UTC日期时间格式

JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:578Unicorn
edited November 2018 inHelp
Hi,

I'm trying to import a column of times in the format
1 2014-09-01T14:30:12:123Z
2 2014-09-01T14:32:13:33Z
3 2014-09-01T14:12:12:523Z
4 2014-09-01T14:23:12:323Z
Currently RapidMiner uses SimpleDateFormat for the date formattinghttp://docs.oracle.com/javase/6/docs/api/java/text/SimpleDateFormat.html
but I'm struggling to work out how to implement it for the above values. It's the ISO 8601 datetime format so I had thought it would be pretty straightforwardhttp://www.w3.org/TR/NOTE-datetime

I have tired the patterns
yyyy-MM-dd'T'HH:mm:ss.SSSZ (this works for 1,3 & 4 but fails on 2)
yyyy-MM-dd'T'HH:mm:ss.SZ (this fails on all)
yyyy-MM-dd'T'HH:mm:ss.Sz (this fails on all)

Anyone have any ideas?

Answers

  • JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:578Unicorn
    I've now realised that
    yyyy-MM-dd'T'HH:mm:ss.S works however I have a new value that's throwing errors.

    2014-07-08T14:20:52Z

    In this example as the event happened exactly on .000 the server has removed all the numbers that it deems unnecessary.
    I need to therefore work out how to be a bit more flexible with this. Arrgh!! )
  • JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:578Unicorn
    Okay, I've solved this for my needs and added an additional converter to first import the dates as nominal and then loop through them to concat the data as necessary for SimpleDateFormat formatting.

    I'm leaving this open for now as this only works for UTC dates so if you have any other dates (or a better way of doing it than two loop attributes) let me know below.





    <宏/ >






    <关键参数="text" value="Date
2014-09-01T14:30:12.123Z
2014-09-01T14:32:13.33Z
2014-09-01T14:12:12.5Z
2014-09-01T14:23:12.323Z
2014-07-08T14:20:52Z
2014-07-08T14:20Z
2014-07-08T14Z
2014-07-08Z"/>


    <关键参数="encoding" value="UTF-8"/>


    <关键参数="date_format" value="yyyy-MM-dd'T'HH:mm:ss.SSSZ"/>

    <关键参数="time_zone" value="UTC"/>
    <关键参数="locale" value="English (United Kingdom)"/>
    <关键参数="encoding" value="UTF-8"/>

    <关键参数="0" value="DateField.true.nominal.attribute"/>




    <关键参数="DoubleDate" value="DateField"/>



    <连接from_op = from_写文档port="file" to_op="Read CSV" to_port="file"/>








    Select the 'Date fields stored as nominal to convert' using the Select Attributes operator here.
    NOTE: This only works for dates stored as UTC as it removes the timezone code 'Z'. 
    <关键参数="attribute_filter_type" value="subset"/>
    <关键参数="attributes" value="DateField|DoubleDate|"/>
    <关键参数="include_special_attributes" value="true"/>



    <关键参数="%{loop_attribute}" value="replace(%{loop_attribute},"Z","")"/>
    <关键参数="%{loop_attribute}" value="if(length(%{loop_attribute})==19,concat(%{loop_attribute},".000"),DateField)"/>
    <关键参数="%{loop_attribute}" value="if(length(%{loop_attribute})==16,concat(%{loop_attribute},":00.000"),DateField)"/>
    <关键参数="%{loop_attribute}" value="if(length(%{loop_attribute})==13,concat(%{loop_attribute},":00:00.000"),DateField)"/>
    <关键参数="%{loop_attribute}" value="if(length(%{loop_attribute})==10,concat(%{loop_attribute},"T00:00:00.000"),DateField)"/>










    Select the 'Date fields stored as nominal to convert' using the Select Attributes operator here.
    NOTE: This only works for dates stored as UTC as it removes the timezone code 'Z'.

    Annoyingly this is a two part conversion as the loop will try to execute again if you convert from nominal to date within the process and so two loops is the best alternative thus far. 
    <关键参数="attribute_filter_type" value="subset"/>
    <关键参数="attributes" value="DateField|DoubleDate|"/>
    <关键参数="include_special_attributes" value="true"/>


    <关键参数="attribute_name" value="%{loop_attribute}"/>
    <关键参数="date_type" value="date_time"/>
    <关键参数="date_format" value="yyyy-MM-dd'T'HH:mm:ss.S"/>
    <关键参数="time_zone" value="UTC"/>
    <关键参数="locale" value="English (United Kingdom)"/>

















Sign InorRegisterto comment.