def _download_file(download_url: str,
downloader_job: DownloaderJob,
target_file_path: str,
force_ftp: bool = False) -> bool:
""" Download file dispatcher. Dispatches to the FTP or Aspera downloader """
# SRA files have Apsera downloads.
if 'ftp.sra.ebi.ac.uk' in download_url and not force_ftp:
# From: ftp://ftp.sra.ebi.ac.uk/vol1/fastq/ERR036/ERR036000/ERR036000_1.fastq.gz
# To: [email protected]:/vol1/fastq/ERR036/ERR036000/ERR036000_1.fastq.gz
download_url = download_url.replace('ftp://', 'era-fasp@')
download_url = download_url.replace('ftp', 'fasp')
download_url = download_url.replace('.uk/', '.uk:/')
return _download_file_aspera(download_url,
downloader_job,
target_file_path,
source="ENA")
elif "ncbi.nlm.nih.gov" in download_url and not force_ftp:
# Try to convert old-style endpoints into new-style endpoints if possible
try:
if 'anonftp' in download_url:
accession = download_url.split('/')[-1].split('.sra')[0]
new_url = get_fasp_sra_download(accession)
if new_url:
download_url = new_url
except Exception:
pass
return _download_file_aspera(download_url,
downloader_job,
target_file_path,
source="NCBI")
else:
return _download_file_ftp(download_url, downloader_job,
target_file_path)
def _build_ncbi_file_url(run_accession: str):
""" Build the path to the hypothetical .sra file we want """
accession = run_accession
first_three = accession[:3]
first_six = accession[:6]
# Prefer the FASP-specific endpoints if possible..
download_url = get_fasp_sra_download(run_accession)
if not download_url:
# ..else, load balancing via coin flip.
if random.choice([True, False]):
download_url = NCBI_DOWNLOAD_URL_TEMPLATE.format(
first_three=first_three,
first_six=first_six,
accession=accession)
else:
download_url = NCBI_PRIVATE_DOWNLOAD_URL_TEMPLATE.format(
first_three=first_three,
first_six=first_six,
accession=accession)
return download_url