Cheetah: Python-Powered Template Engine

I wrote:

> BTW, did the change to #compiler-settings (to allow one
> line #compiler useNameMapper = 0 directives) ever get made?

Ignore this question...I've been using it all along.  Duh...

Had a brain fart while reading the Cheetah User Guide, which doesn't
mention the change. Sorry.

...Edmund.

>Mike wrote:
>
>> They are placed in a dictionary, self._globalSetVars .
>> There's also a local variable globalSetVars shadowing it.
>>
>> (Note that we reserve the right to change the implementation
>> in a future version.)
>
>Oh, OK. I suppose another way of getting at the variables in question is
to just turn on the >NameMapper when I need to get at the global variable.
This way, I don't have to worry about >implementation changes to Cheetah.

As a follow-up... the neatest way to get at a global variable with the
NameMapper off seems to be to #set a local variable to it *before* turning
off the NameMapper. i.e.:

#set $localVar = $globalVar
#compiler useNameMapper 0
$localVar

...Edmund.

On Fri, Jun 21, 2002 at 03:01:40PM -0400, Edmund Lian wrote:
> 
> Mike wrote:
> 
> > They are placed in a dictionary, self._globalSetVars .
> > There's also a local variable globalSetVars shadowing it.
> >
> > (Note that we reserve the right to change the implementation
> > in a future version.)
> 
> Oh, OK. I suppose another way of getting at the variables in question is to
> just turn on the NameMapper when I need to get at the global variable. This
> way, I don't have to worry about implementation changes to Cheetah.

That sounds like a better way.  Why are you turning off the NameMapper
anyway?  Tavis put the option in in case somebody needed it off for some
reason, but I can't think of any reason to ever turn it off.

> BTW, did the change to #compiler-settings (to allow one line #compiler
> useNameMapper = 0 directives) ever get made?

Yes, it was added in 0.9.13b1.  But the syntax is:

        #compiler commentStartToken = '//'
        // a comment
        #compiler reset
        // no longer a comment

Hmm, it's not documented yet.

I don't remember what we were going to do with #compiler-settings,
whether it was going to stay or disappear.  #settings is still slated
for disappearance.

-- 
-Mike (Iron) Orr, iron@...  (if mail problems: mso@...)
   http://iron.cx/     English * Esperanto * Russkiy * Deutsch * Espan~ol

Mike wrote:

> Why are you turning off the NameMapper anyway?  Tavis put the option
> in in case somebody needed it off for some reason, but I can't think
> of any reason to ever turn it off.

Tavis is a wise man... it appears I'm one of those somebodies!  :-)

I had to turn it off to get my form handling code to work properly. The
code makes use of non-default argument function calls, Python 2.2 property
method calls, etc. I the NameMapper can't handle these properly. Turning it
off allows me precise control over things.

For example, in a Cheetah template I write:


#from PageTemplate import PageTemplate
#from WebFormDefinitions import ThisForm
#extends PageTemplate

#def mainPanel
#compiler useNameMapper = 0
#set $f = ThisForm($self.$transaction) ## create a form instance

## place form's fields and values, as well as error message
## if any; HTML autogenerated from field's definition and error
## checks.
<form method=post action="">
$f.username $f.username.error<br>
$f.weight $f.weight.error
<br>
$f.submit  ## place a submit button
</form>



WebFormDefinitions contains:

class ThisForm(WebForm):
    def createForm(self):
        self.addField(Text('username'))
        self.addField(Decimal('weight', precision=2, minimum=0,
maximum=300))
        self.addField(Button('submit', defaultValue="OK"))

    def processForm(self):
        # do stuff


The form presents itself as widgets that do basic validation, range checks,
etc. There's some tricky stuff going in in terms of use of __call__
methods, etc. to get this, and the NameMapper gets confused.

...Edmund.

On Fri, Jun 21, 2002 at 03:46:28PM -0400, Edmund Lian wrote:
> 
> Mike wrote:
> 
> > Why are you turning off the NameMapper anyway?  Tavis put the option
> > in in case somebody needed it off for some reason, but I can't think
> > of any reason to ever turn it off.
> 
> Tavis is a wise man... it appears I'm one of those somebodies!  :-)
> 
> I had to turn it off to get my form handling code to work properly. The
> code makes use of non-default argument function calls, Python 2.2 property
> method calls, etc. I the NameMapper can't handle these properly. Turning it
> off allows me precise control over things.

If you have the time and patience, I'd like to see how exactly
NameMapper miscompiles them.  I'd need to know the placeholder, what it
compiles to, and what it should compile to.  Perhaps we can identify
some bugs in NameMapper or warnings for the documentation.

> #compiler useNameMapper = 0
> #set $f = ThisForm($self.$transaction) ## create a form instance
> 
> ## place form's fields and values, as well as error message
> ## if any; HTML autogenerated from field's definition and error
> ## checks.
> <form method=post action="">
> $f.username $f.username.error<br>
> $f.weight $f.weight.error
> <br>
> $f.submit  ## place a submit button

I assume these are the placeholders it messes up on.  That 
"$self.$transaction" looks suspicious.  

-- 
-Mike (Iron) Orr, iron@...  (if mail problems: mso@...)
   http://iron.cx/     English * Esperanto * Russkiy * Deutsch * Espan~ol

Mike wrote:

> If you have the time and patience, I'd like to see how exactly
> NameMapper miscompiles them.  I'd need to know the placeholder, what it
> compiles to, and what it should compile to.  Perhaps we can identify
> some bugs in NameMapper or warnings for the documentation.

Sure... I'll do what I can...


> I assume these are the placeholders it messes up on.  That
> "$self.$transaction" looks suspicious.

$self.$transaction is what I had to use to get the transaction object
passed into the form with the NameMapper off.

How should I try to trace the points of failure? I'm thinking that as a
first step, I should try compiling the same form with and without the
NameMapper to see what the differences are? Is this the right approach?

...Edmund.

On Fri, Jun 21, 2002 at 04:26:24PM -0400, Edmund Lian wrote:
> How should I try to trace the points of failure? I'm thinking that as a
> first step, I should try compiling the same form with and without the
> NameMapper to see what the differences are? Is this the right approach?

You know which placeholders are giving the wrong answer or generating
errors.  Those are the ones we care about.  Since at this level I don't
think the context matters, you can just put all the questionable
syntaxes in a separate template, one per line, with an
"#errorCatcher Echo" at the top to avoid the NotFound error.

As for $self.$transaction, what do these print?

$repr($transaction)
$repr($transaction())
$repr($self.transaction)
$repr($self.transaction())
$repr(${self}.transaction)
$repr(${self}.transaction())
$repr(${self}.$transaction)
$repr(${self}.$transaction())

The transaction object itself is self.transaction(), but any of the
first four syntaxes should give you that if autocalling is on.  I wonder
if the 'transaction' variable is being masked by another variable
'transaction' somewhere in the searchList.

-- 
-Mike (Iron) Orr, iron@...  (if mail problems: mso@...)
   http://iron.cx/     English * Esperanto * Russkiy * Deutsch * Espan~ol

Hey all,

What is the best way to loop over a set of data and only use one template?

Here's what I do now:

-- begin sample.py --

from Cheetah.Template import Template
import smtplib

mailer = smtplib.SMTP('mydomain.com')
emailTemplate = Template (source=open('massMailer.txt').read())
emailAddresses = [...] # long list of emails

for email in emailAddresses:
    emailTemplate.addToSearchList ({'EmailAddress':email})
    mailer.sendmail ('me@...', email, str(emailTemplate)
    del template._searchList[-1] # ?!?

-- end sample.py

For speediness, I don't want to re-instantiate a new template every time I want to send out an email (I'm sending to nearly 4,000 and the list is growing every week).

The line where I delete _searchList[-1], I know, is really bad coding.  But, I couldn't find any other way to reset the search list so I could re-use it to send the email.  Is there a way to set the search list anew every time I want to use it?

Thanks,
Tracy


--
Tracy S. Ruggles ~ tr@...
OutputLinks ~ 888/999.2674 x TR
"Your Output Industry Portal"

On June 22, 2002 09:38 am, Tracy Ruggles wrote:
> Hey all,
>
> What is the best way to loop over a set of data and only use one templa=
te?
> Here's what I do now:
>
> -- begin sample.py --
>
> from Cheetah.Template import Template
> import smtplib
>
> mailer =3D smtplib.SMTP('mydomain.com')
> emailTemplate =3D Template (source=3Dopen('massMailer.txt').read())
> emailAddresses =3D [...] # long list of emails
>
> for email in emailAddresses:
>     emailTemplate.addToSearchList ({'EmailAddress':email})
>     mailer.sendmail ('me@...', email, str(emailTemplate)
>     del template._searchList[-1] # ?!?
>
> -- end sample.py

how about this:

from Cheetah.Template import Template
import smtplib

mailer =3D smtplib.SMTP('mydomain.com')
emailTemplate =3D Template (source=3Dopen('massMailer.txt').read())
emailAddresses =3D [...] # long list of emails
aDict =3D {}
emailTemplate.addToSearchList(aDict)
for email in emailAddresses:
     aDict['EmailAddress'] =3D email
     mailer.sendmail('me@...', email, str(emailTemplate))

On Sat, Jun 22, 2002 at 11:38:35AM -0500, Tracy Ruggles wrote:
> What is the best way to loop over a set of data and only use one template?
> 
> For speediness, I don't want to re-instantiate a new template every
> time I want to send out an email (I'm sending to nearly 4,000 and the
> list is growing every week).
> Here's what I do now:
> 
> The line where I delete _searchList[-1], I know, is really bad coding.
> But, I couldn't find any other way to reset the search list so I could
> re-use it to send the email.  Is there a way to set the search list
> anew every time I want to use it?

You don't need to replace the search *list*, you just need to update
those certain variables (or in this case, one variable).  You can
either save a reference to the dictionary in an ordinary Python 
variable and modify the dictionary in place between e-mails, or skip the
dictionary entirely and set the variables directly in the template
object.  

You can gain startup speed improvement by using a precompiled
template.  
	cheetah compile -p massMailer.txt >massMailer.py
	

-- begin sample.py --

from massMailer import massMailer
import smtplib

mailer = smtplib.SMTP('mydomain.com')
emailTemplate = massMailer()
emailAddresses = [...] # long list of emails

for email in emailAddresses:
    emailTemplate.EmailAddress = email
    mailer.sendmail ('me@...', email, str(emailTemplate) )

-- end sample.py

-- 
-Mike (Iron) Orr, iron@...  (if mail problems: mso@...)
   http://iron.cx/     English * Esperanto * Russkiy * Deutsch * Espan~ol

even better!

On June 22, 2002 10:02 am, Mike Orr wrote:
> On Sat, Jun 22, 2002 at 11:38:35AM -0500, Tracy Ruggles wrote:
> > What is the best way to loop over a set of data and only use one
> > template?
> >
> > For speediness, I don't want to re-instantiate a new template every
> > time I want to send out an email (I'm sending to nearly 4,000 and the
> > list is growing every week).
> > Here's what I do now:
> >
> > The line where I delete _searchList[-1], I know, is really bad coding=
=2E
> > But, I couldn't find any other way to reset the search list so I coul=
d
> > re-use it to send the email.  Is there a way to set the search list
> > anew every time I want to use it?
>
> You don't need to replace the search *list*, you just need to update
> those certain variables (or in this case, one variable).  You can
> either save a reference to the dictionary in an ordinary Python
> variable and modify the dictionary in place between e-mails, or skip th=
e
> dictionary entirely and set the variables directly in the template
> object.
>
> You can gain startup speed improvement by using a precompiled
> template.
> =09cheetah compile -p massMailer.txt >massMailer.py
>
>
> -- begin sample.py --
>
> from massMailer import massMailer
> import smtplib
>
> mailer =3D smtplib.SMTP('mydomain.com')
> emailTemplate =3D massMailer()
> emailAddresses =3D [...] # long list of emails
>
> for email in emailAddresses:
>     emailTemplate.EmailAddress =3D email
>     mailer.sendmail ('me@...', email, str(emailTemplate) )
>
> -- end sample.py