aboutsummaryrefslogtreecommitdiff
path: root/plip/exchange/webservices.py
diff options
context:
space:
mode:
authorNavan Chauhan <navanchauhan@gmail.com>2020-07-02 20:48:33 +0530
committerNavan Chauhan <navanchauhan@gmail.com>2020-07-02 20:48:33 +0530
commit4be08f7bdd77991e9e453c1cda863c3f20c338d5 (patch)
tree083e8e91622221185a28fd50754abc2f86b1df43 /plip/exchange/webservices.py
initial commit
Diffstat (limited to 'plip/exchange/webservices.py')
-rw-r--r--plip/exchange/webservices.py54
1 files changed, 54 insertions, 0 deletions
diff --git a/plip/exchange/webservices.py b/plip/exchange/webservices.py
new file mode 100644
index 0000000..0c8cd3e
--- /dev/null
+++ b/plip/exchange/webservices.py
@@ -0,0 +1,54 @@
+import sys
+from urllib.error import HTTPError
+from urllib.request import urlopen
+
+import lxml.etree as et
+
+from plip.basic import logger
+
+logger = logger.get_logger()
+
+
+def check_pdb_status(pdbid):
+ """Returns the status and up-to-date entry in the PDB for a given PDB ID"""
+ url = 'http://www.rcsb.org/pdb/rest/idStatus?structureId=%s' % pdbid
+ xmlf = urlopen(url)
+ xml = et.parse(xmlf)
+ xmlf.close()
+ status = None
+ current_pdbid = pdbid
+ for df in xml.xpath('//record'):
+ status = df.attrib['status'] # Status of an entry can be either 'UNKWOWN', 'OBSOLETE', or 'CURRENT'
+ if status == 'OBSOLETE':
+ current_pdbid = df.attrib['replacedBy'] # Contains the up-to-date PDB ID for obsolete entries
+ return [status, current_pdbid.lower()]
+
+
+def fetch_pdb(pdbid):
+ """Get the newest entry from the RCSB server for the given PDB ID. Exits with '1' if PDB ID is invalid."""
+ pdbid = pdbid.lower()
+ logger.info(f'checking status of PDB-ID {pdbid}')
+ state, current_entry = check_pdb_status(pdbid) # Get state and current PDB ID
+
+ if state == 'OBSOLETE':
+ logger.info(f'entry is obsolete, getting {current_entry} instead')
+ elif state == 'CURRENT':
+ logger.info('entry is up-to-date')
+ elif state == 'UNKNOWN':
+ logger.error('invalid PDB-ID (entry does not exist on PDB server)')
+ sys.exit(1)
+ logger.info('downloading file from PDB')
+ # get URL for current entry
+ # @todo needs update to react properly on response codes of RCSB servers
+ pdburl = f'http://www.rcsb.org/pdb/files/{current_entry}.pdb'
+ try:
+ pdbfile = urlopen(pdburl).read().decode()
+ # If no PDB file is available, a text is now shown with "We're sorry, but ..."
+ # Could previously be distinguished by an HTTP error
+ if 'sorry' in pdbfile:
+ logger.error('no file in PDB format available from wwPDB for the given PDB ID.')
+ sys.exit(1)
+ except HTTPError:
+ logger.error('no file in PDB format available from wwPDB for the given PDB ID')
+ sys.exit(1)
+ return [pdbfile, current_entry]