SPIRES formats via Inspire

Tcb has been working on formatting for a few months on and off, currently it is March 2008

Scope/Purpose

my goal is to reproduce at the very least the SPIRES formatting for default search results. This touches on several pieces of inspire.

Modules/Areas needing work

format_elements

bfe_INSPIRE_

I'm using several custom elements labeled as "INSPIRE" because they are dedicated to reproducing the quirks of the SPIRES formatting and unlikely to be useful elsewhere. These are all getting committed to inspire cvs

Citation linking

Need to mess with cited by to get the linkage right.

Keywords display

get this working a la inspire.cern.ch

bfe_

I've also modified some regular invenio elements (and CERN specific elems) for my purposes, these are also in inspire CVS.

bfe_field

Added some filtering abilites in order to display only fields and subfields that match certain criteria. Not clear that this is better than doing this in custom bfes.

format_templates

Default_HTML_brief.bft

Committed in INSPIRE CVS

webstyle_templates

webstyle_templates_inspire.py

Also managed in inspire cvs

This template needs to be updated to work with invenio v0.99 It includes inspire.css as well.

Tibor and Marko to work on this

bibconvert/data mapping

Working with the formats brings up some data related questions:

DOI/Journal info

currently we store both DOI and journal info in 773a How do we distinguish and display? answer is that we need to use p,v,d,n for the other stuff, or if we prefer to store in one subfield, store it in something different than 773a

SPIRES2MARC.xsl, and SPIRES output adjusted for this

URL names

spires labels urls in pre-url. This information was never imported. Really we should replace the url-lookups with names that are hard coded in the records? Or should we use a dynamic url generator for journals/dois etc, and then store other links separately in 856u with display names?

bfe_INSPIRE_links

the above two problems are resolved in bfe_INSPIRE_links.py, to some extent.

Dates

we need to decide on a basic storage format. At one time we decided on barebones 20081102 but I think the hyphenated verion of this is better, need to have a convention for precision only at the month level.

it would be nice to change this in the dateutils module itself (the month or year only precision) We should also change the datetext output to allow for both EU and US style dates (can we do this with i18n? a config setting?)

Original version

Fall 07 Jerome and other CERN folks implemented a SPIRES format example. IT was described and dissected here and more information is available in earlier versions of this twiki page.

Suggestions for improvement from original version (Fall 07)

• display jnl pub info (marc 773)
• display conf pubnote CPBN (no marc tag yet?)
• distinguish between DESY, free (author) and automatically generated keywords and use different notations in formats
• display only year (marc 269) if day/month not available - suppress trailing zeroes
• add "p" after page nr
• no display of DESY-record-nr DRN (internal use) (marc 088, subfield 9 = DESY)

• html brief:
  • no display of keywords option if fulltext not available
  • no display of rep nrs (marc 037) after "eprint:" unless subfield s = "arxiv" (leads to wrong reference link)

• detailed record:
  • use FIELD-CODE and TYPE-CODE (no marc tag yet?) instead of PPF-SUBJECT (marc 650)
  • display DOI
  • smaller font for Meeting Note (marc 111a) (smaller than title font)
  • remove double titles (display only marc 245, not 246)
  • publication notes for books (DPBN, marc 500) should not appear as "Note" (-> Imprint?). Ideally should be split up and entered into the resp marc fields (260). DPBN example: Reinbek, Germany: Rowohlt (2005) 189 p

Institutions

• country acronyms to be replaced by full name via KB

• detailed record:
  • inside the box:
    • "Theses in HEP"
  • display longitude/latitude + map to the right
  • display public note (680i)
  • display historical data (6781a)
  • display parent/offspring links (510)