def test_sessionDatabaseAssign(self):
session = Session()
session.database = "db_test1"
self.assertEqual(session.database, "db_test1",
"session.database assignment failed")
session.database = "db_test2"
self.assertEqual(session.database, "db_test2",
"session.database re-assignment failed")
def build_architecture(data=None):
global rebuild, session, serv, db, dbPath
global editStore, authStore, instStore, userStore, xmlp
global docStoreConfigStore
session = Session()
session.database = 'db_hubedit'
session.environment = 'apache'
# session.user = None
serv = SimpleServer(session,
os.path.join(cheshire3Root,
'configs',
'serverConfig.xml'
)
)
db = serv.get_object(session, 'db_hubedit')
dbPath = db.get_path(session, 'defaultPath')
editStore = db.get_object(session, 'editingStore')
userStore = db.get_object(session, 'hubAuthStore')
instStore = db.get_object(session, 'institutionStore')
docStoreConfigStore = db.get_object(session, 'documentStoreConfigStore')
authStore = db.get_object(session, 'adminAuthStore')
xmlp = db.get_object(session, 'LxmlParser')
rebuild = False
def process_update(self, req):
self.version = "1.1"
self.operationStatus = "fail"
if (not req.version):
diag = Diagnostic7()
diag.message = "Mandatory 'version' parameter not supplied"
diag.details = 'version'
raise diag
config = req.config
db = config.parent
req._db = db
session = Session()
session.environment = "apache"
session.database = db.id
if req.operation == "info:srw/operation/1/create":
# Do Create
self.handle_create(session, req)
elif req.operation == "info:srw/operation/1/replace":
# Do Replace
self.handle_replace(session, req)
elif req.operation == "info:srw/operation/1/delete":
# Do Delete
self.handle_delete(session, req)
elif req.operation == "info:srw/operation/1/metadata":
# Do Metadata update
self.handle_metadata(session, req)
else:
# Barf
diag = SRWDiagnostics.Diagnostic1()
diag.details = "Unknown operation: %s" % req.operation
self.diagnostics = [diag]
def build_architecture(data=None):
# data argument provided for when function run as clean-up - always None
global session, serv, db, dbPath, docParser, \
fullTxr, fullSplitTxr, \
ppFlow, \
rebuild
# globals line 1: re-establish session; maintain user if possible
if (session):
u = session.user
else:
u = None
session = Session()
session.database = 'db_ead'
session.environment = 'apache'
session.user = u
serv = SimpleServer(session, os.path.join(cheshirePath, 'cheshire3', 'configs', 'serverConfig.xml'))
db = serv.get_object(session, 'db_ead')
dbPath = db.get_path(session, 'defaultPath')
docParser = db.get_object(session, 'LxmlParser')
# globals line 4: transformers
fullTxr = db.get_object(session, 'htmlFullTxr')
fullSplitTxr = db.get_object(session, 'htmlFullSplitTxr')
# globals line 5: workflows
ppFlow = db.get_object(session, 'preParserWorkflow'); ppFlow.load_cache(session, db)
rebuild = False
def build_architecture(data=None):
global session, serv, db, qf, xmlp, recordStore, sentenceStore, paragraphStore, resultSetStore, articleTransformer, kwicTransformer
session = Session()
session.environment = 'apache'
session.user = None
serv = SimpleServer(session,
os.path.join(cheshire3Root, 'configs', 'serverConfig.xml')
)
session.database = 'db_' + databaseName
db = serv.get_object(session, session.database)
qf = db.get_object(session, 'defaultQueryFactory')
xmlp = db.get_object(session, 'LxmlParser')
recordStore = db.get_object(session, 'recordStore')
articleTransformer = db.get_object(session, 'article-Txr')
kwicTransformer = db.get_object(session, 'kwic-Txr')
def build_architecture(data=None):
global session, serv, db, qf, xmlp, recordStore, resultSetStore, idxStore, articleTransformer, kwicTransformer, proxExtractor, simpleExtractor, adf, fimi2, rule, arm, vecTxr, vectorStore, armTableTxr
session = Session()
session.environment = 'apache'
session.user = None
serv = SimpleServer(session,
os.path.join(cheshire3Root, 'configs', 'serverConfig.xml')
)
session.database = 'db_' + databaseName
db = serv.get_object(session, session.database)
qf = db.get_object(session, 'defaultQueryFactory')
xmlp = db.get_object(session, 'LxmlParser')
recordStore = db.get_object(session, 'recordStore')
resultSetStore = db.get_object(session, 'resultSetStore')
simpleExtractor = db.get_object(session, 'SimpleExtractor')
proxExtractor = db.get_object(session, 'ProxExtractor')
articleTransformer = db.get_object(session, 'article-Txr')
kwicTransformer = db.get_object(session, 'kwic-Txr')
idxStore = db.get_object(session, 'indexStore')
sys.stderr.write("for help use --help\n")
sys.stderr.flush()
return 2
except Error as e:
lgr.log_lvl(session, 40, str(e))
if debug:
raise
return 1
# Build environment...
session = Session()
serv = SimpleServer(
session,
os.path.join(cheshire3Root,
'configs',
'serverConfig.xml'
)
)
session.database = 'db_hubedit'
db = serv.get_object(session, 'db_hubedit')
lgr = db.get_path(session, 'defaultLogger')
authStore = db.get_object(session, 'hubAuthStore')
superAuthStore = db.get_object(session, 'hubSuperAuthStore')
xmlp = db.get_object(session, 'LxmlParser')
if __name__ == "__main__":
sys.exit(main())
if db.get_setting(session, 'SRW') or db.get_setting(session, 'srw'):
db._cacheProtocolMaps(session)
map = db.protocolMaps.get('http://www.loc.gov/zing/srw/', None)
# check that there's a path and that it can actually be requested from this handler
if (map is not None) and \
(map.databaseUrl.startswith((handlerUrl + '/', handlerUrl[1:] + '/'))):
map2 = db.protocolMaps.get('http://www.loc.gov/zing/srw/update/', None)
configs[map.databaseUrl] = {'http://www.loc.gov/zing/srw/' : map,
'http://www.loc.gov/zing/srw/update/' : map2}
elif (map is not None):
apache.log_error("Database URL ({0}) does not match handler URL ({1}); will not handle database {{2}}".format(map.databaseUrl, handlerUrl, db.id), apache.APLOG_WARNING)
else:
# too many dbs to cache in memory
for dbid, conf in serv.databaseConfigs.iteritems():
db = serv.get_object(session, dbid)
session.database = dbid
if db.get_setting(session, 'SRW') or db.get_setting(session, 'srw'):
db._cacheProtocolMaps(session)
pmap = db.protocolMaps.get('http://www.loc.gov/zing/srw/', None)
if (pmap is not None) and (pmap.databaseUrl.startswith((handlerUrl + '/', handlerUrl[1:] + '/'))):
configs[pmap.databaseUrl] = (dbid, {'http://www.loc.gov/zing/srw/': pmap.id})
pmap2 = db.protocolMaps.get('http://www.loc.gov/zing/srw/update/', None)
if pmap2 is not None:
configs[pmap.databaseUrl][1].update({'http://www.loc.gov/zing/srw/update/': pmap2.id})
elif (pmap is not None):
apache.log_error("Database URL ({0}) does not match handler URL ({1}); will not handle database {{2}}".format(pmap.databaseUrl, handlerUrl, dbid), apache.APLOG_WARNING)
# remove cached db object
try:
del serv.objects[dbid]
except KeyError:
pass
## count words in books, and list titles
## used to create booklist
import os
import re
from lxml import etree
import json
from cheshire3.document import StringDocument
from cheshire3.internal import cheshire3Root
from cheshire3.server import SimpleServer
from cheshire3.baseObjects import Session
session = Session()
session.database = 'db_dickens'
serv = SimpleServer(session,
os.path.join(cheshire3Root, 'configs', 'serverConfig.xml'))
db = serv.get_object(session, session.database)
qf = db.get_object(session, 'defaultQueryFactory')
resultSetStore = db.get_object(session, 'resultSetStore')
idxStore = db.get_object(session, 'indexStore')
list_books = [
'BH', 'BR', 'DC', 'DS', 'ED', 'GE', 'HT', 'LD', 'MC', 'NN', 'OCS', 'OMF',
'OT', 'PP', 'TTC', 'AgnesG', 'Antoni', 'arma', 'cran', 'Deronda',
'dracula', 'emma', 'frank', 'jane', 'Jude', 'LadyAud', 'mary', 'NorthS',
'persuasion', 'pride', 'sybil', 'Tess', 'basker', 'Pomp', 'mill', 'dorian',
'Prof', 'native', 'alli', 'Jekyll', 'wwhite', 'vanity', 'VivianG', 'wh'
]
titles = {
# Create the parser for the "remove" command
parser_remove = subparsers.add_parser('remove',
help='Remove an existing user')
parser_remove.add_argument('username',
type=str,
nargs='*',
help='Username of the user(s) to remove')
parser_remove.set_defaults(func=remove_user)
# Build environment...
session = Session()
serv = SimpleServer(
session,
os.path.join(cheshire3Root,
'configs',
'serverConfig.xml'
)
)
session.database = 'db_ead'
db = serv.get_object(session, 'db_ead')
xmlp = db.get_object(session, 'LxmlParser')
authStore = db.get_object(session, 'hubAuthStore') # Editors
superAuthStore = db.get_object(session, 'adminAuthStore') # Hub Staff
instStore = db.get_object(session, 'institutionStore') # Institutions
if __name__ == '__main__':
sys.exit(main())
def test_sessionDatabaseAssign(self):
session = Session()
session.database = "db_test1"
self.assertEqual(session.database, "db_test1", "session.database assignment failed")
session.database = "db_test2"
self.assertEqual(session.database, "db_test2", "session.database re-assignment failed")
#!/usr/bin/python
import sys
import os
from cheshire3.baseObjects import Session
from cheshire3.server import SimpleServer
from cheshire3.internal import cheshire3Root
# Build environment...
session = Session() # a Session - used to store
print cheshire3Root
serv = SimpleServer(session, os.path.join(cheshire3Root, 'configs', 'serverConfig.xml'))
session.logger = serv.get_path(session, 'defaultLogger') # a logger
db = serv.get_object(session, 'db_tdo_index') # the Database
session.database = db.id
#qf = db.get_object(session, 'defaultQueryFactory')
def testVec():
recordStore = db.get_object(session, 'recordStore')
rec = recordStore.fetch_record(session, 1)
idx= db.get_object(session, 'idx-topic')
vec = idx.fetch_vector(session, rec)
map2 = db.protocolMaps.get(
'http://www.loc.gov/zing/srw/update/', None)
configs[map.databaseUrl] = {
'http://www.loc.gov/zing/srw/': map,
'http://www.loc.gov/zing/srw/update/': map2
}
elif (map is not None):
apache.log_error(
"Database URL ({0}) does not match handler URL ({1}); will not handle database {{2}}"
.format(map.databaseUrl, handlerUrl,
db.id), apache.APLOG_WARNING)
else:
# too many dbs to cache in memory
for dbid, conf in serv.databaseConfigs.iteritems():
db = serv.get_object(session, dbid)
session.database = dbid
if db.get_setting(session, 'SRW') or db.get_setting(session, 'srw'):
db._cacheProtocolMaps(session)
pmap = db.protocolMaps.get('http://www.loc.gov/zing/srw/', None)
if (pmap is not None) and (pmap.databaseUrl.startswith(
(handlerUrl + '/', handlerUrl[1:] + '/'))):
configs[pmap.databaseUrl] = (dbid, {
'http://www.loc.gov/zing/srw/':
pmap.id
})
pmap2 = db.protocolMaps.get(
'http://www.loc.gov/zing/srw/update/', None)
if pmap2 is not None:
configs[pmap.databaseUrl][1].update(
{'http://www.loc.gov/zing/srw/update/': pmap2.id})
elif (pmap is not None):
sys.path.insert(1, os.path.join(cheshirePath, 'cheshire3', 'code')) from cheshire3.baseObjects import Session from cheshire3.server import SimpleServer from cheshire3.document import StringDocument from cheshire3 import exceptions as c3errors from cheshire3.web.www_utils import read_file # import customisable variables #from localConfig import * # Build environment... session = Session() serv = SimpleServer(session, os.path.join(cheshirePath, 'cheshire3', 'configs', 'serverConfig.xml')) session.database = 'db_ead' db = serv.get_object(session, 'db_ead') lgr = db.get_path(session, 'defaultLogger') recordStore = db.get_object(session, 'recordStore') authStore = db.get_object(session, 'eadAuthStore') compStore = db.get_object(session, 'componentStore') clusDocFac = db.get_object(session, 'clusterDocumentFactory') clusDb = serv.get_object(session, 'db_ead_cluster') clusRecordStore = clusDb.get_object(session, 'eadClusterStore') xmlp = db.get_object(session, 'LxmlParser') lockfilepath = db.get_path(session, 'defaultPath') + '/indexing.lock'
def groupDist(dist): hits = sum(dist.values()) occs=0 for v in dist: occs += int(v) * int(dist[v]) for i in [1,2,3]: print "%s\t%s\t%0.2f" % (i, dist[i], float(dist[i])/float(hits) * 100.0) fourPlus=0 for i in range(4,max(dist.keys())): try: fourPlus += dist[i] except: continue print "4+\t%s\t%0.2f" % (fourPlus, float(fourPlus)/float(hits) * 100.0) print "\n%i occurrences in %i articles" % (occs,hits) session = Session() serv = SimpleServer(session, "../../configs/serverConfig.xml") db = serv.get_object(session, 'db_news') session.database = 'db_news' idxStore = db.get_object(session, 'indexStore') recStore = db.get_object(session, 'recordStore')
# separate file containing display configs + some HMTL for table rows etc.
from clic.dickens.web.dickensWebConfig import *
from clic.dickens.web.dickensSearchHandler import SearchHandler
from clic.dickens.web.dickensBrowseHandler import BrowseHandler
cheshirePath = os.environ.get('HOME', '/home/cheshire')
logPath = os.path.join(cheshirePath, 'clic', 'www', databaseName, 'logs', 'searchHandler.log')
htmlPath = os.path.join(cheshirePath, 'clic', 'www', databaseName, 'html')
session = Session()
session.environment = 'apache'
session.user = None
serv = SimpleServer(session, os.path.join(cheshire3Root, 'configs', 'serverConfig.xml'))
session.database = 'db_dickens'
db = serv.get_object(session, session.database)
authStore = db.get_object(session, 'authStore')
# Discover objects...
def handler(req):
global db, htmlPath, logPath, cheshirePath, xmlp, recordStore
form = FieldStorage(req)
try:
dir = req.uri[1:].rsplit('/')[1]
except IndexError:
return apache.HTTP_NOT_FOUND
remote_host = req.get_remote_host(apache.REMOTE_NOLOOKUP)
lgr = FileLogger(logPath, remote_host)
# lgr.log(req.uri)
