start refactoring into new parser, with dataclass types

2 files changed, 79 insertions, 3 deletions

diff --git a/tests/test_grobid2json.py b/tests/test_grobid2json.py
index 6e3dac2..ed5d996 100644
--- a/tests/test_grobid2json.py
+++ b/tests/test_grobid2json.py
@@ -1,10 +1,12 @@
 import xml
 import json
 import pytest
-from grobid_tei_xml.grobid2json import teixml2json
 
+from grobid_tei_xml import teixml2json, parse_document_xml, GrobidDocument, GrobidCitation
+from grobid_tei_xml.types import *
 
-def test_small_xml():
+
+def test_teixml2json_small_xml():
 
     with open('tests/files/small.xml', 'r') as f:
         tei_xml = f.read()
@@ -13,6 +15,80 @@ def test_small_xml():
 
     assert teixml2json(tei_xml) == json_form
 
+    assert parse_document_xml(tei_xml).to_dict() == json_form
+
+def test_teixml2json_small_xml():
+
+    with open('tests/files/small.xml', 'r') as f:
+        tei_xml = f.read()
+
+    doc = parse_document_xml(tei_xml)
+    expected = GrobidDocument(
+        grobid_version='0.5.1-SNAPSHOT',
+        grobid_timestamp='2018-04-02T00:31+0000',
+        language_code='en',
+        header=GrobidHeader(
+            title="Dummy Example File",
+            authors=[
+                GrobidAuthor(
+                    name="Brewster Kahle",
+                    given_name="Brewster",
+                    surname="Kahle",
+                    affiliation=GrobidAffiliation(
+                        department="Faculty ofAgricultrial Engineering",
+                        laboratory="Plant Physiology Laboratory",
+                        institution="Technion-Israel Institute of Technology",
+                        address=GrobidAddress(
+                            post_code="32000",
+                            settlement="Haifa",
+                            country="Israel",
+                        ),
+                    )
+                ),
+                GrobidAuthor(
+                    name="J Doe",
+                    given_name="J",
+                    surname="Doe",
+                ),
+            ],
+            journal=GrobidJournal(
+                name="Dummy Example File. Journal of Fake News. pp. 1-2. ISSN 1234-5678",
+            ),
+            date="2000",
+        ),
+        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.",
+        citations=[
+            GrobidCitation(
+                index=0,
+                id="b0",
+                authors=[
+                    GrobidAuthor(
+                        name="A Seaperson",
+                        given_name="A",
+                        surname="Seaperson"
+                    )
+                ],
+                date="2001",
+                journal="Letters in the Alphabet",
+                title="Everything is Wonderful",
+                volume="20",
+                pages="1-11",
+            ),
+            GrobidCitation(
+                index=1,
+                id="b1",
+                authors=[],
+                date="2011-03-28",
+                journal="The Dictionary",
+                title="All about Facts",
+                volume="14",
+            ),
+        ],
+    )
+
+    assert doc == expected
+
 
 def test_invalid_xml():
 
diff --git a/tests/test_grobid_unstructured.py b/tests/test_grobid_unstructured.py
index b203b30..91b7398 100644
--- a/tests/test_grobid_unstructured.py
+++ b/tests/test_grobid_unstructured.py
@@ -1,6 +1,6 @@
 import pytest
 
-from grobid_tei_xml.grobid_unstructured import transform_grobid_ref_xml
+from grobid_tei_xml.grobid2json import transform_grobid_ref_xml
 
 
 def test_transform_grobid_ref_xml():