diff --git a/data/corpus-simple-50.json b/data/corpus-simple-50.json index 1dd92d9..b593a41 100644 --- a/data/corpus-simple-50.json +++ b/data/corpus-simple-50.json @@ -290,7 +290,8 @@ "publication_year": 2014, "source": "JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES", "affiliations": [ - "Royal Netherlands Meteorol Inst, NL-3730 AE De Bilt, Netherlands" + "Royal Netherlands Meteorol Inst, NL-3730 AE De Bilt, Netherlands", + "Royal Netherlands Meteorol Inst, POB 201, NL-3730 AE De Bilt, Netherlands" ], "countries": [ "Netherlands" @@ -435,7 +436,8 @@ "publication_year": 2014, "source": "JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES", "affiliations": [ - "Lawrence Livermore Natl Lab, Program Climate Model Diag & Intercomparison, Livermore, CA USA" + "Lawrence Livermore Natl Lab, Program Climate Model Diag & Intercomparison, Livermore, CA USA", + "Space & Naval Warfare Syst Command Syst Ctr, Atmospher Propagat Branch, San Diego, CA 92152 USA" ], "countries": [ "USA" @@ -554,7 +556,8 @@ "NASA, Goddard Inst Space Studies, New York, NY 10025 USA", "Seoul Natl Univ, Sch Earth & Environm Sci, Seoul, South Korea", "Japan Agcy Marine Earth Sci & Technol, Yokohama, Kanagawa, Japan", - "Univ Reading, Natl Ctr Atmospher Sci, Reading, Berks, England" + "Univ Reading, Natl Ctr Atmospher Sci, Reading, Berks, England", + "Max Planck Inst Meteorol, Bundesstr 55, D-20146 Hamburg, Germany" ], "countries": [ "Germany", diff --git a/extract-fields.ini b/extract-fields.ini index f8fb67e..5b8e56d 100644 --- a/extract-fields.ini +++ b/extract-fields.ini @@ -15,6 +15,13 @@ [JSONParse] +[assign] +path = tmp.reprint_addresses +value = get('static_data.fullrecord_metadata.reprint_addresses.address_name', []).castArray().map(name => name.address_spec.full_address) + +path = tmp.addresses +value = get('static_data.fullrecord_metadata.addresses.address_name', []).castArray().map(name => name.address_spec.full_address) + [replace] path = uri value = get("UID") @@ -31,9 +38,8 @@ path = source value = get('static_data.summary.titles.title').filter(title => title.type === "source").map(title => title.content).join('|') -; TODO: ajouter les adresses de reprint_addresses (au même niveau que addresses) path = affiliations -value = get('static_data.fullrecord_metadata.addresses.address_name', []).castArray().map(name => name.address_spec.full_address) +value = get('tmp.addresses', []).concat(_.get(self, 'tmp.reprint_addresses')).remove(null).uniq() path = countries value = get('static_data.fullrecord_metadata.addresses.address_name', []).castArray().map("address_spec.country", []).uniq()