Merge branch 'master' of mus.cacr.caltech.edu:htsworkflow

[htsworkflow.git] / scripts / htsw-record-runfolder

diff --git a/scripts/htsw-record-runfolder b/scripts/htsw-record-runfolder
new file mode 100755 (executable)
index 0000000..288ec1a
--- /dev/null
+++ b/scripts/htsw-record-runfolder
@@ -0,0 +1,123 @@
+#!/usr/bin/env python
+
+from htsworkflow.util.hdquery import get_hd_serial_num
+from htsworkflow.frontend import settings
+
+from optparse import OptionParser
+import os
+import re
+import sys
+import urllib2
+import urlparse
+
+runfolder_pattern = re.compile(r'[0-9]{6}_[-A-Za-z\d]+_\d+_(?P<flowcell>[A-Z\d]+)\.tgz')
+
+def extract_flowcell(runfolder_name):
+    path, basename = os.path.split(runfolder_name)
+    match = runfolder_pattern.match(basename)
+    if match is not None:
+        return match.group('flowcell')
+    else:
+        return None
+    
+def construct_parser():
+    """
+    """
+    msg = "usage: %prog [-d </dev/sdX> | -s <serial_number] [-f <flowcell>] [archived dirs]"
+    parser = OptionParser()
+    parser.add_option('-u', '--url', default=None,
+                      help="Alternate url for marking archived flowcells")
+    parser.add_option("-f", "--flowcell",  type="string", help="flowcell being archived")
+    parser.add_option("-d", "--device", type="string",
+                      help="device flowcell is being archived to")
+    parser.add_option("-s", "--serial", type="string", help="serial num. of archive device")
+    parser.add_option("-v", "--verbose", action="store_true", default=False)
+    
+    return parser
+
+
+def update_db(root_url, flowcells, serial, debug=False):
+    """
+    Creates link between flowcell and storage device over http
+    """
+    for fc in flowcells:
+        url = urlparse.urljoin(root_url, '%s/%s/' % (fc, serial))
+        
+        req = urllib2.Request(url)
+        try:
+            response = urllib2.urlopen(req)
+        except urllib2.URLError, e:
+            print 'ERROR - HTTP OUTPUT (Return Code: %s); use -v/--verbose for more details.' % (e.code)
+            if debug:
+                print e.read()
+            sys.exit(e.code)
+        
+        print "DB Update of %s & %s succeeded" % (fc, serial)
+        print response.read()
+    
+
+def process_args(parser):
+    """
+    returns flowcell and serial#
+    """
+    options, args = parser.parse_args()
+    
+    msg = []
+    
+    # Only provide device or serial
+    if options.device is not None and options.serial is not None:
+        parser.error("Please provide only --device or --serial.\n"\
+                     "The serial number is extracted automatically if the"\
+                     "device is provided.")
+
+    # allow user to override the default destination URL
+    if options.url is not None:
+        root_url = options.url
+    else:
+        root_url = settings.LINK_FLOWCELL_STORAGE_DEVICE_URL
+
+    # if device and serial missing:
+    if options.device is None and options.serial is None:
+        parser.error('One of --device or --serial is required')
+
+    flowcells = []
+    
+    # sanitize args    
+    for runfolder in args:
+        flowcell_id = extract_flowcell(runfolder)
+        if flowcell_id is None:
+            parser.error('archive names must look like YYMMDD_MACHINE_RUN_FLOWCELLID.tgz\n'\
+                         '(got %s)' % (runfolder,))
+        else:
+            flowcells.append(flowcell_id)
+            
+    if options.flowcell is not None:
+        flowcells.append(options.flowcell)
+        
+    if len(flowcells) == 0:
+        parser.error('please specify a  --flowcell or list of runfolder archives\n'\
+                     'for archival. I need something to do.')
+
+    # Update db records
+    if options.device is not None:
+        serial = get_hd_serial_num(options.device)
+        update_db(root_url, flowcells, serial=serial, debug=options.verbose)
+    elif options.serial is not None:
+        update_db(root_url, flowcells, serial=options.serial, debug=options.verbose)
+    else:
+        msg ="FATAL should not happen error occured; i.e. the best kind!"
+        raise ValueError, msg
+    
+    
+
+def main():
+    """
+    """
+    parser = construct_parser()
+    process_args(parser)
+    
+    #print "Database Updated."
+    sys.exit(0)
+
+if __name__ == '__main__':
+    main()