Compound conditionals in case when statements? Syntax?

[Randy Kramer]

I need (or want ;-) to do something like the following:

when ((/^---\+\+ (.*)/) and (new_record == true))

I've tried a lot of variations, but either this just won't work, or I haven't
managed to guess the proper syntax. (I've also tried googling and searching
in the pickaxe(2).)

Aside: Maybe the compound won't work because one is what the pickaxe calls a
condition and the other is a comparison? But, I'm not sure which is which.

If this can't work, I'll try (I've been trying) a nested if statement, but
I'll ask about that in my next post. ;-)

Randy Kramer

 

[Michael Fellinger]

I need (or want ;-) to do something like the following:

when ((/^---\+\+ (.*)/) and (new_record == true))

I've tried a lot of variations, but either this just won't work, or I haven't
managed to guess the proper syntax. (I've also tried googling and searching
in the pickaxe(2).)

Aside: Maybe the compound won't work because one is what the pickaxe calls a
condition and the other is a comparison? But, I'm not sure which is which.

If this can't work, I'll try (I've been trying) a nested if statement, but
I'll ask about that in my next post. ;-)

this can't work for the simple reason that the objects given at 'when'
are being sent the === message with the object in 'case'. In your
'when' it's the result of the stuff in closures.
This generally is the job of 'if', just keep in mind that case/when is
for matching, if/else is for conditionals.
One roundabout way exists to make this work though:

class Proc
def ===(obj)
self[obj]
end
end

case 'foo'
when lambda{|e| new_record and /^---\+\+ (.*)/ === e }
puts :aye_match
else
puts :sorry_you_lose
end

This is generally only advised if your middle name is some form of
(Advent|Dang)erous, I won't be able to take responsibility for all
resulting damage. But also note that new_record being true or false
makes it simple to work with it without first comparing true/false
with true, the result won't change, true cannot become any more true.

a = true
# true
puts "hey" if a == true
hey
# nil
puts "hey" if a
hey

Hope this helps a little bit. Have fun with ruby!

^ manveru

 

[MenTaLguY]

I need (or want ;-) to do something like the following:

when ((/^---\+\+ (.*)/) and (new_record == true))

When you see a case clause like this:

case obj
when foo
# a
when bar
# b
else
# c
end

Picture it like this:

if foo === obj
# a
elsif bar === obj
# b
else
# c
end

(the order of arguments to === is important)

So, when you write:

when ((/^---\+\+ (.*)/) and (new_record == true))

It's really like writing:

if ((/^---\+\+ (.*)/) and (new_record == true)) === obj

... which isn't quite what you want.

Using a nested if statement might be the simplest solution.

Incidentally, it isn't usually necessary to write:

new_record == true

If new_record is true, the result will be true anyway, and
if it is false, the result will be false anyway. You can
just use:

new_record

...instead. Similarly, instead of either:

new_record != true
new_record == false

...you can simply write one of:

!new_record
not new_record

(! and not mean the same thing)

-mental

 

[Peña, Botp]

From: Randy Kramer [mailto:[email protected]]=20
# Subject: Compound conditionals in case when statements? Syntax?
#=20
# I need (or want ;-) to do something like the following:
# when ((/^---\+\+ (.*)/) and (new_record =3D=3D true))

cases come in two forms,

1 classic switch form, like
=20
case sw
when foo
#..
when bar
#..
else
#..
end

this form is interpreted using if is like (as already explained by =
mentalguy),

if foo=3D=3D=3Dsw
#..
elsif bar=3D=3D=3Dsw
#..
else
#..
end


2 conditional form (similar to nested ifs), like

case
when foo
#...
when bar and condition1
#...
when condition2 or condition3
#..
else
#..
end

this form is interpreted using if like,

if foo
#..
elsif bar and condition1
#..
elsif condition2 or condition3
#..
else
#..
end

I would reckon fr your post that you may want the second form. So =
something like,

case
when ((/^---\+\+ (.*)/) and new_record
#..
end

But i'm not totally sure since you didn't post any sample code. but =
hey, if your code does not fit the cases, then by all means, do the =
ifs..

hth.

kind regards -botp

 

[John Joyce]

I need (or want ;-) to do something like the following:

when ((/^---\+\+ (.*)/) and (new_record == true))

I've tried a lot of variations, but either this just won't work,
or I haven't
managed to guess the proper syntax. (I've also tried googling and
searching
in the pickaxe(2).)

Aside: Maybe the compound won't work because one is what the
pickaxe calls a
condition and the other is a comparison? But, I'm not sure which
is which.

If this can't work, I'll try (I've been trying) a nested if
statement, but
I'll ask about that in my next post. ;-)

this can't work for the simple reason that the objects given at 'when'
are being sent the === message with the object in 'case'. In your
'when' it's the result of the stuff in closures.
This generally is the job of 'if', just keep in mind that case/when is
for matching, if/else is for conditionals.
One roundabout way exists to make this work though:

class Proc
def ===(obj)
self[obj]
end
end

case 'foo'
when lambda{|e| new_record and /^---\+\+ (.*)/ === e }
puts :aye_match
else
puts :sorry_you_lose
end

This is generally only advised if your middle name is some form of
(Advent|Dang)erous, I won't be able to take responsibility for all
resulting damage. But also note that new_record being true or false
makes it simple to work with it without first comparing true/false
with true, the result won't change, true cannot become any more true.

a = true
# true
puts "hey" if a == true
hey
# nil
puts "hey" if a
hey

Hope this helps a little bit. Have fun with ruby!

^ manveru

Adventerous? That's a strange middle name...

 

[Robert Klemme]

this can't work for the simple reason that the objects given at 'when'
are being sent the === message with the object in 'case'. In your
'when' it's the result of the stuff in closures.
This generally is the job of 'if', just keep in mind that case/when is
for matching, if/else is for conditionals.

See Pena's comment: there is another form of 'case' which can be nicely
used to solve this.

Kind regards

robert

 

[Robert Klemme]

From: Randy Kramer [mailto:[email protected]]
# Subject: Compound conditionals in case when statements? Syntax?
#
# I need (or want ;-) to do something like the following:
# when ((/^---\+\+ (.*)/) and (new_record == true))

cases come in two forms,

1 classic switch form, like

case sw
when foo
#..
when bar
#..
else
#..
end

this form is interpreted using if is like (as already explained by mentalguy),

if foo===sw
#..
elsif bar===sw
#..
else
#..
end


2 conditional form (similar to nested ifs), like

case
when foo
#...
when bar and condition1
#...
when condition2 or condition3
#..
else
#..
end

this form is interpreted using if like,

if foo
#..
elsif bar and condition1
#..
elsif condition2 or condition3
#..
else
#..
end

I would reckon fr your post that you may want the second form. So something like,

case
when ((/^---\+\+ (.*)/) and new_record
#..
end

Maybe rather:

case
when new_record && /^---\+\+ / =~ s

Note the matching operator, otherwise $_ will be matched. Note also,
that it is superfluous to match the rest of the string unless the
content need to be further processed. I'd probably also reverse the
order since I assume new_record is a local variable and that test is
likely faster than the regexp matching.

Kind regards

robert

 

[Peña, Botp]

From: Robert Klemme [mailto:[email protected]]=20
# case
# when new_record && /^---\+\+ / =3D~ s

oops, yes. That's what i get on pasting without thinkig (nor testing
=20
# Note the matching operator, otherwise $_ will be matched.
# Note also, that it is superfluous to match the rest of the
# string unless the content need to be further processed. =20
# I'd probably also reverse the order since I assume=20
# new_record is a local variable and that test is=20
# likely faster than the regexp matching.

indeed and many thanks.
kind regards -botp

 

[kim.toms]

I use:
case true
when a == b && c == d
when a != b && c == d
...
end

 

[Randy Kramer]

--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: Multipart/Mixed;
boundary="Boundary-00=_jWkFH5Zrh6K2Hdc"

--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

From: Robert Klemme [mailto:[email protected]]=20
# case
# when new_record && /^---\+\+ / =3D~ s

Thanks to all who replied!

I've tried the suggested approaches and variations that I could think of, a=
nd=20
still no luck.

I've attached the following small files in case anyone wants to try to see =
if=20
they can get it to work:
* askconvert: the code--at the point in question you will find two=20
commented out lines and a comment that will be fairly self explanatory (I=20
think)
* test.txt: a test file
* test.aml.txt: a file containing the desired result
* utest: a bash script for doing a sort of poor man's unit test--this ma=
y=20
not work for you unless you make some modifications (like to the ./askconve=
rt=20
line)

The unit test works by running the program and then doing a diff between=20
text.aml (the output from the program) and test.aml.txt. Note that the=20
current date and time are incorporated in two of the header lines, the From=
=20
and the Date: line, so those will always show up as different. So, I=20
visually inspect the diff for these two criteria:

* the only lines in the diff related to From and Date lines with differe=
nt=20
dates
* all titles in the diff should show up as variations of "Primary"

The other thing that will show up is the minor spacing issue mentioned belo=
w.

Someday, I might write code (in that unit test file) to, one way or another=
,=20
ignore those dates so they don't show up in the diff. (I might process=20
test.aml and test.aml.std to replace the real date/times with dummys that a=
re=20
all the same.)

Note that there is one minor linespace issue remaining in the program that =
I=20
just haven't been able to resolve so far. It is so minor that I'm going to=
=20
ignore it for now, but, you can spot it if you watch carefully when you run=
=20
the diff.

Randy Kramer

oops, yes. That's what i get on pasting without thinkig (nor testing
=20
# Note the matching operator, otherwise $_ will be matched.
# Note also, that it is superfluous to match the rest of the
# string unless the content need to be further processed. =20
# I'd probably also reverse the order since I assume=20
# new_record is a local variable and that test is=20
# likely faster than the regexp matching.
=20
indeed and many thanks.
kind regards -botp
=20
=20

--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: application/x-ruby;
name="askconvert"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="askconvert"

#! /usr/bin/env ruby

# This is iter(ation)-4 (in progress) of ruby program askconvert to convert (initially) askRhk03 formatted files to askRhk04 format. (Later versions may perform additional conversions.) This version does work only for the hard coded conversion of file test.txt to test.aml. Iter-5 will presumably address getting the filenames from the command line. Iter-4 is sort of a break--code cleanup and maybe some comments--much as it is against my religion, I'm sort of trying to see to what extent this is (or can be made) "self-documenting"--I do have plenty of documentation (some pretty rough) in other files like askconvert.dev.aml.

File.open("test.aml", "w") do |f|
first_record = true
new_record = false
titles = Array.new
File.open("test.txt").each do |line|
case line
when /^\x80\x81\x82\x83$/
new_record = true
# when new_record && /^---\+\+ / =~ line
# titles.push $1.strip
# the previous two lines could replace the next five lines if the previous two lines worked (or I could change the case statement to an if, elsif, elsif ... (or whatever)--case just seemed more "elegant" somehow--I'll wait and see if anybody can show me how to get those two lines to work.
when /^---\+\+ (.*)/
if new_record
titles.push $1.strip
else
f.puts line.rstrip + "\n" end
else
if new_record then
# I could take the next 14 lines and "move" them to a module / method named something like "process_header"--I'd probably also learn something by doing so. I think I'm "learned out" for the moment. ;-)
unless first_record then f.puts "morF\n\n"
else first_record = false end
if titles.length == 3 then
title = titles.pop
titles.reverse!
else
title = titles.shift end
f.puts "From \"" + title + "\" " + %x{date -u +"%a %b %d %T %Z %Y"} + "Date: " + %x{date +"%D %H:%M %P %Z"} + "Subject: " + title + "\n"
unless titles.empty?
f.puts "\n"
titles.each { |title| if title.strip.length > 0 then f.puts "T: " + title + "\n" end}
f.puts "\n" end
titles.clear
new_record= false
else
f.puts line.rstrip + "\n" end
end
end
f.puts "morF\n"
end

--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: application/vnd.sun.xml.draw.template;
name="test.aml.std"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
filename="test.aml.std"

=46rom "Primary" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

=46rom "Primary" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary

T: Secondary

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

=46rom "Primary1" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary1

T: Secondary1
T: Tertiary1

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

=46rom "Primary1" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary1

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

=46rom "Primary1" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary1

T: Secondary1

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

=46rom "Primary2" Wed Oct 17 17:02:59 UTC 2007
Date: 10/17/07 13:02 pm EDT
Subject: Primary2

T: Secondary2
T: Tertiary2
T: Fourtiary2 ;-)

=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

morF

--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: text/plain;
charset="iso-8859-1";
name="test.txt"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
filename="test.txt"

=80=81=82=83
=2D--++ Primary
=2D--++=20
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

=80=81=82=83
=2D--++ Primary
=2D--++ Secondary
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

=80=81=82=83
=2D--++ Tertiary1
=2D--++ Secondary1
=2D--++ Primary1
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

=80=81=82=83
=2D--++=20
=2D--++=20
=2D--++ Primary1
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

=80=81=82=83
=2D--++=20
=2D--++ Secondary1
=2D--++ Primary1
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.

=80=81=82=83
=2D--++ Primary2
=2D--++ Secondary2
=2D--++ Tertiary2
=2D--++ Fourtiary2 ;-)
=20
=46ile test.txt, for testing.

A few lines, to start.

Goodbye.

=2D--++ Another heading

The heading above this should not disappear.


--Boundary-00=_jWkFH5Zrh6K2Hdc
Content-Type: text/plain;
charset="iso-8859-1";
name="utest"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="utest"

#! /usr/bin/env bash


# the cd is helpful for me--others will have to adjust it appropriately (or delete/comment it out)--also consider what to do with ./askconvert

#cd /rhk/code/w/askRhk/askconvert

# to make sure diff is looking at a new version of test.aml
rm -f test.aml

/askconvert
# which currently runs the conversion on test.txt producing test.aml--later it will need the filenames on the command line

diff test.aml test.aml.std | less

--Boundary-00=_jWkFH5Zrh6K2Hdc--
--Boundary-00=_jWkFH5Zrh6K2Hdc--

 

[Randy Kramer]

--Boundary-00=_SqkFHVRI2sVWoto
Content-Type: Multipart/Mixed;
boundary="Boundary-00=_SqkFHVRI2sVWoto"

--Boundary-00=_SqkFHVRI2sVWoto
Content-Type: text/plain;
charset="iso-8859-6"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Oops, I had a dumb mistake in one of those two commented out lines in
askconvert. Fixing them doesn't solve the problem, but if you don't fix them
you won't get the program to work.

Attached is a revised copy of askconvert.

Sorry about that!
Randy Kramer

I've tried the suggested approaches and variations that I could think of, and
still no luck.

--Boundary-00=_SqkFHVRI2sVWoto
Content-Type: application/x-ruby;
name="askconvert"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="askconvert"

#! /usr/bin/env ruby

# This is iter(ation)-4 (in progress) of ruby program askconvert to convert (initially) askRhk03 formatted files to askRhk04 format. (Later versions may perform additional conversions.) This version does work only for the hard coded conversion of file test.txt to test.aml. Iter-5 will presumably address getting the filenames from the command line. Iter-4 is sort of a break--code cleanup and maybe some comments--much as it is against my religion, I'm sort of trying to see to what extent this is (or can be made) "self-documenting"--I do have plenty of documentation (some pretty rough) in other files like askconvert.dev.aml.

File.open("test.aml", "w") do |f|
first_record = true
new_record = false
titles = Array.new
File.open("test.txt").each do |line|
case line
when /^\x80\x81\x82\x83$/
new_record = true
# when new_record && /^---\+\+ (.*)/ =~ line
# titles.push $1.strip
# the previous two lines could replace the next five lines if the previous two lines worked (or I could change the case statement to an if, elsif, elsif ... (or whatever)--case just seemed more "elegant" somehow--I'll wait and see if anybody can show me how to get those two lines to work.
when /^---\+\+ (.*)/
if new_record
titles.push $1.strip
else
f.puts line.rstrip + "\n" end
else
if new_record then
# I could take the next 14 lines and "move" them to a module / method named something like "process_header"--I'd probably also learn something by doing so. I think I'm "learned out" for the moment. ;-)
unless first_record then f.puts "morF\n\n"
else first_record = false end
if titles.length == 3 then
title = titles.pop
titles.reverse!
else
title = titles.shift end
f.puts "From \"" + title + "\" " + %x{date -u +"%a %b %d %T %Z %Y"} + "Date: " + %x{date +"%D %H:%M %P %Z"} + "Subject: " + title + "\n"
unless titles.empty?
f.puts "\n"
titles.each { |title| if title.strip.length > 0 then f.puts "T: " + title + "\n" end}
f.puts "\n" end
titles.clear
new_record= false
else
f.puts line.rstrip + "\n" end
end
end
f.puts "morF\n"
end

--Boundary-00=_SqkFHVRI2sVWoto--
--Boundary-00=_SqkFHVRI2sVWoto--

 

[Randy Kramer]

and still no luck.

Ok, for the record I got it to work--here's a little working test program:

#! /usr/bin/env ruby

new_record = true
line = "---++ tasty"

case
when (new_record and /^---\+\+ (.*)/ =~ line)
puts "a title if at the beginning of a record: " + $1
when /^---\+\+ (.*)/ =~ line
puts "a level 2 heading if not at the beginning of a record: " + $1
else
puts "neither a title nor a level 2 heading"
end

My mistake was leaving "line" after "case" when I tried to switch to the
alternate form of case statement. Sometimes (or more often ;-) I can be
pretty dense.

(And, of course, the other thing I finally did was make a small test snippet
of the code for experimentation.)

Randy Kramer