def convert(inFilePath, destDir, targetFormat, dropTables=False):
# Output file is name of input file with the
# .json extension replaced by .sql
outFullPath = buildOutputFileName(inFilePath, destDir)
# Log file will go to <destDir>/../TransformLogs, the file being named j2s_<inputFileName>.log:
logDir = os.path.join(destDir, '..', 'TransformLogs')
if not os.access(logDir, os.W_OK):
try:
os.makedirs(logDir)
except OSError:
# Log dir already exists:
pass
logFile = os.path.join(logDir, 'j2s_%s.log' % os.path.basename(inFilePath))
# Create an instance of JSONToRelation, taking input from the given file:
# and pumping output to the given output path:
if targetFormat == 'csv':
outputFormat = OutputDisposition.OutputFormat.CSV
elif targetFormat == 'sql_dump':
outputFormat = OutputDisposition.OutputFormat.SQL_INSERT_STATEMENTS
else:
outputFormat = OutputDisposition.OutputFormat.SQL_INSERTS_AND_CSV
outSQLFile = OutputFile(outFullPath, outputFormat,
options='wb') # overwrite any existing sql file
jsonConverter = JSONToRelation(InURI(inFilePath),
outSQLFile,
mainTableName='EdxTrackEvent',
logFile=logFile,
progressEvery=10000)
try:
jsonConverter.setParser(
EdXTrackLogJSONParser(jsonConverter,
'EdxTrackEvent',
replaceTables=dropTables,
dbName='Edx',
progressEvery=10000))
except Exception as e:
with open(logFile, 'w') as fd:
fd.write(
"In json2sql: could not create EdXTrackLogJSONParser: %s" %
` e `)
# Try to delete the .sql file that was created when
# the OutputFile instance was made in the JSONToRelation
# instantiation statement above:
try:
outSQLFile.remove()
except Exception as e:
pass
sys.exit(1)
jsonConverter.convert()
