I have following tuple H1 and I want to strsplit its $0 into tuple.However I always get an error message:
DUMP H1:
(item32;item31;,1)
m = FOREACH H1 GENERATE STRSPLIT($0, ";", 50);
ERROR 1000: Error during parsing. Lexical error at line 1, column 40.
Encountered: after : "\";"
Anyone knows what's wrong with the script?
There is an escaping problem in the pig parsing routines when it encounters this semicolon.
You can use a unicode escape sequence for a semicolon: \u003B
. However this must also be slash escaped and put in a single quoted string. Alternatively, you can rewrite the command over multiple lines, as per Neil's answer. In all cases, this must be a single quoted string.
H1 = LOAD 'h1.txt' as (splitme:chararray, name);
A1 = FOREACH H1 GENERATE STRSPLIT(splitme,'\\u003B'); -- OK
B1 = FOREACH H1 GENERATE STRSPLIT(splitme,';'); -- ERROR
C1 = FOREACH H1 GENERATE STRSPLIT(splitme,':'); -- OK
D1 = FOREACH H1 { -- OK
splitup = STRSPLIT( splitme, ';' );
GENERATE splitup;
}
A2 = FOREACH H1 GENERATE STRSPLIT(splitme,"\\u003B"); -- ERROR
B2 = FOREACH H1 GENERATE STRSPLIT(splitme,";"); -- ERROR
C2 = FOREACH H1 GENERATE STRSPLIT(splitme,":"); -- ERROR
D2 = FOREACH H1 { -- ERROR
splitup = STRSPLIT( splitme, ";" );
GENERATE splitup;
}
Dump H1;
(item32;item31;,1)
Dump A1;
((item32,item31))
Dump C1;
((item32;item31;))
Dump D1;
((item32,item31))
STRSPLIT on a semi-colon is tricky. I got it to work by putting it inside of a block.
raw = LOAD 'cname.txt' as (name,cname_string:chararray);
xx = FOREACH raw {
cname_split = STRSPLIT(cname_string,';');
GENERATE cname_split;
}
Funny enough, this is how I originally implemented my STRSPLIT() command. Only after trying to get it to split on a semicolon did I run into the same issue.