SAS raw data read--line pointers with inconsistent lines and embedded headers - file-io

Question

The raw data I want to read not only runs across several different lines but also doesn't have the same number of lines per record. To further complicate things it also has headers that appear in the middle of the file after every page that ruin everything.
Time: 1:47pm Item Master Report For 06/06/2013 Report: GMRIMMSB
Item Type: Nonstock
Item Asset Inven Dsp ---Order--- ---Primary---- Substute Contract Hazd Count
Stat Class Class Unt Unit Conv Loc Vendor Manufacturer Nbr Item Nbr Number Flag Cycle
------------------------------------------------------------------------------------------------------------------------------------
ITEM 20049 TEST PNEUMONIA S LATEX ZL22 (30859001)
A 0173 6 PK PK 1 NSL 2431 R30859001
Vendor 1: 2431 FISHER SCIENTIFIC COMPANY 2: 2658 REMEL
3: 536 ABBOTT LABS - DIAGNOSTIC DIVISION 4: 1404 MUREX DIAGNOSTICS INC.
ITEM 20051 ANTIGEN BACTER. WELLCOGEN ZL26 B1901-51
A 0173 6 PK PK 1 NSL 2431 30859602
Vendor 1: 2431 FISHER SCIENTIFIC COMPANY 2: 3804 CARDINAL HEALTH-ALLEGIANCE
3: 2658 REMEL 4: 536 ABBOTT LABS - DIAGNOSTIC DIVISION
5: 1404 MUREX DIAGNOSTICS INC.
ITEM 20053 FILM DUPLICATING 10X12
I 0173 14 BX BX 1 NSX 1335 112010
Vendor 1: 1335 AGFA CORPORATION
ITEM 20055 FILM HTU 10 X 12
I 0173 14 BX BX 1 NSX 1335 094010
Vendor 1: 1335 AGFA CORPORATION
ITEM 20056 FILM HTU 8 X 10
I 0173 14 BX BX 1 NSX 1335 094008
Vendor 1: 1335 AGFA CORPORATION
ITEM 20057 SOL AXSYM FLUIDIES CHECK (09A3401)
A 0173 119 BX BX 1 NSL 536
Vendor 1: 536 ABBOTT LABS - DIAGNOSTIC DIVISION
ITEM 20058 FILM DUPLICATING 8 X 10
I 0173 14 BX BX 1 NSX 1335 112008
Vendor 1: 1335 AGFA CORPORATION
ITEM 20059 FILM HTU 14 X 17
I 0173 14 BX BX 1 NSX 1335 094014
Vendor 1: 1335 AGFA CORPORATION
Item Asset Inven Dsp ---Order--- ---Primary---- Substute Contract Hazd Count
Stat Class Class Unt Unit Conv Loc Vendor Manufacturer Nbr Item Nbr Number Flag Cycle
------------------------------------------------------------------------------------------------------------------------------------
ITEM 20060 FILM HTU 30 X 35
I 0173 14 BX BX 1 NSX 1335 094030
Vendor 1: 1335 AGFA CORPORATION
ITEM 20061 FILM HTU 14 X 14
I 0173 14 BX BX 1 NSX 1335 094001
Vendor 1: 1335 AGFA CORPORATION
Here is the code I have working with (in SAS studio)....
libname mylib '/folders/myfolders/';
data myfile;
length itm $ 4 itemnum 5 itemdesc $ 40 inac $ 2 assetcl $ 4 invcl 3 dspunit $ 2
ordunit $ 2 convr 4 loc $ 4 vndnum 4 manufnum $ 20 vendinfo $ 80;
infile '/folders/myfolders/ItemstrSM.txt' missover;
input #1 itm $ itemnum itemdesc $ &
#2 inac $ assetcl $ invcl dspunit $ ordunit $ convr loc $ vndnum manufnum
#3 vendinfo & $ ;
run;
proc print data=myfile noobs;
run;

Answer 1

If you do not need to routinely handle lots of this file, you can simply use a text editor to solve the problem.
Replace the page head to blank.
Replace '\nITEM' to '###ITEM'
Replace '\n' to space
Replace '###' to '\n'
Now you have one entry per line in the text file.

Answer 2

One workaround you can use to deal with the repeated header rows is to use something along the lines of the following before your main input statement:
input #;
if _infile_ = "Header row text" then delete;
This loads the whole line into _infile_ without populating any variables and holds it for processing by your main input statement (provided that it isn't a header row).

Answer 3

Assuming the same formating you've shown us there continues, and that there are no more than 6 vendors for each item:
data test(drop=itm);
length itm $4 itemnum 8 itemdesc $132
inac $1 assetcl invcl 8 dspunit ordunit $2 convr 8 loc $3 vndnum 8 manufnum $132;
infile "c:\users\c41928\documents\egreadin.txt" missover lrecl=32767 truncover;
input itm $ #;
if itm="ITEM" then
do;
input itemnum itemdesc $ /
inac $ assetcl $ invcl dspunit $ ordunit $ convr loc $ vndnum manufnum /
vendinfo1 $132.;
if index(vendinfo1,"2:") then
input itm $ #;
if itm="3:" then
do;
input #1 vendinfo2 $132.;
input itm $ #;
if itm = "5:" then
do;
input #1 vendinfo3 $132.;
output;
end;
else output;
end;
else output;
end;
else delete;
run;
If there is a unknown, and possibly infinite, number of vendors, then an itterative do loop would be more suitable. Otherwise, if there is a known number of vendors higher than 6, more do loops could be added.
Anyway, it's not very pretty but it works.