def file_of_published_article(self, filename):
eif_filename_without_path = os.path.basename(filename)
article_info = ArticleInfo(eif_filename_without_path)
version = article_info.get_version_from_zip_filename()
update_date = article_info.get_update_date_from_zip_filename()
if version != None and update_date != None:
return True
return False
class TestArticleStructure(unittest.TestCase):
@unpack
@data({'input': 'elife-07702-vor-r4.zip', 'expected': None},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected':'2012-10-15T00:00:00Z'})
def test_get_update_date_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_update_date_from_zip_filename()
self.assertEqual(result, expected)
@unpack
@data({'input': 'elife-07702-vor-r4.zip', 'expected': None},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected': '1'})
def test_get_version_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_version_from_zip_filename()
self.assertEqual(result, expected)
def get_article_xml_key(bucket, expanded_folder_name):
files = bucket.list(expanded_folder_name + "/", "/")
for bucket_file in files:
key = bucket.get_key(bucket_file.key)
filename = key.name.rsplit('/', 1)[1]
info = ArticleInfo(filename)
if info.file_type == 'ArticleXML':
return key, filename
return None
def process_key(self, key, cdn_path):
# determine filename (without folder) and obtain ArticleInfo instance
filename = key.name.rsplit('/', 1)[1]
info = ArticleInfo(filename)
# see if we have any formats available for the file_type of this file
formats = self.get_formats(info.file_type)
if formats is not None:
# generate images for relevant formats
fp = self.get_file_pointer(key)
self.generate_images(formats, fp, info, cdn_path)
def get_xml_file_name(self, settings, expanded_folder_name, xml_bucket, version):
files = self._get_bucket_files(settings, expanded_folder_name, xml_bucket)
for filename in files:
info = ArticleInfo(filename)
if info.file_type == 'ArticleXML':
if version is None:
return filename
v_number = '-v'+ version + '.'
if v_number in filename:
return filename
return None
def do_activity(self, data=None):
try:
session = Session(self.settings)
version = session.get_value(data['run'], 'version')
filename = session.get_value(data['run'], 'filename_last_element')
article_structure = ArticleInfo(filename)
version_date, error = self.get_version(
self.settings, article_structure, article_structure.article_id,
version)
if error is not None:
self.logger.error(error)
self.emit_monitor_event(
self.settings, article_structure.article_id, version,
data['run'], self.pretty_name, "error", " ".join(
("Error Looking up version article",
article_structure.article_id, "message:", error)))
return activity.activity.ACTIVITY_PERMANENT_FAILURE
self.emit_monitor_event(
self.settings, article_structure.article_id, version,
data['run'], self.pretty_name, "end", " ".join(
("Finished Version Lookup for article",
article_structure.article_id, "version:", version)))
session.store_value(data['run'], 'update_date', version_date)
return activity.activity.ACTIVITY_SUCCESS
except Exception as e:
self.logger.exception(
"Exception when trying to Lookup next version")
self.emit_monitor_event(
self.settings, article_structure.article_id, version,
data['run'], self.pretty_name, "error", " ".join(
("Error looking up version for article",
article_structure.article_id, "message:", str(e))))
return activity.activity.ACTIVITY_PERMANENT_FAILURE
def test_get_version_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_version_from_zip_filename()
self.assertEqual(result, expected)
def find_xml_filename_in_map(self, file_name_map):
for old_name, new_name in file_name_map.iteritems():
info = ArticleInfo(new_name)
if info.file_type == 'ArticleXML':
return new_name
class TestArticleStructure(unittest.TestCase):
@unpack
@data({'input': 'elife-07702-vor-r4.zip', 'expected': None},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected':'2012-10-15T00:00:00Z'})
def test_get_update_date_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_update_date_from_zip_filename()
self.assertEqual(result, expected)
@unpack
@data({'input': 'elife-07702-vor-r4.zip', 'expected': None},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected': '1'})
def test_get_version_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_version_from_zip_filename()
self.assertEqual(result, expected)
@unpack
@data(
{'input': 'elife-07702-vor-r4.zip', 'expected': 'ArticleZip'},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected': 'ArticleZip'},
{'input': 'elife-00666-v1.pdf', 'expected': 'Other'},
{'input': 'elife-00666-v1.xml', 'expected': 'ArticleXML'},
{'input': 'elife-00666-app1-fig1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-app1-fig1-figsupp1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-app2-video1.mp4', 'expected': 'Other'},
{'input': 'elife-00666-box2-fig1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-code1-v1.xml', 'expected': 'Other'},
{'input': 'elife-00666-data1-v1.xlsx', 'expected': 'Other'},
{'input': 'elife-00666-fig1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig2-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig2-figsupp1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig2-figsupp2-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig3-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig3-v10.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig3-figsupp1-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig3-video1.mp4', 'expected': 'Other'},
{'input': 'elife-00666-fig4-v1.tif', 'expected': 'Figure'},
{'input': 'elife-00666-fig4-code1-v1.xlsx', 'expected': 'Other'},
{'input': 'elife-00666-figures-v1.pdf', 'expected': 'FigurePDF'},
{'input': 'elife-00666-inf001-v1.jpeg', 'expected': 'Inline'},
{'input': 'elife-00666-repstand1-v1.pdf', 'expected': 'Other'},
{'input': 'elife-00666-resp-fig1-v1.png', 'expected': 'Figure'},
{'input': 'elife-00666-resp-video1.mp4', 'expected': 'Other'},
{'input': 'elife-00666-supp1-v1.csv', 'expected': 'Other'},
{'input': 'elife-00666-supp2-v2.tif', 'expected': 'Other'},
{'input': 'elife-00666-supp2-v3.docx', 'expected': 'Other'},
{'input': 'elife-00666-table3-data1-v1.xlsx', 'expected': 'Other'},
{'input': 'elife-00666-video1.mp4', 'expected': 'Other'},
{'input': 'elife-00666-video1-data1-v1.xlsx', 'expected': 'Other'},
)
def test_get_file_type_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.file_type
self.assertEqual(result, expected)
@unpack
@data(
{'input': 'elife-07702-vor-r4.zip', 'expected': False},
{'input': 'elife-00013-vor-v1-20121015000000.zip', 'expected': False},
{'input': 'elife-00666-v1.pdf', 'expected': False},
{'input': 'elife-00666-v1.xml', 'expected': False},
{'input': 'elife-00666-app1-fig1-v1.tif', 'expected': True},
{'input': 'elife-00666-app1-fig1-figsupp1-v1.tif', 'expected': True},
{'input': 'elife-00666-app2-video1.mp4', 'expected': False},
{'input': 'elife-00666-box2-fig1-v1.tif', 'expected': True},
{'input': 'elife-00666-code1-v1.xml', 'expected': False},
{'input': 'elife-00666-data1-v1.xlsx', 'expected': False},
{'input': 'elife-00666-fig1-v1.tif', 'expected': True},
{'input': 'elife-00666-fig2-v1.tif', 'expected': True},
{'input': 'elife-00666-fig2-figsupp1-v1.tif', 'expected': True},
{'input': 'elife-00666-fig2-figsupp2-v1.tif', 'expected': True},
{'input': 'elife-00666-fig3-v1.tif', 'expected': True},
{'input': 'elife-00666-fig3-figsupp1-v1.tif', 'expected': True},
{'input': 'elife-00666-fig3-video1.mp4', 'expected': False},
{'input': 'elife-00666-fig4-v1.tif', 'expected': True},
{'input': 'elife-00666-fig4-code1-v1.xlsx', 'expected': False},
{'input': 'elife-00666-figures-v1.pdf', 'expected': False},
{'input': 'elife-00666-inf001-v1.jpeg', 'expected': False},
{'input': 'elife-00666-repstand1-v1.pdf', 'expected': False},
{'input': 'elife-00666-resp-fig1-v1.png', 'expected': True},
{'input': 'elife-00666-resp-video1.mp4', 'expected': False},
{'input': 'elife-00666-supp1-v1.csv', 'expected': False},
{'input': 'elife-00666-table3-data1-v1.xlsx', 'expected': False},
{'input': 'elife-00666-video1.mp4', 'expected': False},
{'input': 'elife-00666-video1-data1-v1.xlsx', 'expected': False},
{'input': 'elife-00666-supp1-v1.tif', 'expected': False}
)
def test_article_figure(self, input, expected):
self.assertEqual(article_structure.article_figure(input), expected)
def test_get_original_files(self):
files = ['elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v10.tif',
'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx',
'elife-07702-vor-r4.zip',
'elife-07398-media1.jpg']
expected = ['elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v10.tif',
'elife-00666-inf001-v1.jpg',
'elife-00666-table3-data1-v1.xlsx']
self.assertListEqual(article_structure.get_original_files(files), expected)
def test_get_media_file_images(self):
files = ['elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx',
'elife-07702-vor-r4.zip',
'elife-00666-video2.jpg',
'elife-07398-media1.jpg']
expected = ['elife-00666-video2.jpg',
'elife-07398-media1.jpg']
self.assertListEqual(article_structure.get_media_file_images(files), expected)
def test_get_figures_for_iiif(self):
"Only .tif of original figures"
files = ['elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v1.jpg',
'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx',
'elife-07702-vor-r4.zip',
'elife-6148691793723703318-fig10-v1.gif',
'elife-9204580859652100230-fig2-data1-v1.xls',
'elife-00666-video2.jpg',
'elife-07398-media1.jpg']
expected = ['elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-video2.jpg',
'elife-07398-media1.jpg']
self.assertListEqual(article_structure.get_figures_for_iiif(files), expected)
# see https://github.com/elifesciences/elife-continuum-documentation/blob/master/file-naming/file_naming_spec.md
def test_get_figures_pdfs(self):
files = ['elife-07398-media1.jpg',
'elife-00666-figures-v1.pdf',
'elife-00353-v1.pdf',
'elife-00353-v1.xml',
'elife-18425-figures-v2.pdf']
expected = ['elife-00666-figures-v1.pdf',
'elife-18425-figures-v2.pdf']
self.assertListEqual(article_structure.get_figures_pdfs(files), expected)
@data(u'elife-15224-fig1-figsupp1.tif',
u'elife-15224-resp-fig1.tif', u'elife-15224-figures.pdf',
u'elife-15802-fig9-data3.docx', u'elife-11792.mp4',
u'elife-00005-media1-code1.wrl')
def test_is_video_file_false(self, filename):
result = article_structure.is_video_file(filename)
self.assertFalse(result)
@data(u'elife-11792-media2.mp4', u'elife-15224-fig1-figsupp1-media.tif', u'elife-11792-video1.mp4',
u'elife-99999-resp-media1.avi', u'elife-00005-media1.mov')
def test_is_video_file_true(self,filename):
result = article_structure.is_video_file(filename)
self.assertTrue(result)
@data(u'elife-15224-fig1-figsupp1.tif')
def test_file_parts(self, filename):
prefix, extension = article_structure.file_parts(filename)
self.assertEqual(prefix, u'elife-15224-fig1-figsupp1')
self.assertEqual(extension, u'tif')
def test_get_videos(self):
files = [u'elife-13273-fig1-v1.tif', u'elife-13273-fig2-figsupp1-v1.tif', u'elife-13273-fig2-figsupp2-v1.tif', u'elife-13273-fig2-figsupp3-v1.tif', u'elife-13273-fig2-v1.tif', u'elife-13273-fig3-data1-v1.xlsx', u'elife-13273-fig3-figsupp1-v1.tif', u'elife-13273-fig3-figsupp2-v1.tif', u'elife-13273-fig3-figsupp3-v1.tif', u'elife-13273-fig3-figsupp4-v1.tif', u'elife-13273-fig3-figsupp5-v1.tif', u'elife-13273-fig3-v1.tif', u'elife-13273-fig4-figsupp1-v1.tif', u'elife-13273-fig4-v1.tif', u'elife-13273-fig5-data1-v1.xlsx', u'elife-13273-fig5-figsupp1-v1.tif', u'elife-13273-fig5-v1.tif', u'elife-13273-fig6-data1-v1.xlsx', u'elife-13273-fig6-data2-v1.xlsx', u'elife-13273-fig6-figsupp1-v1.tif', u'elife-13273-fig6-figsupp2-v1.tif', u'elife-13273-fig6-v1.tif', u'elife-13273-fig7-v1.tif', u'elife-13273-fig8-v1.tif', u'elife-13273-fig9-v1.tif', u'elife-13273-figures-v1.pdf', u'elife-13273-media1.mp4', u'elife-13273-v1.pdf', u'elife-13273-v1.xml']
result = article_structure.get_videos(files)
self.assertListEqual(result, [u'elife-13273-media1.mp4'])
def test_pre_ingest_assets(self):
files = ['elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v1.jpg',
'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx',
'elife-07702-vor-r4.zip',
'elife-6148691793723703318-fig10-v1.gif',
'elife-9204580859652100230-fig2-data1-v1.xls',
'elife-00666-video2.jpg',
'elife-07398-media1.jpg',
'elife-00666-figures-v1.pdf',
'elife-18425-figures-v2.pdf',
'elife-13273-media1.mp4']
expected = ['elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-video2.jpg',
'elife-07398-media1.jpg',
'elife-13273-media1.mp4',
'elife-00666-figures-v1.pdf',
'elife-18425-figures-v2.pdf']
self.assertItemsEqual(article_structure.pre_ingest_assets(files), expected)
def test_get_file_type_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.file_type
self.assertEqual(result, expected)
def do_activity(self, data=None):
"""
Do the work
"""
run = data['run']
if self.logger:
self.logger.info('data: %s' % json.dumps(data, sort_keys=True, indent=4))
info = S3NotificationInfo.from_dict(data)
storage_context = StorageContext(self.settings)
session = Session(self.settings)
filename_last_element = session.get_value(run, 'filename_last_element')
# zip name contains version information for previously archived zip files
article_structure = ArticleInfo(filename_last_element)
article_id = article_structure.article_id
session.store_value(run, 'article_id', article_id)
session.store_value(run, 'file_name', info.file_name)
if self.logger:
self.logger.info("Expanding file %s" % info.file_name)
version = session.get_value(run, 'version')
status = article_structure.status
if status is None or (status != 'vor' and status != 'poa'):
self.logger.error("Name '%s' did not match expected pattern for status" %
filename_last_element)
return activity.activity.ACTIVITY_PERMANENT_FAILURE # status could not be determined, exit workflow.
article_version_id = article_id + '.' + version
session.store_value(run, 'article_version_id', article_version_id)
session.store_value(run, 'run', run)
session.store_value(run, 'status', status)
self.emit_monitor_event(self.settings, article_id, version, run, "Expand Article", "start",
"Starting expansion of article " + article_id)
try:
# download zip to temp folder
tmp = self.get_tmp_dir()
local_zip_file = self.open_file_from_tmp_dir(filename_last_element, mode='wb')
storage_resource_origin = self.settings.storage_provider + "://" + info.bucket_name + "/" + info.file_name
storage_context.get_resource_to_file(storage_resource_origin, local_zip_file)
local_zip_file.close()
# extract zip contents
folder_name = path.join(article_version_id, run)
content_folder = path.join(tmp, folder_name)
makedirs(content_folder)
with ZipFile(path.join(tmp, filename_last_element)) as zf:
zf.extractall(content_folder)
upload_filenames = []
for f in listdir(content_folder):
if isfile(join(content_folder, f)) and f[0] != '.' and not f[0] == '_':
upload_filenames.append(f)
self.check_filenames(upload_filenames)
bucket_folder_name = article_version_id + '/' + run
for filename in upload_filenames:
source_path = path.join(content_folder, filename)
dest_path = bucket_folder_name + '/' + filename
storage_resource_dest = self.settings.storage_provider + "://" + self.settings.publishing_buckets_prefix + \
self.settings.expanded_bucket + "/" + dest_path
storage_context.set_resource_from_filename(storage_resource_dest, source_path)
self.clean_tmp_dir()
session.store_value(run, 'expanded_folder', bucket_folder_name)
self.emit_monitor_event(self.settings, article_id, version, run, "Expand Article",
"end", "Finished expansion of article " + article_id +
" for version " + version + " run " + str(run) +
" into " + bucket_folder_name)
except Exception as e:
self.logger.exception("Exception when expanding article")
self.emit_monitor_event(self.settings, article_id, version, run, "Expand Article",
"error", "Error expanding article " + article_id +
" message:" + e.message)
return activity.activity.ACTIVITY_PERMANENT_FAILURE
return True
class TestArticleStructure(unittest.TestCase):
@unpack
@data({
'input': 'elife-07702-vor-r4.zip',
'expected': None
}, {
'input': 'elife-00013-vor-v1-20121015000000.zip',
'expected': '2012-10-15T00:00:00Z'
})
def test_get_update_date_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_update_date_from_zip_filename()
self.assertEqual(result, expected)
@unpack
@data({
'input': 'elife-07702-vor-r4.zip',
'expected': None
}, {
'input': 'elife-00013-vor-v1-20121015000000.zip',
'expected': '1'
})
def test_get_version_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_version_from_zip_filename()
self.assertEqual(result, expected)
@unpack
@data(
{
'input': 'elife-07702-vor-r4.zip',
'expected': 'ArticleZip'
},
{
'input': 'elife-00013-vor-v1-20121015000000.zip',
'expected': 'ArticleZip'
},
{
'input': 'elife-00666-v1.pdf',
'expected': 'Other'
},
{
'input': 'elife-00666-v1.xml',
'expected': 'ArticleXML'
},
{
'input': 'elife-00666-app1-fig1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-app1-fig1-figsupp1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-app2-video1.mp4',
'expected': 'Other'
},
{
'input': 'elife-00666-box2-fig1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-code1-v1.xml',
'expected': 'Other'
},
{
'input': 'elife-00666-data1-v1.xlsx',
'expected': 'Other'
},
{
'input': 'elife-00666-fig1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig2-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig2-figsupp1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig2-figsupp2-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig3-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig3-v10.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig3-figsupp1-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig3-video1.mp4',
'expected': 'Other'
},
{
'input': 'elife-00666-fig4-v1.tif',
'expected': 'Figure'
},
{
'input': 'elife-00666-fig4-code1-v1.xlsx',
'expected': 'Other'
},
{
'input': 'elife-00666-figures-v1.pdf',
'expected': 'FigurePDF'
},
{
'input': 'elife-00666-inf001-v1.jpeg',
'expected': 'Inline'
},
{
'input': 'elife-00666-repstand1-v1.pdf',
'expected': 'Other'
},
{
'input': 'elife-00666-resp-fig1-v1.png',
'expected': 'Figure'
},
{
'input': 'elife-00666-resp-video1.mp4',
'expected': 'Other'
},
{
'input': 'elife-00666-supp1-v1.csv',
'expected': 'Other'
},
{
'input': 'elife-00666-supp2-v2.tif',
'expected': 'Other'
},
{
'input': 'elife-00666-supp2-v3.docx',
'expected': 'Other'
},
{
'input': 'elife-00666-table3-data1-v1.xlsx',
'expected': 'Other'
},
{
'input': 'elife-00666-video1.mp4',
'expected': 'Other'
},
{
'input': 'elife-00666-video1-data1-v1.xlsx',
'expected': 'Other'
},
)
def test_get_file_type_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.file_type
self.assertEqual(result, expected)
@unpack
@data({
'input': 'elife-07702-vor-r4.zip',
'expected': False
}, {
'input': 'elife-00013-vor-v1-20121015000000.zip',
'expected': False
}, {
'input': 'elife-00666-v1.pdf',
'expected': False
}, {
'input': 'elife-00666-v1.xml',
'expected': False
}, {
'input': 'elife-00666-app1-fig1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-app1-fig1-figsupp1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-app2-video1.mp4',
'expected': False
}, {
'input': 'elife-00666-box2-fig1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-code1-v1.xml',
'expected': False
}, {
'input': 'elife-00666-data1-v1.xlsx',
'expected': False
}, {
'input': 'elife-00666-fig1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig2-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig2-figsupp1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig2-figsupp2-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig3-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig3-figsupp1-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig3-video1.mp4',
'expected': False
}, {
'input': 'elife-00666-fig4-v1.tif',
'expected': True
}, {
'input': 'elife-00666-fig4-code1-v1.xlsx',
'expected': False
}, {
'input': 'elife-00666-figures-v1.pdf',
'expected': False
}, {
'input': 'elife-00666-inf001-v1.jpeg',
'expected': False
}, {
'input': 'elife-00666-repstand1-v1.pdf',
'expected': False
}, {
'input': 'elife-00666-resp-fig1-v1.png',
'expected': True
}, {
'input': 'elife-00666-resp-video1.mp4',
'expected': False
}, {
'input': 'elife-00666-supp1-v1.csv',
'expected': False
}, {
'input': 'elife-00666-table3-data1-v1.xlsx',
'expected': False
}, {
'input': 'elife-00666-video1.mp4',
'expected': False
}, {
'input': 'elife-00666-video1-data1-v1.xlsx',
'expected': False
}, {
'input': 'elife-00666-supp1-v1.tif',
'expected': False
})
def test_article_figure(self, input, expected):
self.assertEqual(article_structure.article_figure(input), expected)
def test_get_original_files(self):
files = [
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v10.tif', 'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx', 'elife-07702-vor-r4.zip',
'elife-07398-media1.jpg'
]
expected = [
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v10.tif', 'elife-00666-inf001-v1.jpg',
'elife-00666-table3-data1-v1.xlsx'
]
self.assertListEqual(article_structure.get_original_files(files),
expected)
def test_get_media_file_images(self):
files = [
'elife-00666-fig2-figsupp2-v1.tif', 'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx', 'elife-07702-vor-r4.zip',
'elife-00666-video2.jpg', 'elife-07398-media1.jpg'
]
expected = ['elife-00666-video2.jpg', 'elife-07398-media1.jpg']
self.assertListEqual(article_structure.get_media_file_images(files),
expected)
def test_get_figures_for_iiif(self):
"Only .tif of original figures"
files = [
'elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v1.jpg', 'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx', 'elife-07702-vor-r4.zip',
'elife-6148691793723703318-fig10-v1.gif',
'elife-9204580859652100230-fig2-data1-v1.xls',
'elife-00666-video2.jpg', 'elife-07398-media1.jpg'
]
expected = [
'elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif', 'elife-00666-video2.jpg',
'elife-07398-media1.jpg'
]
self.assertListEqual(article_structure.get_figures_for_iiif(files),
expected)
# see https://github.com/elifesciences/elife-continuum-documentation/blob/master/file-naming/file_naming_spec.md
def test_get_figures_pdfs(self):
files = [
'elife-07398-media1.jpg', 'elife-00666-figures-v1.pdf',
'elife-00353-v1.pdf', 'elife-00353-v1.xml',
'elife-18425-figures-v2.pdf'
]
expected = ['elife-00666-figures-v1.pdf', 'elife-18425-figures-v2.pdf']
self.assertListEqual(article_structure.get_figures_pdfs(files),
expected)
@data(u'elife-15224-fig1-figsupp1.tif', u'elife-15224-resp-fig1.tif',
u'elife-15224-figures.pdf', u'elife-15802-fig9-data3.docx',
u'elife-11792.mp4', u'elife-00005-media1-code1.wrl')
def test_is_video_file_false(self, filename):
result = article_structure.is_video_file(filename)
self.assertFalse(result)
@data(u'elife-11792-media2.mp4', u'elife-15224-fig1-figsupp1-media.tif',
u'elife-11792-video1.mp4', u'elife-99999-resp-media1.avi',
u'elife-00005-media1.mov')
def test_is_video_file_true(self, filename):
result = article_structure.is_video_file(filename)
self.assertTrue(result)
@data(u'elife-15224-fig1-figsupp1.tif')
def test_file_parts(self, filename):
prefix, extension = article_structure.file_parts(filename)
self.assertEqual(prefix, u'elife-15224-fig1-figsupp1')
self.assertEqual(extension, u'tif')
def test_get_videos(self):
files = [
u'elife-13273-fig1-v1.tif', u'elife-13273-fig2-figsupp1-v1.tif',
u'elife-13273-fig2-figsupp2-v1.tif',
u'elife-13273-fig2-figsupp3-v1.tif', u'elife-13273-fig2-v1.tif',
u'elife-13273-fig3-data1-v1.xlsx',
u'elife-13273-fig3-figsupp1-v1.tif',
u'elife-13273-fig3-figsupp2-v1.tif',
u'elife-13273-fig3-figsupp3-v1.tif',
u'elife-13273-fig3-figsupp4-v1.tif',
u'elife-13273-fig3-figsupp5-v1.tif', u'elife-13273-fig3-v1.tif',
u'elife-13273-fig4-figsupp1-v1.tif', u'elife-13273-fig4-v1.tif',
u'elife-13273-fig5-data1-v1.xlsx',
u'elife-13273-fig5-figsupp1-v1.tif', u'elife-13273-fig5-v1.tif',
u'elife-13273-fig6-data1-v1.xlsx',
u'elife-13273-fig6-data2-v1.xlsx',
u'elife-13273-fig6-figsupp1-v1.tif',
u'elife-13273-fig6-figsupp2-v1.tif', u'elife-13273-fig6-v1.tif',
u'elife-13273-fig7-v1.tif', u'elife-13273-fig8-v1.tif',
u'elife-13273-fig9-v1.tif', u'elife-13273-figures-v1.pdf',
u'elife-13273-media1.mp4', u'elife-13273-v1.pdf',
u'elife-13273-v1.xml'
]
result = article_structure.get_videos(files)
self.assertListEqual(result, [u'elife-13273-media1.mp4'])
def test_pre_ingest_assets(self):
files = [
'elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif',
'elife-00666-fig2-figsupp2-v1.jpg', 'elife-00666-inf001-v1.jpg',
'elife-00666-inf001-v1-80w.jpg',
'elife-00666-table3-data1-v1.xlsx', 'elife-07702-vor-r4.zip',
'elife-6148691793723703318-fig10-v1.gif',
'elife-9204580859652100230-fig2-data1-v1.xls',
'elife-00666-video2.jpg', 'elife-07398-media1.jpg',
'elife-00666-figures-v1.pdf', 'elife-18425-figures-v2.pdf',
'elife-13273-media1.mp4'
]
expected = [
'elife-00666-app1-fig1-figsupp1-v1.tif',
'elife-00666-fig2-figsupp2-v1.tif', 'elife-00666-video2.jpg',
'elife-07398-media1.jpg', 'elife-13273-media1.mp4',
'elife-00666-figures-v1.pdf', 'elife-18425-figures-v2.pdf'
]
self.assertItemsEqual(article_structure.pre_ingest_assets(files),
expected)
def test_get_version_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.get_version_from_zip_filename()
self.assertEqual(result, expected)
def test_get_file_type_from_zip_filename(self, input, expected):
self.articleinfo = ArticleInfo(input)
result = self.articleinfo.file_type
self.assertEqual(result, expected)