def insertSeaFlowCruiseTraj():
server = 'Rainier'
tableName = 'tblCruise_Trajectory'
rawFilePath = cfgv.rep_allSeaFlowCruises_raw
os.chdir(rawFilePath)
sfl_cruise_list = glob.glob('*.sfl*')
usecols_sfl = ['DATE', 'LAT', 'LON']
for cruise in sfl_cruise_list:
prefix = cruise[:-8] + '_traj'
rawFileName = cruise
path = rawFilePath + rawFileName
exportBase = cfgv.opedia_proj + 'db/dbInsert/export/'
export_path = '%s%s.csv' % (exportBase, prefix)
print(cruise)
Cruise_ID = iF.findID_CRUISE(cruise[:-8])
df = pd.read_csv(cruise, sep='\t', usecols=usecols_sfl)
df['DATE'] = pd.to_datetime(df['DATE'], format='%Y-%m-%dT%H:%M:%S')
df['Cruise_ID'] = Cruise_ID
df.rename(columns={
'DATE': 'time',
'LAT': 'lat',
'LON': 'lon'
},
inplace=True)
df = df[['Cruise_ID', 'time', 'lat', 'lon']]
df = ip.removeMissings(['time', 'lat', 'lon'], df)
df = ip.NaNtoNone(df)
df = ip.colDatatypes(df)
df = ip.convertYYYYMMDD(df)
df = ip.removeDuplicates(df)
df.to_csv(export_path, index=False)
ip.sortByTimeLatLon(df, export_path, 'time', 'lat', 'lon')
print('export path: ', export_path)
# print(export_path,tableName)
iF.toSQLbcp(export_path, tableName, server)
def insertAMTCruiseTraj():
server = 'Rainier'
tableName = 'tblCruise_Trajectory'
usecols = ['Cruise_name', 'time', 'lat', 'lon']
rawFilePath = cfgv.rep_AMT_cruises_raw + 'amt/'
rawFileName = 'master_AMT.csv'
path = rawFilePath + rawFileName
exportBase = cfgv.opedia_proj + 'db/dbInsert/export/'
os.chdir(rawFilePath)
df = pd.read_csv(rawFilePath + rawFileName, sep=',', usecols=usecols)
for Cruise_name in df['Cruise_name'].unique():
export_path = '%s%s.csv' % (exportBase, Cruise_name)
print(Cruise_name)
cruise_df = df[df['Cruise_name'] ==
Cruise_name] #selects only df of cruise
Cruise_ID = iF.findID_CRUISE(Cruise_name[0:3] + Cruise_name[-2:])
cruise_df['Cruise_ID'] = Cruise_ID
cruise_df = ip.removeMissings(['time', 'lat', 'lon'], cruise_df)
cruise_df = ip.convertYYYYMMDD(cruise_df)
cruise_df = ip.NaNtoNone(cruise_df)
cruise_df = ip.colDatatypes(cruise_df)
cruise_df = ip.convertYYYYMMDD(cruise_df)
cruise_df = ip.removeDuplicates(cruise_df)
cruise_df = cruise_df[['Cruise_ID', 'time', 'lat', 'lon']]
cruise_df.to_csv(export_path, index=False)
ip.sortByTimeLatLon(cruise_df, export_path, 'time', 'lat', 'lon')
print('export path: ', Cruise_name + export_path)
iF.toSQLbcp(export_path, tableName, server)
def insertSeaFlowCruiseSalinity():
server = 'Rainier'
tableName = 'tblCruise_Salinity'
rawFilePath = cfgv.rep_allSeaFlowCruises_raw
os.chdir(rawFilePath)
sfl_cruise_list = glob.glob('*.sfl*')
usecols_sfl = ['DATE', 'LAT', 'LON', 'SALINITY']
for cruise in sfl_cruise_list:
prefix = cruise[:-8] + '_temp'
rawFileName = cruise
path = rawFilePath + rawFileName
exportBase = cfgv.opedia_proj + 'db/dbInsert/export/'
export_path = '%s%s.csv' % (exportBase, prefix)
print(cruise)
Cruise_ID = iF.findID_CRUISE(cruise[:-8])
df = pd.read_csv(cruise, sep='\t', usecols=usecols_sfl)
df['DATE'] = pd.to_datetime(df['DATE'], format='%Y-%m-%dT%H:%M:%S')
df['DEPTH'] = 5.0
df['Cruise_ID'] = Cruise_ID
df.rename(columns={
'DATE': 'time',
'LAT': 'lat',
'LON': 'lon',
'DEPTH': 'depth',
'SALINITY': 'salinity'
},
inplace=True)
df = df[['Cruise_ID', 'time', 'lat', 'lon', 'depth', 'salinity']]
df = ip.removeMissings(['time', 'lat', 'lon', 'depth'], df)
df = df[pd.to_numeric(df['salinity'], errors='coerce').notnull()]
df = ip.NaNtoNone(df)
df = ip.colDatatypes(df)
df = ip.convertYYYYMMDD(df)
df = ip.removeDuplicates(df)
print(df.head())
if df.empty:
print(cruise +
' had no salinity values. Not inserted into database')
else:
df.to_csv(export_path, index=False)
ip.sortByTimeLatLon(df, export_path, 'time', 'lat', 'lon')
print('export path: ', export_path)
# print(export_path,tableName)
iF.toSQLbcp(export_path, tableName, server)
def insertAMTCruiseTemperature():
server = 'Rainier'
tableName = 'tblCruise_Temperature'
usecols = ['Cruise_name', 'time', 'lat', 'lon', 'temp', 'temp_flag']
rawFilePath = cfgv.rep_AMT_cruises_raw + 'amt/'
rawFileName = 'master_AMT.csv'
path = rawFilePath + rawFileName
exportBase = cfgv.opedia_proj + 'db/dbInsert/export/'
os.chdir(rawFilePath)
df = pd.read_csv(rawFilePath + rawFileName, sep=',', usecols=usecols)
for Cruise_name in df['Cruise_name'].unique():
export_path = '%s%s%s.csv' % (exportBase, Cruise_name, tableName)
print(Cruise_name)
cruise_df = df[df['Cruise_name'] ==
Cruise_name] #selects only df of cruise
Cruise_ID = iF.findID_CRUISE(Cruise_name[0:3] + Cruise_name[-2:])
cruise_df['Cruise_ID'] = Cruise_ID
cruise_df = cruise_df[(cruise_df['temp_flag'] != 'N')
& (cruise_df['temp_flag'] != 'S') &
(cruise_df['temp_flag'] != 'M') &
(cruise_df['temp_flag'] != 'L')]
cruise_df = ip.removeMissings(['time', 'lat', 'lon'], cruise_df)
cruise_df = ip.convertYYYYMMDD(cruise_df)
cruise_df = ip.colDatatypes(cruise_df)
cruise_df = ip.convertYYYYMMDD(cruise_df)
cruise_df = ip.removeDuplicates(cruise_df)
cruise_df = ip.renameCol(cruise_df, 'temp', 'temperature')
cruise_df = cruise_df[[
'Cruise_ID', 'time', 'lat', 'lon', 'temperature'
]]
cruise_df = cruise_df.dropna(subset=['temperature'])
cruise_df = ip.NaNtoNone(cruise_df)
if cruise_df.empty:
print(Cruise_name +
' had no temperature values. Not inserted into database')
else:
cruise_df.to_csv(export_path, index=False)
ip.sortByTimeLatLon(cruise_df, export_path, 'time', 'lat', 'lon')
print('export path: ', export_path)
iF.toSQLbcp(export_path, tableName, server)
