From bd508b4ffbab3ea56134f753a9746ff9d582fde3 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Wed, 25 Sep 2019 18:02:43 -0700
Subject: test of GROBID client

---
 python/tests/test_grobid.py | 53 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 python/tests/test_grobid.py

diff --git a/python/tests/test_grobid.py b/python/tests/test_grobid.py
new file mode 100644
index 0000000..fca234a
--- /dev/null
+++ b/python/tests/test_grobid.py
@@ -0,0 +1,53 @@
+
+import pytest
+import struct
+import responses
+
+from sandcrawler import GrobidClient
+
+
+FAKE_PDF_BYTES = b"%PDF SOME JUNK" + struct.pack("!q", 112853843)
+
+with open('tests/files/23b29ea36382680716be08fc71aa81bd226e8a85.xml', 'rb') as f:
+    REAL_TEI_XML = f.read()
+
+@responses.activate
+def test_grobid_503():
+
+    client = GrobidClient(host_url="http://localhost:8070")
+
+    status = b'{"status": "done broke due to 503"}'
+    responses.add(responses.POST,
+        'http://localhost:8070/api/processFulltextDocument', status=503,
+        body=status)
+
+    resp = client.process_fulltext(FAKE_PDF_BYTES)
+
+    # grobid gets POST 1x times
+    assert len(responses.calls) == 1
+
+    assert resp['status_code'] == 503
+    assert resp['status'] == "error"
+    print(resp)
+    assert False
+
+@responses.activate
+def test_grobid_503():
+
+    client = GrobidClient(host_url="http://localhost:8070")
+
+    responses.add(responses.POST,
+        'http://localhost:8070/api/processFulltextDocument', status=200,
+        body=REAL_TEI_XML, content_type='text/xml')
+
+    resp = client.process_fulltext(FAKE_PDF_BYTES)
+
+    # grobid gets POST 1x times
+    assert len(responses.calls) == 1
+
+    assert resp['status_code'] == 200
+    assert resp['status'] == "success"
+    print(type(resp['tei_xml']))
+    print(type(REAL_TEI_XML))
+    assert resp['tei_xml'] == REAL_TEI_XML.decode('utf-8')
+    #assert resp['tei_xml'].split('\n')[:3] == REAL_TEI_XML.split('\n')[:3]
-- 
cgit v1.2.3