

It is indexed with this difference: the JOYO kanji are distinct in the left-hand pane.

What I need now are suggestions from anyone learning Japanese as to how to improve this "browse" page. I already have pages in which you can search the whole dictionary or these 9"chapters".

Why in parts? Because it takes time and memory to load. I will build an "AJAX" style version using the async Curl worker framework after the Android and iOS versions are available.

Needed: someone using a Mac Pro o give feedback when I release the local "desktop" off-line version.

Next major effort: adding mnemonics and user "notes" (persistent annotations.)

[mmmason8967]

It seems to work very well. The only request I'd make is for the kanji in the list in the sidebar to be in a bigger font, as some of the more complex kanji degrade quite badly。For example,　票, ひょう, although to be fair it doesn't look much better at the font size used on this forum (but it is better).

I'm using Opera 12.02 on Windows 7.

マイケル

[rshiplett]

Thanks - I have made a request at communities.curl.com as there is no API to access the font size in that left-side pane (at least not in the public documentation at www.curl.com). The alternative is to build my own tree+scroller pane from scratch (which I may yet do.)

Thanks for saying which browser you run ... might I ask which font you are seeing ? I can add one you like to the style list. At the moment the applet is looking for HanaMinA or Arial Unicode MS.

thanks again,

Robert

[mmmason8967]

It's very probably Arial Unicode MS because I have that font installed but I don't have HanaMinA.　And the Roman text looks exactly like Arial.

I can't tell for sure because the tools I normally rely on (Dragonfly in Opera, Firebug in Firefox) just show me the styles for the underlying HTML page, which is the one you see if you don't have CURL installed. They won't show me the styles used in the CURL applet.

Most of the kanji are just about legible--actually, they're probably fine if you're fairly skilled in Kanji (which I'm not). It's the ones with lots of horizontal lines that suffer: if you need to show 8 horizontal lines using 11 pixels or less, it's going to look ugly...

マイケル

[rshiplett]

I have some alternate views of edict2 at http://kanji.aule-browser.com such as http://www.aule-browser.com/kanji/edict ... -curl.html

Some issues remain with the search in that applet (trade-offs in my effort to keep it small and fast) but I hope the font is better.

Would you have a moment to check out that one ?

Btw, there is a sub-domains quirk with my Curl site license such that www must be used up front in the URL for the Curl applets and not just kanji.

thanks

Robert

[mmmason8967]

The kanji on the edict2-dbl-kanji-grid-curl page are pretty clear apart from some odd ones, all of which seem to be the ones with lots of horizontal lines. I guess there isn't going to be a good way to get around that.

I tried one of the 'ugly' kanji, 擤, in Notepad, increasing the font size until it got clear. It became reasonable at 18 points and perfectly clear at 26 points. Interestingly, 24 points is pretty ugly (much worse than 20 point or 22 point), so I guess TrueType hinting is causing some problems as well.

マイケル

[rshiplett]

Hi

Part 2 has arrived at http://www.aule-browser.com/kanji/#browse

The next 7 MUST be generated by a script, as the work is exhausting (14-16 hours edits of 14.000+ lines per file !)

I will do the script in Object Icon from http://code.google.com/p/objecticon or in Unicon,which ever runs best (Icon does backtracking and is string-oriented expression-based language.) Rebol3 would have been good for Unicode string parsing, but it looks to be toast - abandoned by Carl Sassenrath ? Getting the script right might be 6 - 8 hrs work ... after I get a good build of one of the other Icon on linux. And I have a PC to assemble from parts ... finally going Intel i7 Ivy Bridge.

( I refuse to work in Perl and try not to rely on Regex )

[rshiplett]

Hi

I have added two 'visual' Contents sections to Part 1 and Part 2 at http://www.aule-browser.com/kanji/index.html#browse - they allow a more effective CTRL-f search as there are links to position yourself in the dictionary relative to the character for the entries that you want - and for now, are a quick check that a given kanji has no entry in that section (although it may occur in other expressions as other than the first character.)

If you have a moment - do they seem helpful to you ? They are WORK to create until I have a script to parse the data files.

thanks

Robert

[mmmason8967]

I've had a very quick play around and yes, the contents section does look useful. One of the interesting things for me is that the kanji are clearly in some kind of order but I haven't a clue what it is. Is there a standard way of sorting kanji, and if so, how does it work? I kind of expected there'd be a sort order that would start with simple kanji and then get more and more complicated, but that obviously isn't the case.

I can see why you want to write a script -- these pages are huge! I honestly cannot imagine how you've managed to edit the first two by hand…

マイケル

[natsukoy9313]

That's true, マイケルsan...I also gave a look at the site and it's really huge!!
Robert-san, you're doing an amazing job there!!

Natsuko(奈津子)
Team JapanesePod101.com

[rshiplett]

I kind of expected there'd be a sort order that would start with simple kanji

At http://www.aule-browser.com/kanji/kanjidic2-joyo.html I think that you will see the kanjidic2 entry order, UCS and Henshall order.

If I convert each kanji segment to an object (as they are in my "grid" version of an edict2 viewer, then I can control the order with any given view.

Some of the pages have "school grade" grouping.

[rshiplett]

I kind of expected there'd be a sort order that would start with simple kanji

In file 3, these are the first Joyo kanji in their order in the file:

捜 掃 挿 操 早 曹 争 痩 相 窓
総 草 荘 葬 藻 装 走 送 遭 霜
騒 像 臓 蔵 贈 造 促 側 則 即


There is SOME order ... so that should reflect in the left-pane disebar where they stand-out a bit with their short list of Eng equivalents.

R

[natsukoy9313]

Thank you very much for explanation, Robert-san

Natsuko(奈津子),
Team JapanesePod101.com

[rshiplett]

Thank you very much for explanation, Robert-san

Natsuko(奈津子),
Team JapanesePod101.com

btw

All 9 sections now have Contents so I can build a Search pane before I finish with the indexing ...
http://www.aule-browser.com/kanji/index.html#browse

The Contents web page should be updated in about an hour (after breakfast ! )

PS
I ended up using a Curl script for the contents, as I can stub-and-test so fast in that IDE

[rshiplett]

Would you have a moment to look at the revised format of Part ! before I proceed with indexes for Parts 3 - 9 ?

http://www.aule-browser.com/kanji/index.html#browse

thanks

Robert