Arrays and Strings

[mbutler755]

I have been summoned with the task of taking a tab delimited file and converting that file so it can be read by a third-party program. I am trying to think of the best way to do this, but I am having a hard time. Here is a view of the raw data that must be altered:

fkstoreid	HeaderMark	B1	RegNo	dateofbusiness	NoRegEntries	Tips	OverRings	Voids$	Voids#	NetSales	DP1_2Guests	DP3Guests	DP4Guests	DetailMark	fkdaypartid	RecType	GL	B2	Sales
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	2	0	41005	NULL	2220.25
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	3	0	41005	NULL	9170.19
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	4	0	41005	NULL	670.75
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	2	0	42100	NULL	52.03
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	3	0	42100	NULL	398.34
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	4	0	42100	NULL	137.94
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	2	0	43000	NULL	78.92
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	3	0	43000	NULL	311.93
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	4	0	43000	NULL	163.74
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	2	0	44000	NULL	10.98
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	3	0	44000	NULL	44.23
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	4	0	44000	NULL	12
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	2	0	41950	NULL	247.33
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	3	0	41950	NULL	867.89
1061	H	NULL	1	01/01/2008	NULL	NULL	NULL	113.26	35	12785.47	144	531	37	D	4	0	41950	NULL	38.54
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	2	0	41005	NULL	1998.09
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	3	0	41005	NULL	5950
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	4	0	41005	NULL	449.91
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	2	0	42100	NULL	65.58
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	3	0	42100	NULL	402.47
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	4	0	42100	NULL	94.51
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	2	0	43000	NULL	51.08
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	3	0	43000	NULL	376.61
1061	H	NULL	1	01/02/2008	NULL	NULL	NULL	129.78	42	9168.35	143	361	23	D	4	0	43000	NULL	65.79

So, each date will have a header (denoted by "H"). With this table every record has an "H." This is not the way it needs to be. I need to compare dates and when they differ, create a new header and make the rest of the data (if any) a detail (denoted by "D").

I am toiling with the best way of doing this. So far, I am not having much luck. I have used arrays quite successfully in Java, but I am not sure that this is the best way to go here. I would have to get how many lines there are and read each line. Make the changes, and then go on to the next line. I think scanning the document line by line using a loop is the best way. However, I am not sure.

I am in need of some opinions. What do you think I should?

[stanav]

Create a datatable consisting of all the columns in that file then use either of these methods:
1. Use a TextFieldParser to read the file and fill in the datatable.
2. Use a streamreader and read each line, split it (by the tab character) into an array, then add a new datarow to your datatable with that data.

[Shaggy Hiker]

The datatable was what came first to my mind, as well, but I noted the bit about doing a comparison and changing things. An alternative would be to create a class that has a member for each of the fields in the table. You could then add a constructor to the class that took a string, split the string on Tab, then filled the fields appropriately. You could then have a List (of T) of this class type to hold all the items. The advantage to this approach appears to be just the organization, and you could do the same thing with a datatable, with the manipulation coming between splitting the string and adding the fields to the new datarow. To some extent it matters what you want to do with the data afterwards, which we don't know.

[Dnereb]

As I understand your question...
You can have diffrent headers based on a date and details linked to a date.
sounds like you should need two objects:
a header object.
a detail object.

depending on your architecture you can opt to use two tables in a dataset to read the data in, with a relation between these(1 header table as parent and a details table as child)

Or an businessObject Header (a class) containing a list of details (another class) that has an attribute <serializable>

in both cases yoou can serialize them to an XML file.
a third party just needs the XML scheme to write code against. but an example file will do just as well.

[Shaggy Hiker]

What does the third party app want? I read your question as meaning that the third-party app wants data in a rigidly defined format that is already set in stone, while Dnereb read it as the third-party app has yet to be written. Both seem reasonable, but which one is right really affects the answer.