Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.java.programmer > #33971 > unrolled thread

Stream question

Back to article view | Back to comp.lang.java.programmer

Contents

  Stream question Rhino <no_offline_contact@example.com> - 2016-03-15 14:11 -0400
    Re: Stream question Sebastian <news@seyweiler.dyndns.org> - 2016-03-15 21:21 +0100
      Re: Stream question Rhino <no_offline_contact@example.com> - 2016-03-16 00:44 -0400

#33971 — Stream question

I'm new to streams but very intrigued with the possibilities.

I'm trying to write a stream that does grouping, counting and summing at 
the same time. The data involved is actually quite simple but writing 
the streaming statement I need is proving challenging and I'm not seeing 
anything really helpful in Google searches.

First, let me describe my data, then I'll show you how I've solved 
two-thirds of the problem. Perhaps you can tell me how to fit in the 
missing piece.

The data is ticket sales from a company that sells concert tickets. Each 
sale consists of an agency code, an order number, order date and the 
number of tickets sold. Therefore, it looks like this:

AgencyCode  OrderNumber OrderDate  TicketsSold
----------  ----------- ---------  -----------
TW          111111  2016-03-01          4
TW          111112  2016-03-01          2
CP          201000  2016-03-01          3
TW          111113  2016-03-01          8
CP          201001  2016-03-02          2
EL          300001  2016-03-01          4
AS          400000  2016-03-02          2

What I'm trying to get out of this data is a summary showing the total 
number of orders for each agency code and the total number of tickets 
sold for that same agency code. Therefore, the values I want to get for 
this particular set of data is:

AgencyCode  Orders   TicketsSold
TW               3            14
CP               2             5
EL               1             4
AS               1             2

I've got the grouping working and also the number of tickets sold. It's 
just the counting of the orders that I'm trying to get.

Here's how I got the tickets sold by agency:
	     Map<String, Integer> salesByAgency
	         = ticketOrders.stream()
	             .collect(Collectors.groupingBy(TicketSale::getAgencyCode,
	                     Collectors.summingInt(TicketSale::getTicketsSold)));

TicketSale is the class that holds a single ticket order. My collection, 
ticketOrders, is a LinkedHashSet holding a bunch of TicketSale records.

How do I adjust what I have to get the number of orders for each agency 
code?


-- 
Rhino

[toc] | [next] | [standalone]

#33975

Am 15.03.2016 19:11, schrieb Rhino:
[example snipped]
>
> How do I adjust what I have to get the number of orders for each agency
> code?
>

Two possibilities come to my mind.

(A) This answers your question of how to adjust what you have.

On Feb 22nd I posted in this group in the thread "stream filter twice", 
see 
https://groups.google.com/d/msg/comp.lang.java.programmer/zSqNcCNDCRI/mmeBA3w0DQAJ

In that post I discuss the possibility of using a so-called "combining 
collector" from the library jool (https://github.com/jOOQ/jOOL). In your 
case, such a collector would apply two downstream collectors 
simultaneously and return a pair of results:

// untested
Map<String, Tuple2<Integer, Long>> salesByAgency
            = ticketOrders.stream()
                .collect(groupingBy(TicketSale::getAgencyCode,
                         Tuple.collectors(
                            summingInt(TicketSale::getTicketsSold),
                            counting())));

If you don't want the dependency on the library, it is easy to write the 
collector yourself, see the post cited.

Notes:
1. Questions like this involving processing data from the same stream 
source multiple times in different ways come up very often. Jool 
provides a very neat way to deal with this need. Excellent library.

2. Besides summing, you can also do any other reduction using 
Collectors.reducing() as one of the downstream collectors.


(B) This does not adjust what you have, but is a different (and much 
simpler) way to do it.

Map<String, IntSummaryStatistics> salesByAgency
     = ticketOrders.stream()
         .collect(groupingBy(TicketSale::getAgencyCode,
                 Collectors.summarizingInt(TicketSale::getTicketsSold)));

IntSummaryStatistics has methods getCount() and getSum().

-- Sebastian

[toc] | [prev] | [next] | [standalone]

#33982

On 3/15/2016 4:21 PM, Sebastian wrote:
> Am 15.03.2016 19:11, schrieb Rhino:
> [example snipped]
>>
>> How do I adjust what I have to get the number of orders for each agency
>> code?
>>
>
> Two possibilities come to my mind.
>
> (A) This answers your question of how to adjust what you have.
>
> On Feb 22nd I posted in this group in the thread "stream filter twice",
> see
> https://groups.google.com/d/msg/comp.lang.java.programmer/zSqNcCNDCRI/mmeBA3w0DQAJ
>
>
> In that post I discuss the possibility of using a so-called "combining
> collector" from the library jool (https://github.com/jOOQ/jOOL). In your
> case, such a collector would apply two downstream collectors
> simultaneously and return a pair of results:
>
> // untested
> Map<String, Tuple2<Integer, Long>> salesByAgency
>             = ticketOrders.stream()
>                 .collect(groupingBy(TicketSale::getAgencyCode,
>                          Tuple.collectors(
>                             summingInt(TicketSale::getTicketsSold),
>                             counting())));
>
> If you don't want the dependency on the library, it is easy to write the
> collector yourself, see the post cited.
>
> Notes:
> 1. Questions like this involving processing data from the same stream
> source multiple times in different ways come up very often. Jool
> provides a very neat way to deal with this need. Excellent library.
>
> 2. Besides summing, you can also do any other reduction using
> Collectors.reducing() as one of the downstream collectors.
>
I have to admit I'm not keen on using an extra library. It just seems 
like something else to learn. (And maintain. And install everywhere my 
program runs.) I'm sure it's as great a library as you say it is but 
your second solution is the one I prefer, at least so far. As I learn 
more, I may develop a preference for Jool ;-)
>
> (B) This does not adjust what you have, but is a different (and much
> simpler) way to do it.
>
> Map<String, IntSummaryStatistics> salesByAgency
>      = ticketOrders.stream()
>          .collect(groupingBy(TicketSale::getAgencyCode,
>                  Collectors.summarizingInt(TicketSale::getTicketsSold)));
>
> IntSummaryStatistics has methods getCount() and getSum().

I really like this! It's nice and simple to look at and when I ran it in 
debug, it worked very nicely and produced the result I wanted. The only 
negative is that it returns two statistics that I don't need and I'm 
probably paying something for that but the volumes of data I am 
processing are very low so I don't see any extra expense as significant. 
(It might be if I were processing millions of records but I'm only going 
to have dozens most of the time.)

Thank you VERY much for this solution!

I've got a lot to learn but this will help me get off to a good start as 
I try to understand streams and stream handling.

-- 
Rhino

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.java.programmer

csiph-web