source: TI07-MOLES/trunk/PythonCode/wsgi/DocumentRetrieve.py @ 2487

from eXistInterface import ndg_eXist
from ndg_xqueries import *
from xml.dom import minidom 
from stripped_xqueries import strip_de_xquery

# The MOLES document retrieval is a python port of :
# TI07-MOLES/trunk/JavaCode/returnmolesxmldb/ndg/services/returnmoles/Main.java
# Note that ndgRetreive essentially provides test cases for this code.

debug=1

def queryReplace(xquery,repository,localID,targetCollection):
    ''' Used to modify the vanilla xqueries to a query which gets the actual requested documents '''
    xquery=xquery.replace('RepositoryID',repository,1)
    xquery=xquery.replace('LocalID',localID,1)
    xquery=xquery.replace('TargetCollection', targetCollection)
    return xquery
    
class DocumentRetrieve (ndg_eXist):
    ''' This class provides a document retrieval service via the NDG exist interfaces '''
    def __init__(self,repository):

        ndg_eXist.__init__(self,db=repository)
        self.repository=repository

    def get(self,repository,schema,localID,format='NDG-B0',targetCollection='/db/discovery/moles'):

        # We are making the assumption for now that everything is stored as 
        # MOLES documents and that we can retrieve from MOLES anything we like
        # in another format.
        if schema == 'CSML':
            
            # temporary expedient ...  should really use Kev's code ...
            raise TypeError('DocumentRetrieve code does not yet know how to find specific CSML Documents')
        
        elif (schema[0:5] == 'NDG-B' or schema[0:3]=='DIF' or schema[0:4] == 'MDIP'):
            
            # find out what type of object actually exists of this sort
            # 0 - None, 1 - Activity, 2 - DPT, 3 - ObsStn, 4 - DE
            # if the schema is a DIF, we expect to find a DE from the DIF ingestion to MOLES ...
            
            # the following xquery can be used to generate a listing of the database
            # contents ...
            # xquery=listingQuery

            # default code has targetCollection='/db/ndg_B_metadata, this gets us the right one ...
            xquery=ObjectTypeQuery
            xquery=queryReplace(xquery,repository,localID,targetCollection)
         
            id,summary=self.executeQuery(xquery)
            hits=summary['hits']
            print 'bnl',targetCollection,xquery
            
            if hits!=1: raise ValueError,'%s documents returned for uri %s:%s:%s (in %s)'%(hits,repository,'NDG-B0',localID,targetCollection)
            
            # check output formats
            allowed = ['DIF','DC','ISO19139','NDG-B0','NDG-B1','MDIP']
            if format not in allowed: raise TypeError,'Invalid document output format [%s]'%format
            
            # get output type 
            r=self.retrieve(id,0,{})
            
            # typical response looks like this:
            #        <objectType xmlns="http://ndg.nerc.ac.uk/moles">1</objectType>
            xml=minidom.parseString(r)
            objectType=int(xml.getElementsByTagNameNS(
                            "http://ndg.nerc.ac.uk/moles", "objectType").item(0).firstChild.data)
            if debug: print 'objectType=%s'%objectType
            
            if format.find('NDG')==-1 and objectType!=4:
                msg={1:'Activity',2:'Data Production Tool',3:'Observation Station'}[objectType]
                raise TypeError,'Document URI type [%s] not valid for output format [%s]'%(msg,format)
            
            #select the correct xquery
            xquery={'DIF':DIFQuery,
                    'DC':DublinCoreDEQuery,
                    'ISO19139':ISO19139Query,
                    'NDG-B0':MOLESQuery,
                    'MDIP':MDIPQuery,
                    'NDG-B1':{
                        4:StubBDEQuery,#strip_de_xquery,#
                        3:StubBObsStnQuery,
                        2:StubBDPTQuery,
                        1:StubBActQuery}[objectType]
                     }[format]
                     
            xquery=queryReplace(xquery,repository,localID,targetCollection)
            try:
                id,summary=self.executeQuery(xquery)
            except Exception,e:
                print xquery
                raise Exception,e
            hits=summary['hits']
            # should only be the one document in the result set
            if hits!=1: 
                if debug:
                    f=open('xquery.%s.fails.xq'%format,'w')
                    f.write(xquery)
                    f.close()
                raise ValueError,'Actual Document Query returned [%s] hits - internal error!'%hits
             
            # now let's get it and return it 
            r=self.retrieve(id,0,{})

        else:
            raise TypeError('Unknown Schema "%s" in URI'%schema)
        self.sessionRelease(id)
        return r