query.py

query.py

from urllib import urlopen,quote
from urlparse import urljoin

from tms.parser import Record, TMSParser
from settings import TMS_SERVER_ENTRY_POINT
from settings import AUTHOR_SERVER_ENTRY_POINT
from settings import LOG_TMS_REQUESTS

def indent_xml(xml, indent=' '*4):
    """Indent and return XML."""
    from xml.dom.minidom import parseString
    return parseString(xml).toprettyxml(indent)

class Connection(object):
    def __init__(self, server=TMS_SERVER_ENTRY_POINT,author_server=AUTHOR_SERVER_ENTRY_POINT):
        if server:
            self.server = server
        if author_server:
            self.author_server = author_server

class QueryProcessor(object):
    " Fires actual requests to the TMS backend and receives xml responses "
        
    def __init__(self, connection=Connection()):
        self.connection = connection

    def find(self,*args, **kwargs):
        if len(kwargs.keys()) == 1:
            k = kwargs.keys()[0]
            v = kwargs[k]
            return self.cit(query='dc:%s=%s' % (k,v))
        else:
            raise NotImplementedError('only one key=value pair is supported')

    def get_author(self,authorname,action='get', command='search',fields='*',start=None,end=None):
        """ Fires an C-IT query to get Author Biografie """
        if not start:
            range = '*'
        else:
            range = '%s-%s' % (str(start),str(end))
        url = self.connection.author_server + 'action=%s&command=%s&' \
                                       'query=displayname=%s&' \
                                       'fields=%s&range=%s' % (action,command,unicode(authorname).encode('ascii','ignore'),fields,range)
        response = urlopen(quote(url, '/:=')).read()
        return response

    def cit(self,action='get',
            command='search',query='*=*',fields='*',start=None,end=None, sort=None):
        """ Fires an C-IT query """
        if not start:
            range = '*'
        else:
            range = '%s-%s' % (str(start),str(end))
        url = self.connection.server + 'action=%s&command=%s&' \
                                       'query=%s&' \
                                       'fields=%s&range=%s' % (action,command,unicode(query).encode('ascii','replace'),fields,range)
        if sort:
            url = '%s&sort=%s' % (url,sort)
        # added url quoting martijng 04/01/2010

        #if flag is true we print tms request to stderr
        if LOG_TMS_REQUESTS:
            import sys
            print >>sys.stderr, 'TMS REQUEST:',quote(url, "/:=")

        response = urlopen(quote(url, '/:=')).read()
        return response

    def sru(self,version='1.1',query=None,maximum_records=20,startrecord=1,*args,**kwargs):
        """ Fires an SRU query """
        kwargs['maximumRecords']= maximum_records
        kwargs['startrecord'] = startrecord
        kwargs['version'] = quote(version)
        kwargs['query'] = quote(query)
        #fixed elements in request
        kwargs['operation'] = 'searchRetrieve'
        kwargs['recordSchema'] = 'dc'
        kwargs['sortKeys']= 'dc:identifier'
        #full url of request
        url = self.connection.server + '?operation=%(operation)s&version=%(version)s&' \
                                       'query=%(query)s&' \
                                       'maximumRecords=%(maximumRecords)d&recordSchema=%(recordSchema)s&' \
                                       'startrecord=%(startrecord)d&sortKeys=%(sortKeys)s' % kwargs
        response = urlopen(url).read()
        return response