Your company here — click to reach over 10,000 unique daily visitors

dictfmt - Man Page

formats a DICT protocol dictionary database


dictfmt  -c5|-t|-e|-f|-h|-j|-p [options]  basename
dictfmt  -i|-I [options]


dictfmt takes a file, FILE, on stdin, and creates a dictionary database named basename.dict, that conforms to the DICT protocol.  It also creates an index file named basename.index. By default, the index is sorted according to the C locale, and only alphanumeric characters and spaces are used in sorting, however this may be changed with the --locale and --allchars options.  ( basename is commonly chosen to correspond to the basename of FILE , but this is not mandatory.)  

Unless the database is extremely small, it is highly recommended that basename.dict be compressed with /usr/bin/dictzip to create basename.dict.dz. (dictzip is included in the dictd source package.)    

FILE may be in any of the several formats described by  the format options -c5, -t, -e, -f, -h, -j, -p, -i or -I. Exactly one of  these options must be given.

dictfmt prepends several headers are to the .dict file.  The 00-database-url header gives the value of the -u option as the URL of the site from which the original database was obtained.  The 00-database-short header gives the value of the -s option as the short name of the dictionary.  (This "short name" is the identifying name given by the "dict- D" option.)  If the -u and/or -s options are omitted, these values will be shown as "unknown", which is undesirable for a publicly distributed database.

The date of conversion (formatting) is given in the 00-database-info header.  All text in the input file prior to the first headword (as defined by the appropriate formatting option) is appended to this header.  All text in the input file following a headword, up to the next headword, is copied unchanged to the .dict file.

Formatting Options


FILE is formatted with headwords preceded by 5 or more underscore characters (_) and a blank line.  All text until the next headword is considered the definition.  Any leading `@' characters are stripped out, but the file is otherwise unchanged. This option was written to format the CIA WORLD FACTBOOK 1995.


-c5, --without-info and --without-headword options are implied. Use this option, if an input database comes from dictunformat utility.


FILE is in html format, with the headword tagged as bold.  (<B>headword - </B>)
This option was written to format EASTON'S 1897 BIBLE DICTIONARY.  A typical entry from Easton is:

<A NAME="T0000005">
<B>Abagtha - </B>
one of the seven eunuchs in Ahasuerus's court (Esther 1:10; 2:21).

This is converted to:
  one of the seven eunuchs in Ahasuerus's court (Esther 1:10; 2:21).

The heading "<A NAME="T0000005"> is omitted, and the headword `Abagtha' is indexed.

NOTE: This option should be used with caution.  It removes several html tags (enough to format Easton properly), but not all.  The Makefile that was originally written to format dict-easton uses sed scripts to modify certain cross reference tags.  It may be necessary to pipe the input file through a sed script, or hack the source of dictfmt in order to properly format other html databases.


FILE is formatted with the headwords starting in column 0, with the definition indented at least one space (or tab character) on subsequent lines.   The third line starting in column 0 is taken as the first headword , and the first two lines starting in column 0 are treated as part of the 00-database-info header.  This option was written to format the F.O.L.D.O.C.


FILE is formatted with the headwords starting in column 0, followed by a comma, with the definition continuing on the same line.  All text before the first single character line is included in 00-database-info header, and lines with only one character are omitted from the .dict file.   The first headword is on the line following the first single character line. The headword is indexed; the text of the file is not changed.  This option was written to format HITCHCOCK'S BIBLE NAMES DICTIONARY.


FILE is formatted with headwords starting in col 0, enclosed in colons, followed by the definition. The colons surrounding the headword are removed, and the headword is indexed.  Lines beginning with '*', '=', or '-' are also removed. All text before the first headword is included in the headers.  This option was written to format the JARGON FILE.  
NOTE: Some recent versions of the JARGON FILE had three blanks inserted before the first colon at each headword.  These must be removed before processing with dictfmt.  (sed scripts have been used for this purpose. ed, awk, or perl scripts are also possible.)


FILE is formatted with `%h' in column 0, followed by a blank, followed by the headword, optionally followed by a line containing `%d' in column 0.  The definition starts on the following line.  The first line beginning ´%h´ and any lines beginning '%d' are stripped from the .dict file, and '%h ' is stripped from in front of the headword.  All text before the first headword is included in the headers. The second line beginning '%h' is taken as the first headword. This option was written to format Jay Kominek's elements database.  

-i -I

These two options are different from all other formatting options. They are intended to resort (according to dictd requirement) an .index file given on stdin. That is .dict file is not generated at all. Only resorting is made. Three- or four-column .index like input is expected. -i expects decimal offset and length, while -I expects them in base64 format.


-u url

Specifies the URL of the site from which the raw database was obtained. If this option is specified, 00-database-url headword and appropriate definition will be ignored.

-s name

Specifies the name and, optionally, the version and date, of the database.  (If this contains spaces, it must be quoted.) If this option is specified, 00-database-short headword and appropriate definition will be ignored.


display license and copyright information


display version information


output debugging information


display a help message

--locale locale

Specifies the locale used for sorting.  If no locale is specified, the "C" locale is used. For using UTF-8 mode, --utf8 is needed.


generates database in 8-bit mode, see --locale option also.
Note: This option is deprecated. Use it for creating 8-bit (non-UTF8) dictionaries only. In order to create UTF-8 dictionary, use --utf8 option instead.


If specified, UTF-8 database is created.


Specifies that all characters should be used for the search, by default only alphabetic, numeric characters and spaces are put to .index file and therefore are used in search. Creates the special entry 00-database-allchars.


makes the search case sensitive. Creates the special entry 00-database-case-sensitive.

--headword-separator sep

sets the headword separator, which allows several words to have the same definition.  For example, if ´--headword-separator %%%' is given, and the input file contains ´autumn%%%fall', both 'autumn' and 'fall' will be indexed as  headwords, with the same definition.

--index-data-separator sep

sets the index/data separator, which allows to set the first and fourth columns of .index file independently. That is the first column can be treated as an index column (where the MATCH command searches) and the fourth column as a result column (where the MATCH gets things to be returned), and they (1-st and 4-th columns) are completely independant of each other. The default value for this separator is ASCII symbol " \034".


multiple headwords will be written on separate lines in the .dict file.  For use with '--headword-separator.


When --utf-8 is specified headwords are lowercased and non-alphanumeric characters are removed from it before saving to .index file in order to simplify the search. When --index-keep-orig option is used fourth column is created (if necessary) in .index file, and contains an original headword which is returned by MATCH command. This option may be useful to prevent converting " AT&T" to " ATT" or to keep proper nouns with uppercased first letter.


headwords will not be included in .dict file


header will not be copied to DB info entry


URL will not be copied to DB info entry


time of creation will not be copied to DB info entry


By default dictfmt creates a special entry 00-database-dictfmt-X.Y.Z that contains (in .dict file) dictfmt version in format dictfmt-X.Y.Z. This option suppresses this.


DB info entry will not be created. This may be useful if 00-database-info headword is expected from stdin (dictunformat outputs it).

--columns columns

By default dictfmt wraps strings read from stdin to 72 columns. This option changes this default. If it is set to zero or negative value, wrapping is off.

--default-strategy strategy

Sets the default search strategy for the database. It will be used instead of strategy '.'. Special entry 00-database-default-strategy is created for this purpose. This option may be useful, for example, for dictionaries containing mainly phrases but the single words. In any case, use this option if you are absolutely sure what you are doing.

--mime-header mime_header

When client sends OPTION MIME command to the dictd , definitions found in this database are prepended by the specified MIME header. Creates the special entry 00-database-mime-header.


dictfmt was written by Rik Faith (faith@cs.unc.edu) as part of the dict-misc package. dictfmt is distributed under the terms of the GNU General Public License.  If you need to distribute under other terms, write to the author.


This manual page was written by Robert D. Hilliard <hilliard@debian.org> .  

See Also

dict(1), dictd(8), dictzip(1), dictunformat(1), http://www.dict.org, RFC 2229

Referenced By

dictd(8), dictfmt_index2suffix(1), dictfmt_index2word(1), dictunformat(1).

25 December 2000