In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

I want to extract the name and the kB's of swap.

For this I use something like:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}
set statusF [open /proc/1002/status]
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]

Is that a good idea, or could I better implement it in a different
way?

--
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Cecil Westerhof <[email protected]> wrote:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

I want to extract the name and the kB's of swap.

For this I use something like:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}
set statusF [open /proc/1002/status]
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]

Is that a good idea, or could I better implement it in a different
way?

That's one way, but not the only way. Another way is something like
this:

proc procget {file} {
set fd [open $file RDONLY]

while {[gets $fd line] != -1} {
switch -regexp -matchvar mv -- $line {
{^Name:[[:blank:]]+([^\n]+)} { set name [lindex $mv 1] }
{^VmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB} {
set swap [lindex $mv 1]
close $fd
return [list $name $swap]
}
}
}
close $fd
error "Did not find expected data"
}

There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

jtyler <[email protected]> wrote:

On 2/20/2022 4:17 PM, Rich wrote:

There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

If better means easiest, then this comes to mind:

$ rlwrap -pred tclsh
% set line {}
% catch {set line [exec grep -s VmSwap /proc/1351/status]}
0
% set line
VmSwap: 44852 kB
% lindex $line 1

This does rely on the contents of the status file in proc not being
someday modified to include a character that is special to the Tcl
implicit string to list parsing code.

Relying on implicit string to list conversion is a source of weird
errors, years later, at 3:17am, while you are away for a vacation, that
is dependent on the fact that at that exact time, a "special character"
(to the string to list parsing code) appeared in the data being
consumed.

I.e., it is an "input data dependent bug".

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2/20/2022 4:17 PM, Rich wrote:

There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

If better means easiest, then this comes to mind:


$ rlwrap -pred tclsh
% set line {}
% catch {set line [exec grep -s VmSwap /proc/1351/status]}
0
% set line
VmSwap: 44852 kB
% lindex $line 1
44852
% set line {}
% catch {set line [exec grep -s VmSwap /proc/1351noexist/status]}
1
% set line
%

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Monday, February 21, 2022 at 1:17:30 AM UTC+1, Rich wrote:

Cecil Westerhof wrote:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

I want to extract the name and the kB's of swap.

For this I use something like:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}
set statusF [open /proc/1002/status]
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]
[...]

[...]
There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

A few suggestions (beyond RDONLY that Rich suggested):
* there is [fileutil::cat]
* you have not stated which process - /proc/[pid]/status ?
* there is [string cat] instead of multiple [append]s
* depending on future plans, you might want to parse the key-value-pairs to a dictionary ...

# this is my take from the console - maintainable code looks different ;-) package require fileutil
set status [concat {*}[lmap line [split [::fileutil::cat /proc/[pid]/status] \n] {
if {[regexp {^([^:]+):\s+(.*)$} $line -> name value]} {list $name $value} continue
}]]
puts [dict get $status VmSwap] ;# still with the unit attached

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Rich <[email protected]d> writes:

Cecil Westerhof <[email protected]> wrote:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

I want to extract the name and the kB's of swap.

For this I use something like:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n} >> set statusF [open /proc/1002/status]
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]

Is that a good idea, or could I better implement it in a different
way?

That's one way, but not the only way. Another way is something like
this:

proc procget {file} {
set fd [open $file RDONLY]

while {[gets $fd line] != -1} {
switch -regexp -matchvar mv -- $line {
{^Name:[[:blank:]]+([^\n]+)} { set name [lindex $mv 1] }
{^VmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB} {
set swap [lindex $mv 1]
close $fd
return [list $name $swap]
}
}
}
close $fd
error "Did not find expected data"
}

I had to look twice, but that works.

There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

For the moment I will keep with mine version. That is more of my
taste. ;-)


But when the input-file can become large, your way would probably be
better.

--
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am 21.02.22 um 00:48 schrieb Cecil Westerhof:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

Is that a good idea, or could I better implement it in a different
way?

Use the System, Luke!

set swapinfo [ exec grep VmSwap: /proc/$PID/status ]


you work in a known system ( linux : has grep .. )
and it is obvious what the code wants to achieve.
and kiss : one line.

Uwe

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

heinrichmartin <[email protected]> writes:

On Monday, February 21, 2022 at 1:17:30 AM UTC+1, Rich wrote:

Cecil Westerhof wrote:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

I want to extract the name and the kB's of swap.

For this I use something like:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}
set statusF [open /proc/1002/status]
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]
[...]

[...]
There are also multiple other ways.

As to which is /better/, I suspect that is going to be largely a matter
of taste.

A few suggestions (beyond RDONLY that Rich suggested):

Yeah, I already added that. Is the default, but it makes things
clearer.

* there is [fileutil::cat]

I will look into that.

* you have not stated which process - /proc/[pid]/status ?

It was hard-coded. But see end of the post.

* there is [string cat] instead of multiple [append]s

I look into that also.

* depending on future plans, you might want to parse the
key-value-pairs to a dictionary ...

At the moment that looks like overkill.


At the moment I have:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}


foreach file [glob /proc/*/status] {
try {
set statusF [open $file RDONLY]
} on error msg {
# Process does not exist anymore
if {${::errorCode} == {POSIX ENOENT {no such file or directory}}} {
continue
}
puts "errorInfo: ${::errorInfo}"
puts "errorCode: ${::errorCode}"
puts "msg: ${msg}"
exit
}
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]
if {${result} && ($swap != 0)} {
puts "$name: $swap"
}
}

--
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Cecil Westerhof <[email protected]> writes:

* there is [fileutil::cat]

I will look into that.

[snip]

* there is [string cat] instead of multiple [append]s

I look into that also.

[snip]

At the moment I have:
set regex {}
append regex {^Name:[[:blank:]]+([^\n]+)\n}
append regex {.*}
append regex {\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n}

foreach file [glob /proc/*/status] {
try {
set statusF [open $file RDONLY]
} on error msg {
# Process does not exist anymore
if {${::errorCode} == {POSIX ENOENT {no such file or directory}}} {
continue
}
puts "errorInfo: ${::errorInfo}"
puts "errorCode: ${::errorCode}"
puts "msg: ${msg}"
exit
}
set statusStr [read ${statusF}]
close ${statusF}
set result [regexp ${regex} ${statusStr} match name swap]
if {${result} && ($swap != 0)} {
puts "$name: $swap"
}
}

Changed it to:
package require fileutil


set regex [string cat \
{^Name:[[:blank:]]+([^\n]+)\n} \
{.*} \
{\nVmSwap:[[:blank:]]+([[:digit:]]+)[[:blank:]]+kB\n} \
]


foreach file [glob /proc/*/status] {
try {
set statusStr [::fileutil::cat $file]
} on error msg {
# Process does not exist anymore
if {${::errorCode} == {NONE}} {
continue
}
puts "errorInfo: ${::errorInfo}"
puts "errorCode: ${::errorCode}"
puts "msg: ${msg}"
exit
}
set result [regexp ${regex} ${statusStr} match name swap]
if {${result} && ($swap != 0)} {
puts "$name: $swap"
}
}

When a file does not exist ::fileutil::cat generates the error NONE.
Personally I find that a bit strange.

--
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Uwe Klein <[email protected]> writes:

Am 21.02.22 um 00:48 schrieb Cecil Westerhof:

In linux you find in /proc/PID/status:
Name: exim4
.
.
.
VmSwap: 980 kB

Is that a good idea, or could I better implement it in a different
way?

Use the System, Luke!

set swapinfo [ exec grep VmSwap: /proc/$PID/status ]

you work in a known system ( linux : has grep .. )
and it is obvious what the code wants to achieve.
and kiss : one line.

But it does not what I want: retrieve exactly name and swap.
And could also be relatively expensive.

--
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am 21.02.22 um 17:17 schrieb Uwe Klein:

Am 21.02.22 um 12:49 schrieb Cecil Westerhof:

set swapinfo [ exec grep VmSwap: /proc/$PID/status ]

% set swapinfo [ exec egrep VmSwap:\|Name /proc/6239/status ]
Name: lyx
VmSwap: 45868 kB
%

a Billion ways to go from there :-)

Uwe

% time { set swapinfo [ exec egrep VmSwap:\|Name /proc/6239/status ]} 100 4527.38 microseconds per iteration
%
Uwe

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am 21.02.22 um 12:49 schrieb Cecil Westerhof:

set swapinfo [ exec grep VmSwap: /proc/$PID/status ]

% set swapinfo [ exec egrep VmSwap:\|Name /proc/6239/status ]
Name: lyx
VmSwap: 45868 kB
%

a Billion ways to go from there :-)

Uwe

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

* Uwe Klein <[email protected]>
| % time { set swapinfo [ exec egrep VmSwap:\|Name /proc/6239/status ]} 100
| 4527.38 microseconds per iteration

% time { set swapinfo [ exec egrep VmSwap:\|Name /proc/3640/status ]} 100 2132.79 microseconds per iteration

proc swapinfo {} {
set res [list]
set pid [pid]
set fd [open /proc/$pid/status]
while {[gets $fd line] >= 0} {
if {[string range $line 0 4] eq "Name:"
|| [string range $line 0 6] eq "VmSwap:"} {
lappend res $line
if {[llength $res] == 2} {
break
}
}
}
close $fd
return $res
}

% time swapinfo 100
59.26 microseconds per iteration

:-)

R'

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am 21.02.22 um 18:54 schrieb Ralf Fassel:

* Uwe Klein <[email protected]>
| % time { set swapinfo [ exec egrep VmSwap:\|Name /proc/6239/status ]} 100 | 4527.38 microseconds per iteration

% time { set swapinfo [ exec egrep VmSwap:\|Name /proc/3640/status ]} 100 2132.79 microseconds per iteration

different hardware

proc swapinfo {} {
set res [list]
set pid [pid]
set fd [open /proc/$pid/status]
while {[gets $fd line] >= 0} {
if {[string range $line 0 4] eq "Name:"
|| [string range $line 0 6] eq "VmSwap:"} {
lappend res $line
if {[llength $res] == 2} {
break
}
}
}
close $fd
return $res
}

% time swapinfo 100
59.26 microseconds per iteration

touche

:-)

R'

G!
Uwe

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

using an inline glob style search of a list of lines from the status file seems faster than swapinfo on my machine, and to my eye is easy to read and modify.


proc getkv {file {key VmSwap:*}} {

set fd [open $file r]
try {
set lines [split [read $fd] \n]
} finally {close $fd}

set name [lsearch -inline -glob $lines Name:*]
if {"" eq $name} {
error "no Name: line in $file"
}

set val [lsearch -inline -glob $lines $key]
if {"" eq $val} {
error "No $key line found in $file"
}

list [getval $name] [getval $val]
}


this trims any leading or trailing white space from whatever is after the first ":"

proc getval {str} {
string trim [string range $str [string first ":" $str]+1 end]
}

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)