Microsoft Access find field in table with known value - sql

Question

We have a fairly large Oracle database that we are able to connect via Microsoft Access and ODBC with read-only access. We work with a front-end system that does not match the structure behind the scenes and often I need to query the system via Microsoft Access. The problem is that we are not provided any documentation as to structure and the structure needs serious attention. Searching for the field that I need is very time consuming.
With our front end, I'm able to view the values that I want to query, and I know the key fields, but I need to find the field that contains the known value.
If I have a record where I know the value of field "A", and have the value of field "X", is it possible to query field "X"?
Front end shows
Student ID: 12345678
Payments: 23456
Back end
TechID: 12345678
???: 23456
Can I query "???"

Answer 1

You can do this by iterating over the collection of tables, and for each table, the collection of fields.
Open Database
Get all Tables
For Each Table
Get all Fields
For Each Field
If Field type is text ... and
If Field size is not TOO Long ...
Search for string
If found, write to a results bucket
Next
Next
Here is an example of code for cataloging tables (source here)
Public Function GenerateDataDictionary(aDataDictionaryTable As String)
'*** Usage: GenerateDataDictionary("MyDataDictionaryTable")
'*** Extracts the information about the tables for the data dictionary
'*** and inserts it to a table named aDataDictionaryTable
Dim tdf As TableDef, fldCur As Field, colTdf As TableDefs
Dim rstDatadict As Recordset
Dim i As Integer, j As Integer, k As Integer
Set rstDatadict = CurrentDb.OpenRecordset(aDataDictionaryTable)
Set colTdf = CurrentDb.TableDefs
'Go through the database and get a tablename
For Each tdf In CurrentDb.TableDefs
'Do what you want with the table names here.
rstDatadict.AddNew
rstDatadict.Update
rstDatadict.AddNew
rstDatadict![Table] = tdf.NAME
rstDatadict![Field] = "----------------------------"
rstDatadict![Display] = "----------------------------"
rstDatadict![Type] = ""
rstDatadict.Update
rstDatadict.AddNew
rstDatadict![Table] = "Table Description:"
For j = 0 To tdf.Properties.Count - 1
If tdf.Properties(j).NAME = "Description" Then
rstDatadict![Field] = tdf.Properties(j).Value
End If
Next j
rstDatadict.Update
rstDatadict.AddNew
rstDatadict.Update
For i = 0 To tdf.Fields.Count - 1
Set fldCur = tdf.Fields(i)
rstDatadict.AddNew
rstDatadict![Table] = tdf.NAME
rstDatadict![Field] = fldCur.NAME
rstDatadict![Size] = fldCur.Size
Select Case fldCur.Type
Case 1
FieldDataType = "Yes/No"
Case 4
FieldDataType = "Number"
Case 8
FieldDataType = "Date"
Case 10
FieldDataType = "String"
Case 11
FieldDataType = "OLE Object"
Case 12
FieldDataType = "Memo"
Case Else ' Other values.
FieldDataType = fldCur.Type
End Select
rstDatadict![Type] = FieldDataType
For j = 0 To tdf.Fields(i).Properties.Count - 1
If fldCur.Properties(j).NAME = "Description" Then
rstDatadict![DESCRIPTION] = fldCur.Properties(j).Value
End If
If fldCur.Properties(j).NAME = "Caption" Then
rstDatadict![Display] = fldCur.Properties(j).Value
End If
If fldCur.Properties(j).NAME = "Rowsource" Then
rstDatadict![LookupSQL] = fldCur.Properties(j).Value
End If
Next j
rstDatadict.Update
Next i
Debug.Print " " & tdf.NAME
Next tdf
End Function
You can catalog your findings in Access by making a table of field-names which joins to a table of table-names. Then your searches are based on the catalog instead of raw collections.
I reverse-engineered the schema for MAS 90 (with JobOps add-in) this way. There's no map, but I had a read-only ODBC connection which I used in precisely the way you propose. The purchasing accountant would give me a distinctive Product Number and I'd run it through this comprehensive engine. Over time I succeeded in distilling 700 tables comprising 18k fields down to 20 tables and a few hundred fields. That allowed us to export our data.

Answer 2

The answer to your question is simple. No, you cannot do that.
There are two solutions that I can think of. The first is to manually concatenate all the values together and then look for the row that contains the value. This is imperfect, but might work:
select *
from (select t.*, ('|'""col1||'|'||col2+'|' . . .||'|') as allcols
from t
) t
where instr('|23456|', allcols) > 0
This would find any row that has that value in a column. Probably close enough for what you want.
The second is to use UNPIVOT to do essentially the same thing.
I would strongly suggest that you invest a little bit of time to find the mapping between the fields, and then create a view in Oracle that has the field names as seen in the application. It sounds like this would save you a lot of effort in the medium term.