NAME

hxcite-mkbib - expand references and create bibliography

SYNOPSIS

hxcite-mkbib [ -b base ] [ -p pattern ] [ -s separator ] bibfile [ file ]

DESCRIPTION

The hxcite-mkbib commands copies file to standard output, looking for strings of the form "[[label]]" and for a template for a bibliography. The label may not include white space and the double pair of square brackets must enclose the label without any spaces in between. If hxcite-mkbib finds the label in the bibfile, the string is replaced by the pattern. The pattern can include certain variables. If the label is not found in bibfile, it is left unchanged.

The default pattern replaces the string with a hyperlink, but if the -p option is used, the replacement can be any pattern. The input doesn't even have to be HTML.

The file consists of three parts:

preamble: The preamble is the part up to the first occurrence of %{. The preamble is copied to the output once (with bracketed labels ("[[label]]") expanded). The character % is treated specially. To create a single % in the output, there must be two in the preamble (%%). All other occurrences of % followed by another letter are not copied, but are collected into a string called the "sort order." and used to sort the entries, as explained below.
template: The template starts with %{L: and ends with a matching %}. The text in between is copied as often as there are bibliographic entries in bibfile that correspond to bracketed labels in file. Variables in the template are replaced by the corresponding field in the bibliographic entry: all occurrences of %x will be replaced by the field %x of the entry. Parts of the text may be enclosed in %{x: and %}. This means that the text in between should only be output if the current entry has a field x. Text that is enclosed in %{!x: and %} will only be output if the entry does not have a field x. Both kinds of conditional sections may also be nested.
postamble: The text after the %} is copied unchanged to the output, after all bibliographic entries have been processed.

By default bibliographic entries are copied to the output in the order of the labels in file, except that labels that occur more than once are only used once. If the preamble contains occurrences of %x (where x is neither "%" nor "{") then these together determine the sort order. E.g., if the preamble contains %A%D then the entries will be sorted first on field A (author) and then on field D (date).

Here is an example of a file that creates a bibliography in HTML format:

<html>
<title>Bibliography</title>
... text with [[references]] here...
<!--%A%D sorted on author, then date -->
<dl>
%{L:
<dt id="%L">%{A:A%}%{!A:%{E:E%}%{!E:%{Q:Q%}%{!Q:-%}%}%}</dt>
<dd>%{B:"%T"
  in: %{E:%E (eds)
  %}<cite>%B.</cite>%{V: %V.%}
  %}%{J:"%T"
  in: %{E:%E (eds)
  %}<cite>%J.</cite>%{V: %V.%}%{N: %N.%}%{P: pp. %P.%}
  %}%{!B:%{!J:<cite>%T.</cite>
  %}%}%{I:%I.
  %}%{D:%D.
  %}%{C:%C.
  %}%{R:%R.
  %}%{S:%S.
  %}%{O:%O
  %}%{U:<a href="%U">%U</a>
  %}</dd>
%}
</dl>
</html>

This template starts with four lines of preamble, including the sort string %A%D on line 3. The sort string itself will not be output, but the rest of the comment will.

From the line %{L: to the line %} is the template. E.g., the line that starts with <dt id=... contains a complex conditional text that prints the authors (%A) if there are any, otherwise the editors (%E) if there are any, otherwise the institution that is the author (%Q), if any, and a dash otherwise. Note how the parts are nested, Most of the text is inside %{!A:...%}, meaning that that part will only be effective if there is no author field (%A).

The final two lines are the postamble and will simply be copied unchanged.

A bibliographic entry that looks like this in bibfile:

%L Java
%A Gosling, James
%A Joy, Bill
%A Steele, Guy
%T The Java language specification
%D 1998
%I Addison-Wesley
%U http://java.sun.com/docs/books/jls/index.html

will be printed by the template above as:

<dt id="Java">Gosling, James; Joy, Bill; Steele, Guy</dt>
<dd><cite>The Java language specification.</cite>
  Addison-Wesley.
  1998.
  <a href="http://java.sun.com/docs/books/jls/index.html">http://java.sun.com/docs/books/jls/index.html</a>
  </dd>

OPTIONS

The following options are supported:

-p pattern

Specifies the pattern by which the string [[label]] is replaced. The pattern may include the variables %b (which will be replaced by the value of the -b option) and %L (which will be replaced by the label). The default pattern is

<a href="%b#%L" rel="biblioentry">[%L]</a>

-b base

Sets the value for the %b variable in the pattern. Typically this is set to a relative or absolute URL. By default this value is an empty string.

-s separator

If there are multiple authors or editors in an entry, their names will be listed with a separator in between. By default the separator is "; " (i.e., a semicolon and a space). With this option the separator can be changed.

OPERANDS

The following operands are supported:

bibfile: The name of a bibliographic database must be given. It must be a file in refer(1) format and every entry must have at least a %L field, which is compared to the bracketed labels. (Entries without such a field will be ignored.)
file: The name of the input file is optional. If absent, hxmkbib(1) will read the template from stdin.

DIAGNOSTICS

The following exit values are returned:

0: Successful completion.
> 0: An error occurred. Usually this is because a file could not be opened or because the %{ and %} pairs are not properly nested. Very rarely it may also be an out of memory error. Some of the possible error messages:
missing ':' in pattern: hxmkbib found a %{ but the second or third letter after it was not a colon.
no '%{' in template file: The template file is unusable, because it contains no template.
unbalanced %{..%} in pattern: There are more %{ than %}.

BUGS

Sorting is primitive: the program doesn't parse dates or names and simply sorts "Jan 2000" under the letter "J" and "Albert Camus" under the letter "A". For the moment the only work-around is to put names in the bibfile as "Camus, Albert".

The program simply lists all authors or editors. There is no way to generate an "et. al." after the third one. The work-around is to put the "et. al." in the bibfile. Putting commas between the first authors and the word "and" before the final one is also not possible.

The program doesn't try to interpret names of authors or editors and they cannot be reformatted. It is impossible to write a name that is specified as "Sartre, Jean-Paul" in the bibfile as "J. Sartre" or as "Jean-Paul Sartre" in the output.

There is no way to suppress a period after a field if the field already ends with a period. E.g., the template "%{A:A.%}" may generate "A. Person Jr.." if the author is "A. Person Jr." The only option is to either not put periods in the bibfile or not put periods in the template.

Entries in the bibfile can only be used if they have a %L (label) field. The program cannot find entries by searching for keywords, like refer(1).

hxmkbib will replace any ampersands (&) and less-than (<) and greater-than (>) signs that occur in the bibfile by their XML entities & < > on the assumption that the template is HTML/XML. This may not be appropriate for other formats.

hxcite-mkbib is a (bash) shell script that calls hxcite(1) and hxmkbib(1), and is therefore not portable to all platforms.