source: hdldemo/trunk/hdldemo.py @ 558

"""
Demonstration handle dump for CMIP/ESGF files ..

 -h: print this message;
 -v: print version;
 -t: run a test;
 -f <file name>: examine file, print path to replacement if this file is obsolete, print path to sibling files.
"""

import string, collections, os
import ncq3
try:
 import urllib
 from urllib import request
except:
 import requests

from testdata import *

class phandle(object):
  
  def __init__(self, hdlDict, k='values'):
    self.h = hdlDict
    self.d = {}
    try:
      for r in hdlDict[k]:
        self.d[r['type']] = r['data']
    except:
      print ( hdlDict[k] )
      raise


class ghandle(object):
  def __init__(self,url):
     try:
       fh = request.urlopen( url )
       msg = eval( fh.read() )
     except:
       r = requests.get( url )
       msg = r.json()
     assert type( msg ) == type( {} ), 'Response of wrong type'
     for k in ['responseCode', 'handle']:
       assert k in msg, 'Required key %s not found: %s' % (k, str( msg.keys() ) )

     ###print ( 'url read ok' )
     self.msg = msg

class main(object):
  knownargs0 = ['-h','-v','-t']
  knownargs1 = ['-f']
  def __init__(self, args):
    self.htmpl = 'http://hdl.handle.net/api/handles/%s'
    self.version = '0.01.01'
    self.args = args
    self.parseArgs()
    if self.d.get( '-v', False ):
      print ( self.version )
      return

    if self.d.get( '-h', False ):
      print (self.version)
      print ( __doc__ )
      return

    if self.d.get( '-t', False ):
      self.runTest()
      return

    if '-f' in self.d:
      fn = self.d['-f']
      self.dumpF(fn)

  def dumpF(self,fn):
    assert os.path.isfile( fn ), 'File %s not found' % fn
    f = ncHead( fn )
    thisid = string.replace(f.ga['tracking_id'], 'hdl:999999', '10876.test' )
    url = self.htmpl % thisid
    g = ghandle( url )
    self.p = phandle(  g.msg )
    gp = ghandle( self.htmpl % self.p.d['parent']['value'] )
    self.pp = phandle(  gp.msg )
    if 'replaced_by' in self.p.d:
      print '******* OBSOLETE FILE ******'
      pr = phandle(  ghandle( self.htmpl % self.p.d['replaced_by']['value'] ).msg )
      print 'REPLACED BY: ',pr.d['URL']['value']
    else:
      print 'File is current'

    cl = eval( self.pp.d['children']['value'] )
    thisfound=False
    print '------- SIBLINGS -------'
    for c in cl:
      if c == thisid:
        thisfound = True
      else:
        gc = ghandle( self.htmpl %  c )
        pc = phandle( gc.msg )
        if 'replaced_by' in pc.d:
          pcr = phandle(  ghandle( self.htmpl % pc.d['replaced_by']['value'] ).msg )
          print '>>>', pcr.d['URL']['value']
        elif 'URL' in pc.d:
          print pc.d['URL']['value']
        else:
          print 'Sibling url not found',pc.d.keys()

  def runTest(self):
    f = ncHead( fn1 )
    url = self.htmpl % f.ga['tracking_id']
    url = string.replace( url, 'hdl:999999', '10876.test' )
    print url
    g = ghandle( url )
    self.p = phandle(  g.msg )
    expected= ['creation_date', 'aggregation_level', 'HS_ADMIN', '10320/loc', 'checksum', 'URL', 'parent']
    ## 'replaced_by' if obsolete
    for k in expected:
      assert k in self.p.d, 'Expected handle content key %s not found:: %s' % (k,str(self.p.d.keys()))
    assert 'tracking_id' in self.p.d or 'trackingID' in self.p.d, 'No tracking id found: %s' % str(self.p.d.keys())
    for k in expected:
      print ('%s: %s' % (k,self.p.d[k]))

    print ('PARSING PARENT ..... ' )
    print ( self.htmpl % self.p.d['parent']['value'] )
    g = ghandle( self.htmpl % self.p.d['parent']['value'] )
    self.pp = phandle(  g.msg )
    for k in self.pp.d.keys():
      print ('%s: %s' % (k,self.pp.d[k]))
    #'isReplacedBy' if obsolete
    expected= ['creation_date', 'aggregation_level', 'HS_ADMIN', '10320/loc', 'checksum', 'URL', 'children', 'tracking_id']

  def parseArgs(self):
    self.d = {}
    kn = self.knownargs0 + self.knownargs1
    xx = []
    al = self.args[1:]
    while len(al) > 0:
      a = al.pop(0)
      if a not in kn:
        xx.append(a)
      elif a in self.knownargs1:
         self.d[a] = al.pop(0)
      else:
         self.d[a] = True
    if len(xx) > 0:
      print ('ARGUMENTS NOT RECOGNISED: %s' % str(xx) )
        
class ncHead(object):
  def __init__(self, fn):
    nc0 = ncq3.open( fn )
    nc0.getDigest()
    self.ga = {}
    for a in nc0.alla:
      self.ga[a.name] = a.value
      
      
if __name__ == "__main__":
  import sys
  m = main( sys.argv )