This is an answer to a question about Talend that was posted on StackOverflow. I wasn’t able to post as many pictures as I needed due to house rules in place, so I have moved the whole thing here and linked back to it. The general gist of the question was that there was a mahoosive table that would have eaten too much memory if its entire contents were dragged into Talend, so the user wanted to be able to take some values from a different table, string them out into a list, pop them into a context variable and then squirt them back into another SQl query so that he ended up with a smaller set of results to work with. With me so far? OK, read on!
This should be possible. I’m not working in MySQL but I have something roughly equivalent here that I think you should be able to adapt to your needs.
As you can see, I’ve got some data coming out of the table and getting filtered by tFilterRow_1 to only show the rows I’m interested in.
The next step is to limit it to just the field I want to use in the variable. I’ve used tMap_3 rather than a tFilterColumns because the field I’m using is a string and I wanted to be able to concatenate single quotes around it but if you’re using an integer you might not need to do that. And of course if you have a lot of repetition you might also want to get a tUniqueRows in there as well to save a lot of unnecessary repetition
The next step is the one that does the magic. I’ve got a list like this:
etc, and I want to turn it into ‘A1′,’A2′,’B1′,’B2’ so I can slot it into my where clause. For this, I’ve used tAggregateRow_1, selecting “list” as the aggregate function to use.
Next up, we want to take this list and put it into a context variable (I’ve already created the context variable in the metadata – you know how to do that, right? If not, here’s a quick rundown). Use another tMap component, feeding into a tContextLoad widget. tContextLoad always has two columns in its schema, so map the output of the tAggregateRows to the “value” column and enter the name of the variable in the “key”. In this example, my context variable is called MyList
Now your list is loaded as a text string and stored in the context variable ready for retrieval. So open up a new input and embed the variable in the sql code like this
“SELECT distinct MY_COLUMN
from MY_SECOND_TABLE where the_selected_row in (“+
It should be as easy as that, and when I whipped it up it worked first time, but let me know if you have any trouble and I’ll see what I can do.
I had a tricky problem a while ago and nobody seemed to know how to do it so when I worked it out, I thought it might be fun to post a how-to here for other people to crib from and take the credit. Wait, is this such a great idea? Oh well, never mind, here goes…
The challenge is to take a group of scanned pages from a document management system and prepare them for migration into Servelec Corelogic’s Frameworki/Mosaic product. The documents are scanned on a page-by-page basis as TIFFs, and the objective is to merge the pages into a single file, either as TIFFs or as PDFs in a new folder, with the paths held in a database table. In this example, I’ve used nConvert, which is largely free, although if you use it commercially you should buy a license. There’s another free program that I believe can do the same job, although I haven’t specifically tried it – namely Irfanview.
The general strategy is:
- List the where they’re stored in the file system or EDRMS
- Use t-sql or pl/sql to write a command line function to grioup all the individual files (pages) together and merge them into a single file in the file system
- Pass the location of the new file to the import process.
Starting in Talend Open Studio, the first step is to create as new job using the tFileList component as the starting point, to get a list of files in the folder you’re interested in.
Use an iterator to connect to the next step- a tFileProperties component, which you can use to get the file properties of each file in turn. Check the image below for the format to use. You can use this to store the details of all the files in a table called – in this example – FILE_SILESYSTEM.
To move to the next stage, I’ve used a T-SQL function to create a shell-command that does two things: first, create a new folder for the files to live in, and second to invoke a third party app called nConvert to merge the pages into a single file. In the command below, you can see the “md” command being used to create the folder. nConvert- a third party app – can then be called to either merge the files or to merge and conver them to pdfs.
cmd /c cd C:/test/smart_files/ &
md ID &
cd ID &
md 64398 &
nconvert -multi -out tiff -c 5 -o C:/test/smart_files/ID/64398/164994_v1.tif U:/00707000/00706853.tif U:/00707000/00706854.tif U:/00707000/00706855.tif U:/00707000/00706856.tif U:/00707000/00706857.tif U:/00707000/00706858.tif U:/00707000/00706859.tif U:/00707000/00706860.tif U:/00707000/00706861.tif U:/00707000/00706862.tif U:/00707000/00706863.tif U:/00707000/00706864.tif U:/00707000/00706865.tif U:/00707000/00706866.tif U:/00707000/00706867.tif U:/00707000/00706868.tif U:/00707000/00706869.tif U:/00707000/00706870.tif U:/00707000/00706871.tif U:/00707000/00706872.tif U:/00707000/00706873.tif U:/00707000/00706874.tif >>C:/test/output.txt
In the example above, I’m just merging them but it’s simple to merge them as a pdf by just chainging the format to
The content of the table can then be split in two; first, the bult of the table can be passed to the import process. The last column – containing the output of the T-SQL function is stored in the final column of a table and the output passed to a shell command using a tMap component:
into an iterator….
The iterator then passes the output of the function into a shell command and merges the files into a single file in the specified folder.
You now have a list of merged files in a format the import process can understand and a folder containing the merged files, all stored in the place in which the import process expects to find them. It should be straightforward to simply run the load procedure and scoop up the merged file into Mosaic.
It’s that time of year again, and project managers up and down the country are wondering what to put in their team members’ stockings. Well, have no fear, here’s my must-have gift-giving guide for the Data Guru who has everything.
1. A Better Computer
Is your data migration lead’s brow furrowed? Does he spend hours staring at his screen clenching and unclenching his fists as the record count ticks from 100 to 200 on a 10,000,000 record load? This might be a sign that the refurbished Pentium III laptop or the virtual box accessed through a dumb terminal that you thought would be so much more cost-effective than a new Dell wasn’t such a good choice after all.
As data flies in and out of it, headed for the target database, both of the test machine’s kilobytes fill up immediately and it starts furiously swapping to keep up. The lights dim, the smell of burning fills the air, development computer fails to respond to mouse-clicks, the screen fades to grey. This is when that lovely, christmassy scarlet colour can be seen in the cheeks of your colleague.
Why not log in to the purchasing portal and order a better computer? What it costs you will be more than made up in fees as work gets done more quickly and doesn’t spill over into extra days and evening work.
2. Talend Data Integration Suite
OK, so Open Studio is the best £0.00 you’ve ever spent, but there’s a whole other level of greatness!
3. The Force Awakens Tie-in Poster
The power of the Force (AKA The Disney Corp) has reached into the world of data migration, producing a system even more powerful than PDMv2, and now you can buy inspirational posters based on the movie script to help motivate your data migration lead to fight the power of the dark side.
4. Another Spreadsheet
This one is a perennial favourite, and ultimately what most data migration professionals are given every year. We’ve all seen this heart-warming yuletide scene: Late December, a few scant weeks before go-live, and the project team are pulling on their coats, ready to go down the pub for their Christmas do. As if suddenly remembering something, one of the BAs turns and says
“Oh by the way, I’ve just emailed you a spreadsheet the business have told me about. It has mission-critical data on it and they absolutely can’t go live without it. Merry Christmas!”
…and with that they are all gone, leaving the vision of a slowly turning egg-timer reflected in the tears of – one assumes – pure joy, streaming down the data migrator’s face.
Happy Christmas…. And remember, we’re making a list, we’re checking it twice….
Last days of the migration project mantra:
We recently introduced our ten-year-old daughter to the Indiana Jones movies. Even the fourth one, but let’s not talk about that. At the very end of the original Raiders of the Lost Ark, there’s a scene in which the Ark of the Covenant is boxed up and placed in a warehouse surrounded by what looks like tens of thousands of identical boxes. The modern equivalent of its desert resting place is not an underground tomb, guarded by snakes and poison darts, but a total immersion in endless, bureaucratic sameness. Now, I don’t know if you’ve ever tried to implement a “Spreadsheet Amnesty” but if you have, you’ll know it is exactly like that. Ex. Act. Ly.
For the uninitiated, a spreadsheet amnesty is an essential part of any data migration project. Essentially, the problem you have is that the old system you’re setting out to replace is total rubbish. After all, if it wasn’t, you wouldn’t be there. the staff are clever people with a job to do, and they can’t be hampered by this bad system, so they’ve created all manner of spreadsheets, access databases, and who-knows-what to record all their day-to-day working data in. If you ignore it, you’ll be starting out with an incomplete system, missing key data. So, you have to ask everyone to identify their data sources so you can scoop them all in and use them to plug gaps in the main database. Sometimes, you might come across political impediments; in all likelihood, internal business teams have been waging a war of attrition against these spreadsheets for years, and I’ve heard of cases where, when they get onto the project board there’s been a desire to exert some influence to “punish” the offenders by ruling these contraband data sources as invalid and out of scope. If you go down that route, you are condemning the project to repeat the mistakes of the past. Hence the name “amnesty”. No blame. Everyone is welcome, and so is their data.
But politics aside, when you go out and trawl through the spreadsheets, it can be like searching for a needle in a large stack of identical needles. Or, if you prefer, a sacred relic in a warehouse full of fake sacred relics. The trouble is, there are often hundreds of spreadsheets which often seem to be capturing slightly different view of the same data, and a lot of the time, it’s hard to pick out the ones that have unique data being used to drive a specific business process, not just a mish-mash of items drawn from other places. I thought it would be helpful to list out a few questions that are worth asking when deciding which box to pry open in your search for the ark.
So you visit the team room, and ask them to show you the spreadsheets they’re using day-to-day in addition to recording in the case recording system, which for the sake of argument, I’ll refer to as “Disappointech”. As each one is produced you ask:
Is this a report from Disappointech?
Is the answer is yes, it’s not a data migration source. Whatever is in it must be in the core system already. As an aside though, if this is used as part of business-as-usual, you need to make sure the business analysts who are configuring the software are aware of it, and that the new system will produce an equivalent report.
Is there anything here that’s not in Disappointech?
If the answer is no, the team might just have been compiling or copy/pasting the information into a spreadsheet for ease of use. Again, there’s nothing here that you need to migrate but it’s worth thinking how the new system can save them having to do such a tedious, time-consuming chore. Maybe a report could help?
Is it information you need to have in the new system?
OK, so we’ve established that this is unique information that can’t be found in Disappointech and only exists in the spreadsheet. But does it belong in the new system? If it’s about a social worker’s annual leave, say, or contributions to the coffee fund, or the staff Eurovision sweepstake then the answer is probably no and you can move on. More likely, it’s something borderline: information that’s pretty close to the type of data you need to bring in but not quite in scope, and you might have to get a project decision on which side of the fence it lies on so you don’t get into “scope creep”.
Is the data about a specific person?
If the data is aggregated – for example, if it shows – such-and-such a percentage of visits done on time in each month, say, but you can’t identify specific dates for a specific service-user in there then this is not likely to be usable. Most social care systems have to associate each item with a specific client record and it isn’t possible to go from a statistic back to the source data that went into producing the statistic.
What’s the quality like?
Having established that there’s something here that the project needs to have, you’ll need to assess whether it’s tidy enough to import. Is each service user’s Disappointech ID present and correct? Is each item on its own row? Do columns with dates in all have real dates or do some of them have things like “To Be Confirmed” or the dreaded “N/K”? Do the team store all the data in this one spreadsheet or do they have multiple copies, say one for each financial year?
If the quality is low, you’ve got a few choices. you could improvise and write some fairly complicated code that tries to work around the problems. That’s a high-risk strategy because it’s quite likely to need a lot of maintenance as new rows get added to the sheet, introducing new problems. A second option is to ask them to transfer the data to a new template which you can set up with lots of validation so that it has to be filled in in a certain way. Lastly, it’s always worth considering whether all this pain is necessary and whether – if the list is fairly short – they could manually transfer the data by keying it in on day 1, perhaps with a bit of help from someone who knows the new software well. It’s surprising how often this is the best solution for everyone.
OK, I think that covers Excel spreadsheet. I was going to write about how to cope with poorly designed Access Databases as well, but instead, here is a visual metaphor to describe what that feels like:
Earlier this month on BBC Radio 4, File on 4 broadcast an investigation into the state of provision for Care Leavers in the UK, ahead of the National Audit Office’s report on the same subject, which is due out later in the summer. I’ll link to the audio at the bottom of the page in case you’re interested.
The whole thing was deeply saddening, just because of the subject matter: young people – children really – who find themselves, through no fault of their own, in a situation where they have had to be taken away from home and placed in public care. The stories are many and varied, and of course children’s experiences differ wildly. Many will end up with utterly lovely carers, many will return home in a few days or weeks, but whatever the situation, it’s hard to imagine ever being able to come through that kind of separation completely emotionally unscathed, especially when it lasts for years and involves a series of placements in homes of varying quality. The programme focused on people at the upper end of that age group, in mid-late teens, when they seem more independent, able to cope even with being temporarily housed in a B&B instead of a carer’s home, when perhaps people around them are starting to forget that they are needy and hurt and to see them as troublemakers or delinquent in some way. There was a real sense of damage done and lives dislocated in the stories the interviewees told of life in the fringes of the care system.
For those of us who work in and around the social care sector, we know that social work teams are incredibly hard-working and committed to their role. It’s not the sort of job you go into for the money and the glory (just ask the tabloids…) so if that’s a person’s chosen career you can usually assume there’s an underlying core of dedication there. The programme raised some interesting points though, both for social workers themselves about case management, and for policy makers about how resources are allocated to these services, because, after all, you can only stretch a team and a budget so far, no matter how well-managed it might be and how dedicated the individual staff members. I can’t pretend to have any answers on this of course.
What did strike me, as a data person, was the figures that came up in the discussion, comparing numbers of care leavers in suitable accommodation across various local authorities, The numbers vary wildly. Now, I’ve been around a bit and I’ve seen how some of this data is collected across over maybe a dozen local authorities in my time, to one degree or another, and I didn’t place a lot of significance in the numbers. Here are a few general impressions, picked up from the whirl of past experience:
Cases like this are usually managed by a specialist team, maybe serving all looked after children, or maybe even more specialised on care leavers from the age of fourteen or so, (when preparation for adulthood begins in earnest) up to about twenty-five in some cases, depending on the situation. Often the teams are keeping their eye on the ball just fine, down at the local level in Looked After Reviews or Pathway Plan Reviews, where a young person’s situation is discussed. However, the government can’t really assimilate that level of detail. All it wants at year end for its SSDA903 Return are a few codes relating to whether or not the authority is in touch with the young person, whether they are properly housed, in employment and so on. Getting those codes is where the problem lies. A really green performance manager (and I must say, I’ve never met one) might just assume whatever they get out of the central case recording system is the gospel. It isn’t though, because in their creaky old legacy systems, either there’s nowhere in that system to record the codes needed (that’s often why I’m theer: to help replace it with something better) or else they aren’t well understood. In rare cases where they are fully recorded, they might be out of date or else other recording or interpretation errors have crept in as the social worker tries to relate the building they have just placed a child in and the list of options in the dropdown menu before them. More hard-nosed performance managers, with a detached, world-weary look, will chase people to update the system and trust that these entreaties are acted on, and of course there will be others who just give up and send round questionnaires to be filled in offline and sent back. So, in short, when I hear that Council X has 100% suitable accommodation and Council Y has only 35%, I don’t necessarily assume that Council Y is worse than Council Y, because there’s a complicated network of factors in play, and the difference might come down, largely, to the efficiency of the data-gathering machine. It tends to be that some areas of reporting are more scrutinised than others, and it might be that with more emphasis on Care Leavers recently in the press, both Council X and Council Y will dig into the data more deeply and apply thumbscrews to a couple of deputy managers to get more clarity on the numbers, as certainly happens when other figures (timeliness of Child Protection Reviews, say) dip below a hundred percent.
What lessons can I draw from this? Well, as someone involved in preparation for new systems, I guess there are two things. Firstly, with this kind of data, never assume it’s held in the system; it’s a prime example of items to be addressed during the data migration project’s “Spreadsheet Amnesty”. It needs to be gathered in and assimilated into the body of the migration. Secondly, as far as possible, new systems should be designed in such a way that social workers don’t have to spend all their time “feeding the beast”. The recording system should be easy to complete, with no hidden backwaters, and all the relevant statistical data should be extracted from normal case work instead of needing extra recording. Why should I care about the poor social worker doing a little extra work? Well, because if the team is overstretched already by government cuts, ever hour spent at a terminal is an hour less spent with the care leaver, helping them sort out a bed to sleep in, a college place, or whatever it might be. Whichever way you slice it, that has to be a good thing, both for the care leaver, for the taxpayer and for the social worker. Social workers have feelings too, you know!
Simple case recording also enables more consistency. Whether or not you agree with the idea of government returns and their role in a centralised command-and-control system, it doesn’t help anyone if certain councils are named and shamed on File on 4, not because they are serving their customers poorly, but because stats are being collected haphazardly. Now, of course, I haven’t studied the councils named in the report. I’ve worked at a couple, but not recently enough to be able to say anything specific, and I’m certainly not here to take sides for or against them. All I know is my data-geek sense was tingling when I heard that particular part of the documentary, and that’s what made me sit down and plan this blog post.
Audio: File on 4 – Abandoned to Their Fate
(Can’t see the audio player? Try this link)
Next month the National Audit Office is due to report on the outcomes for young people leaving care. There are claims that, under financial pressure, local authorities are pushing too many teenagers into independent living before they’re ready. File on 4 investigates new figures that suggest many young care leavers are failing to cope – with large numbers ending up in custody, homeless, sexually exploited or pregnant. Social services chiefs say the welfare of care-leavers must be a key priority for the new government. But who holds them to account when they fail those they are meant to have looked after? And, with more cuts on the way, can the system cope? Fran Abrams reveals how hands-off caring can have tragic consequences.