home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!gatech!usenet.ins.cwru.edu!news.ysu.edu!psuvm!auvm!UMSLVMA.BITNET!NASSER
- Message-ID: <SAS-L%93010517160891@UGA.CC.UGA.EDU>
- Newsgroups: bit.listserv.sas-l
- Date: Tue, 5 Jan 1993 16:13:21 CST
- Reply-To: David Nasser <NASSER@UMSLVMA.BITNET>
- Sender: "SAS(r) Discussion" <SAS-L@UGA.BITNET>
- From: David Nasser <NASSER@UMSLVMA.BITNET>
- Subject: Reading Tricky Flat File
- Lines: 93
-
- Content : Question(s) / Inquiry
- Summary : How to read VeryStrange gummint file?
- System(s) : Large IBM
- Name : David Nasser
-
- The file in question is:
- "Thrift Financial Report Quarterly, June 1992"
- supplied on tape from:
- US Office of Thrift Supervision, Wash., DC
- National Technical Info Service
- US Dept. of Commerce
-
- The doc. for this file is very, *very* sparse: there is no record layout
- as such (that we could find). The data in this file represents
- financial info (i.e. assets, net income, etc.) on Savings and Loan
- Associations (SNLs) reported to one or more federal agencies.
-
- The record length is 128 bytes. All data is character. There are >=180
- records per SNL. The first 79 b. of selected records for the first SNL
- follow:
-
- G0B192060034904SALEM CO-OP BK 3 SOUTH BROADWAY
- G0C1920600349030790000ROCKINGHAM 07416033015
- G0D192060034933010054110803N NNNNNPAAAAAAAAA 00 PPPPPPPPPPPPPXPPPNPP920731
- G0I1920600349SC10 +0000073064SC110 +0000000781SC120 +0000006202SC132 +0000
- G0I1920600349SC150 +0000000000SC162 +0000027884SC166 +0000000000SC170 +0000
- G0I1920600349SC190 +0000000466SC198 +0000000000SC20 +0000000000SC210 +0000
- G0I1920600349SC220 +0000000000SC223 +0000000000SC226 +0000000000SC23 +0000
- G0I1920600349SC24 +0000001251SC240 +0000000600SC250 +0000062805SC253 +0000
- (172 similar "G0I" records omitted here) +0000
- 20S1920600349CSS020348053S.C.B., INC. 3 SOUTH BROADWAY
- 20S2920600349CSS020348053SALEM NH000003079
- 20S3920600349CSS020348053020187470SALEM CO-OPERATIVE BANK 10001A24
- 20S4920600349CSS020348053120+0000000102130+0000000000140+0000000102150+00000000
-
- The first 3 recs (starting with "GOB", "G0C", "G0D" resp.) contain
- data for identifiers (i.e. name, address, etc.) and are not a problem.
- The following 177 "G0I" recs have:
- ID info in the first 13 bytes
- Six repetitions of:
- A 7 b. field with a variable name
- A 1 b. field with a sign associated with:
- A 10 b. field with a data value
- 7 bytes of apparent garbage at the end of the record
-
- They Have Embedded The Variable NAMES In (nearly) Every Record (some-
- thing I have never _ever_ seen before)!
-
- For instance, var name SC10 (Cash less valuation allowance) can be read
- from bytes 14-20 of the first G0I rec. Similarly, the positive sign and
- the actual data value (+0000073064) can be read from bytes 21-31.
- SC10=73064. Piece of cake, eh? eh? There are 6(177)=1062 of these
- beauties.
-
- After the 177 G0I recs are a variable number of 20S recs. These must
- be flushed because there are no data definitions (in the non-existent
- record layout).
-
- It bears mentioning that we also use NTIS bank files which have
- conventional record layout with 1 rec/bank, integer binary data rep.,
- etc. The G0I recs for 1 SNL occupy 128(177)=22,656 bytes. If the SNL
- file were designed like the bank files (they were, previous to 1990),
- the same data would occupy 6(177)(4)=4248 bytes, a savings of around
- 80%. Say, did someone (i.e. Codd, Date) say something re conservation
- of storage being the *essence* of good database design?? Naw, they
- were probably just talking about &normalization& (hiccup!).
-
- Two Questions:
- 1.) Does anyone happen to have any particulars re this file? How was
- it *intended* to be read? Is there a relatively new language
- (i.e. COBOL II or ?) which is engineered to read both variable
- names _and_ values from the same file? Inquiring minds ....
- 2.) Does it make sense to try to read / support this data with SAS,
- given that we dont necessarily have 40 daze/40 nites to hand-code
- everything (which will probably change with the next qrtrly file)?
-
- Howard, are you out there???
-
- TIA. All suggestions welcome.
-
- Unrelated:
- Has anyone been watching SI dev./testing/support of DEC Alpha AXP?
- I will take this (very brief) moment to tip my hat in the direction
- of Goodnight/SI.
-
- Prosit,
- David
-
- "I'm sittin' here wonderin',
- would a matchbox hold my clothes?
- I ain't got so many matches,
- but I got so far to go."
- from "Matchbox Blues", Blind Lemon Jefferson, maybe 1927.
-