def store_terms(self, session, index, hash, record):
if self.tempChunks:
if not hash:
return
# Make sure you know what you're doing
storeid = record.recordStore
if not isinstance(storeid, types.IntType):
storeid = self.storeHashReverse[storeid]
docid = long(record.id)
for k in hash.values():
try:
text = k['text'].encode('utf-8')
except:
print text
text = ""
if not text:
continue
lineList = [text,
str(docid),
str(storeid),
str(k['occurences'])
]
try:
lineList.append(nonTextToken.join(map(str,
k['positions'])
)
)
except KeyError:
# non prox
pass
if not text or not text[0].isalnum():
tf = "other"
else:
tf = text[0].lower()
try:
outh = self.outFiles[index][tf]
except:
if session.task:
fname = '_'.join([self.id,
index.id,
tf,
session.task,
'TEMP']
)
else:
fname = '_'.join([self.id, index.id, tf, 'TEMP'])
fname = os.path.join(self.tempPath, fname)
outh = file(fname, 'w')
self.outFiles[index][tf] = outh
outh.write(nonTextToken.join(lineList) + "\n")
return
if self.outFiles.has_key(index):
BdbIndexStore.store_terms(self, session, index, hash, record)
else:
raise NotImplementedError()
def store_terms(self, session, index, hash, record):
if self.tempChunks:
if not hash:
return
# Make sure you know what you're doing
storeid = record.recordStore
if not isinstance(storeid, types.IntType):
storeid = self.storeHashReverse[storeid]
docid = long(record.id)
for k in hash.values():
try:
text = k['text'].encode('utf-8')
except:
print text
text = ""
if not text:
continue
lineList = [
text,
str(docid),
str(storeid),
str(k['occurences'])
]
try:
lineList.append(
nonTextToken.join(map(str, k['positions'])))
except KeyError:
# non prox
pass
if not text or not text[0].isalnum():
tf = "other"
else:
tf = text[0].lower()
try:
outh = self.outFiles[index][tf]
except:
if session.task:
fname = '_'.join(
[self.id, index.id, tf, session.task, 'TEMP'])
else:
fname = '_'.join([self.id, index.id, tf, 'TEMP'])
fname = os.path.join(self.tempPath, fname)
outh = file(fname, 'w')
self.outFiles[index][tf] = outh
outh.write(nonTextToken.join(lineList) + "\n")
return
if self.outFiles.has_key(index):
BdbIndexStore.store_terms(self, session, index, hash, record)
else:
raise NotImplementedError()
def __init__(self, session, config, parent):
BdbIndexStore.__init__(self, session, config, parent)
self.tempChunks = self.get_setting(session, 'tempChunks')
uri = self.get_path(session, 'srbServer')
uri = uri.encode('utf-8')
uri = uri.strip()
if not uri:
raise ConfigFileException("No srbServer to connect to.")
else:
info = parseSrbUrl(uri)
for (a, v) in info.items():
setattr(self, a, v)
if (isinstance(parent, Database)):
sc = parent.id + "/" + self.id
else:
sc = self.id
self.subcollection = "cheshire3/" + sc
self.connection = None
self._connect()
def __init__(self, session, config, parent):
BdbIndexStore.__init__(self, session, config, parent)
self.tempChunks = self.get_setting(session, 'tempChunks')
uri = self.get_path(session, 'srbServer')
uri = uri.encode('utf-8')
uri = uri.strip()
if not uri:
raise ConfigFileException("No srbServer to connect to.")
else:
info = parseSrbUrl(uri)
for (a, v) in info.items():
setattr(self, a, v)
if (isinstance(parent, Database)):
sc = parent.id + "/" + self.id
else:
sc = self.id
self.subcollection = "cheshire3/" + sc
self.connection = None
self._connect()
def __init__(self, session, config, parent):
BdbIndexStore.__init__(self, session, config, parent)
self.switchingClass = IrodsSwitchingBdbConnection
self.vectorSwitchingClass = IrodsSwitchingBdbConnection
self.coll = None
self.cxn = None
self.env = None
self.host = self.get_setting(session, 'irodsHost', '')
self.port = self.get_setting(session, 'irodsPort', 0)
self.user = self.get_setting(session, 'irodsUser', '')
self.zone = self.get_setting(session, 'irodsZone', '')
self.passwd = self.get_setting(session, 'irodsPassword', '')
self.resource = self.get_setting(session, 'irodsResource', '')
self.allowStoreSubDirs = self.get_setting(session, 'allowStoreSubDirs', 1)
# And open irods
self._open(session)
def commit_indexing(self, session, index):
if self.tempChunks:
temp = self.tempPath
keys = self.outFiles[index].keys()
for f in self.outFiles[index].values():
f.flush()
f.close()
del self.outFiles[index]
sort = self.get_path(session, 'sortPath')
if hasattr(session, 'task'):
task = session.task
else:
task = None
if hasattr(session, 'phase'):
load = 0
else:
load = 1
sfiles = []
for k in keys:
if task:
fn = '_'.join([self.id, index.id, k, task])
else:
fn = '_'.join([self.id, index.id, k])
tf = os.path.join(temp, fn + "_TEMP")
sf = os.path.join(temp, fn + "_SORT")
cmd = "%s -f %s -o %s" % (sort, tf, sf)
f = commands.getoutput(cmd)
os.remove(tf)
if load:
self.commit_indexing2(session, index, sf)
else:
sfiles.append(sf)
return sfiles
else:
BdbIndexStore.commit_indexing(self, session, index)
def commit_indexing(self, session, index):
if self.tempChunks:
temp = self.tempPath
keys = self.outFiles[index].keys()
for f in self.outFiles[index].values():
f.flush()
f.close()
del self.outFiles[index]
sort = self.get_path(session, 'sortPath')
if hasattr(session, 'task'):
task = session.task
else:
task = None
if hasattr(session, 'phase'):
load = 0
else:
load = 1
sfiles = []
for k in keys:
if task:
fn = '_'.join([self.id, index.id, k, task])
else:
fn = '_'.join([self.id, index.id, k])
tf = os.path.join(temp, fn + "_TEMP")
sf = os.path.join(temp, fn + "_SORT")
cmd = "%s -f %s -o %s" % (sort, tf, sf)
f = commands.getoutput(cmd)
os.remove(tf)
if load:
self.commit_indexing2(session, index, sf)
else:
sfiles.append(sf)
return sfiles
else:
BdbIndexStore.commit_indexing(self, session, index)
def __init__(self, session, config, parent):
BdbIndexStore.__init__(self, session, config, parent)
if irods is None:
raise MissingDependencyException(self.objectType,
'irods (PyRods)'
)
self.switchingClass = IrodsSwitchingBdbConnection
self.vectorSwitchingClass = IrodsSwitchingBdbConnection
self.coll = None
self.cxn = None
self.env = None
self.host = self.get_setting(session, 'irodsHost', '')
self.port = self.get_setting(session, 'irodsPort', 0)
self.user = self.get_setting(session, 'irodsUser', '')
self.zone = self.get_setting(session, 'irodsZone', '')
self.passwd = self.get_setting(session, 'irodsPassword', '')
self.resource = self.get_setting(session, 'irodsResource', '')
self.allowStoreSubDirs = self.get_setting(session,
'allowStoreSubDirs',
1)
# And open iRODS
self._open(session)
def begin_indexing(self, session, index):
if not self.tempChunks:
return BdbIndexStore.begin_indexing(self, session, index)
temp = self.get_path(session, 'tempPath')
if not os.path.isabs(temp):
temp = os.path.join(self.get_path(session, 'defaultPath'),
temp)
self.tempPath = temp
if (not os.path.exists(temp)):
try:
os.mkdir(temp)
except:
raise(ConfigFileException('TempPath does not exist and is '
'not creatable.'))
elif (not os.path.isdir(temp)):
raise(ConfigFileException('TempPath is not a directory.'))
# Make temp files on demand, in hash
self.outFiles[index] = {}
def begin_indexing(self, session, index):
if not self.tempChunks:
return BdbIndexStore.begin_indexing(self, session, index)
temp = self.get_path(session, 'tempPath')
if not os.path.isabs(temp):
temp = os.path.join(self.get_path(session, 'defaultPath'),
temp)
self.tempPath = temp
if (not os.path.exists(temp)):
try:
os.mkdir(temp)
except:
raise (ConfigFileException(
'TempPath does not exist and is '
'not creatable.'))
elif (not os.path.isdir(temp)):
raise (ConfigFileException('TempPath is not a directory.'))
# Make temp files on demand, in hash
self.outFiles[index] = {}
def __init__(self, session, config, parent):
BdbIndexStore.__init__(self, session, config, parent)
raise MissingDependencyException(self.objectType, 'irods (PyRods)')
