def __init__(self):
dir_path = os.path.join(
os.path.dirname(os.path.realpath(__file__)),
"udpipe_models"
)
self.serverfiles = ServerFiles(self.server_url)
self.localfiles = LocalFiles(dir_path, serverfiles=self.serverfiles)
def ensure_local(index_url, file_path, local_cache_path,
force=False, progress_advance=None):
localfiles = LocalFiles(local_cache_path,
serverfiles=ServerFiles(server=index_url))
if force:
localfiles.download(*file_path, callback=progress_advance)
return localfiles.localpath_download(*file_path, callback=progress_advance)
class UDPipeModels:
server_url = "https://file.biolab.si/files/udpipe/"
def __init__(self):
dir_path = os.path.join(
os.path.dirname(os.path.realpath(__file__)),
"udpipe_models"
)
self.serverfiles = ServerFiles(self.server_url)
self.localfiles = LocalFiles(dir_path, serverfiles=self.serverfiles)
def __getitem__(self, language):
file_name = self._find_file(language_to_name(language))
return self.localfiles.localpath_download(file_name)
@property
def model_files(self):
try:
return self.serverfiles.listfiles()
except ConnectionError:
return self.localfiles.listfiles()
def _find_file(self, language):
return next(filter(lambda f: file_to_name(f).startswith(language),
map(lambda f: f[0], self.model_files)))
@property
def supported_languages(self):
return list(map(lambda f: file_to_language(f[0]), self.model_files))
def ensure_local(domain, filename, force=False, progress_advance=None):
localfiles = LocalFiles(local_cache_path(),
serverfiles=ServerFiles(server=INDEX_URL))
if force:
localfiles.download(domain, filename, callback=progress_advance)
return localfiles.localpath_download(
domain, filename, callback=progress_advance)
def tag_list():
"""List of available tags and their pretty-print with indices"""
server_url = "http://butler.fri.uni-lj.si/datasets/"
PATH = os.path.join(data_dir(), "datasets")
local_files = LocalFiles(PATH, serverfiles=ServerFiles(server=server_url))
local_info = local_files.serverfiles.allinfo()
nested_tags = [i["tags"] for i in local_info.values() if i["tags"]]
all_tags = sorted(list(set(itertools.chain(*nested_tags))))
w = max(len(t) for t in all_tags)
n = int(75 / (w + 5))
s = [
"{:>3}-{:<{width}}".format(i, t, width=w)
for i, t in enumerate(all_tags)
]
c = "\n".join(["".join(s[x:x + n]) for x in range(0, len(s), n)])
return all_tags, c
import numpy as np
from serverfiles import LocalFiles, ServerFiles
from Orange.data import Table, Domain, StringVariable, DiscreteVariable
from Orange.data import filter as table_filter
from Orange.misc.environ import data_dir
from orangecontrib.bioinformatics.widgets.utils.data import TableAnnotation
domain = 'geo'
_local_cache_path = os.path.join(data_dir(), domain)
_all_info_file = os.path.join(_local_cache_path, '__INFO__')
_server_url = 'http://download.biolab.si/datasets/geo/'
pubmed_url = 'http://www.ncbi.nlm.nih.gov/pubmed/{}'
server_files = ServerFiles(server=_server_url)
local_files = LocalFiles(_local_cache_path, serverfiles=server_files)
def is_cached(gds_id):
return os.path.exists(os.path.join(_local_cache_path, gds_id + '.tab'))
def info_cache(f):
"""Store content of __INFO__ file locally."""
def wrapper():
if not os.path.isfile(_all_info_file) or os.path.getsize(_all_info_file) == 0:
with open(_all_info_file, 'w') as fp:
json.dump(list(server_files.allinfo().items()), fp)
def list_remote(self):
# type: () -> Dict[Tuple[str, str], dict]
client = ServerFiles(server=self.INDEX_URL)
return client.allinfo()
def list_remote(server: str) -> Dict[Tuple[str, ...], dict]:
client = ServerFiles(server)
return client.allinfo()
