Translation Table
Omnidex translates 8-bit extended ASCII characters from position 160
to 255 in the ASCII collating sequence. A translation table allows customization
of these translations and supports any of the 8-bit character sets like
Greek8 or Kana8.
Translations occur when the indexes are built and during all subsequent
searches on that index. For instance, when the indexes are being built,
if the translation table defines the character é to be translated
to e, the word entrée will be stored as entree, in the Omnidex
indexes. Furthermore, a search for both entrée and entree against
those indexes, will qualify any records that contained the word entrée.
The T or Translate option is a database-level indexing option that applies
a custom translation table to all Omnidex indexes in the installation
that do not have the ;NT (No Translate) option installed on them.
The ;NT (No Translate) prevents indexed values from being upshifted. When
excluding words for No Translate indexes, it will be necessary to include
the different case sensitive spellings of the excluded word. For example,
if “CO” is an excluded word, the keyword “Co”
is still indexed for any No Translate indexes unless "Co" is
included in the excluded words list.
Because the translation table affects the contents of the Omnidex indexes,
the translation table should be loaded before building the indexes. Otherwise,
the indexes will have to be rebuilt after the translation table is loaded.
DISC recommends
that the following special characters NOT be redefined:
+ (plus)
, (comma)
- (minus)
: (colon)
* (asterisk)
? (question mark)
$ (dollar sign)
& (ampersand)
^ (carat)
< (less than)
> (greater than)
= (equal)
These characters are used internally by Omnidex for Boolean operators,
ranges, wild cards, and so on.
Create a Translation Table
To create a translation table, create an ASCII text file with 255 lines,
one for each character in the ASCII collating sequence.
- For each character to be translated to a different ASCII counterpart,
place the desired ASCII character in column 1 of the line that corresponds
to the 8-bit character's position in the collating sequence.
- To leave a character untranslated, enter a backslash ( \ ). A backslash
( \ ) in a line means the corresponding character is not translated
at all, even if it is an 8-bit character or in lower case. Therefore,
a character cannot be redefined to be blank or a backslash.
- To use the default translation for that character, leave the line
blank.
- The first 32 ASCII characters, which are control codes, cannot be
redefined. They are translated to blanks regardless of what is specified
in lines 1 through 32 of the translation table.
- Although some characters may represent the combination of two letters,
note that they can only be translated to one ASCII character. Thus “ß”
(the German equivalent of “ss” or “sz”) can
only be “s” or “z” but not “ss”
or “sz”.
- The translation of these characters cannot be overridden.
For example, the Greek8 letter “F”, which Omnidex translates
as a Roman “O” (upper case) by default, occupies position
214 in the ASCII collating sequence. To define “F” as a Roman
“F”, type the ASCII character F in column 1 of line 214. To
translate the lower case “phi”( f )to “f”, type
an f on the appropriate line in collating sequence (here, line 230). Lines
211 through 230 of the translation table might look like this.
.
.
211
212
213
214 F
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230 f
.
.
Loading a Translation Table
To load the translation table, enter T at the Cmd: prompt. DBINSTAL will
prompt for the file that contains the translation table. The sample listing
below shows how to use the TRANSLATE command to load a translation table
named “ttable”:
Cmd: T
Translation table file: ttable
Table updated
To restore the translation table to defaults, create a blank file of
255 lines, and load it using the TRANSLATE command as shown above, or
reinstall the Omnidex indexes without specifying a translation table.
The translation table must be reloaded for an Omnidex environment whenever
the Omnidex indexes are reinstalled on the tables in the environment catalog.
Also, the indexes must be reloaded after loading a new translation table.
Top
|