What are the actual end of line characters in that file?
Read the file as fixed length records (so the end of line character are treated as just ordinary characters) and look at it.
The LIST statement in a SAS data step can help with this.
So this data step will display the first 500 bytes of the file. If any of the characters are CR ('0D'x) or LF ('0A'x) then the LIST statement will show the the hex codes for ALL of the characters in the 100 byte fixed length record.
data _null_;
infile 'myfile.txt' lrecl=100 recfm=f obs=5 ;
input;
list;
run;
The first question you need to answer is. What characters appear the end of the real records. If it the unix standard LF character? The Windows/DOS standard CR and LF pair? Is the deprecated standard used by the original Macintosh of just a CR?
The second quest is do the extra line breaks in the middle of a record use the exact same characters?
If you are lucky the real lines end with CRLF and the extra breaks you are seeing are caused by just CR or just LF. IF that is true then SAS will read the file properly if you tell it to use CRLF as the end of line characters.
data want;
infile 'myfile.txt' dsd dlm=';' truncover termstr=crlf;
input (var1-var8) (:$30.);
run;
Otherwise you are in trouble because the values that have the embedded end of line characters are not quoted. The standard definition of delimited files says that embedded delimiters or end of line characters need to have the values quoted. SAS does not directly support files with embedded end of line characters, but when the value is quoted it is simple to just count the quotes to know whether or not the end of line is occurring inside a quoted string. Such as with this macro %replace_crlf()
So if your file falls into neither of those patterns then you can come close to fixing the problem by counting delimiters, but that will not work right when the embedded end of line characters appear in the last field on the line. So you might need to do some review or manual touch up to get the file converted into something that can be read.
So let's try it on your example.
First let's convert the text you posted back into a file using PARMCARDS.
parmcards4;
PPP;00001;Aa001zz;123;A71 tetstst;a4jgtt;gtrall;grrrr
PTP;0002;App01;412; Azer
( test agerebdb);;;
PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff
PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff
;;;;
Now let's look at the file:
124 options generic;
125 data _null_;
126 infile example termstr=crlf;
127 input;
128 list;
129 run;
NOTE: The infile EXAMPLE is:
(system-specific pathname),
(system-specific file attributes)
RULE: ----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8----+----9----+----0
1 PPP;00001;Aa001zz;123;A71 tetstst;a4jgtt;gtrall;grrrr 53
2 PTP;0002;App01;412; Azer 24
3 ( test agerebdb);;; 20
4 PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff 51
5 PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff 51
NOTE: 5 records were read from the infile (system-specific pathname).
The minimum record length was 20.
The maximum record length was 53.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds
Let's read it in an keep track of the number of semicolons. When we hit 7 (or more!) then let's write a CRLF pair. Otherwise let's just write a bare LF. I have used $200 character variable. So the maximum line it can handle is 200 bytes. But a data step can handle character strings of up to 32,767 bytes. If the lines are longer than that you will need to use a different method.
filename fixed temp;
data _null_;
infile example truncover length=ll termstr=crlf end=eof;
file fixed recfm=n;
input line $char200. ;
length=ll;
count+countc(line,';');
if _n_>1 then do;
if count>7 and _n_>1 then do;
put '0D0A'x ;
count=countc(line,';');
end;
else put '0A'x ;
end;
put line $varying200. length;
if eof then put '0D0A'x;
run;
Now let's read in the resulting file the same way to look at it.
166 data _null_;
167 infile fixed termstr=crlf;
168 input;
169 list;
170 run;
NOTE: The infile FIXED is:
(system-specific pathname),
(system-specific file attributes)
RULE: ----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8----+----9----+----0
1 PPP;00001;Aa001zz;123;A71 tetstst;a4jgtt;gtrall;grrrr 53
2 CHAR PTP;0002;App01;412; Azer. ( test agerebdb);;; 45
ZONE 555333333477333333324767022276772666766662333
NUMR 040B0002B10001B412B01A52A08045340175252429BBB
3 PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff 51
4 PTP;erfff;Zerrrr;847;derrr;55zer;dfererereer;erffff 51
NOTE: 4 records were read from the infile (system-specific pathname).
The minimum record length was 45.
The maximum record length was 53.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds
Now if you don't want that linefeed in the middle of the fifth field on the second line you could change the ELSE statement that is writing it to write something else, or nothing at all.
Click image to register for webinar
