Entries in special character (2)

How do I parse a ctrl-A delimited string?

Short answer: "\001" vs data

Use the vs (vector from scalar) function:

q)fstring
"f1=va\001f2=vb\001f3=vc"
q)"\001" vs fstring
"f1=va"
"f2=vb"
"f3=vc"
q)

We can combine vs with /: (each right) to break up each component:

q)"=" vs/: "\001" vs fstring
"f1" "va"
"f2" "vb"
"f3" "vc"
q)

Note that you cannot put any space between vs and /:. If you did, that would be a comment.

We might next turn fstring's contents into a table using a combination of flip and ! (dict):

q)flip "=" vs/: "\001" vs fstring
"f1" "f2" "f3"
"va" "vb" "vc"
q)`field`value ! flip "=" vs/: "\001" vs fstring
field| "f1" "f2" "f3"
value| "va" "vb" "vc"
q)flip `field`value ! flip "=" vs/: "\001" vs fstring
field value
-----------
"f1"  "va" 
"f2"  "vb" 
"f3"  "vc"
q)

This is almost what we want. Usually, though, you want the field names to be symbols rather than strings. We can do that by applying (@) the $ (cast) operator to (only) the values of the field column:

q)columns: @[flip "=" vs/: "\001" vs fstring; 0; `$]
q)columns
f1   f2   f3  
"va" "vb" "vc"
q)flip `field`value ! columns
field value
-----------
f1    "va" 
f2    "vb" 
f3    "vc" 
q)

However, there is a shortcut we can take to deconstruct the string into a table using one of the many variants of 0::

q)flip `field`value ! "S=\001" 0: fstring
field value
-----------
f1 "va"
f2 "vb"
f3 "vc"
q)

Not only is this last example more succinct, it's much faster:

q)\t do[100000;
        flip `field`value !
                 @[flip "=" vs/: "\001" vs fstring; 0; `$]]
485
q)\t do[100000; flip `field`value ! "S=\001" 0: fstring]
110
q)

See also How do I build a ctrl-A delimited string?

How do I build a ctrl-A delimited string?

Short answer: "\001" sv ("one"; "two"; "three")

To build delimited strings in general, use the sv (scalar from vector) function:

q)", " sv ("Hello"; "world!")
"Hello, world!"
q)

As in many programming languages, the syntax for non-printable ASCII characters is borrowed from C, although only a subset of C's sequences is supported:

        \n      newline
        \r      carriage return
        \t      tab

(q also supports a few printable escape sequences: \\, \" and \'). However, as in C, you can express any ASCII code in q by following a backslash with the code's 3 digit octal representation:

q)"\054\040" sv ("Hello"; "world!")
"Hello, world!"
q)

Since ctrl-A is ASCII 1, we need to use \001:

q)fstring: "\001" sv ("f1=va"; "f2=vb"; "f3=vc")
q)fstring
"f1=va\001f2=vb\001f3=vc"

We can use 0: to write fstring to a file, and use an external program to confirm the result:

q)`:test 0: enlist fstring
`:test
q)\\
$ hexdump test
0000000 66 31 3d 76 61 01 66 32 3d 76 62 01 66 33 3d 76
0000010 63 0a                                          
0000012

To break fstring up into its parts, see How do I parse a ctrl-A delimited string?