Merge branch 'bnewbold-grobid-tei-xml' into 'master'

use grobid_tei_xml for grobid unstructured lookups

See merge request webgroup/fuzzycat!9

1 files changed, 32 insertions, 26 deletions

diff --git a/tests/test_grobid_unstructured.py b/tests/test_grobid_unstructured.py
index dd69936..cf71f91 100644
--- a/tests/test_grobid_unstructured.py
+++ b/tests/test_grobid_unstructured.py
@@ -1,39 +1,43 @@
 import pytest
 
-from fuzzycat.grobid_unstructured import grobid_api_process_citation, grobid_parse_unstructured, grobid_ref_to_release, transform_grobid_ref_xml
+from grobid_tei_xml import parse_citation_xml
+from grobid_tei_xml.types import GrobidBiblio, GrobidAuthor
+
+from fuzzycat.grobid_unstructured import grobid_api_process_citation, grobid_parse_unstructured, grobid_ref_to_release
 
 
 def test_grobid_ref_to_release():
 
-    d = {
-        'title':
-        "some title",
-        'doi':
-        '10.1234/5678',
-        'journal':
-        'some journal',
-        'authors': [
-            {
-                'name': 'ahab sailor',
-                'given_name': 'ahab',
-                'surname': 'sailor'
-            },
-            {
-                'name': 'mary jane',
-                'given_name': 'mary',
-                'surname': 'jane'
-            },
+    d = GrobidBiblio(
+        title="some title",
+        doi='10.1234/5678',
+        journal='some journal',
+        authors=[
+            GrobidAuthor(
+                full_name='ahab sailor',
+                given_name='ahab',
+                surname='sailor',
+            ),
+            GrobidAuthor(
+                full_name='mary jane',
+                given_name='mary',
+                surname='jane'
+            ),
         ],
-    }
+    )
     r = grobid_ref_to_release(d)
-    assert r.title == d['title']
-    assert r.ext_ids.doi == d['doi']
-    assert r.extra['container_name'] == d['journal']
-    assert r.contribs[0].surname == d['authors'][0]['surname']
-    assert r.contribs[1].raw_name == d['authors'][1]['name']
+    assert r.title == d.title
+    assert r.ext_ids.doi == d.doi
+    assert r.extra['container_name'] == d.journal
+    assert r.contribs[0].surname == d.authors[0].surname
+    assert r.contribs[1].raw_name == d.authors[1].full_name
 
 
 def test_transform_grobid_ref_xml():
+    """
+    This used to be a test of the grobid2json file in this repository. Now it
+    is a backwards compatibility test for grobid_tei_xml
+    """
     citation_xml = """
 <biblStruct >
     <analytic>
@@ -83,7 +87,9 @@ def test_transform_grobid_ref_xml():
     </monogr>
 </biblStruct>"""
 
-    d = transform_grobid_ref_xml(citation_xml)
+    citation = parse_citation_xml(citation_xml)
+    assert citation
+    d = citation.to_legacy_dict()
     assert d['title'] == "Mesh migration following abdominal hernia repair: a comprehensive review"
     assert d['authors'][2]['given_name'] == "L"
     assert d['authors'][2]['surname'] == "Taveras"