Biber: A Backend Bibliography Processor For Biblatex

biber
A backend bibliography processor for biblatex

Philip Kime, François Charette Version biber 2.13 (biblatex 3.13)
Philip@kime.org.uk, 20th August 2019
firmicus@ankabut.net
Contents
1. Important Changes 1 3.7. List and Name Separators 39
3.8. Extended Name Format . 39
2. Introduction 3 3.9. Editor Integration . . . . 40
2.1. About . . . . . . . . . . . 3 3.10. BIBTEX macros and the
2.2. Requirements . . . . . . . 4 MONTH field . . . . . . . . . 40
2.3. Compatibility Matrix . . 4 3.11. Biber datasource drivers 41
2.4. License . . . . . . . . . . . 4
3.12. Visualising the Output . 41
2.5. History . . . . . . . . . . . 6
3.13. Tool Mode . . . . . . . . . 43
2.6. Performance . . . . . . . . 9
2.7. Acknowledgements . . . . 9
4. Binaries 49
3. Use 9 4.1. Binary Caches . . . . . . 50
3.1. Options and config file . . 11 4.2. Binary Architectures . . . 51
3.2. Unicode . . . . . . . . . . 32 4.3. Installing . . . . . . . . . . 51
3.3. Input/Output File Loca- 4.4. Building . . . . . . . . . . 52
tions . . . . . . . . . . . . 33
3.4. Logfile . . . . . . . . . . . 33 A. Appendix 55
3.5. Collation and Localisation 34 A.1. Babel/Polyglossia lan-
3.6. Encoding of files . . . . . 36 guage to Locale mapping 55
1. Important Changes
Please see the Changes file which accompanies Biber for the details on changes in
each version. This section is just for important things like incompatible changes
which users should be aware of.
2.13
In tool mode (--tool), Biber now allows datamodel configuration in a user con-
figuration file to add to or override the default datamodel contained in the main
biber-tool.conf file. Previously it was necessary to copy the entire default data
model into a user configuration file and then alter it.
1
2.6
When outputting BibTeX data in tool mode (--tool), Biber now follows a full
internal processing chain involving the data model. In previous versions, BibTeX
output would just output the raw BibTeX input data, only allowing for some re-
formatting options and therefore no tool mode conversions from other formats into
BibTeX format were possible. This change has some normalisation consequences:
• Dates are normalised into DATE fields. Legacy YEAR fields are never output in
BibTeX format data output.
• Fields which are not defined in the data model described in the default biber-tool.conf
are ignored and are neither read nor output. If custom fields are required, they
should be defined in the data model by using a custom tool mode config file
(see below). If you would like to have ignored fields reported on, use the
--validate-datamodel option.
1.9
Biber no longer checks the environment for locales to use for sorting. This was
always rather against the spirit of TeX since it means that the same document might
look different when compiled by different people. However, Biblatex now passes
Babel/Polyglossia language identifiers (or real locale identifiers if you prefer) in the
.bcf and Biber can use these to set the sorting locale globally or on a per-sortscheme
basis. This is better than using environment variables since Babel/Polyglossia are
more LaTeX relevant language environments anyway.
1.8
Various option name changes. Old names are retained for backwards compatibility.
See the output of the --help option.
1.0
The --validate-structure option is now called --validate-datamodel
0.9.9
The output format option --graph has been moved to a new option --output-format.
The option --graph should now be specified as --output-format=dot and the
--dot-include option should be used to specify the elements to include in the
DOT output. For example:
biber --graph=section,field <file>
is now:
biber --output-format=dot --dot-include=section,field <file>
2
1.8
Several option names have changed. Several options have changed names to facilitate
better semantic classification of options. The previous names are supported as legacy
aliases. See the --help output of the Biber command.
0.9.8
The sourcemap option syntax has changed.The syntax was too confusing. It is now
simplified and more powerful. It is uses a sequential processing model to apply
mappings to an entry. See section 3.1.2.
0.9.7
The user config file has a completely new format.The reason for this is that the older
Config::General format could not be extended to deal with more sophisticated
features like per-datasource restrictions. An XML format is much better and in fact
easier to understand. The old format of the map option (now called sourcemap)
was rather confusing because of limitations in the old config file format. Please see
section 3.1.2 and convert your config files to the new format.
0.9.6
Matching of citation keys and datasource entry keys is now case-sensitive. This
is to enforce consistency across the entire BibLaTeX and Biber processing chain.
All of the usual referencing mechanisms in LaTeX are case-sensitive and so is the
matching in BibLaTeX of citations to entries in the .bbl file generated by Biber.
It is inconsistent and messy to enforce case-insensitivity in only Biber’s matching
of citations keys to datasource entry keys. If Biber detects what looks like a case
mismatch between citation keys, it will warn you.
Summary of warnings/errors is now a new format. When Biber finishes writing the
.bbl, it gives a summary count of errors/warnings. It used to do this in the same
format as BibTEX, for compatibility. Now it uses a more consistent and easier to
parse format which matches all other Biber messages. Please note if you need to
support Biber in an external tool. I have updated the notes on AUCTEX support
below to reflect this.
2. Introduction
2.1. About
Biber is conceptually a BIBTEX replacement for Biblatex. It is written in Perl
with the aim of providing a customised and sophisticated data preparation backend
for Biblatex. You do not need to install Perl to use Biber—binaries are provided
for many operating systems via the main TEX distributions (TEXLive, MacTEX,
MiKTEX) and also via download from SourceForge. Functionally, Biber offers a
superset of BIBTEX’s capabilities but is tightly coupled with Biblatex and cannot
3
be used as a stand-alone tool with standard .bst styles. Biber’s primary role is to
support Biblatex by performing the following tasks:
• Parsing data from datasources

• Processing cross-references, entry sets, related entries
• Generating data for name, name list and name/year disambiguation
• Structural validation according to Biblatex data model
• Sorting reference lists
• Outputting data to a .bbl for Biblatex to consume
Biber also has the ability to output different formats than .bbl and can, for
example, output a new BibTeX file which contains only cited entries from the data-
sources (using the --output-format=bibtex option). There is also a ‘<tool’ mode
which operates on datasources instead of individual documents, allowing you to
transform, convert, reformat and generally change the contents of a datasource (see
3.13).
2.2. Requirements
Biber is distributed primarily as a stand-alone binary and is included in TEXLive,
MacTEX and MiKTEX. If you are using any of these distributions, you do not need
any additional software installed to use Biber. You do not need a Perl installation
at all to use the binary distribution of Biber1 .
Biber’s git repository and bug/feature tracker is on github2 . Biber’s documenta-
tion, binary downloads and supporting files are on SourceForge3 Biber is included
into TEXLive, the binaries coming from SourceForge.
2.3. Compatibility Matrix

Biber versions are closely coupled with Biblatex versions. You need to have the
right combination of the two. Biber will warn you during processing if it encounters
information which comes from a Biblatex version which is incompatible. Table 1
shows a compatibility matrix for the recent versions.
2.4. License
Biber is released under the free software Artistic License 2.04
1
If you prefer, you can run Biber as a normal Perl program and doing this does require you to have
a Perl interpreter installed. See section 4.
2
https://github.com/plk/biber
3
http://sourceforge.net/projects/biblatex-biber/
4
http://www.opensource.org/licenses/artistic-license-2.0.php
4
Biber version Biblatex version
2.13 3.13
2.12 3.12
2.11 3.11
2.10 3.10
2.9 3.9
2.8 3.8
2.7 3.7
2.6 3.5, 3.6
2.5 3.4
2.4 3.3
2.3 3.2
2.2 3.1
2.1 3.0
2.0 3.0
1.9 2.9
1.8 2.8x
1.7 2.7
1.6 2.6
1.5 2.5
1.4 2.4
1.3 2.3
1.2 2.1, 2.2
1.1 2.1
1.0 2.0
0.9.9 1.7x
0.9.8 1.7x
0.9.7 1.7x
0.9.6 1.7x
0.9.5 1.6x
0.9.4 1.5x
0.9.3 1.5x
0.9.2 1.4x
0.9.1 1.4x
0.9 1.4x
Table 1: Biber/Biblatex compatibility matrix
5
2.5. History
BIBTEX has been the default (only …) integrated choice for bibliography processing
in TEX for a long time. It has well known limitations which stem from its data
format, data model and lack of Unicode support5 . The .bst language for writing
bibliography styles is painful to learn and use. It is not a general programming
language and this makes it really very hard to do sophisticated automated processing
of bibliographies.
Biblatex was a major advance for LaTeX users as it moved much of the biblio-
graphy processing into LaTeX macros. However, Biblatex still used BIBTEX as a
sorting engine for the bibliography and also to generate various labels for entries.
BIBTEX’s capabilities even for this reduced set of tasks was still quite restricted due
to the lack of Unicode support and the more and more complex programming issues
involved in label preparation and file encoding.
Biber was designed specifically for Biblatex in order to provide a powerful backend
engine which could deal with any required tasks to do with .bbl preparation. Its
main features are:
• Deals with the full range of UTF-8

• Sorts in a completely customisable manner using, when available, CLDR col-
lation tailoring
• Allows for per-entrytype options
• Automatically encodes the .bbl into any supported encoding format6
• Processes all bibliography sections in one pass of the tool
• Output to GraphViz instead of .bbl in order to help visualise complex bibli-
ographies with many crossrefs etc. (see section 3.12)
• Handles UTF-8 citekeys and filenames (given a suitable fully UTF-8 compliant
TEX engine)
• Creates entry sets dynamically and allows easily defined static entry sets, all
processed in one pass
• ‘Syntactic’ inheritance via new @XDATA entrytype and field. This can be
thought of as a field-based generalisation of the BIBTEX @STRING function-
ality (which is also supported).
• ‘Semantic’ inheritance via a generalisation of the BIBTEX crossreference mech-
anism. This is highly customisable by the user—it is possible to choose which
fields to inherit for which entrytypes and to inherit fields under different names
etc. Nested crossreferences are also supported.
• Handles complex auto-expansion and contraction of names and namelists (See
section 4.11.4 of the Biblatex manual for an excellent explanation with ex-
amples, this is quite an impressive feature …)
5
In fact, there is now a Unicode version
6
‘Supported’ here means encodings supported by the Perl Encode module
6
• Extensible modular datasource architecture for ease of adding more datasource
types
• Support for remote datasources
• User-definable mapping and suppression of fields and entrytypes in datasources.
You can use this to, for example, ignore all ABSTRACT fields completely. See
section 3.1.2
• Support for related entries, to enable generic treatment of things like ‘trans-
lated as’, ‘reprinted as’, ‘reprint of’ etc.
• Customisable labels
• Multiple bibliography lists in the same section with different sorting and fil-
tering
• No more restriction to a static data model of specific fields and entrytypes
• Structural validation of the data against the data model with a customisable
validation model
• Tool mode for operations on datasources directly
Figure 1 shows the main functional units of processing in Biber. The most dif-
ficult tasks which Biber performs are the processing of Biblatex’s uniquename and
uniquelist options7 , the sorting of lists8 and the initial data parse and remap
into an internal data model. Biber is getting on for around 20,000 lines of mostly
OO Perl and relies on certain splendid Perl modules such as Unicode::Collate,
Text::BibTeX and XML::LibXML.
It may be useful to know something about the different routes a datasource entry
can take as it passes through Biber.
1. All cited entries which are subsequently found in a datasource are instantiated
in the internal Biber data model.
2. Some uncited entries on which cited entries depend are instantiated in the
internal Biber data model:
• Entries with entrytype @XDATA which are referenced from cited entries.
• Entries mentioned in the CROSSREF or XREF field of a cited entry (unless
they are also cited themselves in which case they are already instantiated
as per item 1 above).
• Clones of entries mentioned as a ‘related’ entry of a cited entry.
• Members of sets, either explicit @SET entrytype entries or dynamic sets.
3. Some uncited but instantiated entries are promoted to cited status so that
they make it into the output:
• Entries instantiated by being members of a set.
7
A rather tricky unbounded loop but with a guaranteed eventual stable exit state.
8
This is multi-field ST sort with an embedded cache for performance.
7
datasources .bcf control file
Decode to UTF-8 Generate uniqueness data
User biber.conf file remap/parse Generate name visibility data
Instantiate dynamic sets and Generate more hashes and

related entries labels
Process XDATA Perform sorting
Process cross-references and

Construct output objects
sets
Validate data model Encode to output encoding
Resolve label* fields

generate hashes
enforce mincrossrefs
output file
Figure 1: Overview of Biber’s main functional units
8
• Entries instantiated by being mentioned as a CROSSREF are promoted to
cited status if CROSSREF’ed or XREF’ed at least mincrosref times.
4. Some of these auto-cited entries have the ‘dataonly’ option set on them so
that Biblatex will only use them for data and will not output them to the
bibliography:
2.6. Performance
Biber can’t really be compared with BIBTEX in any meaningful way performance-
wise. Biber is written in Perl and does a great deal more than BIBTEX which is
written in C. One of Biber’s test cases is a 2150 entry, 15,000 line .bib file which
references a 630 entry macros file with a resulting 160 or so page (A4) formatted
bibliography. This takes Biber just under 30 seconds to process on a reasonable
computer. This is perfectly acceptable, especially for a batch program.
2.7. Acknowledgements
François Charette originally wrote a first modest version of Biber. Philip Kime
joined in the development in 2009 and is largely responsible for making it what it is
today.
3. Use
Firstly, please note that Biber will not attempt to sanitise the content of BIBTEX
datasources. That is, don’t expect it to auto-escape any TEX special characters like
‘&’ or ‘%’ which it finds in, for example, your TITLE fields. It used to do this in
earlier versions in some cases but as of version 0.9, it doesn’t because it’s fraught
with problems and leads to inconsistent expectations and behaviour between dif-
ferent datasource types. In your BIBTEX data sources, please make sure your entries
are legal TEX code.
Running biber --help will display all options and description of each and is the
primary source of usage information. Biber returns an exit code of 0 on success or
2 if there was an error.
Most Biber options can be specified in long or short format. When mentioning
options below, they are referred to as ‘long form|short form’ when an option has
both a long and short form. As usual with such options, when the option requires
an argument, the long form is followed by an equals sign ‘=’ and then the argument,
the short form is followed by a space and then the argument. For example, the
--configfile|-g option can be given in two ways:
9
biber --configfile=somefile.conf
biber -g somefile.conf
With the backend=biber option, Biblatex switches its backend interface and
passes all options and information relevant to Biber’s operation in a control file
with extension .bcf9 . This is conceptually equivalent to the .aux file which LaTeX
uses to pass information to BIBTEX. The .bcf file is XML and contains many options
and settings which configure how Biber is to process the bibliography and generate
the .bbl file.
The usual way to call Biber is simply with the .bcf file as the only argument.
Biblatex always writes the control file with a .bcf extension. Specifying the ‘.bcf’
extension to Biber is optional. Assuming a control file called test.bcf, the following
two commands are equivalent:
biber test.bcf
biber test
Figure 2 is a graphical overview of the data flow for data model information. See
Figure 1 for a more complete overview of Biber’s processing steps.
document data model mapping Biblatex data

datasource
from .bcf model from .bcf
default data model mapping remap parser validation

from .bcf
data model mapping

from Biber config file output file
Figure 2: Model data flow in Biber
9
Biblatex Control File
10
3.1. Options and config file
Biblatex options which Biber needs to know about are passed via the .bcf file. See
Table 2 for the Biblatex options which Biber uses and also for the scopes which are
supported for each option. Biber also has its own options which are set using the
following resource chain, given in decreasing precedence order:
command line options →

biber.conf file →
.bcf file→
Biber hard-coded defaults
Users do not need to care directly about the contents or format of the .bcf file as
this is generated from the options which they specify via Biblatex. The config file is
a place to set commonly used command-line options and also to set options which
cannot be set on the command line.
The configuration file is by default called biber.conf but this can be changed
using the --configfile|-g option. Unless --configfile|-g is used, the config file
is looked for in the following places, in decreasing order of preference:
biber.conf in the current directory →

$HOME/.biber.conf →
$XDG_CONFIG_HOME/biber/biber.conf →
$HOME/Library/biber/biber.conf (Mac OSX only)
$APPDATA/biber.conf (Windows only) →
the output of ‘kpsewhich biber.conf’ (if available on the system)
The config file is XML. Here Below is an example config file which displays the Biber
defaults:
<?xml version="1.0" encoding="UTF-8"?>

<config>
<clrmacros>0</clrmacros>
<collate_options>
<option name="level" value="4"/>
<option name="variable" value="non-ignorable"/>
<option name="normalization" value="prenormalized"/>
</collate_options>
<debug>0</debug>
<decodecharsset>base</decodecharsset>
<dieondatamodel>0</dieondatamodel>
11
Biblatex option Global Per-type Per-entry
alphaothers ✓ ✓
dataonly ✓ ✓
inheritance ✓
labelalpha ✓ ✓
labelalphatemplate ✓ ✓
labeldate ✓ ✓
labeldatespec ✓ ✓
labelnamespec ✓ ✓
labelnumber ✓ ✓
labeltitle ✓ ✓
labeltitleyear ✓ ✓
maxalphanames ✓ ✓ ✓
maxbibnames ✓ ✓ ✓
maxcitenames ✓ ✓ ✓
maxitems ✓ ✓ ✓
minalphanames ✓ ✓ ✓
minbibnames ✓ ✓ ✓
mincitenames ✓ ✓ ✓
minitems ✓ ✓ ✓
presort ✓ ✓ ✓
singletitle ✓ ✓
skipbib ✓ ✓
skiplab ✓ ✓
skiplos ✓ ✓
sortalphaothers ✓ ✓
sortexclusion ✓
sortfirstinits ✓
sorting ✓
uniquelist ✓ ✓ ✓
uniquename ✓ ✓ ✓
useauthor ✓ ✓ ✓
useeditor ✓ ✓ ✓
useprefix ✓ ✓ ✓
usetranslator ✓ ✓ ✓
Table 2: Biblatex options which Biber uses
12
<graph>0</graph>
<input_encoding>UTF-8</input_encoding>
<listsep>and</listsep>
<mincrossrefs>0</mincrossrefs>
<namesep>and</namesep>
<nodieonerror>0</nodieonerror>
<noinit>

<option value="\b\p{Ll}{2}\p{Pd}(?=\S)"/>

<option value="[\x{2bf}\x{2018}]"/>
</noinit>
<nolabel>

<option value="[\p{P}\p{S}\p{C}]+"/>
</nolabel>
<nolog>0</nolog>
<nostdmacros>0</nostdmacros>
<nosort>

<option name="setnames" value="\A\p{L}{2}\p{Pd}(?=\S)"/>

<option name="setnames" value="[\x{2bf}\x{2018}]"/>
</nosort>
<onlylog>0</onlylog>
<others_string>others</others_string>
<ouput_align>0</output_align>
<output_encoding>UTF-8</output_encoding>
<output_fieldcase>upper</output_fieldcase>
<output_format>bbl</output_format>
<output_indent>2</output_indent>
<output_resolve_xdata>0</output_resolve_xdata>
<output_resolve_crossrefs>0</output_resolve_crossrefs>
<output_resolve_sets>0</output_resolve_sets>
<output_safechars>0</output_safechars>
<output_safecharsset>base</output_safecharsset>
<quiet>0</quiet>
<sortcase>true</sortcase>
<sortupper>true</sortupper>
<tool>false</tool>
<trace>false</trace>
<validate_bltxml>0</validate_bltxml>
<validate_config>0</validate_config>
<validate_control>0</validate_control>
<validate_datamodel>0</validate_datamodel>
<wraplines>0</wraplines>
<xsvsep>\s*,\s*</xsvsep>
13
</config>
In practice, the most commonly used options will be set via Biblatex macros in
your document and automatically passed to Biber via the .bcf file. Certain options
apply only to Biber and can only be set in the config file, particularly the more
complex options. Most options are simple tags. Exceptions are the nosort, noinit
and collate-options options which are slightly more complex and can have sub-
options as shown. A much more complex option is the sourcemap option which is
not set by default and which is described in section 3.1.2.
3.1.1. The output-format option

Biber is able to output formats other than .bbl files for Biblatex to consume. It is
also able to output other formats such as DOT for visualisation of entry dependencies
(see section 3.12), the experimental biblatexml XML format, BibTeX .bib files and
an XML version of the .bbl format with extension .bblxml. .bib output is possible
in tool mode, when you are converting an entire datasource file independently of any
particular document (see section 3.13). It is also useful when you want, instead of a
.bbl, a new .bib file containing only the cited entries from a document so that you
can, for example, send a minimally complete package for typesetting to someone.
To do this, you would, after the first LaTeX run, call Biber like this:
biber --output-format=bibtex test.bcf
This would result in a new .bib file called test_biber.bib containing all cited
entries in test.tex, in citation order, formatted according to the various ouput-*
options. You could of course also perform more processing like source mapping (see
section 3.1.2), reencoding (see section 3.6) etc. using more command line options or
a config file.
The .bblxml format for output is an XML version of the .bbl. It cannot be read
by Biblatex but contains the same information as in the .bbl and may be useful
if you want to transform a document bibliography into some other format since
XML is a well-supported transformation format (using, for example, XSLT). By
default, when choosing .bblxml output with the option --output-format=bblxml,
a RelaxNG XML schema is also generated (unless the --no-bblxml-schema is used).
This schema is derived from the active datamodel in the document (passed in the
.bcf from Biblatex) and is placed in the same directory as the .bblxml output file.
The extension of the schema is .rng. The option --validate-bblxml may be used
to validate the .bblxml against the schema.
3.1.2. The sourcemap option

The datasource drivers implement a mapping from datasource entrytypes and fields
into the Biblatex data model. If you want to override or augment the driver mappings
14
you can use the sourcemap option which makes it possible to, for example, have a
datasource with non-standard entrytypes or fields and to have these automatically
mapped into other entrytypes/fields without modifying your datasource. Essentially,
this alters the source data stream which Biber uses to build the internal Biblatex
data model and is an automatic way of editing the datasource as it is read by Biber.
Source mappings can be defined at different ‘levels’ which are applied in a defined
order. See the Biblatex manual regarding these macros:
user-level maps defined with \DeclareSourcemap→

user-level maps defined in the Biber config file (described below)→
style-level maps defined with \DeclareStyleSourcemap→
driver-level maps defined with \DeclareDriverSourcemap
The sourcemap option can only be set in the config file and not on the command
line as it has a complex structure. This option allows you to perform various data-
source mapping tasks which can be useful for pre-processing data which you do not
generate yourself:
• Map datasource entrytypes to different entrytypes.

• Map datasource fields to different fields.
• Add new fields to an entry
• Remove fields from an entry
• Modify the contents of a field using standard Perl regular expression match
and replace.
• Restrict any of the above operations to entries coming from particular data-
sources which you defined in \addresource{} macros.
• Restrict any of the above operations to entries only of a certain entrytype.
There is in fact, more flexibility than the above suggests, examples will show this
below. The format of the sourcemap option section in the config file is described
below, followed by examples which will make things clearer. Items in red are not
literal, they are descriptive meta-values which are explained in the accompanying
text. Items in blue are optional within their parent section or element. The general
structure is:
<sourcemap>
<maps datatype="driver1 " map_overwrite="1|0">
<map1 map_overwrite="1|0"> ... </map1 >
⋮
<mapn map_overwrite="1|0"> ... </mapn >
</maps>
⋮
15
<maps datatype="drivern " map_overwrite="1|0">
<map1 map_overwrite="1|0"> ... </map1 >
⋮
<mapn map_overwrite="1|0"> ... </mapn >
</maps>
</sourcemap>
Here, driver1 …drivern are the names of valid Biber data source drivers (see sec-
tion 3.11). One thing to note here is the map_overwrite attribute. This boolean
attribute determines whether, for this driver mapping section, you may overwrite
existing fields when adding new fields or mapping them. This attribute can be over-
ridden on a per-map basis, see below. A warning will be issued either way saying
whether an existing field will or will not be overwritten. If omitted, it defaults to
‘0’.
The map elements are processed in sequence and contain a number of map_steps
which are also processed in sequence. Each map_step allows you to do a particular
thing or combination of things:
• Change the entrytype of an entry

• Change the name of a field
• Add extra fields the entry
• Change the contents of a field
These facilities are explained in more detail below, with examples. A map element
looks like this:
<map map_overwrite="0|1" map_foreach="loopval">

<per_datasource>datasource</per_datasource>
<per_type>entrytype</per_type>
<per_nottype>entrytype</per_nottype>
<map_step map_type_source="source-entrytype"
map_field_source="source-field"
map_notfield="source-field"
map_type_target="target-entrytype"
map_field_target="target-field"
map_match="match-regexp"
map_nomatch="match-regexp"
map_replace="replace-regexp"
map_field_set="set-field"
map_field_value="set-value"
map_entry_new="newentrykey"
map_entry_newtype="newentrykeytype"
map_entry_entrytarget="newentrykey"
map_append="1"
map_appendstrict="1"
16
map_null="1"
map_entry_null="1"
map_entry_clone="clonekey"
map_origfield="1"
map_origfieldval="1"
map_origentrytype="1"
map_final="1"/>
</map>
• If there are any datasources named in per_datasource elements, this map-

ping only applies to entries coming from the named datasources. There can
be multiple per_datasource elements each specifying one of the datasource
names given in a Biblatex \addbibresource macro.
• If there are any entrytypess named in per_type elements, this mapping only
applies to entries of the named entrytypess.
• If there are any entrytypess named in per_nottype elements, this mapping
only applies to entries not of the named entrytypess.
• The map_overwrite attribute can be used to override the value for this at-
tribute set on the parent maps element. If omitted, it defaults to the parent
maps attribute value.
• The map_foreach attribute loops over all \steps in this \map, setting the spe-
cial variable $MAPLOOP to each of the comma-separated values contained in
loopval. loopval can either be the name of a datafield set defined with Bib-
latex’s \DeclareDatafieldSet, a datasource field which contains a comma-
separated values list or an explicit comma-separated values list itself (and
loopval is determined in that order). This allows the user to repeat a group
of map_steps for each value of loopval. The special variable $MAPUNIQ may
also be used in the map_steps to generate a random unique string. This can be
useful when creating keys for new entries. The special variable $MAPUNIQVAL
may be used the map_steps to refer to the value of the last random unique
string generated with $MAPUNIQ.
Each map_step is looked at in turn and compared with the datasource entry being
processed. A map_step works like this:
• If map_entry_new is set, a new entry is created with the entry key newentrykey
and the entry type newentrykeytype given in the option map_entry_newtype.
This entry is only in-scope during the processing of the current entry and can
be referenced by newentrykey given as the value to map_entrytarget. In
newentrykey, you may use standard Perl regular expression backreferences to
captures from a previous map_match step.
• When a map_field_set step has map_entrytarget set to the entrykey of
an entry created by map_entry_new, the target for the field set will be the
17
map_entrytarget entry rather than the entry being currently processed. This
allows users to create new entries and set fields in them.
• If map_entry_null is set, processing of the map immediately terminates and
the current entry is not created. It is as if it did not exist in the datasource.
Obviously, you should select the entries which you want to apply this to using
prior mapping steps.
• If map_entry_clone is set, a clone of the entry is created with an entry key
clonekey. Obviously this may cause labelling problems in author/year styles
etc. and should be used with care. The cloned entry is in-scope during the
processing of the current entry and can be modified by passing its key as the
value to map_entrytarget. In clonekey, you may use standard Perl regular
expression backreferences to captures from a previous map_match step.
• Change the source-entrytype to target-entrytype, if defined. If map_final
is set then if the entrytype of the entry is not source-entrytype, processing
of this map immediately terminates.
• Change the source-field to target-field, if defined. If map_final is set,
then if there is no source-field field in the entry, processing of this map
immediately terminates.
• If map_notfield is used then only apply the step if the source-field does
not exist.
• If map_match is defined but map_replace is not, only apply the step if the
source-field matches map_match. You can use parentheses as usual to cap-
ture parts of the match and can then use these later when setting a map_field_value.
• map_notmatch is the same as map_match but with the logic reversed.
• Perform a Perl regular expression match and replace on the value of source-field
if map_match and map_replace are defined. You may use (and almost certainly
will want to use) parentheses for back-references in map_replace. Do not quote
the regular expressions in any special (i.e. non-Perly) way—it’s not necessary.
• If map_field_set is defined, then its value is set-field which will be set to
a value specified by further attributes. If map_overwrite is false for this step
and the field to set already exists then the map step is ignored. If map_final
is also set on this step, then processing of the parent map stops at this point.
If map_append is set, then the value to set is appended to the current value of
set-field. map_appendstrict appends only if the set-field is not empty.
The value to set is specified by a mandatory one and only one of the following
attributes:
○ map_field_value — The set-field is set to set-value
○ map_null — The field is ignored, as if it did not exist in the datasource
○ map_origentrytype — The set-field is set to the most recently men-
tioned source-entrytype name.
18
○ map_origfield — The set-field is set to the most recently mentioned
source-field name
○ map_origfieldval — The set-field is set to the most recently men-
tioned source-field value
With BibTeX datasources, you can specify the pseudo-field ‘entrykey’ for source-field
which is the citation key of the entry. Naturally, this ‘field’ cannot be changed (used
as set-field, target-field or changed using map_replace).
Note that for XML datasources like BibLaTeXML, the names of fields and entry-
types are matched in a case sensitive manner. For all other datasource types, entry-
type and field name matching is case insensitive.
Here are some examples:
<map>
<per_datasource>example1.bib</per_datasource>
<per_datasource>example2.bib</per_datasource>
<map_step map_field_set="KEYWORDS" map_field_value="keyw1, keyw2"/>
<map_step map_field_source="ENTRYKEY"/>
<map_step map_field_set="NOTE" map_origfieldval="1"/>
</map>
This would add a KEYWORDS field with value ‘keyw1, keyw2’ and set the NOTE
field to citation key for the entry to all entries which are found in either the
examples1.bib or examples2.bib files. This assumes that the Biblatex source
contains \addresource{example1.bib} and \addresource{example2.bib}.
<map map_overwrite="0">
<map_step map_field_source="TITLE"/>
<map_step map_field_set="NOTE" map_origfieldval="1"/>
</map>
Copy the TITLE field to the NOTE field unless the NOTE field already exists.
<map_step map_field_source="AUTHOR" />
<map_step map_field_set="SORTNAME" map_origfieldval="1" map_final="1"/>
<map_step map_field_source="SORTNAME" map_match="\A(.+?)\s+and.∗" map_replace="$1"/>
</map>
For any entry with an AUTHOR field, try to set SORTNAME to the same as AUTHOR.
If this fails because SORTNAME already exists, stop, otherwise truncate SORTNAME to
just the first name in the name list.
<map_step map_type_source="CHAT" map_type_target="CUSTOMA" map_final="1"/>
<map_step map_field_set="TYPE" map_origentrytype="1"/>
</map>
19
Any @CHAT entrytypes would become @CUSTOMA entrytypes and would automatically
have a TYPE field set to ‘CHAT’ unless the TYPE field already exists in the entry
(because map_overwrite is false). This mapping applies only to entries of type
@CHAT since the first step has map_final set and so if the map_type_source does
not match the entry, processing of this map immediately terminates.
<map>
<per_datasource>examples.bib</per_datasource>
<per_type>ARTICLE</per_type>
<per_type>BOOK</per_type>
<map_step map_field_set="ABSTRACT" map_null="1"/>
<map_step map_field_set="NOTE" map_field_value="Auto-created this field"/>
</map>
Any entries of entrytype ARTICLE or BOOK from the ‘examples.bib’ datasource would
have their ABSTRACT fields removed and a NOTE field added with value ‘Auto-created
this field’.
<map>
<map_step map_field_set="ABSTRACT" map_null="1"/>
<map_step map_field_source="CONDUCTOR" map_field_target="NAMEA"/>
<map_step map_field_source="GPS" map_field_target="USERA"/>
</map>
This removes ABSTRACT fields from any entry, changes CONDUCTOR fields to NAMEA
fields and changes GPS fields to USERA fields
<map>
<map_step map_field_source="PUBMEDID"
map_field_target="EPRINT"
map_final="1"/>
<map_step map_field_set="EPRINTTYPE" map_origfield="1"/>
<map_step map_field_set="USERD"
map_field_value="Some string of things"/>
</map>
Applies only to entries with PUBMED fields and maps PUBMEDID fields to EPRINT fields,
sets the EPRINTTYPE field to ‘PUBMEDID’ and also sets the USERD field to the string
‘Some string of things’.
<map>
<map_step map_field_source="SERIES"
map_match="\A\d∗(.+)"
map_replace="\L$1"/>
</map>
20
Here, the contents of the SERIES field have leading numbers stripped and the re-
mainder of the contents lowercased.
<map>
<map_step map_field_source="TITLE"
map_match="Collected\s+Works.+Freud"
map_final="1"/>
<map_step map_field_set="KEYWORDS" map_field_value="freud"/>
</map>
Here, if for an entry, the TITLE field matches a particular regular expression, we set
a special keyword so we can, for example, make a references section just for certain
items.
<map>
<map_step map_field_source="LISTA" map_match="regexp" map_final="1"/>
<map_step map_field_set="LISTA" map_null="1"/>
</map>
If an entry has a LISTA field which matches regular expression ‘regexp’, then it is
removed.
<map>
<map_step map_field_source="AUTHOR"
map_match="Smith, Bill" map_replace="Smith, William"/>
map_match="Jones, Baz" map_replace="Jones, Barry"/>
</map>
Here, we use multiple match/replace for the same field to regularise some inconstant
name variants. Bear in mind that match/replace processing within a map element is
sequential and the changes from a previous match/replace are already committed.
<map_step map_field_source="AUTHOR" map_match="Doe," map_final="1"/>
<map_step map_field_set="SHORTAUTHOR" map_origfieldval="1"/>
<map_step map_field_set="SORTNAME" map_origfieldval="1"/>
<map_step map_field_source="SHORTAUTHOR"
map_match="Doe,\s*J(?:\.|ohn)(?:[-]*)(?:P\.|Paul)*"
map_replace="Doe, John Paul"/>
<map_step map_field_source="SORTNAME"
</map>
Only applies to entries with an AUTHOR field matching ‘Doe,’. First the AUTHOR field
is copied to both the SHORTAUTHOR and SORTNAME fields, overwriting them if they
21
already exist. Then, these two new fields are modified to canonicalise a particular
name, which presumably has some variants in the datasource.
<map>
<map_step map_field_source="TITLE" map_match="A Title" map_final="1"/>
<map_step map_entry_null="1"/>
</map>
Any entry with a TITLE field matching ‘A Title’ will be completely ignored.
Other datasource types

For datasources other than BIBTEX, (e.g. biblatexml), the source entrytypes and
fields are usually very differently modelled and named.
Here we use a loop to apply a regular expression replacement to several fields:
<maps datatype="bibtex" level="user">

<map map_overwrite="1" map_foreach="author,editor,translator">
<map_step map_field_source="$MAPLOOP" map_match="Smith" map_replace="Jones"/>
</map>
</maps>
3.1.3. The inheritance option

The inheritance option defines the inheritance rules for data inheritance between
entries using, for example, BibTeX’s CROSSREF field. The default setup for this is
defined by Biblatex and is passed in the .bcf file. Defining inheritance rules in the
Biber configuration file is rarely something you would want to do with one notably
exceptional case being when using Biber in tool mode where you might want to
‘materialise’ special inheritance rules (see section 3.13). Here we define the format
of the config file inheritance section, should you need to understand or modify it.
Items in red are not literal, they are descriptive meta-values which are explained in
the accompanying text. Items in blue are optional within their parent section or
element.
<inheritance>
<defaults inherit_all="true|false" override_target="true|false">
<type_pair source="source" target="target"
inherit_all="true|false"
override_target="true|false"/>
⋮
</defaults>
<inherit>
<type_pair source="source" target="target"/>
22
⋮
<field source="source"
target="target"
skip="true|false"
override_target="true|false"/>
⋮
</inherit>
⋮
</inheritance>
• The defaults section specifies the default inheritance rules which are not
otherwise covered by a specific inherit rule. inherit_all specifies that by
default a target inherits all fields from a source. override_target specifies
that by default an existing target field will be overwritten by a source field
it already exists in the target. A type_pair element specifies the defaults
for a particular source and target entrytype combination. source or target
can take the value ‘*’ which is a wildcard representing all possible entrytypes.
• An inherit element specifies how one or more source fields are inherited by
one more source/target pairs which are specified in one or more type_pair
elements within the same inherit element. override_target can be specified
on a per-field basis as can the skip attribute which indicates that a particular
field is not to be inherited by the target.
Here is an example:
<inheritance>
<defaults inherit_all="true" override_target="false">
</defaults>
<inherit>
<type_pair source="mvbook" target="inbook"/>
<type_pair source="mvbook" target="bookinbook"/>
<type_pair source="mvbook" target="suppbook"/>
<type_pair source="book" target="inbook"/>
<type_pair source="book" target="bookinbook"/>
<type_pair source="book" target="suppbook"/>
<field source="author" target="author"/>
<field source="author" target="bookauthor"/>
</inherit>
<inherit>
<type_pair source="*" target="inbook"/>
<type_pair source="*" target="incollection"/>
<field source="*" skip="true"/>
</inherit>
</inheritance>
23
Here we can see that the default is to inherit all fields from the source and not
to override existing target fields if they already exist. Then we see that for some
combinations of sources and targets, the AUTHOR field is inherited from the source
and also the AUTHOR field in the source is inherited as the BOOKAUTHOR field in the
target.
The second inherit element says that INBOOK and INCOLLECTION entries never
inherit the INTRODUCTION field from any source.
In general, it is probably best to copy the default Biblatex inheritance rules and
modify them to your needs. See section 3.13.
3.1.4. The noinit option

The value of the noinit option can only be set in the config file and not on the
command line. This is because the values are Perl regular expressions and would
need special quoting to set on the command line. This can get a bit tricky on
some OSes (like Windows) so it’s safer to set them in the config file. noinit allows
you to ignore parts of a name when generating initials. This is done using Perl
regular expressions which specify what to ignore. You can specific multiple regular
expressions and they will be removed from the name before it is passed to the initials
generating system.
For example, this option can be used to ignore diacritic marks and prefices in
names which should not be considered when sorting. Given (the default):
<noinit>

<option value="\b\p{Ll}{2}\p{Pd}(?=\S)"/>

<option value="[\x{2bf}\x{2018}]"/>
</noinit>
and the BIBTEX datasource entry:
AUTHOR = {{al-Hasan}, ʿAlī},
the initials for the last name will be ‘H’ and not ‘a-H’. The initial for the first name
will be ‘A’ as the diacritic is also ignored. This is tricky in general as you cannot often
determine the difference between a name with a prefix and a hyphenated name with
only, say, two chars in the first part such as ‘Ho-Pun’. You can adjust this option
for your individual cases. By default, only lowercased prefices are looked for so as
to avoid breaking things like ‘Ho-Pun’ where you want the initials to be ‘H.-P.’,
for example. See the Perl regular expression manual page for details of the regular
expression syntax10 .
10
http://perldoc.perl.org/perlre.html
24
3.1.5. The nolabel option
The value of the nolabel option can only be set in the config file and not on the
need special quoting to set on the command line. This can get a bit tricky on some
OSes (like Windows) so it’s safer to set them in the config file. nolabel allows
you to ignore elements of a field when generating labels. This is done using Perl
regular expressions which specify what to ignore. You can specific multiple regular
expressions and they will be removed from a field before it is passed to the label
generating system.
For example, this option can be used to ignore control, punctuation, symbol and
separator characters when generation labels. Given (the default):
<nolabel>

<option value="[\p{P}\p{S}\p{C}]+"/>
</nolabel>
and the BIBTEX datasource entry with default label generation definition (see Bib-
latex documentation for \DeclareLabelalphaTemplate):
AUTHOR = {O'Toole, Alexander},
Then the label for the name will be «OTo07» as the apostrophe is ignored by the
label generation routine. See the Perl regular expression manual page for details of
the regular expression syntax11 .
3.1.6. The nolabelwidthcount option

The value of the nolabelwidthcount option can only be set in the config file and
not on the command line. This is because the values are Perl regular expressions
and would need special quoting to set on the command line. This can get a bit
tricky on some OSes (like Windows) so it’s safer to set them in the config file.
nolabelwidthcount allows you to ignore elements of a field when generating fixed-
width substrings of labels. This is done using Perl regular expressions which specify
what to ignore. You can specific multiple regular expressions and they will be
removed from a field before it is passed to the label generating system.
For example, this option can be used to ignore punctuation characters when gen-
erating substrings for labels. Note that in this example we reset nolabel because
by default this removes punctuation characters. Given:
<nolabel>
11
25
<option value=""/>
</nolabel>
<nolabelwidthcount>
<option value="\p{P}+"/>
</nolabelwidthcount>
and the BIBTEX datasource entry with default label generation definition (see Bib-
latex documentation for \DeclareLabelalphaTemplate):
AUTHOR = {O'Toole, Alexander},
Then the label for the name will be «O’To07» as the apostrophe is ignored by the
substring generation routine. See the Perl regular expression manual page for details
of the regular expression syntax12 .
3.1.7. The sorting option

The sorting option defines the sorting rules for the bibliography lists. Biblatex al-
lows multiple sorting specifications referenced by name as it can print bibliography
information as many times as the user wishes with different filtering and sorting.
This is normally handled by macros in Biblatex which write the XML sorting spe-
cification(s) to the .bcf file for Biber to read but there may be occasions (usually
when using Biber in ‘tool’ mode (see section3.13) when you need to specify the
global sorting specification directly in a Biber config file. This section documents
the XML format of the sorting specification. Items in red are not literal, they are
descriptive meta-values which are explained in the accompanying text. Items in blue
are optional within their parent section or element. See also the nosort option in
section 3.1.8.
<sortingtemplate name="schemename">
<presort type=type>string</presort>
<sortexclusion type=type>
<exclusion>field</exclusion>
⋮
</sortexclusion>
<sort order="n"
final=1
sort_direction="ascending|descending"
sort_case="1|0"
sort_upper="1|0">
<sortitem order="m"
substring_side="left|right"
12
26
substring_width="int"
pad_side="left|right"
pad_width="int"
pad_char="string">field|literal|citeorder</sortitem>
⋮
</sort>
⋮
</sortingtemplate>
Sorting in Biber is a sophisticated procedure of building up a sorting object for

an entry based on a sorting scheme template, the general form of which is shown
above. The sorting routine first traverses every entry in the bibliography list and
generates a sorting object based on the sorting scheme. When this is done, it sorts
the entries according to the sorting objects it has generated for each entry.
A sorting specification must be named with the schemename attribute. In ‘tool’
mode, this must be set to tool. Otherwise, it is a name referenced by a Biblatex
refcontext sorting option. A sorting specification is comprised of a number of sort
elements. Sorting is essentially a process of comparing whatever information is in
the nth sort element collected for every entry (otherwise known as ‘multi-field’
sorting). Within a sort element, there can be any number of sortitem elements
which describe what information to look for in the entry in order to construct this
part of the sorting object; either a field, a literal string or the special ‘citeorder’
pseudo-field.
When generating the sorting information for an entry, within each sort element,
the first sortitem to return a non-empty value for the bibliography entry is used
and the rest of the sortitems in the sort are skipped. A sortitem essentially looks
for a piece of information in the entry and adds this to the sorting object. If it
is looking for a field, then the field must exist in the entry. If it does not, the
sortitem is skipped. If the field does exist, it is added to the sorting object for
the entry after being modified by the attributes of the sortitem.
Once a sortitem has returned the contents of a field, you can use the sub-
string_side (default ‘left’ if any other substring attributes are set) and substring_width
(default ‘4’ if any other substring attributes are set) attributes to truncate the con-
tents of the field by reducing it to a substring taken from the left or right side
of the string and of a number of (UTF-8) characters of your choice. You can also
pad the field with repeated arbitrary characters on either side using the pad_side
(default ‘left’ if any other pad attributes are set), pad_width (default ‘4’ if any other
pad attributes are set) and pad_char (default ‘0’—the digit zero if any other pad
attributes are set) attributes.
A sortitem which is neither a known bibliography sorting field nor the special
‘citeorder’ string is treated as a literal string to add to the sorting object. Naturally,
such a sortitem always ‘finds’ something to add to the sorting object and so it
should never have any other sortitems after it within the sort section as they
27
will never be considered. The ‘citeorder’ sortitem value has a special meaning. It
requests a sort based on the lexical order of the actual citations. For entries cited
in Biblatex within the same citation command like:
\cite{one,two}
there is a distinction between the lexical order and the semantic order. Here ‘one’
and ‘two’ have the same semantic order but a unique lexical order. The semantic
order only matters if you specify further sorting to disambiguate entries with the
same semantic order. For example, this is the definition of the Biblatex none sorting
scheme:
<sortingtemplate>
<presort>mm</presort>
<sort order="1">
<sortitem order="1">citeorder</sortitem>
</sort>
</sortingtemplate>
This sorts the bibliography purely lexically by the order of the keys in the citation
commands. In the example above, it sorts entry ‘one’ before ‘two’. However, suppose
that you consider ‘one’ and ‘two’ to have the same order (semantic order) since they
are cited at the same time and want to further sort these by year. Suppose ‘two’
has an earlier YEAR than ‘one’:
<sortingtemplate>
<sort order="1">
<sortitem order="1">citeorder</sortitem>
</sort>
<sort order="2">
<sortitem order="1">year</sortitem>
</sort>
</sortingtemplate>
This sorts ‘two’ before ‘one’, even though lexically, ‘one’ would sort before ‘two’.
This is possible because the semantic order can be disambiguated by the further
sorting on year. With the standard Biblatex none sorting scheme, the lexical order
and semantic order are identical because there is nothing further to disambiguate
them. This means that you can use ‘citeorder’ just like any other sortitem value,
choosing how to further sort entries cited at the same time (in the same citation
command).
Both sort and sortitem elements have a mandatory order attribute which should
start at ‘1’ and increase for each further element. Order numbers for sortitem
28
elements within a sort element always begin with ‘1’ and don’t increase between
sort elements.
Once a sortitem element has added something to the sorting object (or all
sortitem elements within a sort have been processed, regardless of whether any-
thing was added to the sort object for the entry), some attributes are applied to
the information added and the next sort element is processed. These attributes
on the sort element further determine how any sorting specification added by the
sortitem elements will be used in the sorting.
If the sort element has the final attribute set to ‘1’, then if any sortitem within
the sort returned a non-empty string to add to the sorting object, the construction
of the sorting object for the entry ceases at this point and no more sort elements
are processed. This is used typically to make sure that master sorting keys such as
those specified with the SORTKEY field, if found, are the only thing used to construct
the sorting object. The sort element may further specify that the information at
order ‘n’ should be sorted in ascending order or descending order (default ‘ascend-
ing’), whether case should be considered when sorting (default depends on the Biber
sortcase option which defaults to true) and whether uppercase characters should be
sorted before lower (default depends on the Biber ‘sortupper’ option which defaults
to true).
Finally, there are two special sorting section elements to consider. The presort
element is mandatory and specifies a literal string to add to the very beginning of
all sorting objects for all entries. This is useful when combined with the fact that
you may specify an optional type attribute which specifies a particular entry type
for the presort string specified. Using this mechanism, you can sort, for example, all
ARTICLE entries before all BOOK entries and then all other types of entry:
<sortingtemplate>
<presort type="article">aa</presort>
<presort type="book">bb</presort>
⋮
</sortingtemplate>
This makes it easy to divide a bibliography by type of entry.

The optional sortexclusion element allows you to exclude fields from consider-
ation by sortitem on a per-type basis. For example, if you wanted to ignore the
YEAR field of any REPORT entry types because they are not reliably populated with
data representing a year, you could do:
<sortingtemplate>
⋮
<sortexclusion type="report">year</sortexclusion>
⋮
</sortingtemplate>
29
It is much easier to see how intuitive this all is if you look at a standard sorting
scheme definition. Below is the default Biblatex sorting scheme which appears in
the .bcf when you run Biblatex with no sorting option. This is fully documented
and described in the Biblatex manual along with the LaTeX macros which generate
this XML in the .bcf:
<sortingtemplate>
<sort order="1">
<sortitem order="1">presort</sortitem>
</sort>
<sort order="2" final="1">
<sortitem order="1">sortkey</sortitem>
</sort>
<sort order="3">
<sortitem order="1">sortname</sortitem>
<sortitem order="2">author</sortitem>
<sortitem order="3">editor</sortitem>
<sortitem order="4">translator</sortitem>
<sortitem order="5">sorttitle</sortitem>
<sortitem order="6">title</sortitem>
</sort>
<sort order="4">
<sortitem order="1">sortyear</sortitem>
<sortitem order="2">year</sortitem>
</sort>
<sort order="5">
<sortitem order="1">sorttitle</sortitem>
<sortitem order="2">title</sortitem>
</sort>
<sort order="6">
<sortitem order="1" pad_side="left" pad_width="4" pad_char="0">volume</sortitem>
<sortitem order="2">0000</sortitem>
</sort>
</sortingtemplate>
3.1.8. The nosort option

The value of the nosort option can only be set in the config file and not on the
need special quoting to set on the command line. This can get a bit tricky on some
OSes (like Windows) so it’s safer to set them in the config file. In any case, it’s
unlikely you would want to set them for particular Biber runs; they would more
likely be set as your personal default and thus they would naturally be set in the
30
Set Fields
setnames author
afterword
annotator
bookauthor
commentator
editor
editora
editorb
editorc
foreword
holder
introduction
namea
nameb
namec
shortauthor
shorteditor
translator
settitles booktitle
eventtitle
issuetitle
journaltitle
maintitle
origtitle
title
Table 3: Default Biblatex datafield sets
config file anyway. nosort allows you to ignore parts of a field for sorting. This
is done using Perl regular expressions which specify what to ignore in a field. You
can specify as many patterns as you like for a specific field. Datasource field sets
defined using \DeclareDatafieldSet in Biblatex are also recognised as valid values
and so it is possible to specify nosort regular expressions for arbitrary sets of fields.
Biblatex defines as standard two sets as shown in Table 3.
For example, this option can be used to ignore some diacritic marks and prefices
in names which should not be considered when sorting. Given (the default):
<nosort>

<option name="setnames" value="\A\p{L}{2}\p{Pd}(?=\S)"/>

<option name="setnames" value="[\x{2bf}\x{2018}]"/>
</nosort>
31
and the BIBTEX datasource entry:
AUTHOR = {{al-Hasan}, ʿAlī},
the prefix ‘al-’ and the diacritic ‘ʿ’ will not be considered when sorting. See the Perl
regular expression manual page for details of the regular expression syntax13 .
You may specify any number of option elements. If a nosort option is found for
a specific field, it will override any option for a type which also covers that field.
Here is another example. Suppose you wanted to ignore ‘The’ at the beginning of
a TITLE field when sorting, you could add this to your biber.conf:
<nosort>
<option name="title" value="\AThe\s+"/>
</nosort>
If you wanted to do this for all title fields listed in Table 3, then you would do this:
<nosort>
<option name="settitles" value="\AThe\s+"/>
</nosort>
Note: nosort can be specified for most fields but not for things like dates and
special fields as that wouldn’t make much sense.
3.1.9. The collate-options option

The collate-options option has format similar to nosort. See Section 3.5 for
details about the option, here is an example of a config file setting:
<collate_options>
<option name="level" value="3"/>
<option name="table" value="/home/user/data/otherkeys.txt"/>
</collate_options>
3.2. Unicode
Biber uses NFD UTF-8 internally. All data is converted to NFD UTF-8 when read.
If UTF-8 output is requested (to .bbl for example), the UTF-8 will always be NFC.
13
32
3.3. Input/Output File Locations
3.3.1. Control file
The control file is normally passed as the only argument to Biber. It is searched for
in the following locations, in decreasing order of priority:
Absolute filename →
In the --input-directory, if specified→
In the --output-directory, if specified→
Relative to current directory→
Using kpsewhich, if available
3.3.2. Data sources

Local datasources of type ‘file’ are searched for in the following locations, in decreas-
ing order of priority:
Absolute filename →
In the --input-directory, if specified→
In the --output-directory, if specified→
Relative to current directory→
In the same directory as the control file→
Using kpsewhich for supported formats, if available
Remote file datasources (beginning with http:// or ftp://) are retrieved to a

temp file and processed as normal. Users do not specify explicitly the bibliography
database files; they are passed in the .bcf control file, which is constructed from
the Biblatex ‘\addbibresource{}’ macros.
3.4. Logfile
By default, the logfile for Biber will be named \jobname.blg, so, if you run
biber <options> test.bcf
then the logfile will be called ‘test.blg’. Like the .bbl output file, it will be created
in the --output-directory|-c, if this option is defined. You can override the logfile
name by using the --logfile option:
biber --logfile=lfname test.bcf
33
results in a logfile called ‘lfname.blg’.
Warning: be careful if you are expecting Biber to write to directories which you
don’t have appropriate permissions to. This is more commonly an issue on non-
Windows OSes. For example, if you rely on kpsewhich to find your database files
which are in system TEX directories, you may well not have write permission there
so Biber will not be able to write the .bbl. Use the --output-file|-O option to
specify the location to write the .bbl to in such cases.
3.5. Collation and Localisation

Biber takes care of collating the bibliography for Biblatex. It writes entries to the
.bbl file sorted by a completely customisable set of rules which are passed in the
.bcf file by Biblatex. Biber uses the Perl Unicode::Collate module for collation
which implements the full UCA (Unicode Collation Algorithm). It also has CLDR
(Common Locale Data Repository) tailoring to deal with cases which are not covered
by the UCA.
The locale used for collating a particular field in the bibliography is determined
by the following resource chain which is given in decreasing precedence order:
--collate-options|-c (e.g. -c 'locale => "de_DE"') →

--sortlocale|-l →
Biblatex per-sortset locale option →
Biblatex per-sortscheme locale option→
Biblatex global sortlocale option
The locale will be used to look for a collation tailoring for that locale. It will generate
an informational warning if it finds none. This is not a problem as most standard
collation cases are covered by the standard UCA and many locales neither have nor
need any special collation tailoring.
Biblatex passes sortscheme-specific sorting locales and its global sorting locale in
the .bcf. Biber uses these locales automatically to tailor sorting at various levels
of granularity (see Biblatex docs for the \DeclareSortingScheme macro). Biblatex
can be configured to automatically pass as locale the language setting from Babel
or Polyglossia in which case Biber tries to match this to a sensible locale. See
the Appendix, section A.1 for the mapping. If you want to sort using a specific
locale not listed in A.1, specify this locale exactly in your LaTeX source as the
Biblatex sortlocale option, as the optional argument to \DeclareSortingScheme
macro or as an optional argument to the Biblatex \sort macro according to the
desired granularity. For example, if you want to use traditional Spanish for sorting a
reference list, you need to specify es_ES_trad directly instead of using the ‘spanish’
34
string because the Babel/Polyglossia ‘spanish’ language identifier by default maps
to the modern es_ES locale (which doesn’t include sort tailoring for ‘ch’ in Spanish).
Collation is by default case sensitive. You can turn this off globally using the Biber
option --sortcase=false or from Biblatex using its option
sortcase=false. The option can also be defined per-field so you can sort some
fields case sensitively and others case insensitively. See the Biblatex manual.
By default, Biber collates uppercase before lower. You can reverse this globally for
all sorting using the Biber option --sortupper=false or from
Biblatex by using its option sortupper=false. The option can also be defined
per-field so you can sort some fields uppercase before lower and others lower before
upper. See the Biblatex manual. Be aware though that some locales rightly enforce
a particular setting for this (for example, Danish). You will be able to override it
but Biber will warn you if you do.
There are in fact many options to Unicode::Collate which can tailor the collation
in various ways in addition to the locale tailoring which is automatically performed.
Users should see the the documentation to the module for the various options, most
of which the vast majority of users will never need14 . Options are passed using the
--collate-options|-c option as a single quoted string, each option separated by
comma, each key and value separated by ‘=>’. See examples.
Note: Biber sets the Unicode collation option ‘variable’ to ‘non-ignorable’. Ef-
fectively, this means that punctuation is not ignored when sorting. The default
setting is to ignore such ‘variable weight’ elements. Sorting bibliographies is slightly
more specialised than collating general text and punctuation often matters. In case
you want the UCA default behaviour, see examples. Since Biber always normalises
into NFD when reading data in, no normalisation is requested with Unicode collation
(‘normalization’ option is set to ‘prenormalized’ by default) as this saves some time.
3.5.1. Examples
biber
Call Biber using all settings from the .bcf generated from the LaTeX run. Case
sensitive UCA sorting is performed taking the locale for tailoring from the .bcf if
Biber’s sortlocale option is not used to override the .bcf
biber --sortlocale=de_DE_phonebook
Override any locale setting in the .bcf
biber --sortcase=false
Case insensitive sorting.
biber --sortupper=false --collate-options="backwards => 2"
Collate lowercase before upper and collate French accents in reverse order at UCA
level 2.
14
For details on the various options, see http://search.cpan.org/search?query=Unicode%3A%
3ACollate&mode=all
35
biber --collate-options="variable => 'shifted'"
Use the UCA default setting for variable weight punctuation (which is to ignore it
for sorting, effectively).
3.6. Encoding of files

Biber takes care of re-encoding the datasource data as necessary. In normal use,
Biblatex passes its bibencoding option value to Biber via the .bcf file and this
corresponds to the Biber --input-encoding|e option. Biblatex also passes the
value of its texencoding option (which maps to Biber’s --output-encoding|-E
option) the default value of which depends on which TEX engine and encoding
packages you are using (see Biblatex manual for details).
Biber performs the following tasks:
1. Decodes the datasource into UTF-8 if it is not UTF-8 already

2. Decodes LaTeX character macros in the datasource into UTF-8
3. Encodes the output so that the .bbl is in the encoding that --output-encoding|-E
specifies
4. Warns if it is asked to output to the .bbl any UTF-8 decoded LaTeX character
macros which are not in the --output-encoding|-E encoding. Replaces with
a suitable LaTeX macro
Normally, you do not need to set the encoding options on the Biber command line
as they are passed in the .bcf via the information in your Biblatex environment.
However, you can override the .bcf settings with the command line. The resource
chain for encoding settings is, in decreasing order of preference:
--input-encoding|-e and --output-encoding|-E →

Biber config file →
.bcf control file
3.6.1. LaTeX macro decoding

As mentioned above, Biber always converts as much as possible, including LaTeX
character macros, into UTF-8. This is important for two main reasons. Firstly, this
allows you to have, for example:
@BOOK{key1,
Author = {\"{O}leg Smith}
}
@BOOK{key2,
Author = {Öleg Smith}
}
36
Here, because Biber decodes the macros into UTF-8, it knows that both books are
by the same author because it’s clear that the names are now the same. Secondly, this
allows Biber to output normalised latex macros when a user selects --output-encoding=ascii
etc. This means that the many Biblatex comparison macros used in styles can deal
with comparisons of fields containing macros reliably. The macro to UTF-8 con-
version uses the decoding set specified with the --decodecharsset, see below. To
disable all macro to UTF-8 conversion, you can specify the virtual ‘null’ set as a
value for --decodecharsset or output-safecharsset. This effectively turns off
macro to UTF-8 decoding or encoding respectively.
If you are using PDFLaTeX and \usepackage[utf8]{inputenc}, it is possible
that the UTF-8 characters resulting from Biber’s internal LaTeX character macro
decoding break inputenc. This is because inputenc does not implement all of
UTF-8, only a commonly used subset.
An example–if you had \DJ in your .bib datasource, Biber decodes this correctly
to ‘Đ’ and this breaks inputenc because it doesn’t understand that UTF-8 character.
The real solution here is to switch to a TEX engine with full UTF-8 support like
XETEX or LuaTEX as these don’t use or need inputenc. However, you can also try
the --output-safechars option which will try to convert any UTF-8 chars into
LaTeX macros on output. For information on the --output-safechars option, see
section 3.6.2.
3.6.2. LaTeX macro encoding

The opposite of decoding; converting UTF-8 characters into LaTeX macros. You can
force this with the --output-safechars option which will do a generally good job
of making your .bbl plain ASCII. It can be useful in certain edge cases where your
bibliography introduces characters which can’t be handled by your main document.
See section 3.6.1 above for an example such case.
A common use case for LaTeX macro encoding is when the bibliography datasource
is not ASCII but the .tex file is and so this case is automated for you: if the Biblatex
option ‘texencoding’ (which corresponds to the Biber option ‘--output-encoding|-E’)
is set to an ASCII encoding (‘ascii’ or ‘x-ascii’) and ‘--input-encoding|-e’ is
not ASCII, Biber will automatically set --output-safechars.
Since Biber always decodes into UTF-8 internally, if the --output-encoding|-E
option is not set to UTF-8, Biber will automatically replace any characters which
will not encode in the output encoding with equivalent TeX macros. You will also
receive a warning about this.
See also the biber --help output for the --output-safecharsset and
--decodecharsset options which can customise the set of conversion rules to use.
37
The builtin sets of characters and macros which Biber maps during encoding and
decoding are documented15 .
It is possible to provide a customised encode/decode mapping file using the --recodedata
option. It must adhere to the format of the default data file for reencoding which
is recode_data.xml located in the same Perl install directory as Biber’s Recode.pm
module. Of course it is easier to find this in the Biber source tree. It is most likely
that if you want to use a custom mapping file, you would copy the default file and
edit it, removing some things and perhaps defining some custom recoding sets for
use with --output-safecharsset and --decodecharsset.
Be careful to classify the entries using the correct ‘type’ attribute in the XML file
as this determines how the macro is treated by the code that does the replacement.
Just copy a similar type of macro from the default recoding data file if you are adding
new entries, which is unlikely as the file is quite comprehensive. There is only one
other thing to note. The ‘preferred’ attribute tells Biber to use a specific LaTeX
macro when mapping from UTF-8, just in case there is more than one mapping
from UTF-8 for a particular character. It’s unlikely you will need to use this.
3.6.3. Examples
biber
Set input-encoding and output-encoding from the config file or .bcf
biber --output-encoding=latin2
Encode the .bbl as latin2, overriding the .bcf
biber --output-safechars
Set input-encoding and output-encoding from the config file or .bcf. Force
encoding of UTF-8 chars to LaTeX macros using default conversion set
biber --output-encoding=ascii
Encode the .bbl as ascii, overriding the .bcf. Automatically sets --output-safechars
to force UTF-8 to LaTeX macro conversion
biber --output-encoding=ascii --output-safecharsset=full
Encode the .bbl as ascii, overriding the .bcf. Automatically sets --output-safechars
to force UTF-8 to LaTeX macro conversion using the full set of conversions
biber --decodecharsset=full
Set input-encoding and output-encoding from the config file or .bcf. Use the
full LaTeX macro to UTF-8 conversion set because you have some more obscure
character macros in your .bib datasource which you want to sort correctly
biber --recodedata=/tmp/recode.xml --decodecharsset=special
Specify a user-defined reencoding data file which defines a new reencoding set ‘spe-
cial’ and use this set for decoding LaTeX macros to UTF-8.
15
https://sourceforge.net/projects/biblatex-biber/files/biblatex-biber/2.13/
documentation/utf8-macro-map.html
38
biber -u
Shortcut alias for biber --input-encoding=UTF-8
biber -U
Shortcut alias for biber --output-encoding=UTF-8
3.7. List and Name Separators

With traditional BibTeX, the name and list field separator ‘and’ is hard-coded. The
btparse C library and therefore Biber allows the use of any fixed string, subject to
the same rules as ‘and’ (not at the beginning or end of the name/list, whitespace
must surround the string etc.). This is settable using the options listsep and
namesep, both of which default to the usual ‘and’. You can also change the default
final name in a list which implies ‘et al’. In BibTeX, this is by default the English
‘others’ which is the Biber default also. Don’t try to put any whitespace in these
strings, this is ignored by btparse anyway. Perhaps you prefer your .bib in more
obvious German—set --namesep=und and --others-string=andere and then you
can do:
@BOOK{key,
AUTHOR = {Hans Harman und Barbara Blaupunkt und andere},
}
Bear in mind that these are global settings and apply to all entries in the BibTeX
data format read by Biber. Also bear in mind that this is very unportable as all
BibTeX input/output programs rely on the hard-coded ‘and’ and ‘others’. Hopefully
this will change as these two hard-coded English terms look really rather bad in the
context of multilingual bibliographies.
3.8. Extended Name Format

The parsing rules for BibTeX names are rather archaic and not suited to many
international name formats. Biber supports an extended name format which allows
explicit specification of the parts of names. This allows the use of custom name
parts apart from the four standard BibTeX parts. Extended name formats are
supported in all name fields and can be used along with the usual BibTeX name
format. Recognition of extended name format can be disabled with the Biber option
--noxname in case you do not need the extended format and the auto-detection
causes problems with normal name parsing. The separator = which comes between
the namepart names and values is customisable with the Biber option --xnamesep.
Here is an example:
AUTHOR = {Hans Harman and Simon de Beumont}
AUTHOR = {given=Hans, family=Harman and given=Simon, prefix=de, family=Beumont}
39
These two name specifications are equivalent but the extended format explicitly
names the parts. The supported parts are those specified by the Biblatex data
mode constant nameparts, the default value of which is:
\DeclareDatamodelConstant[type=list]{nameparts}{prefix,family,suffix,given}
As with traditional BibTeX name parsing, initials are automatically generated but
it is also possible to specify these explicitly:
AUTHOR = {given=Jean, prefix=de la, prefix-i=d, family=Rousse}
AUTHOR = {given={Jean Pierre Simon}, given-i=JPS}
Initials are specified by adding the suffix -i to the namepart name. Compound parts
may be protected with braces:
AUTHOR = {given={Jean Pierre}}
If a namepart contains a comma, the whole namepart should be protected with
quotes:
AUTHOR = {"family={Robert and Sons, Inc.}"}
Traditional BibTeX name formats and the extended form may be used together:
AUTHOR = {Hans Harman and given=Simon, prefix=de, family=Beumont}
Per-namelist and per-name options may be specified in the extended name format:
AUTHOR = {namelistopt=true and Hans Harman and
given=Simon, family=Beumont, nameopt=true}
3.9. Editor Integration

Visit http://tex.stackexchange.com/questions/154751/ for a comprehensive
overview on Biber integration in most editors.
3.10. BibTEX macros and the MONTH field

BIBTEX defines macros for month abbreviations like ‘jan’, ‘feb’ etc. Biber also does
this, defining them as numbers since that is what Biblatex wants. In case you are
also defining these yourself (although if you are only using Biblatex, there isn’t much
point), you will get macro redefinition warnings from the btparse library. You can
turn off Biber’s macro definitions to avoid this by using the option --nostdmacros.
Biber will look at any MONTH field in a BIBTEX data source and if it’s not a number
as Biblatex expects (because it wasn’t one of the macros as mentioned above or these
macros were disabled by --nostdmacros), it will try to turn it into the right number
in the .bbl. If you only use Biblatex with your BIBTEX datasource files, you should
probably make any MONTH fields be the numbers which Biblatex expects.
40
3.11. Biber datasource drivers
Biber uses a modular datasource driver model to provide access to supported data-
sources. The drivers are responsible for mapping driver entrytypes and fields to
the Biblatex data model according to a data model specification in the Biblatex file
blx-dm.def. The data model can be changed using Biblatex macros in case you
would like to, for example, use your own entrytype or field names or perhaps have
Biber do some validation on your datasources (see the Biblatex manual).
Data model mapping is an imprecise art and the drivers are the necessarily the
most messy parts of Biber. Most datasource models are not designed with type-
setting in mind and are usually not fine-grained enough to provide the sorts of
information that Biblatex needs. Biber does its best to obtain as much meaningful
information from a datasource as possible. Currently supported datasources drivers
are:
• BIBTEX — BIBTEX data files

• biblatexml — Experimental Biblatex XML format
3.12. Visualising the Output

The option --output-format=dot will cause Biber to write a GraphViz16 .dot
file instead of a .bbl. This file graphs the bibliographic data as it exists after all
processing. You can transform this file using the dot program from GraphViz to
generate a high quality graphical representation of the data in a format of your
choice. A good output format choice with dot is SVG17 which can be viewed in
any modern web browser. This format has the advantage of tooltips and Biber uses
these to give you more information on connections between entries: hover the cursor
on an arrow in the output and it will tell you what it means. To output in SVG,
use this command after installing GraphViz:
dot -Tsvg <file>.dot -o <file>.svg
The --dot-include option takes a comma delimited string as argument. The ele-
ments of this string define the information to include in the .dot output graph. The
valid sub-options are shown in Table 4. If the --dot-include option is not given
then the default setting is implicitly used, which is:
--dot-include=crossref,section,xdata,xref
16
http://www.graphviz.org
17
Scalable Vector Graphics
41
Sub-option Description
crossref Show crossreference relationships
field Show fields within entries
related Show related entries and clones
section Show sections
xdata Show XDATA relationships
xref Show XREF relationships
Table 4: Valid sub-options for the dot-include option
<key> (<entrytype>) <key> (<entrytype>) <key> (<entrytype>)
Cited entry Uncited entry dataonly entry
Section <number> <key> (SET)
Section Entry set
B inherits by CROSSREF from A

A B
B inherits by XREF from A

A B
B inherits by XDATA from A

A B
A is a related entry of B
A B
B is a clone of A
A B
Figure 3: Key to .dot output format
42
3.13. Tool Mode
Biber can run in ‘tool’ mode which is enabled with the --tool command-line only
option. In this mode, Biber is called: biber --tool <datasource>. Tool mode is
a datasource rather than document oriented mode intended for transformations and
modifications of datasources. It does not read a .bcf but instead, it reads all entries
from the file ‘datasource’, applies any changes specified in the command-line options
and Biber config file and writes the resulting datasource out to a new file, defaulting
to a BibTeX file called ‘<datasource>_bibertool.bib’ if the options output-file
and output-format are not specified.
Tool mode is useful if you need to programatically change your datasource using
the semantics provided by Biber or if you would like to convert your data to a
different format. For example, you could choose to reencode your datasource by
turning all UTF-8 characters into LaTeX macros:
biber --tool --output-encoding=ascii file.bib
This would output a copy of file.bib called file_bibertool.bib with all UTF-8
chars changed to LaTeX macros (because when the output is ASCII and the input
encoding is not (it is by default UTF-8), then the --output-safechars option is
automatically enabled). If you utilise the Biber config file, you can set up a com-
plex set of mappings to transform your datasource however you wish in a semantic
manner much more robust than just textual search/replace. You can also use the
--output-resolve meta-option which will process any XDATA fields/entries, entry
aliases and inheritance rules mentioned in the config file (see below).
Sometimes, you might wish to output fields which are BibTeX macros, that is,
you might want this:
@Entrytype{key,
Field = something,
}
instead of this:
@Entrytype{key,
Field = {something},
}
That is, you might not want the output field in braces or quotes as this prevents
BibTeX interpreting the field value as a macro. Use the --output-macro-fields
option to specify a comma-separated list of fields whose values you wish to output
without any BibTeX quoting. You can have spaces between the items in the field
list but then you must enclose the whole option value in quotes. For example, these
two will do the same thing:
43
biber --tool --output-macro_fields=month,publisher
biber --tool --output-macro_fields='month, publisher'
Tool mode also allows some reformatting of the .bib file. The option --tool-fieldcase
can be used to force the entrytype and fieldnames to upper, lower or title case. The
option --tool-indent can be used to customise the indentation of fields. The op-
tion output-align can be used to align all field values neatly. See the Biber --help
output for documentation and defaults. For example, the command:
biber --tool --output-fieldcase=title --output-indent=4 \

--output-align file.bib
results in .bib entries which look like this:
@Entrytype{key,
Author = {...},
Title = {...},
Publisher = {...},
Year = {...},
}
another example:
biber --tool --output-fieldcase=upper --output-indent=2 file.bib
results in entries like this:
@ENTRYTYPE{key,
AUTHOR = {...},
TITLE = {...},
PUBLISHER = {...},
YEAR = {...},
}
Here is an example using the Biber config file to specify all options. This example
uses tool mode to reformat the .bib and also to do some transformations using the
source map functionality. Suppose test.bib contains the following:
@book{book1,
author = {Doe,J.P.},
title = {Ökologische Enterprises},
year = {2013}
}
Further suppose that the biber-tool.conf contains the following:
44
<?xml version="1.0" encoding="UTF-8"?>
<config>
<output_fieldcase>title</output_fieldcase>
<output_encoding>ascii</output_encoding>
<output_safechars>1</output_safechars>
<sourcemap>
<maps datatype="bibtex" map_overwrite="1">
<map_step map_field_source="AUTHOR" map_match="Doe," map_final="1"/>
</map>
</maps>
</sourcemap>
</config>
Now you can run Biber like this:
biber --tool --configfile=biber-tool.conf test.bib
The result will be in test_bibertool.bib and will look like this:

@Book{book1,
Author = {Doe, John Paul},
Title = {\"{O}kologische Enterprises},
Year = {2013},
}
Tool mode is a versatile way of performing many different operations on a .bib file.
By using the config file and tool mode, we have:
• Consistently indented and aligned the entry, normalising fields and entrytype
to title case
• Normalised the AUTHOR field name using regular expressions
• Converted UTF-8 characters to LaTeX macros, and made the output pure
ASCII
If you do not specify any configuration file to use in tool mode, Biber will by default
look for a config file in the usual way (see section 3.1) with the only difference that
if no config file is found, it will use the default biber-tool.conf which is located in
the Biber install tree in the same location as the Config.pm file. This default config
file contains the default Biblatex source mappings for BibTeX datasources and also
the default inheritance rules for CROSSREF processing. This means that when you
use the --output-resolve meta-option, inheritance processing is performed on the
entries and the results of this are ‘materialised’ in the output. For example, consider
a test.bib file:
45
@BOOK{xd1,
AUTHOR = {Edward Ellington},
DATE = {2007},
XDATA = {macmillanalias}
}
@XDATA{macmillan,
IDS = {macmillanalias},
XDATA = {macmillan:pubALIAS, macmillan:loc}
}
@XDATA{macmillan:pub,
IDS = {macmillan:pubALIAS},
PUBLISHER = {Macmillan}
}
@XDATA{macmillan:loc,
LOCATION = {New York and London},
NOTE = {A Note}
}
@BOOK{b1,
TITLE = {Booktitle},
CROSSREF = {mvalias}
}
@MVBOOK{mv1,
IDS = {mvalias},
TITLE = {Maintitle},
SUBTITLE = {Mainsubtitle},
TITLEADDON = {Maintitleaddon}
}
Running Biber as:
biber --tool --output-resolve test.bib
The result of this would be a file test_bibertool.bib with contents:
@BOOK{xd1,
AUTHOR = {Edward Ellington},
DATE = {2007},
NOTE = {A Note},
PUBLISHER = {Macmillan},
}
46
@XDATA{macmillan,
NOTE = {A Note},
}
@XDATA{macmillan:pub,
}
@XDATA{macmillan:loc,
NOTE = {A Note},
}
@BOOK{b1,
MAINSUBTITLE = {Mainsubtitle},
MAINTITLE = {Maintitle},
MAINTITLEADDON = {Maintitleaddon},
TITLE = {Booktitle},
}
@MVBOOK{mv1,
SUBTITLE = {Mainsubtitle},
TITLE = {Maintitle},
TITLEADDON = {Maintitleaddon},
}
Notice here that:
• XDATA references have been resolved completely for entry xd1

• CROSSREF inheritance has been resolved according to the default Biblatex in-
heritance rules for entry b1
• Entry key aliases have been resolved as required in order to perform these
tasks
Tool mode can also be used to convert between datasource formats. For example,
if you wish to covert a BibTeX format data file to the experimental biblatexml
XML format, you can do:
biber --tool --output-format=biblatexml file.bib
This will output a file file_bibertool.bltxml by default. The applicability of the

various output options depends on the output format as shown in table 5 where
dash means that the options has no relevance for the output format.
47
Output format
Option bibtex biblatexml bbl dot
output-align ✓ - - -
output-annotation-marker ✓ - - -
output-named-annotation-marker ✓ - - -
output-indent ✓ ✓ - -
output-field-order ✓ - - -
output-fieldcase ✓ - - -
output-listsep ✓ - - -
output-macro-fields ✓ - - -
output-namesep ✓ - - -
output-resolve-xdata ✓ ✓ - -
output-resolve-crossrefs ✓ ✓ - -
output-resolve-sets ✓ ✓ - -
output-xname ✓ - - -
output-xnamesep ✓ - - -
Table 5: Applicability of the output options
The order of the fields when writing BibTeX data is controlled by the --output-field-order
option. This is a comma-separated list of fields or field classes and fields will be out-
put to entries in the order specified. Any unspecified fields will be output in sorted
order after the specified fields. The field classes are:
names All name fields

lists All non-name list fields
dates All date fields
For the default value, run Biber with the --help option and see the documenta-
tion for the option. --output-listsep, output-namesep and output-xnamesep
can be used to customise separators on output and their default values are the
same as their input option counterparts --listsep, --namesep and --xnamesep.
The option --output-xname can be used to specify that the extended name format
(see section ??) is to be used to output names. --output-annotation-marker
and --output-named-annotation-marker can be used to specify the annotation
markers to write for annotated fields on output. See the Data Annotation feature
documentation in the Biblatex manual.
3.13.1. Customising the Tool Mode Data Model

The default biber-tool.conf contains the default data model specification. You
may add to or override the default data model by using your own config file via
48
the --configfile|-g option. The data model in the default biber-tool.conf
will be read and then any datamodel section in your config file will be read, taking
precedence over the default settings. There is no need to include a complete data
model in your own config file–the biber-tool.conf provides a base. Naturally, for
complex changes to the the default data model, then you may wish to include a
complete data model in your config file.
3.13.2. Customising Tool Mode Inheritance Resolution

The default biber-tool.conf contains, as mentioned above, the default Biblatex
CROSSREF inheritance setup and BibTeX source mappings so that tool mode resolu-
tion works as expected. Of course it is possible to customise these. In Biblatex, this
is accomplished by the \DeclareDataInheritance macros which write appropriate
XML into the .bcf file. Since no .bcf file is used in tool mode, the desired config-
uration must be put into a Biber config file. The source mapping XML specification
is given in section 3.1.2. The inheritance XML specification is given in section 3.1.3.
It is recommended to copy the default biber-tool.conf file, modify this and then
use it as your own biber.conf file or pass it explicitly using the --configfile|-g
option. You can determine the location of the default tool mode config file by using
the --tool-config option which will show you the location of the config file and
exit.
3.13.3. Customising Tool Mode Sorting

A sorting scheme called ‘tool’ can be defined in the config file in order to sort
the entries in tool mode output. See section 3.1.7 for the format of the config file
sorting specification. By default, in tool mode the sorting scheme is the same as the
Biblatex none scheme, that is, no sorting is performed. The sorting locale in tool
mode defaults to ‘en_US’ if you do not use Biber’s sortlocale option.
4. Binaries
Biber is a Perl application which relies heavily on quite a few modules. It is packaged
as a stand-alone binary using the excellent PAR::Packer module which can pack an
entire Perl tree plus dependencies into one file which acts as a stand-alone binary
and is indistinguishable from such to the end user. You can also simply download
the Perl source and run it as a normal Perl program which requires you to have a
working Perl 5.24+ installation and the ability to install the pre-requisite modules.
You would typically only do this if you wanted to keep up with all the bleeding-
edge git commits before they had been packaged as a binary. Almost all users will
not want to do this and should use the binaries from their TEX distribution or
49
downloaded directly from SourceForge in case they need to use a more recent binary
than is included in their TEX distribution.
The binary distributions of Biber are made using the Perl PAR::Packer module.
They can be used as a normal binary but have some behaviour which is worth noting:
• Don’t be worried by the size of the binaries. PAR::Packer essentially con-

structs a self-extracting archive which unpacks the needed files first.
• On the first run of a new version (that is, with a specific hash), they actually
unpack themselves to a temporary location which varies by operating system.
This unpacking can take a little while and only happens on the first run of
a new version. Please don’t kill the process if it seems to take some
time to do anything on the first run of a new binary. If you do, it will
not unpack everything and it will almost certainly break Biber. You will then
have to delete your binary cache (see section 4.1 below) and re-run the Biber
executable again for the first time to allow it to unpack properly.
4.1. Binary Caches

PAR::Packer works by unpacking the required files to a cache location. It only does
this on the first run of a binary by computing a hash of the binary and comparing
it with the cache directory name which contains the hash. So, if you run several
versions of a binary, you will end up with several cached trees which are never used.
This is particularly true if you are regularly testing new versions of the Biber binary.
It is a good idea to delete the caches for older binaries as they are not needed and
can take up a fair bit of space. The caches are located in a temporary location which
varies from OS to OS. The cache name is:
par-<hex_encoded_username>/cache-<hash> (Linux/Unix/OSX)
par-<hex_encoded_username>\cache-<hash> (Windows)
The temp location is not always obvious but these are sensible places to look (where
* can vary depending on username):
• /var/folders/*/*/*/ (OSX, local GUI login shell)

• /var/tmp/ (OSX (remote ssh login shell), Unix)
• /tmp/ (Linux)
• C:\Documents and Settings\<username>\Local Settings\Temp (Windows/Cyg-
win)
• C:\Windows\Temp (Windows)
To clean up, you can just remove the whole par-<hex_encoded_username> direct-
ory/folder and then run the current binary again. You can see the active cache by
50
running biber with the --cache option which will print the current cache location
and exit.
4.2. Binary Architectures

Binaries are available for many architectures, directly on SourceForge. If you want
to run development versions, they are usually only regularly updated for the core
architectures which are not flagged as third-party built above. If you want to regu-
larly run the latest development version, you should probably git clone the relevant
branch and run Biber as a pure Perl program directly.
4.3. Installing
These instructions only apply to manually downloaded binaries. If Biber came with
your TEX distribution just use it as normal.
Download the binary appropriate to you OS/arch18 . Below I assume it’s on your
desktop.
You have to move the binary to somewhere in you command-line or TEX utility
path so that it can be found. If you know how to do this, just ignore the rest of this
section which contains some instructions for users who are not sure about this.
4.3.1. OSX
If you are using the TEXLive MacTEX distribution:
sudo mv ~/Desktop/biber /usr/texbin/

sudo chmod +x /usr/texbin/biber
If you are using the MacPorts TEXLive distribution:
sudo mv ~/Desktop/biber /opt/local/bin/

sudo chmod +x /opt/local/bin/biber
The ‘sudo’ commands will prompt you for your password.
4.3.2. Windows
The easiest way is to just move the executable into your C:\Windows directory since
that is always in your path. A more elegant way is to put it somewhere in your TEX
distribution that is already in your path. For example if you are using MiKTEX:
C:\Program Files\MiKTeX 2.9\miktex\bin\
18
https://sourceforge.net/projects/biblatex-biber
51
4.3.3. Unix/Linux
sudo mv ~/Desktop/biber /usr/local/bin/biber
sudo chmod +x /usr/local/bin/biber
Make sure /usr/local/bin is in your PATH. Search Google for ‘set PATH linux’
if unsure about this. There are many pages about this, for example: http://www.
cyberciti.biz/faq/unix-linux-adding-path/
4.4. Building
Instructions for those who want/need to build an executable from the Perl version.
For this, you will need to have Perl 5.24+ with the following modules (best installed
in this order):
• Module::Build and all dependencies

• All Biber pre-requisites
• PAR::Packer and all dependencies
Biber is very specific in some cases about module versions and sometimes depends
on recent fixes. You can see if you have the Biber Perl dependencies by the usual
Module::Build command:
perl ./Build.PL
run at the root of the Biber Perl distribution directory. Normally, the build proced-
ure for the binaries is as follows19 :
• Get the Biber source tree from SF and put it on the architecture you are
building for
• cd to the root of the source tree
• perl Build.PL (this will check your module dependencies)
• If you are missing dependencies, you will be informed and then you should run
Build installdeps (may need to run this with sudo on Unix-like systems)
• Run the test suite with Build test
• Install with Build install (may need to run this with sudo on Unix-like
systems)
• cd dist/<arch>
• build.sh (build.bat on Windows)
This leaves a binary called ‘biber-<arch>’ (also with a ‘.exe’ extension on Win-
dows/Cygwin) in your current directory. The important part is constructing the
information for the build script. There are two things that need to be configured,
both of which are required by the PAR::Packer module:
19
On Unix-like systems, you may need to specify a full path to the scripts e.g. ./Build
52
1. A list of modules/libraries to include in the binary which are not automatically
detected by the PAR::Packer dependency scanner
2. A list of extra files to include in the binary which are not automatically detected
by the PAR::Packer dependency scanner
To build Biber for a new architecture you need to define these two things as part of
constructing new build scripts:
• Make a new sub-folder in the dist directory named after the architecture you
are building for.
• Copy the biber.files file from an existing build architecture into this direct-
ory.
• For all of the files with absolute pathnames in there (that is, ones we are not
pulling from the Biber tree itself), locate these files in your Perl installation
tree and put the correct path in the file.
• Copy the build script from a similar architecture (i.e. Windows/non-Windows
…) to your new architecture directory.
• Change the --link options to point to where the required libraries reside on
your system.
• Change the --output option to name the resulting binary for your architec-
ture.
• Run the build script
The --link options can be a little tricky sometimes. It is usually best to build
without them once and then run ldd20 on the binary to see which version/location
of a library you should link to. You can also try just running the binary and it
should complain about missing libraries and where it expected to find them. Put
missing library paths into --link options. The --module options are the same for
all architectures and do not need to be modified. On architectures which have or
can have case-insensitive file systems, you should use the build script from either
Windows or OSX as a reference as these include a step to copy the main Biber script
to a new name before packing the binary. This is required as otherwise a spurious
error is reported to the user on first run of the binary due to a name collision when
it unpacks itself.
See the PAR wiki page21 for FAQs and help on building with PAR::Packer. Take
special note of the FAQs on including libraries with the packed binary22 .
20
otool on OSX and depends.exe on Windows
21
http://par.perl.org/wiki/Main_Page
22
http://par.perl.org/wiki/FAQ, section entitled ‘My PAR executable needs some dynamic librar-
ies’
53
4.4.1. Testing a binary build
You can test a binary that you have created by copying it to a machine which
preferably doesn’t have perl at all on it. Running the binary with no arguments
will unpack it in the background and display the help. To really test it without
having LaTeX available, get the two quick test files from SourceForge23 , put them
in a directory and run Biber in that directory like this:
biber --validate-control --convert-control test
This will run Biber normally on the test files plus it will also perform an XSLT
transform on the .bcf and leave an HTML representation of it in the same directory
thus testing the links to the XML and XSLT libraries as well as the BIBTEX parsing
libraries. The output should look something like this (may be differences of Biber
version and locale of course but there should be no errors or warnings).
INFO - This is Biber 2.13

INFO - Logfile is 'test.blg'
INFO - BibLaTeX control file 'test.bcf' validates
INFO - Converted BibLaTeX control file 'test.bcf' to 'test.bcf.html'
INFO - Reading 'test.bcf'
INFO - Found 1 citekeys in bib section 0
INFO - Processing bib section 0
INFO - Looking for BibTeX format file 'test.bib' for section 0
INFO - Found BibTeX data file 'test.bib'
INFO - Decoding LaTeX character macros into UTF-8
INFO - Sorting list 'nyt/global' keys
INFO - No sort tailoring available for locale 'en_GB.UTF-8'
INFO - Sorting list 'shorthands/global' keys
INFO - No sort tailoring available for locale 'en_GB.UTF-8'
INFO - Writing 'test.bbl' with encoding 'UTF-8'
INFO - Output to test.bbl
There should now be these new files in the directory:
test.bcf.html
test.blg
test.bbl
23
https://sourceforge.net/projects/biblatex-biber/files/biblatex-biber/testfiles
54
A. Appendix
A.1. Babel/Polyglossia language to Locale mapping
Language Locale Language Locale Language Locale Language Locale

acadian fr_CA divehi dv_MV latin la_Latn sanskrit sa_IN
american en_US dutch nl_NL latvian lv_LV scottish gd_GB
australian en_AU english en_US lithuanian lt_LT serbian sr_Latn
afrikaans af_ZA esperanto eo_001 lowersorbian dsb_DE serbianc sr_Cyrl
albanian sq_AL estonian et_EE lsorbian dsb_DE slovak sk_SK
amharic am_ET ethiopia am_ET magyar hu_HU slovene sl_SI
arabic ar_001 farsi fa_IR malay id_ID slovenian sl_SI
armenian hy_AM finnish fi_FI malayalam ml_IN spanish es_ES
asturian ast_ES francais fr_FR marathi mr_IN swedish sv_SE
austrian de_AT french fr_FR meyalu id_ID syriac syc
bahasa id_ID frenchle fr_FR mongolian mn_Cyrl tamil ta_IN
bahasai id_ID friulan fur_IT naustrian de_AT telugu te_IN
bahasam id_ID galician gl_ES newzealand en_US thai th_TH
basque eu_ES german de_DE ngerman de_DE thaicjk th_TH
bengali bn_BD germanb de_DE nko ha_NG tibetan bo_CN
bgreek el_GR greek el_GR norsk nb_NO turkish tr_TR
brazil pt_BR hebrew he_IL nynorsk nn_NO turkmen tk_TM
brazilian pt_BR hindi hi_IN occitan oc_FR ukrainian uk_UA
breton br_FR ibygreek el_CY piedmontese pms_IT urdu ur_IN
british en_GB icelandic is_IS pinyin pny UKenglish en_GB
bulgarian bg_BG indon id_ID polish pl_PL uppersorbian hsb_DE
canadian en_US indonesia id_ID polutonikogreek el_GR USenglish en_US
canadien fr_CA interlingua ia_FR portuges pt_PT usorbian hsb_DE
catalan ca_ES irish ga_IE portuguese pt_PT vietnamese vi_VN
coptic cop italian it_IT romanian ro_RO welsh cy_GB
croatian hr_HR japanese ja_JP romansh rm_CH
czech cs_CZ kannada kn_IN russian ru_RU
danish da_DK lao lo_LA samin se_NO
55

Biber: A Backend Bibliography Processor For Biblatex

Uploaded by

Copyright:

Available Formats

Biber: A Backend Bibliography Processor For Biblatex

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Biber: A Backend Bibliography Processor For Biblatex

Uploaded by

Copyright:

Available Formats

biber

A backend bibliography processor for biblatex

biber --graph=section,field <file>

biber --output-format=dot --dot-include=section,field <file>

• Parsing data from datasources

2.3. Compatibility Matrix

Table 1: Biber/Biblatex compatibility matrix

• Deals with the full range of UTF-8

Decode to UTF-8 Generate uniqueness data

User biber.conf file remap/parse Generate name visibility data

Instantiate dynamic sets and Generate more hashes and

Process XDATA Perform sorting

Process cross-references and

Validate data model Encode to output encoding

Resolve label* fields

Figure 1: Overview of Biber’s main functional units

document data model mapping Biblatex data

default data model mapping remap parser validation

data model mapping

Figure 2: Model data flow in Biber

command line options →

biber.conf in the current directory →

<?xml version="1.0" encoding="UTF-8"?>

Table 2: Biblatex options which Biber uses

3.1.1. The output-format option

3.1.2. The sourcemap option

user-level maps defined with \DeclareSourcemap→

• Map datasource entrytypes to different entrytypes.

• Change the entrytype of an entry

<map map_overwrite="0|1" map_foreach="loopval">

• If there are any datasources named in per_datasource elements, this map-

Other datasource types

<maps datatype="bibtex" level="user">

3.1.3. The inheritance option

3.1.4. The noinit option

and the BIBTEX datasource entry:

AUTHOR = {{al-Hasan}, ʿAlī},

AUTHOR = {O'Toole, Alexander},

3.1.6. The nolabelwidthcount option

AUTHOR = {O'Toole, Alexander},

3.1.7. The sorting option

Sorting in Biber is a sophisticated procedure of building up a sorting object for

This makes it easy to divide a bibliography by type of entry.

3.1.8. The nosort option

Table 3: Default Biblatex datafield sets

AUTHOR = {{al-Hasan}, ʿAlī},

3.1.9. The collate-options option

3.3.2. Data sources

Remote file datasources (beginning with http:// or ftp://) are retrieved to a

biber <options> test.bcf

biber --logfile=lfname test.bcf

3.5. Collation and Localisation

--collate-options|-c (e.g. -c 'locale => "de_DE"') →

3.6. Encoding of files

1. Decodes the datasource into UTF-8 if it is not UTF-8 already

--input-encoding|-e and --output-encoding|-E →

3.6.1. LaTeX macro decoding

3.6.2. LaTeX macro encoding

• /var/folders///*/ (OSX, local GUI login shell)