Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #31288 > unrolled thread

What's the tidy/elegant way to protect this against null/empty parameters?

Back to article view | Back to comp.lang.python

Contents

  What's the tidy/elegant way to protect this against null/empty parameters? tinnews@isbd.co.uk - 2012-10-15 12:23 +0100
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Paul Rubin <no.email@nospam.invalid> - 2012-10-15 04:49 -0700
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Chris Rebert <clp2@rebertia.com> - 2012-10-15 04:58 -0700
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Roy Smith <roy@panix.com> - 2012-10-15 08:00 -0400
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Miki Tebeka <miki.tebeka@gmail.com> - 2012-10-15 06:33 -0700
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Terry Reedy <tjreedy@udel.edu> - 2012-10-15 11:45 -0400
    Re: What's the tidy/elegant way to protect this against null/empty parameters? Marco Nawijn <nawijn@gmail.com> - 2012-10-16 03:07 -0700
      Re: What's the tidy/elegant way to protect this against null/empty parameters? tinnews@isbd.co.uk - 2012-10-16 14:44 +0100

#31288 — What's the tidy/elegant way to protect this against null/empty parameters?

I want to fix an error in some code I have installed, however I don't
really want to just bodge it.

The function producing the error is:-

    def get_text(self, idx):               # override !
        node = self.items[idx]

        a= [
            ", ".join(node.tags),
            node.comment,
            node.folderName,
            cd2rd(node.date),
            node.name,
            '[' + self.rating_stars[node.rating] + ']'
            ] [self.select]

        return a


The error occurs when node[] (or at least its members) turn out to be
empty, you get a Traceback that ends with:-

  File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell layout.set_text(self.get_text(thumbnail_num))
  File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", ".join(node.tags),
  TypeError: sequence item 0: expected string, NoneType found

Now its *probably* something higher up the tree causing the problem
(it's only one particular image in 20 thousand or so that breaks
things) but I really want to just get things working.  So, what's the
neatest way to protect the get_text() method from empty data?


-- 
Chris Green

[toc] | [next] | [standalone]

#31290

tinnews@isbd.co.uk writes:
> I want to fix an error in some code I have installed, however I don't
> really want to just bodge it. ...
> Now its *probably* something higher up the tree causing the problem
> (it's only one particular image in 20 thousand or so that breaks
> things) but I really want to just get things working.  

That means you do really want to just bodge it, doesn't it?  (I'm
reading "bodge" as something like "kludge" but maybe it means something
different.)

> So, what's the neatest way to protect the get_text() method from empty
> data?

I'd say the use of None as a picture is already a code smell: figure out
where it is coming from, and fix it.  Or, if you want to just bodge it
(by checking for None and returning the empty string or something)),
then try it and see if it helps.  I'm assuming this is something
noncritical that you're running on your own computer for convenience.  I
wouldn't ship code to a customer without a more careful fix.

[toc] | [prev] | [next] | [standalone]

#31291

On Mon, Oct 15, 2012 at 4:23 AM,  <tinnews@isbd.co.uk> wrote:
> I want to fix an error in some code I have installed, however I don't
> really want to just bodge it.

"bodge". Well, I learned a new word this morning!

> The function producing the error is:-
>
>     def get_text(self, idx):               # override !
>         node = self.items[idx]
>
>         a= [
>             ", ".join(node.tags),
>             node.comment,
>             node.folderName,
>             cd2rd(node.date),
>             node.name,
>             '[' + self.rating_stars[node.rating] + ']'
>             ] [self.select]
>
>         return a
>
>
> The error occurs when node[] (or at least its members) turn out to be
> empty,

To be precise: when node.tags contains one or more `None`s (Python's
equivalent of what other languages call "null" or "nil").
That's what the traceback is saying.

> you get a Traceback that ends with:-
>
>   File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell layout.set_text(self.get_text(thumbnail_num))

Ah, so this is apparently regarding https://code.google.com/p/jbrout/
. Would have been nice not to have had to search and then only locate
it indirectly. Something to consider next time you write in...
Make sure you report your bug upstream!

>   File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", ".join(node.tags),
>   TypeError: sequence item 0: expected string, NoneType found
>
> Now its *probably* something higher up the tree causing the problem
> (it's only one particular image in 20 thousand or so that breaks
> things) but I really want to just get things working.  So, what's the
> neatest way to protect the get_text() method from empty data?

Filter out the `None`s with a generator expression:
    ", ".join(tag for tag in node.tags if tag is not None),

Cheers,
Chris

[toc] | [prev] | [next] | [standalone]

#31292

In article <1b8tk9-un9.ln1@chris.zbmc.eu>, tinnews@isbd.co.uk wrote:

> The function producing the error is:-
> 
>     def get_text(self, idx):               # override !
>         node = self.items[idx]
> 
>         a= [
>             ", ".join(node.tags),
>             node.comment,
>             node.folderName,
>             cd2rd(node.date),
>             node.name,
>             '[' + self.rating_stars[node.rating] + ']'
>             ] [self.select]
> 
>         return a
> 
> 
> The error occurs when node[] (or at least its members) turn out to be
> empty, you get a Traceback that ends with:-
> 
>   File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell 
>   layout.set_text(self.get_text(thumbnail_num))
>   File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", 
>   ".join(node.tags),
>   TypeError: sequence item 0: expected string, NoneType found
> 
> Now its *probably* something higher up the tree causing the problem
> (it's only one particular image in 20 thousand or so that breaks
> things) but I really want to just get things working.  So, what's the
> neatest way to protect the get_text() method from empty data?

Well, you don't describe what get_text() is supposed to return  First, 
you build a list of what I'm guessing are all strings, then you index 
into it and return one of the values.  So, I'm guessing get_text() is 
supposed to return a string.

At a low level, you can certainly fix that by testing to see if 
self.items[idx] returns what you're expecting (an instance of Node?) and 
returning an empty string if it's not:

>     def get_text(self, idx):               # override !
>         node = self.items[idx]
>         if not node:
>             return ""
> 
>         a= [
>             ", ".join(node.tags),
>             node.comment,
>             node.folderName,
>             cd2rd(node.date),
>             node.name,
>             '[' + self.rating_stars[node.rating] + ']'
>             ] [self.select]
> 
>         return a

Whether that makes sense in your program, I have no idea.  What does it 
mean for node to be empty?  Is this a normal occurrence?  If so, then 
your code needs to deal with properly (the suggest above being just one 
possible way).

Or, is it "impossible" for node to be empty?  In that case, that fact 
that it *is* empty is a bug, and the above suggestion will just hide the 
bug for one more level and make it that much harder to figure out what's 
really going on.  What you might really want to do is sprinkle your code 
with "assert node" statements.  This will force your program to crash 
and burn the first time node is empty, which might help you figure out 
why it is.

[toc] | [prev] | [next] | [standalone]

#31300

> I want to fix an error in some code I have installed, ...
Apart from all the reasons why it's bad (see the Python Zen #10). One way to do it is:
    return [i or '' for i in a]

[toc] | [prev] | [next] | [standalone]

#31314

On 10/15/2012 7:23 AM, tinnews@isbd.co.uk wrote:
> I want to fix an error in some code I have installed, however I don't
> really want to just bodge it.
>
> The function producing the error is:-
>
>      def get_text(self, idx):               # override !
>          node = self.items[idx]
>
>          a= [
>              ", ".join(node.tags),
>              node.comment,
>              node.folderName,
>              cd2rd(node.date),
>              node.name,
>              '[' + self.rating_stars[node.rating] + ']'
>              ] [self.select]
>
>          return a
>
>
> The error occurs when node[] (or at least its members) turn out to be
> empty,

This is not the problem.

 > you get a Traceback that ends with:-
>
>    File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell layout.set_text(self.get_text(thumbnail_num))
>    File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", ".join(node.tags),
>    TypeError: sequence item 0: expected string, NoneType found

The specific problem is that node.tags is supposed to be a sequence of 
strings and somehow one instead has None as the first, and probably last 
item. This was likely intended to indicate an empty list, but the way to 
do that is to have a empty list, which would have worked just fine. In 
other words, the likely problem is that node.tags is *not* an empty 
sequence when it should be.

 >>> ','.join([None])
Traceback (most recent call last):
   File "<pyshell#9>", line 1, in <module>
     ','.join([None])
TypeError: sequence item 0: expected str instance, NoneType found
 >>> ','.join([])
''

-- 
Terry Jan Reedy

[toc] | [prev] | [next] | [standalone]

#31374

On Monday, October 15, 2012 1:33:02 PM UTC+2, (unknown) wrote:
> I want to fix an error in some code I have installed, however I don't
> 
> really want to just bodge it.
> 
> 
> 
> The function producing the error is:-
> 
> 
> 
>     def get_text(self, idx):               # override !
> 
>         node = self.items[idx]
> 
> 
> 
>         a= [
> 
>             ", ".join(node.tags),
> 
>             node.comment,
> 
>             node.folderName,
> 
>             cd2rd(node.date),
> 
>             node.name,
> 
>             '[' + self.rating_stars[node.rating] + ']'
> 
>             ] [self.select]
> 
> 
> 
>         return a
> 
> 
> 
> 
> 
> The error occurs when node[] (or at least its members) turn out to be
> 
> empty, you get a Traceback that ends with:-
> 
> 
> 
>   File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell layout.set_text(self.get_text(thumbnail_num))
> 
>   File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", ".join(node.tags),
> 
>   TypeError: sequence item 0: expected string, NoneType found
> 
> 
> 
> Now its *probably* something higher up the tree causing the problem
> 
> (it's only one particular image in 20 thousand or so that breaks
> 
> things) but I really want to just get things working.  So, what's the
> 
> neatest way to protect the get_text() method from empty data?
> 
> 
> 
> 
> 
> -- 
> 
> Chris Green
Hi,

Instead of protecting against empty data, you could just catch the exception, issue a warning and return a default "error" node which is valid. So something like (not tested):

def get_text(self, idx):               # override ! 
        node = self.items[idx] 

        error_a = "A valid, but erroneous representation of a"
        
        try:
            a= [ 
                ", ".join(node.tags), 
                node.comment, 
                node.folderName, 
                cd2rd(node.date), 
                node.name, 
                '[' + self.rating_stars[node.rating] + ']' 
                ] [self.select] 
        except TypeError:
              print 'Oops, something went wrong'
              a = error_a
        # You should always have a valid a here (or another exception has  occured)
        return a 

[toc] | [prev] | [next] | [standalone]

#31401

Marco Nawijn <nawijn@gmail.com> wrote:
> On Monday, October 15, 2012 1:33:02 PM UTC+2, (unknown) wrote:
> > I want to fix an error in some code I have installed, however I don't
> > 
> > really want to just bodge it.
> > 
> > 
> > 
> > The function producing the error is:-
> > 
> > 
> > 
> >     def get_text(self, idx):               # override !
> > 
> >         node = self.items[idx]
> > 
> > 
> > 
> >         a= [
> > 
> >             ", ".join(node.tags),
> > 
> >             node.comment,
> > 
> >             node.folderName,
> > 
> >             cd2rd(node.date),
> > 
> >             node.name,
> > 
> >             '[' + self.rating_stars[node.rating] + ']'
> > 
> >             ] [self.select]
> > 
> > 
> > 
> >         return a
> > 
> > 
> > 
> > 
> > 
> > The error occurs when node[] (or at least its members) turn out to be
> > 
> > empty, you get a Traceback that ends with:-
> > 
> > 
> > 
> >   File "/usr/lib/jbrout/jbrout/listview.py", line 608, in draw_cell layout.set_text(self.get_text(thumbnail_num))
> > 
> >   File "/usr/lib/jbrout/jbrout.py", line 325, in get_text ", ".join(node.tags),
> > 
> >   TypeError: sequence item 0: expected string, NoneType found
> > 
> > 
> > 
> > Now its *probably* something higher up the tree causing the problem
> > 
> > (it's only one particular image in 20 thousand or so that breaks
> > 
> > things) but I really want to just get things working.  So, what's the
> > 
> > neatest way to protect the get_text() method from empty data?
> > 
> > 
> > 
> > 
> > 
> > -- 
> > 
> > Chris Green
> Hi,
> 
> Instead of protecting against empty data, you could just catch the exception, issue a warning and return a default "error" node which is valid. So something like (not tested):
> 
> def get_text(self, idx):               # override ! 
>         node = self.items[idx] 
> 
>         error_a = "A valid, but erroneous representation of a"
>         
>         try:
>             a= [ 
>                 ", ".join(node.tags), 
>                 node.comment, 
>                 node.folderName, 
>                 cd2rd(node.date), 
>                 node.name, 
>                 '[' + self.rating_stars[node.rating] + ']' 
>                 ] [self.select] 
>         except TypeError:
>               print 'Oops, something went wrong'
>               a = error_a
>         # You should always have a valid a here (or another exception has  occured)
>         return a 

That sounds like a reasonable approach, thank you (and all the other ideas).

-- 
Chris Green

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web