From 114c6b611148d2ff499bcea302eee0eca00df647 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 6 Apr 2018 12:39:49 -0700 Subject: small grobid2json test --- mapreduce/tests/files/small.json | 43 +++++++++++++++ mapreduce/tests/files/small.xml | 110 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 153 insertions(+) create mode 100644 mapreduce/tests/files/small.json create mode 100644 mapreduce/tests/files/small.xml (limited to 'mapreduce/tests/files') diff --git a/mapreduce/tests/files/small.json b/mapreduce/tests/files/small.json new file mode 100644 index 0000000..208fb49 --- /dev/null +++ b/mapreduce/tests/files/small.json @@ -0,0 +1,43 @@ +{ + "title": "Dummy Example File", + "authors": [ + {"name": "Brewster Kahle"}, + {"name": "J Doe"} + ], + "journal": { + "name": "Dummy Example File. Journal of Fake News. pp. 1-2. ISSN 1234-5678", + "eissn": null, + "issn": null, + "issue": null, + "publisher": null, + "volume": null + }, + "date": "2000", + "doi": null, + "citations": [ + { "authors": [{"name": "A Seaperson"}], + "date": "2001", + "id": "b0", + "index": 0, + "issue": null, + "journal": "Letters in the Alphabet", + "publisher": null, + "title": "Everything is Wonderful", + "url": null, + "volume": "20"}, + { "authors": [], + "date": "2011-03-28", + "id": "b1", + "index": 1, + "issue": null, + "journal": "The Dictionary", + "publisher": null, + "title": "All about Facts", + "url": null, + "volume": "14"} + ], + "abstract": "Everything you ever wanted to know about nothing", + "body": "Introduction \nEverything starts somewhere, as somebody [1] once said. \n\n In Depth \n Meat \nYou know, for kids. \n Potatos \nQED.", + "acknowledgement": null, + "annex": null +} diff --git a/mapreduce/tests/files/small.xml b/mapreduce/tests/files/small.xml new file mode 100644 index 0000000..78b9ba2 --- /dev/null +++ b/mapreduce/tests/files/small.xml @@ -0,0 +1,110 @@ + + + + + + + GROBID - A machine learning software for extracting information from scholarly documents + + + + + + Dummy Example File + + + + + 2000 + + + + + + BrewsterKahle + + + JDoe + + + + Internet Archive + + + Dummy Example File + + + Dummy Example File. Journal of Fake News. pp. 1-2. ISSN 1234-5678 + + 2000 + + + + + + + + + Fake Data + + + +

Everything you ever wanted to know about nothing

+
+
+
+ + +
Introduction

+Everything starts somewhere, as somebody[1] once said.

+ +
In Depth
+
Meat

+You know, for kids.

+
Potatos

+QED.

+ + +
+ + + + + + Everything is Wonderful + + ASeaperson + + + + Letters in the Alphabet + + 20 + + + + + + + + + All about Facts + + + The Dictionary + + 14 + + + + None + + + +
+
+
+
-- cgit v1.2.3