[GRASSGUI] commands from menus and database table not working

[Glynn Clements]

Michael Barton wrote:

> >     self.cmd_output.write("$ " + command + "\n")
> > TypeError: cannot concatenate 'str' and 'list' objects
> > 
> > Why:
> 
> The problem is not command but "$ " and "\n". They is a strings. Change to
> 
> ['$ '] + command + ['\n']

That will produce a list; does write() accept a list?

But the default string conversion performed by %s probably isn't what
you want. From the leading "$", I'm guessing that you want output in
shell syntax, which is going to need some form of quoting.

OTOH, you could just use Python syntax. Generating Bourne-shell syntax
is arguably wrong for something that's supposed to be cross-platform.

> > method getCmd() of grassTask() class (gui_modules/menform.py, line 215)
> > returns not command string (e.g. "r.info elevation.dem"), but already
> > command list (e.g. ["r.info", "map=elevation.dem"]).
> > 
> > This is in conflict with rest of the code, where ever cmd.split(" ") is
> > called (e.g. wxgui.py line 410 or gui_modules/wxgui_utils.py, line 868).
> 
> cmd.split(" ") is potentially problematic and we should avoid it. For
> example, <some command> map="My map" will end up as
> 
> ['<command>', 'map=My', 'map'].

It doesn't matter for maps, as spaces aren't valid in map names. But
it matters for arbitrary text (titles, etc), filenames, SQL "where"
clauses etc.

> Or <command> query="attribute1>5 and attribute2<15"
> 
> ['<command>', 'query=attribute>5', 'and', 'attribute2<15']
> 
> > 
> > So, what to do? Fix getCmd(), so it returns string or go through whole
> > code and try to fix all cmd.split() calls?
> 
> The latter. We've already got most of them, but there are a few left
> apparently.

What is the purpose of getCmd()? If it's meant to return the command
in a form that can be executed within wxgui, it should return a list.

If it's simply a string to show to the user, the format isn't
critical. OTOH, if it's going to be saved to a script, it needs to be
in the appropriate syntax (sh, csh, cmd.exe, etc).

The main point is that, if you have a function which returns a command
as a string, you need to be absolutely crystal clear on what syntax is
used. If it's Bourne-shell syntax, don't try to execute it via cmd.exe
(and vice-versa). If it's just "human-readable string", don't try to
execute it at all.

Depending upon what you want to do, you might need multiple functions
to return commands in different formats. In that case, I would suggest
having class methods always return commands as lists, and have
separate utility functions for converting such lists to strings.

-- 
Glynn Clements <glynn at gclements.plus.com>