more JALC importer polish

2 files changed, 46 insertions, 6 deletions

diff --git a/python/fatcat_tools/importers/jalc.py b/python/fatcat_tools/importers/jalc.py
index 09b8bd76..e2ccb230 100644
--- a/python/fatcat_tools/importers/jalc.py
+++ b/python/fatcat_tools/importers/jalc.py
@@ -101,28 +101,37 @@ class JalcImporter(EntityImporter):
                 name = eng
                 if not name.find('name'):
                     name = jpn
+                surname = name.find('familyName')
+                if surname:
+                    surname = surname.string
                 contrib = fatcat_client.ReleaseContrib(
                     raw_name=clean(name.find('name').string),
                     given_name=clean(name.find('givenName').string),
-                    surname=clean(name.find('familyName').string),
+                    surname=clean(surname),
                     role='author',
                 )
                 if eng.find('name') and jpn.find('name'):
+                    jpn_surname = jpn.find('familyName')
+                    if jpn_surname:
+                        jpn_surname = jpn_surname.string
                     contrib.extra = {
                         'original_name': {
                             'lang': 'ja',
                             'raw_name': clean(jpn.find('name').string),
                             'given_name': clean(jpn.find('givenName').string),
-                            'surname': clean(jpn.find('familyName').string),
+                            'surname': clean(jpn_surname),
                         }}
                 contribs.append(contrib)
         elif people:
             # TODO: test for this codepath?
             for eng in people:
+                surname = eng.find('familyName')
+                if surname:
+                    surname = surname.string
                 contrib = dict(
                     raw_name=clean(eng.find('name').string),
                     given_name=clean(eng.find('givenName').string),
-                    surname=clean(eng.find('familyName').string),
+                    surname=clean(surname),
                     role='author',
                 )
                 contribs.append(contrib)
@@ -228,10 +237,14 @@ class JalcImporter(EntityImporter):
         # (informally)
         extra['jalc'] = extra_jalc
 
+        title = clean(title)
+        if not title:
+            return None
+
         re = fatcat_client.ReleaseEntity(
             work_id=None,
             title=title,
-            original_title=original_title,
+            original_title=clean(original_title),
             release_type="article-journal",
             release_stage='published',
             release_date=release_date,
diff --git a/python/tests/import_jalc.py b/python/tests/import_jalc.py
index e92c26c0..693e77f4 100644
--- a/python/tests/import_jalc.py
+++ b/python/tests/import_jalc.py
@@ -1,7 +1,7 @@
 
 import json, gzip
 import pytest
-from fatcat_tools.importers import JalcImporter, Bs4XmlFilePusher
+from fatcat_tools.importers import JalcImporter, Bs4XmlFilePusher, Bs4XmlLinesPusher
 from fixtures import api
 from bs4 import BeautifulSoup
 
@@ -43,6 +43,33 @@ def test_jalc_importer(jalc_importer):
     assert counts['skip'] == 0
     assert last_index == jalc_importer.api.get_changelog(limit=1)[0].index
 
+def test_jalc_importer_lines(jalc_importer):
+    last_index = jalc_importer.api.get_changelog(limit=1)[0].index
+    with open('tests/files/jalc_rdf_sample_100.xml', 'r') as f:
+        jalc_importer.bezerk_mode = True
+        counts = Bs4XmlLinesPusher(jalc_importer, f, "<rdf:Description").run()
+    assert counts['insert'] == 93
+    assert counts['exists'] == 0
+    assert counts['skip'] == 0
+
+    # fetch most recent editgroup
+    change = jalc_importer.api.get_changelog_entry(index=last_index+1)
+    eg = change.editgroup
+    assert eg.description
+    assert "jalc" in eg.description.lower()
+    assert eg.extra['git_rev']
+    assert "fatcat_tools.JalcImporter" in eg.extra['agent']
+
+    last_index = jalc_importer.api.get_changelog(limit=1)[0].index
+    with open('tests/files/jalc_rdf_sample_100.xml', 'r') as f:
+        jalc_importer.bezerk_mode = False
+        jalc_importer.reset()
+        counts = Bs4XmlLinesPusher(jalc_importer, f, "<rdf:Description").run()
+    assert counts['insert'] == 0
+    assert counts['exists'] == 93
+    assert counts['skip'] == 0
+    assert last_index == jalc_importer.api.get_changelog(limit=1)[0].index
+
 def test_jalc_xml_parse(jalc_importer):
     with open('tests/files/jalc_lod_sample.xml', 'r') as f:
         soup = BeautifulSoup(f, "xml")
@@ -51,7 +78,7 @@ def test_jalc_xml_parse(jalc_importer):
     print(r.extra)
     assert r.title == "New carbides in the Ni-Ti-Mo-C system"
     assert r.subtitle == None
-    assert r.original_title == "Ｎｉ－Ｔｉ－Ｍｏ－Ｃ系に出現する新炭化物相について"
+    assert r.original_title == "Ni-Ti-Mo-C系に出現する新炭化物相について"
     assert r.publisher == "Japan Society of Powder and Powder Metallurgy"
     assert r.release_type == "article-journal"
     assert r.release_stage == "published"