references: builder should support 999C5d

[jacquerie]

From @jacquerie on April 22, 2017 13:40

Similar to #2245, the builder should support 999C5d and populate reference.document_type from it. For example:

<datafield tag="999" ind1="C" ind2="5">
  <subfield code="e">M. Altarelli</subfield>
  <subfield code="y">2007</subfield>
  <subfield code="m">Technical Design Report No</subfield>
  <subfield code="r">DESY 06-097</subfield>
  <subfield code="o">1</subfield>
  <subfield code="d">report</subfield>
</datafield>

in https://inspirehep.net/record/1593152/export/xme.

Copied from original issue: inspirehep/inspire-next#2275

[jacquerie]

On the other hand, this field must be cleaned of bad data, and we need to know how to map certain fields to our document types:

170782 journal
7521 book
5052 misc
4240 eprint
4230 proc
1581 report
 818 thesis
 289 supplemental-material
   5 1988
   4 1969
   3 1968
   2 Phys.Rev.,D59,086004
   2 1997
   2 1990
   2 1985
   1 translation
   1 mosc
   1 jouranl
   1 SPIRES-9711244
   1 Rev.Mod.Phys.,64,885
   1 Prog.Theor.Exp.Phys.,2015,011C01
   1 Phys.Rev.,D75,054029
   1 Phys.Lett.,B551,360
   1 Ph. D Thesis
   1 PTEP,2012,02B008
   1 Open Appl.Math.J.,I,1
   1 Nuovo Cim.,B37,1
   1 Nucl.Phys.,B28,253
   1 Nature,168,357
   1 Nature,130,363
   1 Living Rev. Solar Phys.,2,1
   1 JHEP
   1 JETP Lett.,86,157
   1 J Algebra Appl.,6,415-441
   1 Gen.Rel.Grav.,40,1653
   1 DOI: http://dx.doi.org/10.1103/PhysRev.50.48
   1 DELPHI Collaboration
   1 Commun.Math.Phys.,277,771
   1 April, 1994
   1 2004
   1 2001
   1 1995
   1 1994
   1 1993
   1 1991
   1 1989
   1 1987
   1 1981
   1 1967
   1 10.1016/j.nima.2010.11.016
   1 10.1016/S0168-9002(03)01368-8
   1 . Sinclair, A. L. Carter, D. Kessler, E. D. Earle, P. Jagam, J. J. Simpson, R. C. Allen, H. H. Chen, P. J. Doe, E. D. Hallman, W. F. Davidson, A. B. McDonald, R. S. Storey, G. T. Ewan, H. B. Mak, and B. C. Robertson,

@michamos, what are we going to do with misc, eprint, and supplemental-material?

[jacquerie]

From @michamos on April 24, 2017 8:41

eprint -> article as we don't distinguish between published/unpublished but only peer reviewed/non-peer reviewed. misc seems to indicate a reference that APS didn't know how to interpret (also footnotes sometimes, as they are mixed with the references in Phys.Rev.) and supplemental-material is for extra information to the paper (appendices that didn't make the length cut). So you can leave them out, as they don't translate to anything in our data model.