Replacing spaces with %20

[Sc0rp]

Ive written a little tcl script that gets the content of a php page Ive set..

everything works fine untill the 'thing' (username in this case) I want to retreive has spaces in it..
cus then the bot somehow replaces the space with a character which the webserver does not understand..
so.. what I want to do is replace a space with %20 which is a space for internet browsers when used in the address.. (duh )
problem is im kinda new at tcl scripting and I dont know how to replace stuff..

so my question is HOW can I replace a space with %20

[strikelight]

Ive written a little tcl script that gets the content of a php page Ive set..

everything works fine untill the 'thing' (username in this case) I want to retreive has spaces in it..
cus then the bot somehow replaces the space with a character which the webserver does not understand..
so.. what I want to do is replace a space with %20 which is a space for internet browsers when used in the address.. (duh )
problem is im kinda new at tcl scripting and I dont know how to replace stuff..

so my question is HOW can I replace a space with %20

Code: Select all

regsub -all " " $text "%20" text

or

Code: Select all

set text [string map {" " "%20"} $text]

I prefer the first method myself..

also, a better place to start before asking for commands, is the tcl manpages.

[user]

Code: Select all

set text [join [split $a " "] %20]

would also work (almost the same as string map cpu-wise and compatible with all tcl versions)

The nice thing about join/split and string map is that they return the result, so there's no need to do the command invocation and then read the value. The nice thing about regsub is that you'll know how many replacements were made and you'll keep your cpu less idle.

btw: I bet you could use + instead of %20 and save some space. (heh)

[strikelight]

Code: Select all

set text [join [split $a " "] %20]

would also work (almost the same as string map cpu-wise and compatible with all tcl versions)

The nice thing about join/split and string map is that they return the result, so there's no need to do the command invocation and then read the value. The nice thing about regsub is that you'll know how many replacements were made and you'll keep your cpu less idle.

btw: I bet you could use + instead of %20 and save some space. (heh)

Code: Select all

% time {set newtext [join [split $text " "] %20]} 100000
33 microseconds per iteration
% time {set newtext [string map {" " %20} $text]} 100000
14 microseconds per iteration
% time {regsub -all " " $text "%20" newtext} 100000
15 microseconds per iteration
% time {set newtext [join [split $text " "] %20]} 100000
30 microseconds per iteration

Are we ever going to stop debating this?

[user]

Interesting numbers...what tcl version? My tests in 8.3.2 were quite different (string map and split/join about the same, regsub 4-5 times slower)

Are we ever going to stop debating this?

I hope not

Try to guess what my first edit was.

[strikelight]

Interesting numbers...what tcl version? My tests in 8.3.2 were quite different (string map and split/join about the same, regsub 4-5 times slower)

Code: Select all

% info patchlevel
8.4.1
% puts "$tcl_platform(os) $tcl_platform(osVersion)"
Windows NT 5.0
% 

[ppslim]

Both "string map" and the split-join method can also count many many replacments where done.

Code: Select all

set len [string length $orig]
set ph "%20"
set new [join [split $orig] $ph]
set count [expr ([string length $new] - $len) / [string length $ph]]

Not as simple or elegant, more CPU intensive.

As for the battle of time, this will vary on text length and system. The battle between Windows and Linux has allways been odd.

Windows has shown on occasions to beat the pants of linux in some commands, but lagged in others. Linux shows consistant replies where windows hasn't.

Note, there was one test missed out.

Code: Select all

set text [string repeat "ssss " 30]
time {set newtext [join [split $text " "] %20]} 1000
time {set newtext [string map {" " %20} $text]} 1000
time {regsub -all " " $text "%20" newtext} 1000
time {set newtext [join [split $text] %20]} 1000

The results in order (cycles reduced)

Code: Select all

92 microseconds per iteration
108 microseconds per iteration
405 microseconds per iteration (ouch)
110 microseconds per iteration

note how using the default marker for "split" rather than " " causes time to increase.

Here are the tests again, with the "string repeat" set to 10

Code: Select all

46 microseconds per iteration
47 microseconds per iteration
151 microseconds per iteration
53 microseconds per iteration

Tcl 8.3.4 - Linux 2.4.21

System level optimisations are not important to a coder, generic speed is. Unless they are gonna use a test suite first to determine the best command for the job.

[Sir_Fz]

may I ask, how do you do these tests ? on what basis do you get the time results ?

[ppslim]

Tcl provides a command called "time".

This will measure the time in microseconds, it takes to complete a command.

You also give it a repeat counter. This will loop and call the command X times. The total time is added together, and devided by the loop value, to give an average time.

Tcl also provides a command line parser, called "tclsh". You can use this to run Tcl script in your everyday tasks, or even as a CGI library. When called on its own, you can use it interactivly.

Through this, you can perform these tests, though you can't test eggdrop scripts, as many of the required commands are not available.

[Sir_Fz]

can you give an example please ? using one of the commands above

[ppslim]

Examples have allready been given.

Code: Select all

time {set test "hello"} 1050

This will run the code

Code: Select all

set test "hello"

1050 times, and tell you how long on average it took to run the command each time.

It will display somthing like

2 microseconds per iteration

[Sir_Fz]

I tried ".tcl time {set test "hello"} 1050" in partyline, it returned: Tcl error: wrong # args: should be "time". what's wrong ?

[egghead]

I tried ".tcl time {set test "hello"} 1050" in partyline, it returned: Tcl error: wrong # args: should be "time". what's wrong ?

Good point. It seems eggdrop has its own definition of time. Could not find it in doc/tcl-commands.doc though.

In any case, you can start tcl on your shell:

system:/usr/home/egghead # tclsh8.3
% time {set test "hello"} 1050
1 microseconds per iteration

[Sir_Fz]

okay thanx egghead I've got it, but may I ask how does the time vary according to how many time a command is repeated.

because when I do "time {set test "hello"} 5" it returns 4 microseconds per iteration ; and when I do "time {set test "hello"} 1050" it returns 1 microseconds per iteration.

so why is it less time to repeat 1050 times than 5 times ?

[user]

Good point. It seems eggdrop has its own definition of time. Could not find it in doc/tcl-commands.doc though.

It's not been part of eggdrop for a while. But there's a proc called "time" in scripts/compat.tcl.