From 6bbdc0df671cb2544ae3e172c004582be578ade0 Mon Sep 17 00:00:00 2001 From: Daniel Ecer <de-code@users.noreply.github.com> Date: Tue, 30 Jan 2018 14:54:33 +0000 Subject: [PATCH] fixed contrib-type at wrong level --- sciencebeam_gym/inference_model/extract_to_xml.py | 6 +++--- sciencebeam_gym/inference_model/extract_to_xml_test.py | 1 + 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/sciencebeam_gym/inference_model/extract_to_xml.py b/sciencebeam_gym/inference_model/extract_to_xml.py index 5a0c70f..7436434 100644 --- a/sciencebeam_gym/inference_model/extract_to_xml.py +++ b/sciencebeam_gym/inference_model/extract_to_xml.py @@ -25,7 +25,7 @@ class Tags(object): class XmlPaths(object): TITLE = 'front/article-meta/title-group/article-title' ABSTRACT = 'front/article-meta/abstract' - AUTHOR = 'front/article-meta/contrib-group/contrib/name' + AUTHOR = 'front/article-meta/contrib-group/contrib' AUTHOR_AFF = 'front/article-meta/contrib-group/aff' class SubTags(object): @@ -33,8 +33,8 @@ class SubTags(object): AUTHOR_GIVEN_NAMES = 'givennames' class SubXmlPaths(object): - AUTHOR_SURNAME = 'surname' - AUTHOR_GIVEN_NAMES = 'given-names' + AUTHOR_SURNAME = 'name/surname' + AUTHOR_GIVEN_NAMES = 'name/given-names' def get_logger(): return logging.getLogger(__name__) diff --git a/sciencebeam_gym/inference_model/extract_to_xml_test.py b/sciencebeam_gym/inference_model/extract_to_xml_test.py index 1d8b92c..223a905 100644 --- a/sciencebeam_gym/inference_model/extract_to_xml_test.py +++ b/sciencebeam_gym/inference_model/extract_to_xml_test.py @@ -140,6 +140,7 @@ class TestExtractedItemsToXml(object): assert xml_root is not None author = xml_root.find(XmlPaths.AUTHOR) assert author is not None + assert author.tag == 'contrib' assert author.attrib.get('contrib-type') == 'author' class TestMain(object): -- GitLab