The program below can generate random data according to some specs (example here is for 2 columns)
It works with a few hundred of thousand lines on my PC (should depend on RAM). I need to scale to dozen of millions row.
How can I optimize the program to write directly to disk ? Subsidiarily how can I "cache" the parsing rule execution as it is always the same pattern repeated 50 Millions times ?
Note: to use the program below, just type generate-blocks and save-blocks output will be db.txt
Rebol[]
specs: [
[3 digits 4 digits 4 letters]
[2 letters 2 digits]
]
;====================================================================================================================
digits: charset "0123456789"
letters: charset "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
separator: charset ";"
block-letters: [A B C D E F G H I J K L M N O P Q R S T U V W X Y Z]
blocks: copy []
generate-row: func[][
Foreach spec specs [
rule: [
any [
[
set times integer! [['digits (
repeat n times [
block: rejoin [block random 9]
]
)
|
'letters (repeat n times [
block: rejoin [ block to-string pick block-letters random 24]
]
)
]
|
[
'letters (repeat n times [block: rejoin [ block to-string pick block-letters random 24]
]
)
|
'digits (repeat n times [block: rejoin [block random 9]]
)
]
]
|
{"} any separator {"}
]
]
to end
]
block: copy ""
parse spec rule
append blocks block
]
]
generate-blocks: func[m][
repeat num m [
generate-row
]
]
quote: func[string][
rejoin [{"} string {"}]
]
save-blocks: func[file][
if exists? to-rebol-file file [
answer: ask rejoin ["delete " file "? (Y/N): "]
if (answer = "Y") [
delete %db.txt
]
]
foreach [field1 field2] blocks [
write/lines/append %db.txt rejoin [quote field1 ";" quote field2]
]
]