CICA 1996 August/CICA2_0896.bin/utils/jreader/jreadr25.doc

Text File | 1995-03-27 | 21KB | 446 lines

                           J R E A D E R

      Japanese Text Reader with Online Dictionary Search & Yomikata Lookup
      ====================================================================

                            Version 2.5


                           (Copyright)
                            J.W. Breen
                           January 1995


                             CONTENTS

                        1. INTRODUCTION

                        2. THIS DOCUMENT

                        3. INSTALLATION

                        4. ENVIRONMENT

                        5. OPERATION

                        6. DICTIONARY SEARCHING

                        7. VERB & ADJECTIVE MODIFICATION

                        8. YOMIKATA SEARCHING

                        9. KANJI INFORMATION

                       10. JREADER ON A PALMTOP

                       11. ADDITIONS TO PREVIOUS VERSION(S)

                       12. AUTHOR'S COMMENT



1. INTRODUCTION

This program provides a PC operating under MS-DOS with the capability to read
and  display a text file containing Japanese characters (kana & kanji),  with
the option of looking up the displayed words in a Japanese/English dictionary
file or in a kanji-to-kana yomikata file.

The Japanese characters in the text files can either be in the EUC,  New-JIS,
Old-JIS  or Shift-JIS codes.  Hankaku codes are supported for Shift-JIS,  but
not for EUC.  Codes which are not supported,  such as NEC-JIS or EUC-hankaku,
can be converted into one of the supported codes  using  a  utility  such  as
JCONV.

Although JREADER is  intended  to  help  non-Japanese  people  read  Japanese
language  text  files,  it can also be used by Japanese to read English text.
Its usefulness in this role is limited  by  the  dictionary,  which  is  more
oriented  to  the Japanese to English mode,  and the fact that the dictionary
search cannot cope with things like English's "strong" verbs (swim/swam/swum,
be/am/are, go/went, etc.).

JREADER is an extension of the  author's  JDIC  (Japanese/English  Dictionary
Display)  program,  which  has  been  designed  specifically  to operate on a
dictionary in the "EDICT" format originally used  by  the  MOKE  (Mark's  Own
Kanji  Editor)  Japanese  text  editor.  As  with  JDIC,  JREADER's operating
environment has been designed to be similar to MOKE's,  and it  can  use  the
same  environment  variables  and control file as MOKE.

The executable code and documentation of JREADER is hereby  released  to  the
public for general use.  It is covered by the author's copyright,  and may be
freely distributed with the proviso that it not be distributed as part  of  a
commercial system without the author's permission.  All usage of this program
is at the user's risk, and there is no warranty on its performance.

All  the  Japanese  displayed is in kana and kanji,  so if you cannot read at
least hiragana and katakana,  this program will not be much use for you.  The
author has NO intention of producing a version using romanized Japanese.


2. THIS DOCUMENT

JREADER  is  an  extension of JDIC,  and shares a similar operating method as
JDIC.  Consequently this document file only includes details of where JREADER
differs  from  JDIC.  Please  make  sure  you  have  and read the appropriate
JDICnn.doc file.


3. INSTALLATION

This program is distributed in a "zoo" archive (jdic25.zoo).  Both  JDIC  and
JREADER share a common operating environment.  Please follow the installation
details in JDIC25.DOC, which is in the "JDIC25.ZOO" file.

In addition, to get the full function from JREADER, you should have the files
WSKTOK.DAT  and  WSKTOK.IND.  These  are the kanji_to_kana file from MOKE and
its index file.  Without them the "y" (yomikata  lookup)  function  will  not
operate. If you are a MOKE user (Version 2.0 or later) you will have them.

The author has produced an expanded form of the WSKTOK.DAT file by adding  in
the  additional entries in EDICT,  plus further entries from the full WNN and
SKK dictionaries.  This is available in the WSKWNN.ZOO  file,  along  with  a
matching WSKTOK.IND index file.

(For the curious,  there is an explanation of these files in an  Appendix  to
JDIC25.DOC.)


4. ENVIRONMENT

JREADER  uses  the  same  environment variables and JDIC.RC/MOKE.RC fields as
JDIC (and MOKE).  These affect things like paths and colours.  See JDIC25.DOC
for details.

JREADER has one special (optional) entry in  the  JDIC.RC/MOKE.RC  file.  The
verb/adjective  deinflection  function  (see  below)  can  be disabled by the
following line in JDIC.RC/MOKE.RC:

jverb   off

The default is for this option to be enabled.


5. OPERATION

(a) LOADING

JREADER is simple to operate. The command-line invocation is:

         jreader <options> text-file(s)

The same -l,  -f,  -v,  -cDIR and -bnn  options  are  used  as  in  JDIC.  In
addition, JREADER uses:

 -sn (3 < n < 8) specifies that the text window is to use n/10 of the screen,
                   The default is n = 7.

 -ddictionary-file  specifies  the file that is to be used as the dictionary,
                   along with an index file  with  an  extension  of  ".jdx".
                   This latter file must be created using the JDXGEN utility.
                   The default is "edict" with "edict.jdx" as the index file,
                   or "jtoe.dct" and "jtoe.jdx", whichever is present.

 -Llogfile  specifies the name of a file to log possible new "edict" entries.
                   The default name is "jreader.log".

 -/search_string specifies a string for which a search is  invoked  when  the
                   file  is  read.  See  the  section  below on searching for
                   strings.  The same options are available as  in  a  string
                   entered from the keyboard, and as well a serach string can
                   be in (EUC coded) kanji or kana.

One or more file names can be provided. MS-DOS wildcards can be used also.

(b) READING FILES

The working screen of JREADER contains two windows.  The upper  displays  the
text being read,  the lower displays control information,  and the dictionary
and yomikata search results.

The lower window also displays a short "help" display when the window is  not
being  used for a regular display.  The help display can be turned off by the
"-v" command-line option and the "verbose off" line in the JDIC.RC  file.  It
can also be toggled on and off by the "o" command.

The first screenful of the text file is displayed when  the  program  starts.
From then on most operation is by single keystroke commands. They are:

<PgDn> reads the next screen of the file.  The  last  line  of  the  previous
screen is repeated as the first line of the next.

<PgUp> reads the previous screen  of  the  file.  The  backspacing  technique
involves backspacing the number of lines on the current screen,  so it should
usually result in the previous screen being displayed,  unless  there  are  a
number of "folded" lines.

<Ctrl-PgUp> restarts the file from the beginning.

<Ctrl-PgDn> skips to the end of the file, and displays the last 10 lines.

<Arrow>  The  four  arrow  keys  can  be  used to position the cursor under a
character which may be used as the start of a key for a dictionary search.  A
down-arrow while on the last line causes the display to scroll down one line,
and an up-arrow on the first line causes an upwards scroll.

<Enter> positions the cursor at the start of the next line.

<End> positions the cursor at the end of the current line.

<Home> positions the cursor at the start of the current line.

<Ctrl-Home> positions the cursor at the start of the screen.

<Ctrl-End> positions the cursor at the last line of the screen.

<Space> triggers a dictionary search using the string of characters beginning
with the one marked by the cursor. (See below.)

<a> the same dictionary search as <space>,  but if the search key begins with
one or more kanji characters, the search will match against any occurrence of
the  character(s)  among  kanji compounds in the dictionary,  instead of just
those at the start of compounds.

</> invokes a prompt for  a  string  of  characters,  the  file  is  searched
forwards, starting at the *second* line on the display, until a line is found
containing that string.  This scan is case sensitive.

There are two special options with this search:

    (i) if the entered string begins with a "\",  the remainder is treated as
    a hexadecimal coding of one or more kanji or kana. If the first character
    of the code is a "k",  the coding is treated as Kuten-encoded,  and if it
    is an "s",  it  is  treated  as  Shift-JIS.  For  example,  \k3214  is  a
    Kuten-encoded kanji and \s82a4 is Shift-JIS encoded kana,  while \3b7a is
    a JIS encoded kanji.

    (Note that it is possible to obtain an incorrect match on occasions  when
    using  this option,  particularly when searching for a single kanji.  The
    scan uses a simple "strstr" function,  which  is  not  sensitive  to  the
    boundaries of individual kanji or kana,  and thus may find a match on the
    the combination of the second byte of one character, and the first of the
    next.)

    (ii) if the first character is a "?", the *previous* search is repeated.

Note that an initial search string can be entered as a  command-line  option.
In all cases the search can be abandoned by pressing the Esc key.

<c> triggers a search similar to the "/" command,  except the  key  is  taken
from  the  screen,  starting  at  the cursor position.  You are asked for the
length of the key,  which may be up to 9  characters  long  (kana,  kanji  or
ASCII). You may repeat the search using the "/" command with the "?" option.

<l> logs the character string marked by the cursor  to  a  file  (default  is
"jreader.log").  The  logged  data is in "edict" format,  i.e.  `kanji [kana]
/english .../',  with the logged characters being  inserted  in  the  `kanji'
field.  You  will be prompted for the string length (up to 9 characters).  If
you respond with Enter,  and the cursor is on a Kanji,  all the kanji in  the
compound  will  be  logged.  You are also given the option of adding up to 50
characters of English to the logged entry.  (The main purpose of the  logging
function is to generate a file of Japanese words which are not  currently  in
the dictionary file.  This file can be edited later, the yomikata and English
translation  added  or  modified,  and  the  entries  included  in  the  full
dictionary.)

<y> invokes a scan of the "WSKTOK.DAT" file to find the yomikata of the kanji
compound starting with the character at the cursor.  [This option only  works
if  the  "WSKTOK.DAT"  and  "WSKTOK.IND" files are available,  i.e.  you need
either to be a MOKE (2.0 or later) user,  or you need to  have  obtained  the
files  separately  from  the  "WSKWNN.ZOO"  archive.]  The  longest  matching
sequence is displayed,  and you are given the option of  logging  this  entry
(kanji and kana) to the "JREADER.LOG" file, along with up to 50 characters of
English.  In  combination  with the <l> option above,  this option provides a
useful way of building up the dictionary file.

<n>  looks up and displays various details about the character at the cursor.
If the character is kana or ASCII,  the JIS or hexadecimal code is displayed.
For  kanji,  the  information  displayed  is the JIS code in hex,  the Nelson
number, the Halpern number, the Radical number (Bushu), the stroke count, the
on and kun readings, the English meaning(s) and a number of other information
fields.  This function requires the "KINFO.DAT"  file  to  be  present.  (See
JDIC25.DOC and KANJIDIC.DOC for further information.)

<s>  skips  ahead  in  the  text  file  to a line starting with "Article:" or
"Subject:".  This is to simplify reading a file containing  several  Japanese
news items.

<k>  skips  the cursor to the start of the next Kanji compound.  If Automatic
Lookup mode is active,  the dictionary is searched for  this  compound.  (See
below)

<w> skips the cursor to the start of the next of the next English word.  i.e.
the first slphabetic character after a non-alphabetic.

<f>  initiates the opening of either the next file on the command line,  or a
totally new file. You are prompted for more details.

<m> displays the next window of dictionary matches (if any).

<d>  displays  a status report of the files in use,  the position in the file
being read,  the buffer usage,  and the state of user configurable  switches.
Note  that  the  line position is not always accurate if there have been some
PgUps,  and particularly if the Ctrl-PgDn skip_to_EOF option has  been  used,
which case the line count is set to 9999.

<j> jump ahead a number of lines. There is a prompt asking for the number.

<v> toggles the verb deinflection function between enabled and disabled.

<b> toggles the automatic blanking of the lower window.  Normally the display
on the lower window is left there until the next search, log, etc. is carried
out.  Some users prefer not to have such displays present.  The  <b>  command
toggles on and off a function which  will  blank  the  lower  window  on  any
keystroke following a search.

<o>  toggles  the  production of the help display in the lower window.  (When
this option is in use,  it over-rides the operation of the automatic blanking
of the lower window.)

<F1> Displays a summary of the keyboard commands.

<F2> Toggles Automatic Lookup mode (See <k> above.)


6. DICTIONARY SEARCHING

The dictionary search is similar to the one used in JDIC, except that the key
is  taken  from  the  text being displayed,  rather than from keyboard input.
Thus the search can be on keys consisting of kanji compounds, as well as kana
and ascii.

Starting with the character marked by the cursor,  the longest match is found
and  displayed,  followed by the next longest,  and so on.  Usually the first
match is the one you want.  The dictionary display is identical  to  that  in
JDIC,  except that each line is preceded by the number of matched characters.
If there are more matched lines than fit in the window, pressing "m" displays
the next window-full.


7. VERB & ADJECTIVE MODIFICATION

When  a  dictionary  search  is initiated for text which consists of a single
kanji followed by two or more kana,  JREADER checks to see if it one  of  the
common verb or adjective conjugations or inflections, and if so, examines the
dictionary  using  the derived "plain" or "dictionary form" of the word.  The
user may then proceed with a normal search.  The inflection details used  are
in  the  file  "VCONJ",  which  may  be modified by the user.  Note that this
feature can be disabled by setting "jverb off" in the  JDIC.RC/MOKE.RC  file,
or  by  omitting the VCONJ file.  It can also be turned on or off dynamically
with the "v" command.

This function is not highly sophisticated,  and will not always  produce  the
right  result,  particularly when handling the more obscure grammatical forms
which use the "-masu stem" of verbs.  It is correct, however, over 95% of the
time,  and eliminates the problem of having the dictionary entry matching the
selected text only appearing about 20 or 30 lines down the display.


8. YOMIKATA SEARCHING

The  "WSKTOK.DAT"  file  contains  thousands  of  kanji  compounds with their
readings in kana.  It is sorted,  and indexed on the first byte of the  first
character  in  the "WSKTOK.IND" file.  JREADER seeks into and scans this file
for the longest matching sequence of characters.  Only one such  compound  is
displayed.  The  present author has expanded the original MOKE file,  and the
expanded version is available in the WSKWNN.ZOO archive.


9. KANJI INFORMATION

The  kanji  information  displayed  by  the  <n>  command  is  in  the   file
"KINFO.DAT".   KINFO.DAT  is  built  from  the  "KANJIDIC"  file.   See   the
KANJIDIC.DOC file for the full details on this information,  and the Appendix
to JDIC25.DOC for the structure of KINFO.DAT.


10. JREADER ON A PALMTOP

JREADER  can  be  used successfully on the tiny HP100LX Palmtop (and probably
other emerging PCs of this type.) See JDIC25.DOC for more  details  of  this.
The author operates JREADER on a Palmtop by:

(a) installing it in the Application Manager as a call to a batch file,  i.e.
the "Path" box contains:  "a:\kanji\jrbat.bat|350".  Note that the "|" is the
upside-down "!".

(b) creating a batch file (JRBAT.BAT) containing the following lines:

        @echo off
        input File Name(s) for JREADER? :
        jreader -f -s6 %ANS%

The  "input.com"  utility,  which  is  in  the JDICPALM.ZOO archive,  is a PD
program which enables a text string (e.g.  a  file  name)  to  be  passed  to
JREADER via the "ANS" environment variable.

11. ADDITIONS TO PREVIOUS VERSION(S)

V1.1 - Yomikata lookup, TAB expansion,  Shift-JIS reading,  PgUp for previous
screen.

V2.0  -  Larger  Help Screen,  double-Escape to exit,  "n" command to look up
Nelson,  etc.  information,  alternative font  files  and  dictionary  names,
multiple  input  files,  file  restart,  single-line scrolling,  text search,
paging of font and index files, capability of handling a dictionary up to 1.5
Mbytes.

V2.1 - Adds the ability to match a kanji with any occurrence  of  it  in  the
dictionary (the <a> function).

V2.2  -  Removes  the 1.5Mbyte restriction on dictionary size.  Tidies up the
kanji display (<n> option).

V2.3 - Added  the  verb/adjective  deinflector  facility,  the  <j>  and  <d>
options,  the  Kuten  field in the kanji display.  Enabled the display of the
last 4 JIS2 kanji when using the K16JIS2.FNT file.  Added the  JDIC.RC  file.
The   -cDIR  command-line  option.   Improved  the  search  speed,   and  the
line-folding in the dictionary and kanji display.

V2.4  -  compressed  the display,  including introducing user-selectable font
spacing,  and rearranging the lower window to enable better operation on  CGA
displays  (e.g.  the HP Palmtop.) Added the <b> blanking of the lower screen,
and the "Searching ..." message.  Added the handling of  half-width  kana  in
SJIS  files.  For  text searching,  added the <c> option,  the "\" setting of
JIS, SJIS and Kuten,  the command-line option,  and the "?" repeat.  Expanded
the "d" display, and fixed the erroneous line counts.  Added the help display
in the lower window.

V2.5 - the EDICT Extension file facility <e>.


12. AUTHOR'S COMMENT

JREADER is to me a natural extension of JDIC,  and further exploits the  fast
dictionary  scanning technique used therein.  It also has been written with a
need in mind.  I had been using Mark Edwards' excellent VIEW and MOKE to read
fj.* news (using the SNUZ news reader.) I was frustrated by the  slowness  of
the  English  lookup  in  MOKE (a sequential read of the entire file) and its
refusal to add a compound to the dictionary if it was not in  the  kanji/kana
henkan  file.  Also  both  MOKE  and  VIEW require precise delineation of the
search string using several keystrokes.  This  can  result  in  several  slow
attempts  to  find  meanings for portions of a kanji compound.  What I wanted
was something friendlier and  faster  in  a  reading  environment,  with  the
capability of providing updates to my EDICT dictionary.

From this grew JREADER,  and it has turned out to be a very powerful Japanese
text  reader,  with  many  devoted  users  around the world.  (JREADER's code
actually formed the basis of the code for XJDIC,  the Unix X11 port of  JDIC,
because  XJDIC  provides virtually all of JREADER's functionality through the
kterm cut_and_paste facility.) To my delight,  the compilers  of  the  Walnut
Creek "East Asian Text Processing" CDROM  sought  my  permission  to  include
JREADER as the default Japanese text reader.

As  with  the JDIC program,  I am grateful to the many beta-testers,  and the
people who have suggested operational improvements, many of which I have been
able to incorporate.

As ever, comments and suggestions are welcome.

Jim Breen ([e-mail])
Department of Robotics & Digital Technology
Monash University
Melbourne, Australia
Nov 1991 - March 1995