def test_esearch_sra_withhistory():
esearch_results = entrez.esearch(DB,
QUERY,
userhistory=True,
retmax=RETMAX,
api_key=API_KEY)
assert len(esearch_results.ids) == RETMAX
assert esearch_results.webenv != ""
assert esearch_results.query_key != ""
def download_biosample_xml(biosample_ids, defaults):
logger.info(f"BioSample - Downloading XML Records")
for i, ids in enumerate(chunked(biosample_ids, BATCH_SIZE)):
start, end = i * BATCH_SIZE, (i + 1) * BATCH_SIZE
logger.info(f"BioSample - Downloading XML Records [Batch {start:,}-{end:,}]")
query = "+OR+".join(ids)
esearch_result = entrez.esearch("biosample", query, **defaults)
webenv = esearch_result.webenv
query_key = esearch_result.query_key
count = esearch_result.count
for accn, xml in biosample_efetch(webenv, query_key, count, defaults):
root = xml_to_root(xml)
doc = parsers_biosample_xml.parse_biosample(root)
yield doc
time.sleep(1)
def check_sra_for_updated_ids(query, collection, defaults):
logger.info(f"SRA - Querying SraMongo for last update")
last_srx_update = get_sramongo_last_srx_update(collection)
logger.info(f"SRA - Querying SRA for: {query}")
esearch_result = entrez.esearch("sra", query, **defaults)
webenv = esearch_result.webenv
query_key = esearch_result.query_key
count = DEBUG_SIZE if _DEBUG else esearch_result.count
logger.info("SRA - Checking for Updates")
ids_to_update = []
for esummary_result in sra_esummary(webenv, query_key, count, defaults):
srx = esummary_result.accn
if esummary_result.update_date != last_srx_update.get(srx, None):
ids_to_update.append(esummary_result.id)
logger.info(f"SRA - {len(ids_to_update):,} IDs to Update")
return ids_to_update
def small_esearch_results() -> entrez.EsearchResult:
return entrez.esearch(DB, QUERY, retmax=RETMAX, api_key=API_KEY)