home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Collection of Tools & Utilities
/
Collection of Tools and Utilities.iso
/
tex
/
jdic23.zip
/
README.J23
< prev
next >
Wrap
Text File
|
1993-07-01
|
4KB
|
93 lines
CHANGES IN JDIC & JREADER V2.3
- tidier (I think) line folding in the dictionary displays. They now fold on
word breaks.
- Expansion of the kanji display to show the Kuten and Halpern codes. This
has meant using a second line in the display.
- a change to the binary search of the dictionary which noticeably speeds up
JDIC and marginally speeds up JREADER. The old method used a binary search
until it found a key match, and then scanned sequentially backwards up the
index/dictionary until it got to the start of the matches. If a very short
key, such as "t" was given, it was *very* slow. The programs now engage a
second binary search to find the top of the list.
- the introduction of their own "jdic.rc" file and JDICRC environment
variable. moke.rc and MOKERC will continue to work, but the purpose of these
are to enable jdic/jreader to be seen to be quite independant of MOKE.
- the option to specify the path of the moke/jdic.rc file on the command
line. This is to assist people running JDIC/JREADER from a file server.
- enabling the display of the last 4 JIS 0208-1983 kanji.
JDIC
The main changes to JDIC are:
- the function to find kanji by Kuten or Halpern codes. The Kuten is easy,
as it is another representation of the JIS code. I have used Alt-F9 to select
this, as it is indeed an alternative to F9. For the Halpern codes, it
necessitated incorporating the Halpern codes in the main KINFO.DAT structure,
thus expanding it, and creating a new KINFOHAL.IDX index. Note that this
means that V2.3 needs the new KINF*.* files. They will not work with the
V2.2 files. All the various bushu/stroke/etc screens in JDIC still show the
Nelson numbers. I baulked at redoing all that. Alt-F8 invokes the Halpern
search, as it is an alternative to the Nelson (F8) search.
- The function to look up a kanji from the display. Again, several people
have asked for this. Alt-F10 brings up a little marker on the left of the
display which can be positioned on the desired line. Press enter to select
the kanji on that line. If there is more than one, the kanji are displayed at
the bottom of the screen for further selection.
- the filter function. Several people have asked for a facility to obscure
the display of the place and person names. Others have suggested breaking
edict up for this reason. My solution is to leave the dictionary intact, and
to build in a user configurable/selectable filter system. There are 3 types
of filters:
o show only those entries which contain one of the specified strings.
o do not show this entry if it contains one of the strings
o as above, but only if it has a single English field
[the reason for the third category is to avoid supression if the word
has a real meaning, eg [shimizu] /spring water/Shimizu (pn)/]
The filter details are defined in the moke.rc file. Here is an extract from
my moke.rc:
jdicfilt 0 2 on "Exclude proper names" (pl, (pn pl) pn)
jdicfilt 1 1 off "Exclude all place names" (pl, pl)
jdicfilt 2 1 off "Exclude all person names" (pn pn)
jdicfilt 3 0 off "Show only person names" (pn pn)
jdicfilt 4 0 off "Show only place names" (pl, pl)
jdicfilt 5 0 off "Show only place & person names" (pl, (pn pl) pn)
JREADER
The main change is the verb/adjective de-inflector function. (This came from
a still-born "translation aid" program (JTRAID), which I worked on about 6
months ago.)
This function will attempt to work out the "dictionary form" of verbs and
adjectives before proceeding with the dictionary scan. It can be turned off
by a jdic.rc entry if it bugs you. It also can be toggled on or off with the
"v" command.
Other options allow you to skip forward a specified number of lines, and
display the current line number.
Jim Breen
July 1993