worker.py 文件源码

def get_report_data(year, quarter):
    ''' invoke tushare get_report_data() with csv output
    brief: to improve data integrality, we repeatedly do these actions in a row,
           call API -> append to file -> drop duplicates
    args: year, quarter
    returns: csv format data containing the whole martket report in specific year, quarter
    json fomat, df.to_json(year+'q'+quarter+'.json', orient='index')
    '''

    # profit
    print "[%s] profit %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.profit.csv'
    df = ts.get_profit_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    save_to_file(filename, df)

    # operation
    print "[%s] operation %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.operation.csv'
    df = ts.get_operation_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    save_to_file(filename, df)

    # growth
    print "[%s] growth %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.growth.csv'
    df = ts.get_growth_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    save_to_file(filename, df)

    # debtpaying
    print "[%s] debtpaying %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.debtpaying.csv'
    df = ts.get_debtpaying_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    save_to_file(filename, df)

    # cashflow
    print "[%s] cashflow %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.cashflow.csv'
    df = ts.get_cashflow_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    save_to_file(filename, df)

    # main report
    print "[%s] main %sq%s" %(datetime.now().strftime("%H:%M:%S.%f"), year, quarter)
    filename = PREFIX + '/' + year + 'q' + quarter + '.csv'
    df = ts.get_report_data(int(year), int(quarter)).sort_values(by='code').drop_duplicates()
    print "\n"
    return save_to_file(filename, df)