class Table(object):
def __init__(self, tablename, conf=None):
if conf is None:
conf = HBaseConfiguration()
self.table = HTable(conf, tablename)
def get(self, rowkey):
'''
Retrieves the specific table row as a dictionary, with
full column names (column family:name) as keys, and deserialized
(from protocol buffers) values as values.
If the row does not exist, returns None.
'''
op = Get(Bytes.toBytes(java_str(rowkey)))
row = self.table.get(op)
if row is not None and not row.isEmpty():
return todict(row)
return None
def getVersions(self, rowkey, limit=None, ts=None):
'''
Retrieves up to _limit_ versions of the specified row,
at or before the specified timestamp (_ts_). If _ts_ is None
or not specified, defaults to now.
'''
op = Get(Bytes.toBytes(java_str(rowkey)))
if ts is not None:
op.setTimeRange(0, ts)
if limit is not None:
op.setMaxVersions(limit)
row = self.table.get(op)
versions = []
if row is not None and not row.isEmpty():
for e in row.list():
col = str(java.lang.String(e.getColumn()))
colts = e.getTimestamp()
val = PBUtil.toValue(e.getValue())
versions.append((colts, {col: val}))
return versions
def scan(self, cols, startrow=None, limit=50, filter=None):
rows = []
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
scan = Scan(startrow)
if filter is not None:
scan.setFilter(filter)
for c in cols:
scan.addColumn(c)
scanner = None
cnt = 0
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
if limit is not None and cnt >= limit:
break
rows.append(todict(rec))
cnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return rows
def scan_apply(self,
cols,
startrow=None,
limit=50,
filter=None,
rowfunc=None):
rows = []
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
else:
startrow = Bytes.toBytes(java_str(startrow))
scan = Scan(startrow)
if filter is not None:
scan.setFilter(filter)
for c in cols:
scan.addColumn(c)
cnt = 0
scanner = None
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
if limit is not None and cnt >= limit:
break
rowfunc(rec)
cnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return
def save(self, rowkey, vals, ts=None):
key = java_str(rowkey)
rowup = Put(Bytes.toBytes(key))
if ts is not None:
rowup.setTimestamp(ts)
for k, v in vals.items():
(fam, col) = KeyValue.parseColumn(Bytes.toBytes(java_str(k)))
if isinstance(v, com.google.protobuf.Message):
rowup.add(fam, col, v.toByteArray())
else:
rowup.add(fam, col, PBUtil.toBytes(v))
self.table.put(rowup)
def update(self, startrow, wherefilt, sets):
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
elif isinstance(startrow, java.lang.String):
startrow = Bytes.toBytes(startrow)
updater = ColumnUpdate(sets)
cols = self.families()
cnt = 0
upcnt = 0
scan = Scan(startrow)
for c in cols:
scan.addColumn(c)
if wherefilt is not None:
scan.setFilter(filter)
scanner = None
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
cnt += 1
rowup = updater.getUpdate(rec)
if rowup is not None:
self.table.commit(rowup)
upcnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return upcnt
def delete(self, rowkey):
op = Delete(Bytes.toBytes(java_str(rowkey)))
self.table.delete(op)
def deleteAll(self, rows):
for rec in rows:
if '__key__' in rec:
op = Delete(Bytes.toBytes(java_str(rec['__key__'])))
self.table.delete(op)
def families(self):
return [
coldesc.getNameAsString() + ":"
for coldesc in self.table.getTableDescriptor().getFamilies()
]
class Table(object):
def __init__(self, tablename, conf=None):
if conf is None:
conf = HBaseConfiguration()
self.table = HTable(conf, tablename)
def get(self, rowkey):
'''
Retrieves the specific table row as a dictionary, with
full column names (column family:name) as keys, and deserialized
(from protocol buffers) values as values.
If the row does not exist, returns None.
'''
op = Get( Bytes.toBytes( java_str(rowkey) ) )
row = self.table.get(op)
if row is not None and not row.isEmpty():
return todict(row)
return None
def getVersions(self, rowkey, limit=None, ts=None):
'''
Retrieves up to _limit_ versions of the specified row,
at or before the specified timestamp (_ts_). If _ts_ is None
or not specified, defaults to now.
'''
op = Get( Bytes.toBytes(java_str(rowkey)) )
if ts is not None:
op.setTimeRange( 0, ts )
if limit is not None:
op.setMaxVersions(limit)
row = self.table.get(op)
versions = []
if row is not None and not row.isEmpty():
for e in row.list():
col = str(java.lang.String(e.getColumn()))
colts = e.getTimestamp()
val = PBUtil.toValue( e.getValue() )
versions.append( (colts, {col: val} ) )
return versions
def scan(self, cols, startrow=None, limit=50, filter=None):
rows = []
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
scan = Scan(startrow)
if filter is not None:
scan.setFilter(filter)
for c in cols:
scan.addColumn(c)
scanner = None
cnt = 0
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
if limit is not None and cnt >= limit:
break
rows.append(todict(rec))
cnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return rows
def scan_apply(self, cols, startrow=None, limit=50, filter=None, rowfunc=None):
rows = []
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
else:
startrow = Bytes.toBytes( java_str(startrow) )
scan = Scan(startrow)
if filter is not None:
scan.setFilter(filter)
for c in cols:
scan.addColumn(c)
cnt = 0
scanner = None
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
if limit is not None and cnt >= limit:
break
rowfunc(rec)
cnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return
def save(self, rowkey, vals, ts=None):
key = java_str(rowkey)
rowup = Put( Bytes.toBytes(key) )
if ts is not None:
rowup.setTimestamp(ts)
for k, v in vals.items():
(fam, col) = KeyValue.parseColumn( Bytes.toBytes( java_str(k) ) )
if isinstance(v, com.google.protobuf.Message):
rowup.add(fam, col, v.toByteArray())
else:
rowup.add(fam, col, PBUtil.toBytes(v))
self.table.put(rowup)
def update(self, startrow, wherefilt, sets):
if startrow is None:
startrow = HConstants.EMPTY_START_ROW
elif isinstance(startrow, java.lang.String):
startrow = Bytes.toBytes(startrow)
updater = ColumnUpdate(sets)
cols = self.families()
cnt = 0
upcnt = 0
scan = Scan(startrow)
for c in cols:
scan.addColumn(c)
if wherefilt is not None:
scan.setFilter(filter)
scanner = None
try:
scanner = self.table.getScanner(scan)
for rec in scanner:
cnt += 1
rowup = updater.getUpdate(rec)
if rowup is not None:
self.table.commit(rowup)
upcnt += 1
finally:
if scanner is not None:
try:
scanner.close()
except:
pass
return upcnt
def delete(self, rowkey):
op = Delete( Bytes.toBytes( java_str(rowkey) ) )
self.table.delete( op )
def deleteAll(self, rows):
for rec in rows:
if '__key__' in rec:
op = Delete( Bytes.toBytes( java_str( rec['__key__'] ) ) )
self.table.delete( op )
def families(self):
return [coldesc.getNameAsString()+":" for coldesc in self.table.getTableDescriptor().getFamilies()]
