Of course, if the wrong input mapping is used, a wrong answer is gotten.
Apply
m=:<:+/2 1*(<a.)[EMAIL PROTECTED]&>' ';TAB,LF
instead of
m0=:<:+/2 1*(<a.)[EMAIL PROTECTED]&>' ';'+',LF NB. '+'was used for
test
reasons
then
((1;S;m)&;:-:fw2dl) line
1
With Line=:50000 199 $line I get
ts'(1;S;m);:"1 Line' NB. Boss
0.21547872 4196864
ts'trim"1 Line' NB. Miller
0.20581284 4196288
ts'fw2dl"1 Line' NB. Bron
2.038465 4200128
ts'(1;S;m);: ,Line'
0.17531539 37750208
ts'trim ,Line'
0.17133324 37749568
ts'fw2dl ,Line'
0.44602124 1.0486099e8
The differences in performance are less than you obtain.
As I stated earlier, Miller's solution is leaner than and equivalent with
mine.
It was a pleasure to help you.
R.E. Boss
-----Oorspronkelijk bericht-----
Van: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Namens Dan Bron
Verzonden: dinsdag 6 februari 2007 22:22
Aan: Programming forum
Onderwerp: RE: [Jprogramming] Efficiently converting fixed-width to
delimited
Raul and R.E. Boss,
I just ran some metrics of our algorithms on my real dataset:
fn fmt TS
Name Time Space
-------- ------ ------
canon _ _
bron00 23.547 5.500
boss00 1.341 1.000
miller00 1.000 1.000
see postscript for the definitions of these verbs. Raul's results match
mine, and his timings are so superior that I threw out my code.
R.E. Boss' verb doesn't match the canonical results even in nominal cases:
ws =: 2 : '(m&i.)`(,~ n $"_ 0~ $) } :. ( (n&i.)`(,~ m $"_ 0~
$) })'
NB. 'X' means data, '.' means space,
NB. '=' means TAB, '/' means LF
w =: '.=/' ws (' ',TAB,LF)
NB. Here's the first line from one of my files.
line =: w LF -.~ noun define
XXXXXXXX=.XXXXXXX...=.XXX.XXXXX.=.....XXXX=............=.XXX
XX=....XXXX=............XXX=...XXXXX..=........=...........X
XX=.XXX=..X=.....XXXXXX=...........=..........=........=....
....=.......XXXXXX/
)
_60 ]\ w^:_1 bron00 line
XXXXXXXX=XXXXXXX=XXX.XXXXX=XXXX==XXXXX=XXXX=XXX=XXXXX==XXX=X
XX=X=XXXXXX=====XXXXXX/
_60 ]\ w^:_1 miller00 line
XXXXXXXX=XXXXXXX=XXX.XXXXX=XXXX==XXXXX=XXXX=XXX=XXXXX==XXX=X
XX=X=XXXXXX=====XXXXXX/
_60 ]\ w^:_1 boss00 line
XXXXXXXX=.XXXXXXX...=.XXX.XXXXX.=.....XXXX=............=.XXX
XX=....XXXX=............XXX=...XXXXX..=........=...........X
XX=.XXX=..X=.....XXXXXX=...........=..........=........=....
....=.......XXXXXX/
Thanks to both of you,
-Dan
PS: Verb definitions:
NB. All verbs may assume y ends with an LF
canon =: [: ; [: (dltb@:}: , {:)&.> e.&(10 9{a.) <;.2 ]
bron00 =: verb define
TAB bron00 y
:
s =. y ~: ' '
d =. y e. LF,x
l =. d ([: ; <@(+./\ );.2) s
d =. 1 (0)} d
t =. d ([: ; <@(+./\.);.1) s
e =. l*.t
y =. e#y
)
bossS=. _2]\"1 }.".;._2 (0 : 0)
''' '' x;LF Z ']0
1 1 3 1 4 1
1 0 3 1 4 1
2 0 3 1 4 2
1 2 3 2 4 2
2 2 3 2 4 2
)
boss00 =: (1;bossS;<:+/2 1*(<a.)[EMAIL PROTECTED]&>' ';'+',LF )&;: "1
NB. other SPACE DELIM
millerS=.0 10#:10*".;._2]0 :0
2.1 0.0 1.1
2.2 0.3 1.2
2.0 3.2 1.2
2.2 3.0 1.1
)
miller00=: (1;millerS;(a.=' ')+2*a.e.LF,TAB)&;:
NB. w =: fread jpath '~temp\fw2dl.txt'
(verb define) bron00`boss00`miller00 NB. canon`
TS =: 0 $~ 2,~#y
for_v. y do.
smoutput 'timing ',;v
TS =: (6!:2 '0 0 $',(;v),' W') (<v_index;0) } TS
6!:3]0.5
smoutput 'spacing ',;v
TS =: (7!:2 '0 0 $',(;v),' W') (<v_index;1) } TS
6!:3]0.5
end.
)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
