def get_data_http(self, filename):
"""Yield the data returned from `filename` of `pipeline_context` in manageable chunks."""
url = self.get_url(filename)
try:
infile = request.urlopen(url)
file_size = utils.human_format_number(
self.catalog_file_size(filename)).strip()
stats = utils.TimingStats()
data = infile.read(config.CRDS_DATA_CHUNK_SIZE)
while data:
stats.increment("bytes", len(data))
status = stats.status("bytes")
bytes_so_far = " ".join(status[0].split()[:-1])
log.verbose("Transferred HTTP",
repr(url),
bytes_so_far,
"/",
file_size,
"bytes at",
status[1],
verbosity=20)
yield data
data = infile.read(config.CRDS_DATA_CHUNK_SIZE)
except Exception as exc:
raise CrdsDownloadError("Failed downloading", srepr(filename),
"from url", srepr(url), ":",
str(exc)) from exc
finally:
try:
infile.close()
except UnboundLocalError: # maybe the open failed.
pass
def download_files(self, downloads, localpaths):
"""Serial file-by-file download."""
download_metadata = get_download_metadata()
self.info_map = {}
for filename in downloads:
self.info_map[filename] = download_metadata.get(
filename, "NOT FOUND unknown to server")
if config.writable_cache_or_verbose(
"Readonly cache, skipping download of (first 5):",
repr(downloads[:5]),
verbosity=70):
bytes_so_far = 0
total_files = len(downloads)
total_bytes = get_total_bytes(self.info_map)
for nth_file, name in enumerate(downloads):
try:
if "NOT FOUND" in self.info_map[name]:
raise CrdsDownloadError(
"file is not known to CRDS server.")
bytes, path = self.catalog_file_size(
name), localpaths[name]
log.info(
file_progress("Fetching", name, path, bytes,
bytes_so_far, total_bytes, nth_file,
total_files))
self.download(name, path)
bytes_so_far += os.stat(path).st_size
except Exception as exc:
if self.raise_exceptions:
raise
else:
log.error("Failure downloading file", repr(name), ":",
str(exc))
return bytes_so_far
return 0
def verify_file(self, filename, localpath):
"""Check that the size and checksum of downloaded `filename` match the server."""
remote_info = self.info_map[filename]
local_length = os.stat(localpath).st_size
original_length = int(remote_info["size"])
if original_length != local_length and config.get_length_flag():
raise CrdsDownloadError(
"downloaded file size", local_length,
"does not match server size", original_length)
if not config.get_checksum_flag():
log.verbose("Skipping sha1sum with CRDS_DOWNLOAD_CHECKSUMS=False")
elif remote_info["sha1sum"] not in ["", "none"]:
original_sha1sum = remote_info["sha1sum"]
local_sha1sum = utils.checksum(localpath)
if original_sha1sum != local_sha1sum:
raise CrdsDownloadError(
"downloaded file", srepr(filename),
"sha1sum", srepr(local_sha1sum),
"does not match server sha1sum", srepr(original_sha1sum))
else:
log.verbose("Skipping sha1sum check since server doesn't know it.")
def plugin_download(self, filename, localpath):
"""Run an external program defined by CRDS_DOWNLOAD_PLUGIN to download filename to localpath."""
url = self.get_url(filename)
plugin_cmd = config.get_download_plugin()
plugin_cmd = plugin_cmd.replace("${SOURCE_URL}", url)
plugin_cmd = plugin_cmd.replace("${OUTPUT_PATH}", localpath)
log.verbose("Running download plugin:", repr(plugin_cmd))
status = os.system(plugin_cmd)
if status != 0:
if status == 2:
raise KeyboardInterrupt("Interrupted plugin.")
else:
raise CrdsDownloadError(
"Plugin download fail status =", repr(status),
"with command:", srepr(plugin_cmd))
def download(self, name, localpath):
"""Download a single file."""
# This code is complicated by the desire to blow away failed downloads. For the specific
# case of KeyboardInterrupt, the file needs to be blown away, but the interrupt should not
# be re-characterized so it is still un-trapped elsewhere under normal idioms which try *not*
# to trap KeyboardInterrupt.
assert not config.get_cache_readonly(), "Readonly cache, cannot download files " + repr(name)
try:
utils.ensure_dir_exists(localpath)
return proxy.apply_with_retries(self.download_core, name, localpath)
except Exception as exc:
self.remove_file(localpath)
raise CrdsDownloadError(
"Error fetching data for", srepr(name),
"at CRDS server", srepr(get_crds_server()),
"with mode", srepr(config.get_download_mode()),
":", str(exc)) from exc
except: # mainly for control-c, catch it and throw it.
self.remove_file(localpath)
raise
