BibTeX2HTML |
BibTeX2HTML is a collection of tools for producing automatically HTML documents from bibliographies written in the BibTeX format. It consists in three command line tools:
bibtex2html is a BibTeX to HTML translator. It is invocated as
where the possible options are described below and where file.bib is the name of the BibTeX file, which must have a .bib suffix. If this file is not given, then entries are input from standard input.
Then two HTML documents are created (unless option
-nobibsource is selected or input is standard input, see below) :
bibtex is called on file.bib in order to produce the a LaTeX document, and then this LaTeX document is translated into an HTML document. The BibTeX file file.bib is also parsed in order to collect additional fields (abstract, url, ps, http, etc.) that will be used in the translation.
If input is standard input and option --output is not given,
the first file is output to standard output, and the second is not
created.
The main interest of bibtex2html with respect to a traditional LaTeX to HTML translator is the use of additional fields in the BibTeX database and the automatic insertion of web links.
A link is inserted:
\url LaTeX macro is used in the text; You can insert web link for other fields and/or specify alternative names for the links using the options -f and -nf (see below).
See the result on this example.
If a BibTeX entry contains a field abstract then its contents is quoted right after the bibliography entry , like this .
This behavior may be suppressed with the option --no-abstract.
If you want both versions with and without abstracts, use the option --both. In that case, links named "Abstract" will be inserted from the page without abstracts to the page with abstracts, like this.
If a BibTeX entry contains a field keywords then its contents is displayed after the bibliography entry (and after the abstract if any).
This behavior may be suppressed with the option --no-keywords.
Most of the command line options have a short version of one character (e.g. -r) and an easy-to-remember/understand long version (e.g. --reverse-sort).
specify the title of the HTML file (default is the file name).
give an additional header for the HTML document.
give an additional footer for the HTML document.
use BibTeX style string (plain, alpha, etc.). Default style is plain.
do not print the abstracts (if any).
do not print the keywords (if any).
produce both pages with and without abstracts. If the BibTeX file is foo.bib then the two pages will be respectively foo.html and foo_abstracts.html (The suffix may be different, see option --suffix). Links are inserted from the page without abstracts to the page with abstracts.
do not print the cite keys. Note: this option implicitly suppresses the use of HTML tables to format the entries; to enforce the use of tables, use option -use-table (passing it after option -nokeys on the command line).
use the cite keys from the BibTeX input file (and not the ones generated by the BibTeX style file).
print URLs instead of files’ types.
interpret the macro \url as HeVeA’s one, i.e. with two
arguments, the first one being the url and the second one the text
to print. The default behavior is to interpret the macro \url
as the one from the package url, which has only one
argument (the url itself).
add a web link for that BibTeX field.
similar to -f but specifies the way to display the link (e.g. -nf springer "At Springer’s").
declare that a field must be treated like the abstract field, i.e. is an annotation to be displayed as a text paragraph below the entry.
make a separate web page for each entry. Beware: this option produces as many HTML files as BibTeX entries!
produce a single document, inserting each BibTeX entry (the input) right after its BibTeX output
set the background color of the HTML file (default is none).
set a style sheet file for the HTML document (default is none).
use HTML DL lists instead of HTML tables to format entries.
use Unicode entities for the following macros :
\models \curlyvee \curlywedge \bigcirc \varepsilon
\not{\models}
use HTML entities for the following macros :
\= \Im \Leftarrow \Re \Rightarrow \aleph \ang \angle \approx \ast \cdot \cdots \cong \copyright \cup \dagger \diamond \emptyset \equiv \exists \forall \ge \geq \in \infty \int \land \lang \lceil \le \leftarrow \leftrightarrow \leq \lfloor \longleftarrow \longrightarrow \lor \lozenge \nabla \ne \neg \neq \ni \notin \oplus \otimes \partial \perp \pm \prod \propto \rang \rceil \rfloor \rightarrow \sim \simeq \sqrt \subset \subseteq \sum \supset \supseteq \therefore \times \tm \to \vartheta \vee \wedge \wp
read the LATEX macros in the given file. Note: bibtex2html does not handle macros arguments; arguments are simply discarded.
do not expand the abbreviation strings, leave them in the output file.
Select only keys appearing in filename. To be used manually
or in conjonction with bib2bib.
exclude an particular entry.
sort by date.
sort as BibTeX (usually by author).
unsorted i.e. same order as in .bib file (default).
reverse the sort.
number entries in reverse order (i.e. from n to 1 in plain style).
do not produce a full HTML document but only its body (useful to merge the HTML bibliography in a bigger HTML document).
do not produce the _bib.html file. In that case, no “BibTeX
entry” link are inserted in the HTML file.
give an alternate suffix string for both HTML files and links (default is .html).
give an alternate suffix string for HTML files (default is .html).
give an alternate suffix string for HTML links (default is .html).
specifies the output file. If file is -, then the
standard output is selected.
specify the BibTeX command (default is bibtex -min-crossrefs=1000). May be useful for example if you need to specify the full path of the bibtex command.
print the BibTeX entries on the standard output (one per line), as selected and sorted by bibtex2html. This is useful if you want to use the selection and sorting facilities of bibtex2html in another program. Note: you may need to set also the -q option (quiet) to suppress the usual output.
ignore BibTeX errors.
be quiet.
stop at the first warning.
print a short usage and exit.
print the version and exit.
do not insert the bibtex2html command in the HTML document (default is to insert it as a comment at the beginning).
bib2bib is a tool for extracting some entries from a list of bibliography files. It is invocated as
where the possible options are described below and where file1.bib ⋯ filen.bib are the names of the BibTeX files, which must have a .bib suffix. If no files at all are given on the command line, then input is taken from standard input.
The options allow to specify a filter condition to test against each references read from bib files. The result will be a new BibTeX file containing all the entries of the input files that satisfy the condition. Notice that this output file contains all the necessary informations: each string and each cross-reference needed will be also in that file.
Additionally, bib2bib may output a file containing all the
keys of entries that satisfy the condition. This second file is
suitable for input as option -citefile to bibtex2html.
specify a condition for selecting the entries. The output will retain only the entries that satisfy this condition. If several such condition are given, then only the entries that satisfy all the conditions are selected. The syntax of conditions is given below, notice that it is better to escape shell expansions in that conditions, in other words, you should write conditions between quotes.
specify the filename where the selected entries are output. If not given, it defaults to standard output.
specify the filename where the list of selected keys is output. If not given, this file is not created.
Notice that the two output files above are suitable for use with bibtex2html. A typical use would be
which will produce exactly the HTML file for the selected references.
expand all abbreviations in the output file.
expand all crossrefs in the output file. Notice that the meaning of such an expansion is not completely obvious: it’s better to let bibtex (via bibtex2html) handle the cross-references itself, depending on the style considered.
Notice that bibtex2html itself will expand the strings (by
default, unless you specify the -noexpand option) but not the
cross-references.
prevent generation of extra comments at beginning of output bib file.
remove all occurrences of field f. This option can be used several times to remove several fields.
rename all occurrences of field f1 into f2. This option can be used several times to rename several fields. Beware that if an entry already has both fields f1 and f2, this will result in two fields f2, and BibTeX styles usually take only the first occurrence into account.
Example:
removes all abstract and copyright fields and rename all x-pdf fields into name url.
sorts the entries of the bibliography with respect to the given field f, which may also be $key or $type to refer to the key or to the entry type, as for filter conditions. It may also be $date, to ask for sorting from oldest to newest, as for option -d of bibtex2html.
This option may be used several times to specify a lexicographic order, such as by author, then by type, then by date:
When sorting, the resulting bibliography will always contains the comments first, then the preambles, then the abbreviations, and finally the regular entries. Be warned that such a sort may put cross-references before entries that refer to them, so be cautious.
reverses the sort order.
be quiet.
stop at the first warning.
outputs the bib file as a two-dimensional array in PHP syntax, in file.
A filter condition is a boolean expression that is evaluated against a BibTeX entry to decide whether this entry should be selected. A condition is either:
and
c2 (or c1 & c2) ;
or
c2 (or c1 | c2);
not c (or
! c) ;
exists f
(or ? f) where f is a field name ;
where an expression is either:
$key which corresponds to the key of
an entry.
$type which corresponds to the type
of an entry (ARTICLE, INPROCEEDINGS, etc.). Notice that an entry
type is always written in uppercase letters.
Comparison operators are the usual ones: =, <, >, <=, >= and <>.
The field names are any sequences of lowercase or uppercase letters (but no distinction is made between lowercase and uppercase letters).
Be careful when writing conditions in a shell command: the shell must
not interpret anything in a condition, such as $key. So usually
you need to put conditions inside quote characters that forbid shell
interpretation: single quotes under Unix shell, or double quotes under
Microsoft Windows shell. This is why strings in conditions may be put
indifferently between single or double quotes: you will use the ones
which are not the ones you use to forbid shell interpretation. In
examples below, we will use Unix convention, so under Windows you have
to permute the use of single and double quotes.
Note that within Makefiles you have to escape the $
character in $key or $type (using $$key and
$$type instead, at least with GNU make).
Regular expressions must be put between single or double quotes, and must
follow the GNU syntax of regular expressions, as for example in GNU Emacs. Any
character other than $^.*+?[] matches itself, see
Table 1 for the meaning of the special characters.
.matches any character except newline [..]character set; ranges are denoted with -, as in[a-z]; an initial^, as in[^0-9], complements the set^matches the beginning of the string matched $matches the end of the string matched \rmatches a carriage return \nmatches a linefeed \tmatches a tabulation \bmatches word boundaries \dddmatches character with ASCII code ddd in decimal \charquotes special character char among $^.*+?[]\regexp *matches regexp zero, one or several times regexp +matches regexp one or several times regexp ?matches regexp once or not at all regexp1 \|regexp2alternative between two regular expressions, this operator has low priority against *,+and?\(regexp\)grouping regular expression
Table 1: Syntax of regular expressions
Notice that if several conditions are given with option -c on
the command line, then they are understood as the conjunction of them,
in other words
is equivalent to
Table 2 shows a formal grammar for conditions.
Cond → Cond andCond ∣ CondorCond ∣notCond ∣existsIdCond → Cond &Cond ∣ Cond|Cond ∣!Cond ∣?Id∣ Expr Comp Expr ∣ Expr :String ∣(Cond)Comp → =∣>∣<∣>=∣<=∣<>Expr → Id ∣ String ∣ Int ∣ $key∣$typeId → [ a−zA−Z]+String → "([^"\] ∣\"∣\\)*"∣'([^'\] ∣\'∣\\)*'Integer → [ 0−9]+
Table 2: Syntax of conditions
bib2bib, these are always converted to uppercase, so you may
take this into account when writting conditions (see below).
title : "Computer" may return true if the title
contains the word Computer with a capital letter, whereas
title : "computer" would return false.
$type = "INPROCEEDINGS",
otherwise the condition would be always false.
not (f = "value") and f <> "value" are not equivalent: for an
entry that does not have a field f, the first condition is true
whereas the second is false.
not has higher priority than and, which itself
has higher priority than or. and and or associate to
the left.
>, <, >= and <= may only be
used between integer values. In any other case, a warning is displayed and
the result is false.
"Filli\^atre", "Filli{\^a}tre", "Filli\^{a}tre"
and "Filliâtre" are considered identical and indeed equal to
"Filliâtre". Note that when using such a string as a
regular expression, there is no need to escape the backslash, since
interpretation of LaTeX accenting commands is made before
interpretation into a regexp. Using HTML entities for matching
accented names is thus considered as the safest method.
Here are some examples to help you writing the filter conditions you are interested in.
The following command reads input files biblio1.bib and
biblio2.bib, and select only entries that appeared in 1999 :
bib2bib -oc cite1999 -ob 1999.bib -c 'year=1999' biblio1.bib biblio2.bib
The resulting file cite1999 contains the list of keys
selected. You can then produce the HTML file by
bibtex2html -citefile cite1999 1999.bib
You may also select references appeared after and/or before a given year. For example, references after 1997:
bib2bib -oc citeaft1997 -ob aft1997.bib -c 'year>1997' biblio.bib
or between 1990 and 1995:
bib2bib -oc cite90-95 -ob 90-95.bib -c 'year>=1990 and year<=1995' biblio.bib
The following command reads input files biblio.bib and select
only entries whose (co)author is Donald Knuth:
bib2bib -oc knuth-citations -ob knuth.bib -c 'author : "Knuth"' biblio.bib
More complicated, if you would like to have only the references whose author is Knuth only, you may try
bib2bib -oc knuth-citations -ob knuth.bib \ -c 'author : "^\(Donald \(E. \)?Knuth\|Knuth, Donald \(E.\)?\)$"' biblio.bib
or equivalently but missing the possible “E.”:
bib2bib -oc knuth-citations -ob knuth.bib -c 'author = "Donald Knuth" or author = "Knuth, Donald"' biblio.bib
Any boolean combination of comparison and/or matching are possible. For example, the following command extract the references that appeared since 1995 and have lambda-calculus in their title, with anything between “lambda” and “calculus”:
bib2bib -oc lambda -c 'year >= 1995 and title : "lambda.*calculus"' biblio.bib
for example, it will select a title containing
$\lambda$-calculus.
bib2bib has the effect of merging several bib files into a
single one. This process may result in duplicate entries in the
resulting files, which is considered as erroneous by bibtex.
Of course, this is not really a bug of bib2bib since it is of
your own to take care not having entries with the same key.
However, there are two particular cases when this occurs naturally: when two bib files share common abbreviations, or when they share common cross-references.
In order to make bib2bib behaves correctly in such a case,
it is designed as follows: for repeated abbrevs, the first abbrev is kept and
the others are ignored, and for repeated regular entries, the last entry
is kept and the others are ignored. With this behaviour, everything
works well as soon as repeated abbrevs are really duplicate abbrevs of
the same sentence, and repeated keys are really duplicate
entries.
aux2bib is a tool extracting the BibTeX references from a .aux file (as produced by LATEX) and building the corresponding BibTeX file. It is invocated as
The BibTeX file is written on the standard output.
By default, all entries of the input BibTeX file are translated into
HTML, including cross-references. Since the latter are there, bibtex
will never expand them. If you want them to be expanded, you have to
tell bibtex2html that crossref entries need not be in the resulting
file. To do that you have to use the option -citefile to give
the exact list of entries you want to see. If a cross-reference is
not in that list, then its fields will be expanded into all entries
that cross-refers to it. (Technically, this work because bibtex2html
calls bibtex with option -min-crossrefs=1000 by default.)
bib2bib -oc knuth-citations -ob knuth.bib -c 'author : "Knuth"' biblio.bibI get "Lexical error in condition: Unterminated string". What’s going wrong?
You are probably running bib2bib under Microsoft Windows, hence
you should permute the use of single quotes and double quotes, as
explained in Section 3.2:
bib2bib -oc knuth-citations -ob knuth.bib -c "author : 'Knuth'" biblio.bib
mail to authors , Fri Feb 12 17:46:23 1999
This document was translated from LATEX by HEVEA.