From 560d5f7cc1672f95e2a953ab5908f4205151a703 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Wed, 27 Oct 2021 18:24:19 -0700
Subject: refactor use of grobid_tei_xml

---
 tests/test_grobid2json.py    | 31 ----------------------
 tests/test_grobid_parse.py   | 62 ++++++++++++++++++++++++++++++++++++++++++++
 tests/test_refs_transform.py |  3 +--
 3 files changed, 63 insertions(+), 33 deletions(-)
 delete mode 100644 tests/test_grobid2json.py
 create mode 100644 tests/test_grobid_parse.py

(limited to 'tests')

diff --git a/tests/test_grobid2json.py b/tests/test_grobid2json.py
deleted file mode 100644
index adf36a1..0000000
--- a/tests/test_grobid2json.py
+++ /dev/null
@@ -1,31 +0,0 @@
-from grobid_tei_xml import parse_document_xml
-
-
-def test_grobid_parse() -> None:
-    """
-    This function formerly tested the grobid2json file in this project. Now it
-    tests backwards-compatibility of the grobid_tei_xml library.
-    """
-
-    with open("tests/files/example_grobid.tei.xml", "r") as f:
-        blob = f.read()
-
-    doc = parse_document_xml(blob)
-    obj = doc.to_legacy_dict()
-
-    assert (
-        obj["title"]
-        == "Changes of patients' satisfaction with the health care services in Lithuanian Health Promoting Hospitals network"
-    )
-
-    ref = [c for c in obj["citations"] if c["id"] == "b12"][0]
-    assert ref["authors"][0] == {"given_name": "K", "name": "K Tasa", "surname": "Tasa"}
-    assert ref["journal"] == "Quality Management in Health Care"
-    assert ref["title"] == "Using patient feedback for quality improvement"
-    assert ref["date"] == "1996"
-    assert ref["pages"] == "206-225"
-    assert ref["volume"] == "8"
-    assert (
-        ref["unstructured"]
-        == "Tasa K, Baker R, Murray M. Using patient feedback for qua- lity improvement. Quality Management in Health Care 1996;8:206-19."
-    )
diff --git a/tests/test_grobid_parse.py b/tests/test_grobid_parse.py
new file mode 100644
index 0000000..c0adf9b
--- /dev/null
+++ b/tests/test_grobid_parse.py
@@ -0,0 +1,62 @@
+from grobid_tei_xml import parse_document_xml
+
+
+def test_grobid_parse_legacy() -> None:
+    """
+    This function formerly tested the grobid2json file in this project. Now it
+    tests backwards-compatibility of the grobid_tei_xml library.
+    """
+
+    with open("tests/files/example_grobid.tei.xml", "r") as f:
+        blob = f.read()
+
+    doc = parse_document_xml(blob)
+    obj = doc.to_legacy_dict()
+
+    assert (
+        obj["title"]
+        == "Changes of patients' satisfaction with the health care services in Lithuanian Health Promoting Hospitals network"
+    )
+
+    ref = [c for c in obj["citations"] if c["id"] == "b12"][0]
+    assert ref["authors"][0] == {"given_name": "K", "name": "K Tasa", "surname": "Tasa"}
+    assert ref["journal"] == "Quality Management in Health Care"
+    assert ref["title"] == "Using patient feedback for quality improvement"
+    assert ref["date"] == "1996"
+    assert ref["pages"] == "206-225"
+    assert ref["volume"] == "8"
+    assert (
+        ref["unstructured"]
+        == "Tasa K, Baker R, Murray M. Using patient feedback for qua- lity improvement. Quality Management in Health Care 1996;8:206-19."
+    )
+
+
+def test_grobid_parse() -> None:
+    """
+    Equivalent to test_grobid_parse_legacy(), but using the GrobidDocument type directly
+    """
+
+    with open("tests/files/example_grobid.tei.xml", "r") as f:
+        blob = f.read()
+
+    doc = parse_document_xml(blob)
+
+    assert (
+        doc.header.title
+        == "Changes of patients' satisfaction with the health care services in Lithuanian Health Promoting Hospitals network"
+    )
+
+    assert doc.citations is not None
+    ref = [c for c in doc.citations if c.id == "b12"][0]
+    assert ref.authors[0].given_name == "K"
+    assert ref.authors[0].full_name == "K Tasa"
+    assert ref.authors[0].surname == "Tasa"
+    assert ref.journal == "Quality Management in Health Care"
+    assert ref.title == "Using patient feedback for quality improvement"
+    assert ref.date == "1996"
+    assert ref.pages == "206-225"
+    assert ref.volume == "8"
+    assert (
+        ref.unstructured
+        == "Tasa K, Baker R, Murray M. Using patient feedback for qua- lity improvement. Quality Management in Health Care 1996;8:206-19."
+    )
diff --git a/tests/test_refs_transform.py b/tests/test_refs_transform.py
index 2fc210f..4b1b588 100644
--- a/tests/test_refs_transform.py
+++ b/tests/test_refs_transform.py
@@ -20,8 +20,7 @@ def test_transform_refs_grobid() -> None:
     )
 
     tei_doc = parse_document_xml(blob)
-    tei_dict = tei_doc.to_legacy_dict()
-    refs = refs_from_grobid(dummy_release, tei_dict)
+    refs = refs_from_grobid(dummy_release, tei_doc)
 
     ref = refs[12]
     assert ref.release_ident == "releasedummy22222222222222"
-- 
cgit v1.2.3