BibFormat API

****************************************************************************
** IMPORTANT NOTE: Note that this documentation is an updated version of  **
** an earlier technical draft of BibFormat specifications. Please first   **
** refer to the BibFormat admin guide.                                    **
****************************************************************************

Technical Overview of the new BibFormat
=======================================

Contents:
1. Python API
2. The philosophy behind BibFormat
3. Differences between the old PHP version and the new Pythonic version
4. Migrating from the previous PHP BibFormat version to the new Pythonic version
5. Specifications of the new BibFormat configuration files.


1. Python API

The APIs of bibformat.py consists in these functions:

 def format_record(recID, of, ln=CFG_SITE_LANG, verbose=0,
                   search_pattern=None, xml_record=None, user_info=None,
                   on_the_fly=False):
     """
     Formats a record given its ID (or its XML representation)
     and an output format.

     Returns a formatted version of the record in the specified
     language, with pattern context, and specified output format.
     The function will define by itself which format template must be
     applied.

     Parameters that allow contextual formatting (like 'search_pattern'
     and 'user_info') are useful only when doing on-the-fly
     formatting, or when caching with care (e.g. caching all formatted
     versions of a record for each possible 'ln').

     The arguments are as follows:

               recID -  the ID of the record to format. If ID does not exist
                        the function returns empty string or an error
                        string, depending on level of verbosity.
                        If 'xml_record' parameter is specified, 'recID'
                        is ignored

                  of -  an output format code. If 'of' does not exist as code in
                        output format, the function returns empty
                        string or an error string, depending on level
                        of verbosity.  ;of' is case insensitive.

                  ln -  the language to use to format the record. If
                        'ln' is an unknown language, or translation
                        does not exist, default CFG_SITE_LANG language
                        will be applied whenever possible.
                        Allows contextual formatting.

             verbose -  the level of verbosity in case of errors/warnings
                        0 - Silent mode
                        5 - Prints only errors
                        9 - Prints errors and warnings

      search_pattern -  the pattern used as search query when asked to
                        format this record (User request in web
                        interface). Allows contextual formatting.

          xml_record -  an XML string representation of the record to
                        format.  If it is specified, recID parameter is
                        ignored. The XML must be pasable by BibRecord.

           user_info - allows to grant access to some functionalities
                       on a page depending on the user's
                       priviledges. 'user_info' is the same structure
                       as the one returned by webuser.collect_user_info(req),
                       (that is a dictionary).

          on_the_fly - if False, try to return an already preformatted
                       version of the record in the database.

     """


 Example:
   >> from invenio.bibformat import format_record
   >> format_record(5, "hb", "fr")


 def format_records(recIDs, of, ln=CFG_SITE_LANG, verbose=0, search_pattern=None,
                    xml_records=None, user_info=None, record_prefix=None,
                    record_separator=None, record_suffix=None,
                    prologue="", epilogue="", req=None, on_the_fly=False):
     """
     Returns a list of formatted records given by a list of record IDs or a
     list of records as xml.
     Adds a prefix before each record, a suffix after each record,
     plus a separator between records.

     Also add optional prologue and epilogue to the complete formatted list.

     You can either specify a list of record IDs to format, or a list of
     xml records, but not both (if both are specified recIDs is ignored).

     'record_separator' is a function that returns a string as separator between
     records. The function must take an integer as unique parameter,
     which is the index in recIDs (or xml_records) of the record that has
     just been formatted. For example separator(i) must return the separator
     between recID[i] and recID[i+1]. Alternatively separator can be a single
     string, which will be used to separate all formatted records.
     The same applies to 'record_prefix' and 'record_suffix'.

     'req' is an optional parameter on which the result of the function
     are printed lively (prints records after records) if it is given.
     Note that you should set 'req' content-type by yourself, and send
     http header before calling this function as it will not do it.

     This function takes the same parameters as 'format_record' except for:

               recIDs -  a list of record IDs to format

          xml_records -  a list of xml string representions of the records to
                         format. If this list is specified, 'recIDs' is ignored.

        record_prefix - a string or a function the takes the index of the record
                        in 'recIDs' or 'xml_records' for which the function must
                        return a string.
                        Printed before each formatted record.

     record_separator - either a string or a function that returns string to
                        separate formatted records. The function takes the index
                        of the record in 'recIDs' or 'xml_records' that is being
                        formatted.

        record_prefix - a string or a function the takes the index of the record
                        in 'recIDs' or 'xml_records' for which the function must
                        return a string.
                        Printed after each formatted record

                  req - an optional request object on which formatted records
                        can be printed (for "live" output )

             prologue - a string printed before all formatted records string

             epilogue - a string printed after all formatted records string

           on_the_fly - if False, try to return an already preformatted version
                        of the records in the database
     """


 def get_output_format_content_type(of):
     """
     Returns the content type (eg. 'text/html' or 'application/ms-excel') \
     of the given output format.

     The function takes this mandatory parameter:

     of - the code of output format for which we want to get the content type
     """


 def record_get_xml(recID, format='xm', decompress=zlib.decompress):
     """
     Returns an XML string of the record given by recID.

     The function builds the XML directly from the database,
     without using the standard formatting process.

     'format' allows to define the flavour of XML:
        - 'xm' for standard XML
        - 'marcxml' for MARC XML
        - 'oai_dc' for OAI Dublin Core
        - 'xd' for XML Dublin Core

     If record does not exist, returns empty string.

     The function takes the following parameters:

          recID - the id of the record to retrieve

         format - the XML flavor in which we want to get the record

     decompress _ a function used to decompress the record from the database
    """

The API of the BibFormat Object ('bfo') given as a parameter to
format function of format elements consist in the following
functions. This API is to be used only inside format elements.

 def control_field(self, tag, escape='0'):
    """
    Returns the value of control field given by tag in record.

    If the value does not exist, returns empty string
    The returned value is always a string.

    'escape' parameter allows to escape special characters
    of the field. The value of escape can be:
          0 - no escaping
          1 - escape all HTML characters
          2 - remove unsafe HTML tags (Eg. keep <br />)
          3 - Mix of mode 1 and 2. If value of field starts with
              <!-- HTML -->, then use mode 2. Else use mode 1.
          4 - Remove all HTML tags
          5 - Same as 2, with more tags allowed (like <img>)
          6 - Same as 3, with more tags allowed (like <img>)
          7 - Mix of mode 0 and mode 1. If field_value starts
              with <!-- HTML -->, then use mode 0. Else use mode 1.
          8. Same as mode 1, but also escape double-quotes
          9. Same as mode 4, but also escape double-quotes

    The arguments are:

         tag    -  the marc code of a field
         escape -  1 if returned value should be escaped. Else 0.
                   (see above for other modes)
    """

 def field(self, tag, escape='0'):
    """
    Returns the value of the field corresponding to tag in the
    current record.

    If the value does not exist, returns empty string
    Else returns the same as bfo.fields(..)[0] (see docstring below).

    'escape' parameter allows to escape special characters
    of the field. The value of escape can be:
          0 - no escaping
          1 - escape all HTML characters
          2 - remove unsafe HTML tags (Eg. keep <br />)
          3 - Mix of mode 1 and 2. If value of field starts with
              <!-- HTML -->, then use mode 2. Else use mode 1.
          4 - Remove all HTML tags
          5 - Same as 2, with more tags allowed (like <img>)
          6 - Same as 3, with more tags allowed (like <img>)
          7 - Mix of mode 0 and mode 1. If field_value starts
              with <!-- HTML -->, then use mode 0. Else use mode 1.
          8 - Same as mode 1, but also escape double-quotes
          9 - Same as mode 4, but also escape double-quotes

    The arguments are:

         tag  -  the marc code of a field
         escape -  1 if returned value should be escaped. Else 0.
                   (see above for other modes)
    """


 def fields(self, tag, escape='0', repeatable_subfields_p=False):
    """
    Returns the list of values corresonding to "tag".

    If tag has an undefined subcode (such as 999C5),
    the function returns a list of dictionaries, whoose keys
    are the subcodes and the values are the values of tag.subcode.
    If the tag has a subcode, simply returns list of values
    corresponding to tag.
    Eg. for given MARC:
        999C5 $a value_1a $b value_1b
        999C5 $b value_2b
        999C5 $b value_3b $b value_3b_bis

        >> bfo.fields('999C5b')
        >> ['value_1b', 'value_2b', 'value_3b', 'value_3b_bis']
        >> bfo.fields('999C5')
        >> [{'a':'value_1a', 'b':'value_1b'},
            {'b':'value_2b'},
            {'b':'value_3b'}]
    By default the function returns only one value for each
    subfield (that is it considers that repeatable subfields are
    not allowed). It is why in the above example 'value3b_bis' is
    not shown for bfo.fields('999C5').  (Note that it is not
    defined which of value_3b or value_3b_bis is returned).  This
    is to simplify the use of the function, as most of the time
    subfields are not repeatable (in that way we get a string
    instead of a list).  You can allow repeatable subfields by
    setting 'repeatable_subfields_p' parameter to True. In
    this mode, the above example would return:
        >> bfo.fields('999C5b', repeatable_subfields_p=True)
        >> ['value_1b', 'value_2b', 'value_3b']
        >> bfo.fields('999C5', repeatable_subfields_p=True)
        >> [{'a':['value_1a'], 'b':['value_1b']},
            {'b':['value_2b']},
            {'b':['value_3b', 'value3b_bis']}]
    NOTICE THAT THE RETURNED STRUCTURE IS DIFFERENT.  Also note
    that whatever the value of 'repeatable_subfields_p' is,
    bfo.fields('999C5b') always show all fields, even repeatable
    ones. This is because the parameter has no impact on the
    returned structure (it is always a list).

    'escape' parameter allows to escape special characters
     of the fields. The value of escape can be:
                  0 - no escaping
                  1 - escape all HTML characters
                  2 - remove unsafe HTML tags (Eg. keep <br />)
                  3 - Mix of mode 1 and 2. If value of field starts with
                      <!-- HTML -->, then use mode 2. Else use mode 1.
                  4 - Remove all HTML tags
                  5 - Same as 2, with more tags allowed (like <img>)
                  6 - Same as 3, with more tags allowed (like <img>)
                  7 - Mix of mode 0 and mode 1. If field_value starts
                      with <!-- HTML -->, then use mode 0. Else use mode 1.
                  8 - Same as mode 1, but also escape double-quotes
                  9 - Same as mode 4, but also escape double-quotes

    The arguments are:

          tag  -  the marc code of a field
          escape -  1 if returned value should be escaped. Else 0.
                   (see above for other modes)
          repeatable_subfields_p - if True, returns the list of
                                   subfields in the dictionary @return
                                   values of field tag in record """

 def kb(self, kb, string, default=""):
    """
    Returns the value of the "string" in the knowledge base "kb".

    If kb does not exist or string does not exist in kb,
    returns 'default' string or empty string if not specified

    The arguments are as follows:

          kb  -  the knowledge base name in which we want to find the mapping.
                 If it does not exist the function returns the original
                 'string' parameter value. The name is case insensitive (Uses
                 the SQL 'LIKE' syntax to retrieve value).

      string  -  the value for which we want to find a translation-
                 If it does not exist the function returns 'default' string.
                 The string is case insensitive (Uses the SQL 'LIKE' syntax
                 to retrieve value).

     default  -  a default value returned if 'string' not found in 'kb'.

    """

 def get_record(self):
    """
    Returns the record encapsulated in bfo as a BibRecord structure.
    You can get full access to the record through bibrecord.py functions.
    """

  Example (from inside BibFormat element):
  >> bfo.field("520.a")
  >> 'We present a quantitative appraisal of the physics potential
      for neutrino experiments.'
  >>
  >> bfo.control_field("001")
  >> '12'
  >>
  >> bfo.fields("700.a")
  >>['Alekhin, S I', 'Anselmino, M', 'Ball, R D', 'Boglione, M']
  >>
  >> bfo.kb("DBCOLLID2COLL", "ARTICLE")
  >> 'Published Article'
  >>
  >> bfo.kb("DBCOLLID2COLL", "not in kb", "My Value")
  >> 'My Value'

Moreover you can have access to the language requested for the
formatting, the search pattern used by the user in the web
interface and the userID by directly getting the attribute from 'bfo':

    bfo.lang
    """
    Returns the language that was asked to be used for the
    formatting. Always returns a string.
    """

    bfo.search_pattern
    """
    Returns the search pattern specified by the user when
    the record had to be formatted. Always returns a string.
    """

    bfo.user_info
    """
    Returns a dictionary with information about current user.
    The returned dictionary has the following structure:
        user_info = {
            'remote_ip' : '',
            'remote_host' : '',
            'referer' : '',
            'uri' : '',
            'agent' : '',
            'uid' : -1,
            'nickname' : '',
            'email' : '',
            'group' : [],
            'guest' : '1'
        }
    """

    bfo.uid
    """
    ! DEPRECATED: use bfo.user_info['uid'] instead
    """

    bfo.recID
    """
    Returns the id of the record
    """

    bfo.req
    """
    ! DEPRECATED: use bfo.user_info instead
    """

    bfo.format
    """
    ! DEPRECATED: use bfo.output_format instead
    """

    bfo.output_format
    """
    Returns the format in which the record is being formatted
    """

  Example (from inside BibFormat element):
  >> bfo.lang
  >> 'en'
  >>
  >> bfo.search_pattern
  >> 'mangano and neutrino and factory'


2. The philosophy behind BibFormat

BibFormat is in charge of formatting the bibliographic records that
are displayed to your users. As you potentially have a huge amount of
bibliographic records, you cannot specify manually for each of them
how it should be formatted. This is why you can define rules that will
allow BibFormat to understand which kind of formatting to apply to a given
record. You define this set of rules in what is called an "output
format".

You can have different output formats, each with its own characteristics.
For example you certainly want that when multiple bibliographic records are
displayed at the same time (as it happens in search results), only
short versions are shown to the user, while a detailed record is
preferable when a single record is displayed. You might also want to
let your users decide which kind of output they want. For example you
might need to display HTML for regular web browsing, but would also
give a BibTeX version of the bibliographic reference for direct
inclusion in a LaTeX document.
See section 5.1 to learn how to create or modify output formats.

While output formats define what kind of formatting must be applied,
they do not define HOW the formatting is done. This is the role of the
"format templates", which define the layout and look of a
bibliographic reference. These format templates are rather easy to
write if you know a little bit of HTML (see section 5.2 "Format
templates specifications"). You will certainly have to create
different format templates, for different kinds of records. For
example you might want records that contain pictures to display them,
maybe with captions, while records that do not have pictures limited
to printing a title and an abstract.

In summary, you have different output formats (like 'brief HTML',
'detailed HTML' or 'BibTeX') that call different format templates
according to some criteria.

There is still one kind of configuration file that we have not talked
about: the "format elements". These are the "bricks" that you use in
format templates, to get the values of a record. You will learn to use
them in your format template in section 5.2 "Format templates
specifications", but you will mostly not need to modify them or create
new ones. However if you do need to edit one, read section 5.3 "Format
elements specifications" (And if you know Python it will be easy, as
they are written in Python).

Finally BibFormat can make use of mapping tables called "knowledge
bases". Their primary use is to act like a translation table, to
normalize records before displaying them. For example, you can say
that records that have value "Phys Rev D" or "Physical Review D" for
field "published in" must display "Phys Rev : D." to users. See
section 5.4 to learn how to edit knowledge bases.

In summary, there are three layers.  Output formats:

+-----------------------------------------------------+
|                    Output Format                    | (Layer 1)
|                    eg: HTML_Brief.bfo               |
+-----------------------------------------------------+

call one of several `format templates':

+-------------------------+ +-------------------------+
|     Format Template     | |     Format Template     | (Layer 2)
|     eg: preprint.bft    | |     eg: default.bft     |
+-------------------------+ +-------------------------+

that use one or several format elements:

+--------------+ +----------------+ +-----------------+
|Format Element| |Format Element  | | Format Element  | (Layer 3)
|eg: authors.py| |eg: abstract.py | | eg: title.py    |
+--------------+ +----------------+ +-----------------+


3. Differences between the old PHP version and the new Pythonic version

The most noticeable differences are:

 a) "Behaviours" have been renamed "Output formats".
 b) "Formats" have been renamed "Format templates". They are now
     written in HTML.
 c) "User defined functions" have been dropped.
 d) "Extraction rules" have been dropped.
 e) "Link rules" have been dropped.
 f) "File formats" have been dropped.
 g) "Format elements" have been introduced. They are written in Python,
     and can simulate c), d) and e).
 h)  Formats can be managed through web interface or through
     human-readable config files.
 i)  Introduction of tools like validator and dependencies checker.
 j)  Better support for multi-language formatting.

Some of the advantages are:

 + Management of formats is much clearer and easier (less concepts,
   more tools).
 + Writing formats is easier to learn : less concepts
   to learn, redesigned work-flow, use of existing well known and
   well documented languages.
 + Editing formats is easier: You can use your preferred HTML editor such as
   Emacs, Dreamweaver or Frontpage to modify templates, or any text
   editor for output formats and format elements. You can also use the
   simplified web administration interface.
 + Faster and more powerful templating system.
 + Separation of business logic (output formats, format elements)
   and presentation layer (format templates). This makes the management
   of formats simpler.

The disadvantages are:

 - No backward compatibility with old formats.
 - Stricter separation of business logic and presentation layer:
   no more use of statements such as if(), forall() inside templates,
   and this requires more work to put logic inside format elements.


4. Migrating from the previous PHP BibFormat version to the new Pythonic version

Old BibFormat formats are no longer compatible with the new BibFormat
files. If you have not modified the "Formats" or modified only a
little bit the "Behaviours", then the transition will be painless and
automatic. Otherwise you will have to manually rewrite some of the
formats. This should however not be a big problem. Firstly because the
Invenio installation will provide both versions of BibFormat for
some time. Secondly because both BibFormat versions can run side by
side, so that you can migrate your formats while your server still
works with the old formats.  Thirdly because we provide a migration
kit that can help you go through this process. Finally because the
migration is not so difficult, and because it will be much easier for
you to customize how BibFormat formats your bibliographic data.

Concerning the migration kit it can:
 a) Effortlessly migrate your behaviours, unless they include complex
    logic, which usually they don't.
 b) Help you migrate formats to format templates and format elements.
 c) Effortlessly migrate your knowledge bases.

Point b) is the most difficult to achieve: previous formats did mix
business logic and code for the presentation, and could use PHP
functions. The new BibFormat separates business logic and
presentation, and does not support PHP. The transition kit will try to
move business logic to the format elements, and the presentation to
the format templates. These files will be created for you, includes
the original code and, if possible, a proposal of Python
translation. We recommend that you do not to use the transition kit to
translate formats, especially if you have not modified default
formats, or only modified default formats in some limited places. You
will get cleaner code if you write format elements and format
templates yourself.


5. Specifications of the new BibFormat configuration files.

   BibFormat uses human readable configuration files. However (apart
   from format elements) these files can be edited and managed through
   a web interface.

5.1 Output formats specifications

Output formats specify rules that define which format template
to use to format a record.
While the syntax of output formats is basic, we recommend that you use
the web interface do edit them, to be sure that you make no error.

The syntax of output format is the following one. First you
define which field code you put as the conditon for the rule.
You suffix it with a column. Then on next lines, define the values of
the condition, followed by --- and then the filename of the template
to use:

  tag 980.a:
  PICTURE --- PICTURE_HTML_BRIEF.bft
  PREPRINT --- PREPRINT_HTML_BRIEF.bft
  PUBLICATION --- PUBLICATION_HTML_BRIEF.bft

This means that if value of field 980.a is equal to PICTURE, then we
will use format template PICTURE_HTML_BRIEF.bft. Note that you must
use the filename of the template, not the name. Also note that spaces
at the end or beginning are not considered. On the following lines,
you can either put other conditions on tag 980.a, or add another tag on
which you want to put conditions.

At the end you can add a default condition:

   default: PREPRINT_HTML_BRIEF.bft

which means that if no condition is matched, a format suitable for
Preprints will be used to format the current record.

The output format file could then look like this:

  tag 980.a:
  PICTURE --- PICTURE_HTML_BRIEF.bft
  PREPRINT --- PREPRINT_HTML_BRIEF.bft
  PUBLICATION --- PUBLICATION_HTML_BRIEF.bft

  tag 8560.f:
  .*@cern.ch --- SPECIAL_MEMBER_FORMATTING.bft

  default: PREPRINT_HTML_BRIEF.bft

You can add as many rules as you want. Keep in mind that they are read
in the order they are defined, and that only first rule that
matches will be used.
Notice the condition on tag 8560.f: it uses a regular expression to
match any email address that ends with @cern.ch (the regular
expression must be understandable by Python)

Some other considerations on the management of output formats:
- Format outputs must be placed inside directory
  /etc/bibformat/outputs/ of your Invenio installation.
- Note that as long as you have not provided a name to an output
  THROUGH the web interface, it will not be available as a choice
  for your users in some parts of Invenio.
- You should remove output formats THROUGH the web interface.
- The format extension of output format is .bfo


5.2 Format templates specifications

Format templates are written in HTML-like syntax. You can use the
standard HTML and CSS markup languague to do the formatting. The best
thing to do is to create a page in your favourite editor, and once you
are glad with it, add the dynamic part of the page, that is print the
fields of the records. Let's say you have defined this page:

  <h1>Some title</h1>
  <p><i>Abstract: </i>Some abstract</p>

Then you want that instead of "Some title" and "Some abstract", the
value of the current record that is being displayed is used. To do so,
you must use a format element brick. Either you know the name of the
brick by heart, or you look for it in the elements documentation (see
section 5.3). For example you would find there that you can print the
title of the record by writting the HTML tag <BFE_TITLE /> in your
format template, with parameter 'default' for a default value.

  <h1><BFE_TITLE default="No Title"/></h1>
  <p><BFE_ABSTRACT limit="1" prefix="<i>Abstract: </i>"
  default="No abstract"/></p>

Notice that <BFE_ABSTRACT /> has a parameter "limit" that <BFE_title/>
 had not ("limit" allows to limit the number of sentences of the
abstract, according to the documentation). Note that while format
elements might have different parameters, they always can take the the
three following ones: "prefix" and "suffix", whose values are printed
only if the element is not empty, and "default", which is printed only
if element is an empty string. We have used "prefix" for the abstract,
so that the label "<i>Abstract: </i>" is only printed if the record
has an abstract.

You should also provide these tags in all of your templates:
 -<name>a name for this template in the admin web interface</name>
 -<description>a description to be used in admin web interface for
  this template</description>

Another feature of the templates is the support for multi-languages
outputs. You can include <lang> tags, which contain tags labeled with
the names of the languages supported in Invenio. For example, one
might write:

  <lang><en>A record:</en><fr>Une notice:</fr></lang>
  <h1><BFE_TITLE default="No Title"/></h1>
  <p><BFE_ABSTRACT limit="1" prefix="<i>Abstract: </i>"
  default="No abstract"/></p>

When doing this you should at least make sure that the default
language of your server installation is available in each <lang>
tag. It is the one that is used if the requested language to display
the record is not available. Note that we could also provide a
translation in a similar way for the "No Title" default value inside
<BFE_Title /> tag.

Some other considerations on the use of elements inside templates:
 -Format elements names are not case sensitive
 -Format elements names always start with <BFE_
 -Format elements parameters can contain '<' characters,
  and quotes different from the kind that delimit parameters (you can
  for example have <BFE_Title default='<a href="#">No Title</a>'/> )
 -Format templates must be placed inside the directory
  /etc/bibformat/templates/ of your Invenio installation
 -The format extension of a template is .bft

Trick: you can use the <BFE_FIELD tag="245__a" /> to print the value
of any field 245 $a in your templates.  This practice is however not
recommended because it would necessitate to revise all format
templates if you change meaning of the MARC code schema.

5.3 Format elements specifications

Format elements are the bricks used in format templates to provide the
dynamic contents inside format templates.

For the most basic format elements, you do not even need to write
them: as long as you define `tag names' for MARC tags in the BibIndex
Admin's Manage logical fields interface (database table tag),
BibFormat knows which field must be printed when <BFE_tag_name/> is
used inside a template.

However for more complex processing, you will need to write a format
element. A format element is written in Python. Therefore its file
extension is ".py". The name you choose for the file is the one that
will be used inside format template to call the element, so choose it
carefully such that it is not too long, but self explanatory (you can
prefix the filename with BFE or not, but the element will always be
called with prefix <BFE_ inside templates).  Then you just need to
drop the file in the lib/python/invenio/bibformat_elements/ directory
of your Invenio installation. Inside your file you have to define
a function named "format_element", which takes at least a "bfo"
parameter (bfo for BibFormat Object). The function must return a
string:

  def format_element(bfo):
      out = ""

      return out

You can have as many parameters as you want, as long as you make sure
that parameter bfo is here. Let's see how to define an element that
will print a brief title. It will take a parameter 'limit' that will
limit the number of characters printed. We can provide some
documentation for the elemen in the docstring of the
function.

 def format_element(bfo, limit="10"):
      """
      Prints a short title

      @param limit a limit for the number of printed characters
      """

      out = ""

      return out

Note that we put a default value of 10 in the 'limit' parameter.  To
get some value of a field, we must request the 'bfo' object. For
example we can get the value of field 245.a (field "title"):

 def format_element(bfo, limit="10"):
      """
      Prints a short title

      @param limit a limit for the number of printed characters
      """

      title = bfo.field('245.a')

      limit = int(limit)
      if limit > len(title):
          limit = len(title)

      return title[:limit]

As format elements are written in Python, we have decided not to give
permission to edit elements through the web interface. Firstly for
security reasons. Secondly because Python requires correct indentation,
which is difficult to achieve through a web interface.

You can have access to the documentation of your element through a web
interface. This is very useful when you are writing a format template,
to see which elements are available, what they do, which parameters they
take, what are the default values of parameters, etc. The
documentation is automatically extracted from format elements.
Here follows an sample documentation generated for the element
<BFE_TITLE />:

+--------------------------------------------------------------------------------------------+
|  TITLE                                                                                     |
|  -----                                                                                     |
|  <BFE_TITLE separator="..." prefix="..." suffix="..." default="..." />                  |
|                                                                                            |
|      Prints the title of a record.                                                         |
|                                                                                            |
|      Parameters:                                                                           |
|            separator - separator between the different titles.                             |
|            prefix - A prefix printed only if the record has a value for this element.      |
|            suffix - A suffix printed only if the record has a value for this element.      |
|            default - A default value printed if the record has no value for this element.  |
|                                                                                            |
|       See also:                                                                            |
|            Format templates that use this element                                          |
|            The Python code of this element                                                 |
+--------------------------------------------------------------------------------------------+

The more you provide documentation in the docstring of your elements,
the easier it will be to write format template afterwards.

Some remarks concerning format elements:
 -parameters are always string values
 -if no value is given as parameter in format the template, then the
  value of parameter is "" (emtpy string)
 -the docstring should contain a description, followed by
  "@param parameter: some description for parameter" for each
  parameter (to give description for each parameter
  in element documentation), and @see an_element.py, another_element.py
  (to link to other elements in the documentation). Similar to JavaDoc.
 -the following names cannot be used as parameters:
  "default", "prefix", "suffix" and escape. They can however always be
  used in the format template for any element.

Another important remark concerns the 'escaping' of output of format
elements. In most cases, format elements output is to be used for
HTML/XML. Therefore special characters such as < or & have to be
'escaped', replaced by '<' and '&'. This is why all outputs
produced by format elements are automatically escaped by BibFormat,
unless specified otherwise.  This means that you do not have to care
about meta-data that would break your HTML displaying or XML export
such as a physics formula like 'a < b'. Please also note that value
given in 'prefix', 'suffix' and 'default' parameters are not escaped,
such that you can safely use HTML tags for these.

There are always cases where the default 'escaping' behaviour of
BibFormat is not desired. For example when you explicitely output HTML
text, like links: you do not want to see them escaped.  The first way
to avoid this is to modify the call to your format element in the
format template, by setting the default 'escape' parameter to 0:

 

This is however inconvenient as you have to possibly need to modify a
lot of templates. The other way of doing is to add another function to
your format element, named 'escape':

 def escape_values(bfo):
     """
     Called by BibFormat in order to check if output of this element
     should be escaped.
     """
     return 0

In that way all calls to your format element will produce unescaped
output.  You will have to take care of escaping values "manually" in
your format element code, in order to avoid non valid outputs or XSS
vulnerabilities. There are methods to ease the escaping in your code
described in section 1.
Please also note that if you use this method, your element can still
be escaped if a call to your element from a format template
explicitely specifies to escape value using parameter 'escape'.


5.4 Knowledge bases specifications

Knowledge bases cannot be managed through configuration files.
You can very easily add new bases and mappings using the given web GUI.

 -- End of file --