def test_remove_non_path_from_url(self):
self.assertEqual(remove_non_path_from_url('https://foo/bar'),
'https://foo/bar')
self.assertEqual(remove_non_path_from_url('https://foo//?bar#frag'),
'https://foo')
self.assertEqual(remove_non_path_from_url('/path/to/foo#frag'),
'/path/to/foo')
def workspace_from_url(self, mets_url, dst_dir=None, clobber_mets=False, mets_basename=None, download=False, src_baseurl=None):
"""
Create a workspace from a METS by URL (i.e. clone it).
Sets the mets.xml file
Arguments:
mets_url (string): Source mets URL
dst_dir (string, None): Target directory for the workspace
clobber_mets (boolean, False): Whether to overwrite existing mets.xml. By default existing mets.xml will raise an exception.
download (boolean, False): Whether to download all the files
src_baseurl (string, None): Base URL for resolving relative file locations
Returns:
Workspace
"""
if mets_url is None:
raise ValueError("Must pass 'mets_url' workspace_from_url")
# if mets_url is a relative filename, make it absolute
if is_local_filename(mets_url) and not Path(mets_url).is_absolute():
mets_url = str(Path(Path.cwd() / mets_url))
# if mets_basename is not given, use the last URL segment of the mets_url
if mets_basename is None:
mets_basename = nth_url_segment(mets_url, -1)
# If src_baseurl wasn't given, determine from mets_url by removing last url segment
if not src_baseurl:
last_segment = nth_url_segment(mets_url)
src_baseurl = remove_non_path_from_url(remove_non_path_from_url(mets_url)[:-len(last_segment)])
# resolve dst_dir
if not dst_dir:
if is_local_filename(mets_url):
log.debug("Deriving dst_dir %s from %s", Path(mets_url).parent, mets_url)
dst_dir = Path(mets_url).parent
else:
log.debug("Creating ephemeral workspace '%s' for METS @ <%s>", dst_dir, mets_url)
dst_dir = tempfile.mkdtemp(prefix=TMP_PREFIX)
# XXX Path.resolve is always strict in Python <= 3.5, so create dst_dir unless it exists consistently
if not Path(dst_dir).exists():
Path(dst_dir).mkdir(parents=True, exist_ok=False)
dst_dir = str(Path(dst_dir).resolve())
log.debug("workspace_from_url\nmets_basename='%s'\nmets_url='%s'\nsrc_baseurl='%s'\ndst_dir='%s'",
mets_basename, mets_url, src_baseurl, dst_dir)
self.download_to_directory(dst_dir, mets_url, basename=mets_basename, if_exists='overwrite' if clobber_mets else 'raise')
workspace = Workspace(self, dst_dir, mets_basename=mets_basename, baseurl=src_baseurl)
if download:
for f in workspace.mets.find_files():
workspace.download_file(f)
return workspace
def workspace_from_url(self,
mets_url,
dst_dir=None,
clobber_mets=False,
mets_basename=None,
download=False,
src_baseurl=None):
"""
Create a workspace from a METS by URL (i.e. clone if :py:attr:`mets_url` is remote or :py:attr:`dst_dir` is given).
Arguments:
mets_url (string): Source METS URL or filesystem path
Keyword Arguments:
dst_dir (string, None): Target directory for the workspace. \
By default create a temporary directory under :py:data:`ocrd.constants.TMP_PREFIX`. \
(The resulting path can be retrieved via :py:attr:`ocrd.Workspace.directory`.)
clobber_mets (boolean, False): Whether to overwrite existing ``mets.xml``. \
By default existing ``mets.xml`` will raise an exception.
download (boolean, False): Whether to also download all the files referenced by the METS
src_baseurl (string, None): Base URL for resolving relative file locations
Download (clone) :py:attr:`mets_url` to ``mets.xml`` in :py:attr:`dst_dir`, unless
the former is already local and the latter is ``none`` or already identical to its directory name.
Returns:
a new :py:class:`~ocrd.workspace.Workspace`
"""
log = getLogger('ocrd.resolver.workspace_from_url')
if mets_url is None:
raise ValueError("Must pass 'mets_url' workspace_from_url")
# if mets_url is a relative filename, make it absolute
if is_local_filename(mets_url) and not Path(mets_url).is_absolute():
mets_url = str(Path(Path.cwd() / mets_url))
# if mets_basename is not given, use the last URL segment of the mets_url
if mets_basename is None:
mets_basename = nth_url_segment(mets_url, -1)
# If src_baseurl wasn't given, determine from mets_url by removing last url segment
if not src_baseurl:
last_segment = nth_url_segment(mets_url)
src_baseurl = remove_non_path_from_url(
remove_non_path_from_url(mets_url)[:-len(last_segment)])
# resolve dst_dir
if not dst_dir:
if is_local_filename(mets_url):
log.debug("Deriving dst_dir %s from %s",
Path(mets_url).parent, mets_url)
dst_dir = Path(mets_url).parent
else:
log.debug("Creating ephemeral workspace '%s' for METS @ <%s>",
dst_dir, mets_url)
dst_dir = mkdtemp(prefix=TMP_PREFIX)
# XXX Path.resolve is always strict in Python <= 3.5, so create dst_dir unless it exists consistently
if not Path(dst_dir).exists():
Path(dst_dir).mkdir(parents=True, exist_ok=False)
dst_dir = str(Path(dst_dir).resolve())
log.debug(
"workspace_from_url\nmets_basename='%s'\nmets_url='%s'\nsrc_baseurl='%s'\ndst_dir='%s'",
mets_basename, mets_url, src_baseurl, dst_dir)
self.download_to_directory(
dst_dir,
mets_url,
basename=mets_basename,
if_exists='overwrite' if clobber_mets else 'skip')
workspace = Workspace(self,
dst_dir,
mets_basename=mets_basename,
baseurl=src_baseurl)
if download:
for f in workspace.mets.find_files():
workspace.download_file(f)
return workspace
