class PostgresInsertOutput(PostgresDbOutput):
"""
Output by inserting a single record in a Postgres database table.
Input is a Stetl record (Python dict structure) or a list of records.
Creates an INSERT for Postgres to insert each single record.
When the "replace" parameter is True, any existing record keyed by "key" is
attempted to be UPDATEd first.
NB a constraint is that the first and each subsequent each record needs to contain
all values as an INSERT and UPDATE query template is built once for the columns
in the first record.
consumes=[FORMAT.record_array, FORMAT.record]
"""
# Start attribute config meta
@Config(ptype=str, required=False, default='public')
def table(self):
"""
Table for inserts.
"""
pass
@Config(ptype=bool, required=False, default=False)
def replace(self):
"""
Replace record if exists?
"""
pass
@Config(ptype=str, required=False, default=None)
def key(self):
"""
The key column name of the table, required when replacing records.
"""
pass
# End attribute config meta
def __init__(self, configdict, section, consumes=FORMAT.record):
DbOutput.__init__(self, configdict, section, consumes=[FORMAT.record_array, FORMAT.record])
self.query = None
self.update_query = None
self.db = None
def init(self):
# Connect only once to DB
log.info('Init: connect to DB')
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
def exit(self):
# Disconnect from DB when done
log.info('Exit: disconnect from DB')
self.db.disconnect()
def create_query(self, record):
# We assume that all records do the same INSERT key/values
# See http://grokbase.com/t/postgresql/psycopg/12735bvkmv/insert-into-with-a-dictionary-or-generally-with-a-variable-number-of-columns
# e.g. INSERT INTO lml_files ("file_name", "file_data") VALUES (%s,%s)
query = "INSERT INTO %s (%s) VALUES (%s)" % (
self.cfg.get('table'), ",".join(['%s' % k for k in record]), ",".join(["%s", ] * len(record.keys())))
log.info('query is %s', query)
return query
def create_update_query(self, record):
# We assume that all records do the same UPDATE key/values
# https://stackoverflow.com/questions/1109061/insert-on-duplicate-update-in-postgresql/6527838#6527838
# e.g. UPDATE table SET field='C', field2='Z' WHERE id=3;
query = "UPDATE %s SET (%s) = (%s) WHERE %s = %s" % (
self.cfg.get('table'), ",".join(['%s ' % k for k in record]), ",".join(["%s", ] * len(record.keys())), self.key, "%s")
log.info('update query is %s', query)
return query
def insert(self, record):
res = 0
if self.replace and self.key and self.key in record:
# Replace option: try UPDATE if existing
# https://stackoverflow.com/questions/1109061/insert-on-duplicate-update-in-postgresql/6527838#6527838
values = record.values()
values.append(record[self.key])
res = self.db.execute(self.update_query, values)
# del_query = "DELETE FROM %s WHERE %s = '%s'" % (self.cfg.get('table'), self.key, record[self.key])
# res = self.db.execute(del_query)
if res < 1:
# Do insert with values from the record dict
# only if we did not do an UPDATE (res==0) on existing record.
self.db.execute(self.query, record.values())
self.db.commit(close=False)
def write(self, packet):
# Deal with empty or zero-length data structures (list or dict)
if packet.data is None or len(packet.data) == 0:
return packet
# ASSERT: record data present
# record is Python dict (single record) or list of Python dict (multiple records)
record = packet.data
# Generate INSERT query template once
first_record = record
if type(record) is list and len(record) > 0:
first_record = record[0]
# Create INSERT and optional UPDATE query-templates once
if self.query is None:
self.query = self.create_query(first_record)
if self.replace and self.key and not self.update_query:
self.update_query = self.create_update_query(first_record)
# Check if record is single (dict) or array (list of dict)
if type(record) is dict:
# Do insert with values from the single record
self.insert(record)
# log.info('committed record key=%s' % record[self.key])
elif type(record) is list:
# Multiple records in list
for rec in record:
# Do insert with values from the record
self.insert(rec)
log.info('committed %d records' % len(record))
return packet
class Josene(Device):
def __init__(self):
Device.__init__(self, 'jose')
self.model_query = "SELECT id,parameters,model from calibration_models WHERE predicts = '%s' AND invalid = FALSE ORDER BY timestamp DESC LIMIT 1"
self.state_query = "SELECT state from calibration_state WHERE process = '%s' AND model_id = %d ORDER BY timestamp DESC LIMIT 1"
self.state_insert = "INSERT INTO calibration_state (process, model_id, state) VALUES ('%s', %d, '%s')"
self.sensor_model_names = {
'co': 'carbon_monoxide__air_',
'no2': 'nitrogen_dioxide__air_',
'o3': 'ozone__air_'
}
self.config_dict = None
def init(self, config_dict):
self.config_dict = config_dict
self.process_name = config_dict['process_name']
self.db = PostGIS(config_dict)
self.db.connect()
ids = dict()
parameters = dict()
models = dict()
state = dict()
# Query ANN Calibration Model and its State from DB for each calibrated sensor.
if self.model_query is not None and len(self.sensor_model_names) > 0:
log.info('Getting calibration models and state from database')
for k in self.sensor_model_names:
v = self.sensor_model_names[k]
id, param, model = self.query_model(v)
ids[k] = id
parameters[k] = param
models[k] = model
model_state = self.query_state(id)
state[k] = model_state
else:
log.info('No query for fetching calibration models given or no '
'mapping for calibration models to gas components given.')
# Put Model and State info in the Device definitions.
for k in ids:
SENSOR_DEFS[k]['converter_model']['model_id'] = ids[k]
for k in parameters:
SENSOR_DEFS[k]['converter_model']['running_mean_weights'] = parameters[k]
for k in models:
SENSOR_DEFS[k]['converter_model']['mlp_regressor'] = models[k]
for k, v in state.iteritems():
for device_id, device_state in v.iteritems():
for gas, state in device_state.iteritems():
v[device_id][gas] = RunningMean.from_dict(state)
SENSOR_DEFS[k]['converter_model']['state'] = v
def exit(self):
# Save the calibration state.
for k in self.sensor_model_names:
model = SENSOR_DEFS[k]['converter_model']
self.save_state(model['model_id'], json.dumps(model['state']))
self.db.commit(close=False)
def get_sensor_defs(self):
return SENSOR_DEFS
def raw_query(self, query_str):
self.db.execute(query_str)
db_records = self.db.cursor.fetchall()
log.info('read recs: %d' % len(db_records))
return db_records
def query_model(self, name):
query = self.model_query % name
log.info('Getting calibration model with query: %s' % query)
ret = self.raw_query(query)
if len(ret) > 0:
id, parameters, model = ret[0]
return id, parameters, pickle.loads(model)
else:
log.warn("No model found for %s" % name)
return None, {}, {}
def query_state(self, model_id):
query = self.state_query % (self.process_name, model_id)
log.info('Getting calibration model state with query: %s' % query)
ret = self.raw_query(query)
if len(ret) > 0:
return ret[0][0]
else:
log.warn("No state found for model_id=%d" % model_id)
return {}
def save_state(self, model_id, state):
insert_query = self.state_insert % (self.process_name, model_id, state)
log.info('Inserting calibration model state for process %s model_id=%d' % (self.process_name, model_id))
ret = self.db.execute(insert_query)
if ret != 1:
log.warn('Cannot save state for process %s model_id=%d' % (self.process_name, model_id))
# Get raw sensor value or list of values
def get_raw_value(self, name, val_dict):
val = None
if type(name) is list:
name = name[0]
return self.get_raw_value(name, val_dict)
# name is list of names
# for n in name:
# if n in val_dict:
# if val is None:
# val = []
# val.append(val_dict[n])
else:
# name is single name
if name in val_dict:
val = val_dict[name]
if 'audio' in name:
# We may have audio encoded in 3 bands
bands = [float(val & 255), float((val >> 8) & 255), float((val >> 16) & 255)]
val = bands[0]
return val, name
# Check for valid sensor value
def check_value(self, name, val_dict, value=None):
val = None
if type(name) is list:
# name is list of names
for n in name:
result, reason = self.check_value(n, val_dict, value)
if result is False:
return result, reason
else:
# name is single name
if name not in val_dict and value is None:
return False, '%s not present' % name
else:
if value is not None:
val = value
else:
val = val_dict[name]
if val is None:
return False, '%s is None' % name
if name not in SENSOR_DEFS:
return False, '%s not in SENSOR_DEFS' % name
name_def = SENSOR_DEFS[name]
# Audio inputs: need to unpack 3 bands and check for decibel vals
if 'audio' in name:
bands = [float(val & 255), float((val >> 8) & 255), float((val >> 16) & 255)]
# determine validity of these 3 bands
dbMin = name_def['min']
dbMax = name_def['max']
err_cnt = 0
msg = ''
for i in range(0, len(bands)):
band_val = bands[i]
# accumulate outliers
if band_val < dbMin:
err_cnt +=1
msg += '%s: val(%s) < min(%s)\n' % (name, str(band_val), str(name_def['min']))
elif band_val > dbMax:
err_cnt +=1
msg += '%s: val(%s) > max(%s)\n' % (name, str(band_val), str(name_def['max']))
# Only invalid if all bands outside range
if err_cnt >= len(bands):
return False, msg
return True, '%s OK' % name
if 'min' in name_def and val < name_def['min']:
return False, '%s: val(%s) < min(%s)' % (name, str(val), str(name_def['min']))
if 'max' in name_def and val > name_def['max']:
return False, '%s: val(%s) > max(%s)' % (name, str(val), str(name_def['max']))
return True, '%s OK' % name
# Get location as lon, lat
def get_lon_lat(self, val_dict):
result = (None, None)
if 's_longitude' in val_dict and 's_latitude' in val_dict:
lon = SENSOR_DEFS['longitude']['converter'](val_dict['s_longitude'])
lat = SENSOR_DEFS['latitude']['converter'](val_dict['s_latitude'])
valid, reason = self.check_value('latitude', val_dict, value=lat)
if not valid:
return result
valid, reason = self.check_value('longitude', val_dict, value=lon)
if not valid:
return result
result = (lon, lat)
return result
def write(self, packet):
if packet.data is None:
return packet
gml_doc = packet.data
log.info('inserting features in DB')
db = PostGIS(self.cfg.get_dict())
db.connect()
# print self.to_string(gml_doc, False, False)
# NS = {'base': 'urn:x-inspire:specification:gmlas:BaseTypes:3.2', 'gml': 'http://www.opengis.net/gml/3.2'}
# featureMembers = gml_doc.xpath('//base:member/*', namespaces=NS)
featureMembers = gml_doc.xpath("//*[local-name() = '%s']/*" % self.feature_member_tag)
count = 0
gml_ns = None
for childNode in featureMembers:
if gml_ns is None:
if childNode.nsmap.has_key('gml'):
gml_ns = childNode.nsmap['gml']
else:
if childNode.nsmap.has_key('GML'):
gml_ns = childNode.nsmap['GML']
gml_id = childNode.get('{%s}id' % gml_ns)
feature_type_id = self.feature_type_ids[childNode.tag]
# Find a GML geometry in the GML NS
ogrGeomWKT = None
# gmlMembers = childNode.xpath(".//gml:Point|.//gml:Curve|.//gml:Surface|.//gml:MultiSurface", namespaces=NS)
gmlMembers = childNode.xpath(
".//*[local-name() = 'Point']|.//*[local-name() = 'Polygon']|.//*[local-name() = 'Curve']|.//*[local-name() = 'Surface']|.//*[local-name() = 'MultiSurface']")
geom_str = None
for gmlMember in gmlMembers:
if geom_str is None:
geom_str = etree.tostring(gmlMember)
# no need for GDAL Python bindings for now, maybe when we'll optimize with COPY iso INSERT
# ogrGeom = ogr.CreateGeometryFromGML(str(gmlStr))
# if ogrGeom is not None:
# ogrGeomWKT = ogrGeom.ExportToWkt()
# if ogrGeomWKT is not None:
# break
blob = etree.tostring(childNode, pretty_print=False, xml_declaration=False, encoding='UTF-8')
if geom_str is None:
sql = "INSERT INTO gml_objects(gml_id, ft_type, binary_object) VALUES (%s, %s, %s)"
parameters = (gml_id, feature_type_id, db.make_bytea(blob))
else:
# ST_SetSRID(ST_GeomFromGML(%s)),-1)
sql = "INSERT INTO gml_objects(gml_id, ft_type, binary_object, gml_bounded_by) VALUES (%s, %s, %s, ST_SetSRID( ST_GeomFromGML(%s),%s) )"
parameters = (gml_id, feature_type_id, db.make_bytea(blob), geom_str, self.srid)
if db.execute(sql, parameters) == -1:
log.error("feat num# = %d error inserting feature blob=%s (but continuing)" % (count, blob))
# will fail but we will close connection also
db.commit()
# proceed...
log.info('retrying to proceed with remaining features...')
db = PostGIS(self.cfg.get_dict())
db.connect()
count = 0
count += 1
exception = db.commit()
if exception is not None:
log.error("error in commit")
log.info("inserted %s features" % count)
return packet
class PostgresInsertOutput(PostgresDbOutput):
"""
Output by inserting a single record in a Postgres database table.
Input is a Stetl record (Python dict structure) or a list of records.
Creates an INSERT for Postgres to insert each single record.
When the "replace" parameter is True, any existing record keyed by "key" is
attempted to be UPDATEd first.
NB a constraint is that the first and each subsequent each record needs to contain
all values as an INSERT and UPDATE query template is built once for the columns
in the first record.
consumes=[FORMAT.record_array, FORMAT.record]
"""
# Start attribute config meta
@Config(ptype=str, required=False, default='public')
def table(self):
"""
Table for inserts.
"""
pass
@Config(ptype=bool, required=False, default=False)
def replace(self):
"""
Replace record if exists?
"""
pass
@Config(ptype=str, required=False, default=None)
def key(self):
"""
The key column name of the table, required when replacing records.
"""
pass
# End attribute config meta
def __init__(self, configdict, section, consumes=FORMAT.record):
DbOutput.__init__(self,
configdict,
section,
consumes=[FORMAT.record_array, FORMAT.record])
self.query = None
self.update_query = None
self.db = None
def init(self):
# Connect only once to DB
log.info('Init: connect to DB')
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
def exit(self):
# Disconnect from DB when done
log.info('Exit: disconnect from DB')
self.db.disconnect()
def create_query(self, record):
# We assume that all records do the same INSERT key/values
# See http://grokbase.com/t/postgresql/psycopg/12735bvkmv/insert-into-with-a-dictionary-or-generally-with-a-variable-number-of-columns
# e.g. INSERT INTO lml_files ("file_name", "file_data") VALUES (%s,%s)
query = "INSERT INTO %s (%s) VALUES (%s)" % (
self.cfg.get('table'), ",".join(['%s' % k
for k in record]), ",".join([
"%s",
] * len(record.keys())))
log.info('query is %s', query)
return query
def create_update_query(self, record):
# We assume that all records do the same UPDATE key/values
# https://stackoverflow.com/questions/1109061/insert-on-duplicate-update-in-postgresql/6527838#6527838
# e.g. UPDATE table SET field='C', field2='Z' WHERE id=3;
query = "UPDATE %s SET (%s) = (%s) WHERE %s = %s" % (self.cfg.get(
'table'), ",".join(['%s ' % k for k in record]), ",".join([
"%s",
] * len(record.keys())), self.key, "%s")
log.info('update query is %s', query)
return query
def insert(self, record):
res = 0
if self.replace and self.key and self.key in record:
# Replace option: try UPDATE if existing
# https://stackoverflow.com/questions/1109061/insert-on-duplicate-update-in-postgresql/6527838#6527838
values = record.values()
values.append(record[self.key])
res = self.db.execute(self.update_query, values)
# del_query = "DELETE FROM %s WHERE %s = '%s'" % (self.cfg.get('table'), self.key, record[self.key])
# res = self.db.execute(del_query)
if res < 1:
# Do insert with values from the record dict
# only if we did not do an UPDATE (res==0) on existing record.
self.db.execute(self.query, record.values())
self.db.commit(close=False)
def write(self, packet):
# Deal with empty or zero-length data structures (list or dict)
if packet.data is None or len(packet.data) == 0:
return packet
# ASSERT: record data present
# record is Python dict (single record) or list of Python dict (multiple records)
record = packet.data
# Generate INSERT query template once
first_record = record
if type(record) is list and len(record) > 0:
first_record = record[0]
# Create INSERT and optional UPDATE query-templates once
if self.query is None:
self.query = self.create_query(first_record)
if self.replace and self.key and not self.update_query:
self.update_query = self.create_update_query(first_record)
# Check if record is single (dict) or array (list of dict)
if type(record) is dict:
# Do insert with values from the single record
self.insert(record)
# log.info('committed record key=%s' % record[self.key])
elif type(record) is list:
# Multiple records in list
for rec in record:
# Do insert with values from the record
self.insert(rec)
log.info('committed %d records' % len(record))
return packet
class DeegreeBlobstoreInput(Input):
"""
Read features from deegree Blobstore DB into an etree doc.
produces=FORMAT.etree_doc
"""
# Start attribute config meta
@Config(ptype=int, required=False, default=10000)
def max_features_per_doc(self):
"""
Max features to read from input feature GML stream per internal document.
"""
pass
@Config(ptype=str, required=True, default=None)
def start_container(self):
"""
Tag that starts container.
"""
pass
@Config(ptype=str, required=True, default=None)
def end_container(self):
"""
Tag that ends container.
"""
pass
@Config(ptype=str, required=False, default=False)
def start_feature_tag(self):
"""
XML tag that starts Feature.
"""
pass
@Config(ptype=str, required=False, default=None)
def end_feature_tag(self):
"""
XML tag that ends Feature.
"""
pass
# End attribute config meta
def __init__(self, configdict, section):
Input.__init__(self, configdict, section, produces=FORMAT.etree_doc)
self.cur_feature_blob = None
self.rowcount = 0
# http://www.mkyong.com/regular-expressions/how-to-extract-html-links-with-regular-expression/
self.regex_xlink_href = re.compile(
"\\s*(?i)xlink:href\\s*=\\s*(\"#([^\"]*\")|'#[^']*'|(#[^'\">\\s]+))"
)
self.db = None
self.xlink_db = None
self.buffer = None
self.feature_count = 0
# Reusable XML parser
self.xml_parser = etree.XMLParser(remove_blank_text=True)
def init(self):
pass
def read(self, packet):
if packet.is_end_of_stream():
return packet
if self.db is None:
# First time read
log.info("reading records from blobstore..")
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
sql = self.cfg.get('sql')
self.rowcount = self.db.execute(sql)
self.cur = self.db.cursor
log.info("Read records rowcount=%d" % self.rowcount)
# Init separate connection to fetch objects referenced by xlink:href
self.xlink_db = PostGIS(self.cfg.get_dict())
self.xlink_db.connect()
# Query active
while self.cur is not None:
if self.buffer is None:
self.buffer = self.init_buf()
self.buffer.write(self.start_container)
# Get next blob record
record = self.cur.fetchone()
# End of all records
if record is None:
# End of records: start closing
self.buffer.write(self.end_container)
self.cur = None
self.db.commit()
# Only create doc if there are features in the buffer
if self.feature_count > 0:
self.buffer_to_doc(packet)
packet.set_end_of_doc()
break
else:
# New record: embed feature blob in feature tags and write to buffer
feature_blob = self.write_feature(record)
# If we have local xlinks: fetch the related features as well from the DB and
# output them within the same document (local href resolvable)
# TODO: in some cases we may need to be recursive (xlinks in xlinked features...)
# First construct a single query for all xlinks
xlink_sql = None
for xlink in self.regex_xlink_href.finditer(feature_blob):
gml_id = xlink.group(1).strip('"').strip('#')
# We don't want multiple occurences of the same xlinked feature
if gml_id in self.xlink_ids:
continue
self.xlink_ids.add(gml_id)
if xlink_sql is None:
xlink_sql = "SELECT binary_object from gml_objects where gml_id = '%s'" % gml_id
else:
xlink_sql += "OR gml_id = '%s'" % gml_id
# Should we retrieve and write xlinked features?
if xlink_sql is not None:
# Fetch from DB
self.xlink_db.execute(xlink_sql)
while True:
# Get next blob record
xlink_record = self.xlink_db.cursor.fetchone()
if xlink_record is None:
break
self.write_feature(xlink_record)
# Should we output a doc
if self.feature_count >= self.max_features_per_doc:
# End of records: create XML doc
self.buffer.write(self.end_container)
self.buffer_to_doc(packet)
break
if self.cur is None:
# All records handled: close off
packet.set_end_of_stream()
# log.info("[%s]" % packet.data)
return packet
def write_feature(self, record):
feature_blob = str(record[0])
# Write start-tag, blob element, end-tag
self.buffer.write(self.start_feature_tag)
self.buffer.write(feature_blob)
self.buffer.write(self.end_feature_tag)
self.feature_count += 1
return feature_blob
def init_buf(self):
buffer = StringIO()
buffer = codecs.getwriter("utf8")(buffer)
self.feature_count = 0
self.xlink_ids = set()
return buffer
def buffer_to_doc(self, packet):
# Process/transform data in buffer
self.buffer.seek(0)
try:
packet.data = etree.parse(self.buffer, self.xml_parser)
except Exception as e:
bufStr = self.buffer.getvalue()
if not bufStr:
log.info("parse buffer empty: content=[%s]" % bufStr)
else:
log.error("error in buffer parsing %s" % str(e))
raise
self.buffer.close()
self.buffer = None
class DeegreeBlobstoreInput(Input):
"""
Read features from deegree Blobstore DB into an etree doc.
produces=FORMAT.etree_doc
"""
# Start attribute config meta
@Config(ptype=int, required=False, default=10000)
def max_features_per_doc(self):
"""
Max features to read from input feature GML stream per internal document.
"""
pass
@Config(ptype=str, required=True, default=None)
def start_container(self):
"""
Tag that starts container.
"""
pass
@Config(ptype=str, required=True, default=None)
def end_container(self):
"""
Tag that ends container.
"""
pass
@Config(ptype=str, required=False, default=False)
def start_feature_tag(self):
"""
XML tag that starts Feature.
"""
pass
@Config(ptype=str, required=False, default=None)
def end_feature_tag(self):
"""
XML tag that ends Feature.
"""
pass
# End attribute config meta
def __init__(self, configdict, section):
Input.__init__(self, configdict, section, produces=FORMAT.etree_doc)
self.cur_feature_blob = None
self.rowcount = 0
# http://www.mkyong.com/regular-expressions/how-to-extract-html-links-with-regular-expression/
self.regex_xlink_href = re.compile("\\s*(?i)xlink:href\\s*=\\s*(\"#([^\"]*\")|'#[^']*'|(#[^'\">\\s]+))")
self.db = None
self.xlink_db = None
self.buffer = None
self.feature_count = 0
# Reusable XML parser
self.xml_parser = etree.XMLParser(remove_blank_text=True)
def init(self):
pass
def read(self, packet):
if packet.is_end_of_stream():
return packet
if self.db is None:
# First time read
log.info("reading records from blobstore..")
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
sql = self.cfg.get('sql')
self.rowcount = self.db.execute(sql)
self.cur = self.db.cursor
log.info("Read records rowcount=%d" % self.rowcount)
# Init separate connection to fetch objects referenced by xlink:href
self.xlink_db = PostGIS(self.cfg.get_dict())
self.xlink_db.connect()
# Query active
while self.cur is not None:
if self.buffer is None:
self.buffer = self.init_buf()
self.buffer.write(self.start_container)
# Get next blob record
record = self.cur.fetchone()
# End of all records
if record is None:
# End of records: start closing
self.buffer.write(self.end_container)
self.cur = None
self.db.commit()
# Only create doc if there are features in the buffer
if self.feature_count > 0:
self.buffer_to_doc(packet)
packet.set_end_of_doc()
break
else:
# New record: embed feature blob in feature tags and write to buffer
feature_blob = self.write_feature(record)
# If we have local xlinks: fetch the related features as well from the DB and
# output them within the same document (local href resolvable)
# TODO: in some cases we may need to be recursive (xlinks in xlinked features...)
# First construct a single query for all xlinks
xlink_sql = None
for xlink in self.regex_xlink_href.finditer(feature_blob):
gml_id = xlink.group(1).strip('"').strip('#')
# We don't want multiple occurences of the same xlinked feature
if gml_id in self.xlink_ids:
continue
self.xlink_ids.add(gml_id)
if xlink_sql is None:
xlink_sql = "SELECT binary_object from gml_objects where gml_id = '%s'" % gml_id
else:
xlink_sql += "OR gml_id = '%s'" % gml_id
# Should we retrieve and write xlinked features?
if xlink_sql is not None:
# Fetch from DB
self.xlink_db.execute(xlink_sql)
while True:
# Get next blob record
xlink_record = self.xlink_db.cursor.fetchone()
if xlink_record is None:
break
self.write_feature(xlink_record)
# Should we output a doc
if self.feature_count >= self.max_features_per_doc:
# End of records: create XML doc
self.buffer.write(self.end_container)
self.buffer_to_doc(packet)
break
if self.cur is None:
# All records handled: close off
packet.set_end_of_stream()
# log.info("[%s]" % packet.data)
return packet
def write_feature(self, record):
feature_blob = str(record[0])
# Write start-tag, blob element, end-tag
self.buffer.write(self.start_feature_tag)
self.buffer.write(feature_blob)
self.buffer.write(self.end_feature_tag)
self.feature_count += 1
return feature_blob
def init_buf(self):
buffer = StringIO()
buffer = codecs.getwriter("utf8")(buffer)
self.feature_count = 0
self.xlink_ids = set()
return buffer
def buffer_to_doc(self, packet):
# Process/transform data in buffer
self.buffer.seek(0)
try:
packet.data = etree.parse(self.buffer, self.xml_parser)
except Exception as e:
bufStr = self.buffer.getvalue()
if not bufStr:
log.info("parse buffer empty: content=[%s]" % bufStr)
else:
log.error("error in buffer parsing %s" % str(e))
raise
self.buffer.close()
self.buffer = None
def write(self, packet):
if packet.data is None:
return packet
gml_doc = packet.data
log.info('inserting features in DB')
db = PostGIS(self.cfg.get_dict())
db.connect()
# print self.to_string(gml_doc, False, False)
# NS = {'base': 'urn:x-inspire:specification:gmlas:BaseTypes:3.2', 'gml': 'http://www.opengis.net/gml/3.2'}
# featureMembers = gml_doc.xpath('//base:member/*', namespaces=NS)
featureMembers = gml_doc.xpath("//*[local-name() = '%s']/*" %
self.feature_member_tag)
count = 0
gml_ns = None
for childNode in featureMembers:
if gml_ns is None:
if childNode.nsmap.has_key('gml'):
gml_ns = childNode.nsmap['gml']
else:
if childNode.nsmap.has_key('GML'):
gml_ns = childNode.nsmap['GML']
gml_id = childNode.get('{%s}id' % gml_ns)
feature_type_id = self.feature_type_ids[childNode.tag]
# Find a GML geometry in the GML NS
ogrGeomWKT = None
# gmlMembers = childNode.xpath(".//gml:Point|.//gml:Curve|.//gml:Surface|.//gml:MultiSurface", namespaces=NS)
gmlMembers = childNode.xpath(
".//*[local-name() = 'Point']|.//*[local-name() = 'Polygon']|.//*[local-name() = 'Curve']|.//*[local-name() = 'Surface']|.//*[local-name() = 'MultiSurface']"
)
geom_str = None
for gmlMember in gmlMembers:
if geom_str is None:
geom_str = etree.tostring(gmlMember)
# no need for GDAL Python bindings for now, maybe when we'll optimize with COPY iso INSERT
# ogrGeom = ogr.CreateGeometryFromGML(str(gmlStr))
# if ogrGeom is not None:
# ogrGeomWKT = ogrGeom.ExportToWkt()
# if ogrGeomWKT is not None:
# break
blob = etree.tostring(childNode,
pretty_print=False,
xml_declaration=False,
encoding='UTF-8')
if geom_str is None:
sql = "INSERT INTO gml_objects(gml_id, ft_type, binary_object) VALUES (%s, %s, %s)"
parameters = (gml_id, feature_type_id, db.make_bytea(blob))
else:
# ST_SetSRID(ST_GeomFromGML(%s)),-1)
sql = "INSERT INTO gml_objects(gml_id, ft_type, binary_object, gml_bounded_by) VALUES (%s, %s, %s, ST_SetSRID( ST_GeomFromGML(%s),%s) )"
parameters = (gml_id, feature_type_id, db.make_bytea(blob),
geom_str, self.srid)
if db.execute(sql, parameters) == -1:
log.error(
"feat num# = %d error inserting feature blob=%s (but continuing)"
% (count, blob))
# will fail but we will close connection also
db.commit()
# proceed...
log.info('retrying to proceed with remaining features...')
db = PostGIS(self.cfg.get_dict())
db.connect()
count = 0
count += 1
exception = db.commit()
if exception is not None:
log.error("error in commit")
log.info("inserted %s features" % count)
return packet
class WeewxDbInput(SqliteDbInput):
"""
Reads weewx raw archive records from SQLite.
"""
def __init__(self, configdict, section):
SqliteDbInput.__init__(self, configdict, section)
self.progress_query = self.cfg.get('progress_query')
self.progress_update = self.cfg.get('progress_update')
# Connect only once to DB
log.info('Init: connect to Postgres DB')
self.progress_db = PostGIS(self.cfg.get_dict())
self.progress_db.connect()
def exit(self):
# Disconnect from DB when done
log.info('Exit: disconnect from DB')
self.progress_db.disconnect()
def after_chain_invoke(self, packet):
"""
Called right after entire Component Chain invoke.
Used to update last id of processed file record.
"""
# last_datetime.datetime.fromtimestamp(self.last_id).strftime('%Y-%m-%d %H:%M:%S')
ts_local = time.strftime("%Y-%m-%d %H:%M:%S %Z", time.localtime(self.last_id))
log.info('Updating progress table ts_unix=%d ts_local=%s' % (self.last_id, ts_local))
self.progress_db.execute(self.progress_update % (self.last_id, ts_local))
self.progress_db.commit(close=False)
log.info('Update progress table ok')
return True
def read(self, packet):
# Get last processed id of archive table
self.progress_db.execute(self.progress_query)
progress_rec = self.progress_db.cursor.fetchone()
self.last_id = progress_rec[3]
log.info('progress record: %s' % str(progress_rec))
# Fetch next batch of archive records
archive_recs = self.do_query(self.query % self.last_id)
log.info('read archive_recs: %d' % len(archive_recs))
# No more records to process?
if len(archive_recs) == 0:
packet.set_end_of_stream()
log.info('Nothing to do. All file_records done')
return packet
# Remember last id processed for next query
self.last_id = archive_recs[len(archive_recs)-1].get('dateTime')
packet.data = archive_recs
# Always stop after batch, otherwise we would continue forever
packet.set_end_of_stream()
return packet
class ProgressTracker(Filter):
""""
Filter to track progress of a stream of processed records.
Stores progress (last id, last timestamp etc) in Postgres table.
"""
@Config(ptype=str, required=False, default='localhost')
def host(self):
"""
host name or host IP-address, defaults to 'localhost'
"""
pass
@Config(ptype=str, required=False, default='5432')
def port(self):
"""
port for host, defaults to '5432'
"""
pass
@Config(ptype=str, required=False, default='postgres')
def user(self):
"""
User name, defaults to 'postgres'
"""
pass
@Config(ptype=str, required=False, default='postgres')
def password(self):
"""
User password, defaults to 'postgres'
"""
pass
@Config(ptype=str, required=False, default='public')
def schema(self):
"""
The postgres schema name, defaults to 'public'
"""
pass
@Config(ptype=str, required=False, default='progress')
def table(self):
"""
Table name, defaults to 'progress'.
"""
pass
@Config(ptype=str, required=True)
def progress_update_query(self):
"""
Query to update progress
Required: True
Default: ""
"""
pass
@Config(ptype=str, required=True)
def id_key(self):
"""
Key to select id from record array
Required: True
"""
@Config(ptype=str, default=None, required=False)
def name_key(self):
"""
Key to select name from record array
Required: True
"""
def __init__(self, config_dict, section):
Filter.__init__(self, config_dict, section,
consumes=[FORMAT.record_array, FORMAT.record],
produces=[FORMAT.record_array, FORMAT.record])
self.last_ids = None
self.db = None
def init(self):
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
def invoke(self, packet):
self.last_ids = dict()
if packet.data is None or packet.is_end_of_doc() or packet.is_end_of_stream():
log.info("No packet data or end of doc/stream")
return packet
record_in = packet.data
if type(record_in) is not list:
record_in = [record_in]
for record in record_in:
if self.name_key is not None:
name = record[self.name_key]
else:
name = "all"
if len(record) > 0:
new = record[self.id_key]
self.last_ids[name] = max(self.last_ids.get(name, -1), new)
log.info("Last ids are: %s", str(self.last_ids))
return packet
def after_chain_invoke(self, packet):
"""
Called right after entire Component Chain invoke.
Used to update last id of processed file record.
"""
for name in self.last_ids:
param_tuple = (self.last_ids[name], name)
log.info('Updating progress table with (id=%d, name=%s)' % param_tuple)
self.db.execute(self.progress_update_query % param_tuple)
self.db.commit(close=False)
log.info('Update progress table ok')
else:
log.info('No update for progress table')
return True
class PostgresInsertOutput(PostgresDbOutput):
"""
Output by inserting single record into Postgres database.
Input is a record (Python dic structure) or a Python list of dicts (records).
Creates an INSERT for Postgres to insert each single record.
consumes=FORMAT.record
"""
def __init__(self, configdict, section, consumes=FORMAT.record):
DbOutput.__init__(self, configdict, section, consumes=[FORMAT.record_array, FORMAT.record])
self.query = None
self.db = None
self.key = self.cfg.get('key')
def init(self):
# Connect only once to DB
log.info('Init: connect to DB')
self.db = PostGIS(self.cfg.get_dict())
self.db.connect()
def exit(self):
# Disconnect from DB when done
log.info('Exit: disconnect from DB')
self.db.disconnect()
def create_query(self, record):
# We assume that all records do the same INSERT key/values
# See http://grokbase.com/t/postgresql/psycopg/12735bvkmv/insert-into-with-a-dictionary-or-generally-with-a-variable-number-of-columns
# e.g. INSERT INTO lml_files ("file_name", "file_data") VALUES (%s,%s)
query = "INSERT INTO %s (%s) VALUES (%s)" % (self.cfg.get('table'), ",".join(['%s' % k for k in record]), ",".join(["%s",]*len(record.keys())))
log.info('query is %s', query)
return query
def write(self, packet):
# Deal with empty or zero-length data structures (list or dict)
if packet.data is None or len(packet.data) == 0:
return packet
# ASSERT: record data present
# record is Python dict (single record) or list of Python dict (multiple records)
record = packet.data
# Generate INSERT query template once
first_record = record
if type(record) is list and len(record) > 0:
first_record = record[0]
# Create query once
if self.query is None:
self.query = self.create_query(first_record)
# Check if record is single (dict) or array (list of dict)
if type(record) is dict:
# Do insert with values from the single record
self.db.execute(self.query, record.values())
self.db.commit(close=False)
# log.info('committed record key=%s' % record[self.key])
elif type(record) is list:
# Multiple records in list
for rec in record:
# Do insert with values from the record
self.db.execute(self.query, rec.values())
self.db.commit(close=False)
log.info('committed %d records' % len(record))
return packet
