Re: "fields" in files

[Paul E Condon <pecondon@mesanetworks.net>]

On 20110130_134532, Axel Freyn wrote:
> Hi Lisi,
> On Sun, Jan 30, 2011 at 12:18:39PM +0000, Lisi wrote:
> > On Sunday 30 January 2011 11:35:56 Camaleón wrote:
> > > On Sun, 30 Jan 2011 11:26:16 +0000, Lisi wrote:
> > > > In teh following:
> > > > -rw-r--r-- 1 lisi lisi    19503 2011-01-28 21:12 Etch_repositories.odt
> > > >
> > > > I can see eight pieces of information: permissions, directory or file
> > > > (information that is also at the beginning of the permissions), owner,
> > > > group (or group, owner), size, date, name of file or directory.
> > > >
> > > > What are these called?  I have always called them fields, but it would
> > > > appear that I am wrong to do so.
> > >
> > > directory/file
> > > user perms-group perms-others perms
> > > user-group
> > > size
> > > last modified timestamp
> > > file name
> > >
> > > I call them "file attributes" but not sure if that's the technical name
> > > though :-)
> > 
> > Thanks, Camaleón!  Yes ,that it what they are in this context.  But in more 
> > general terms, like page, chapter, section etc., rather than the names of 
> > that particular page etc. -  what are they called?  I want to use it with cut 
> > if possible, but can't even find if there is a suitable option when I don't 
> > know what they are called.  Column is too narrow (in the physical sense) a 
> > definition.  There are 21 of what cut calls columns in the name of the file 
> > alone in this sample!
> 
> Is that what you want to achieve?
> 
> ls -l | sed -e 's/  */ /g'  | cut -f 8 -d " "
This give the file name field.
An alternative is:
ls -l | tr -s ' ' | cut -f 8 -d ' '

But beware. If the file name contains embedded space(s), it will
give only the leading part up to the first space. To get the whole
file name, use:
ls -l | tr -s ' ' | cut -f 8- -d ' '

If you are interested in processing the date and time fields, be
aware that the detailed format that is used depends on how old 
the file is at the time of execution of 'ls'. I get around this
pandering to human traditions by defining shell variable

TIME_STYLE=+%Y%m%d_%H%M%S 

This combines date and time-of-day into a single field. A file 
of data about files using this, remains correct when reconstructed
at a later date. A full-up geek design decision with no concession
to human frailty would be to use

TIME_STYLE=+%s

This gives seconds since UNIX epoch. Which might be useful if you are
collecting data from computers that are operating in different time
zones.

hth
-- 
Paul E Condon           
pecondon@mesanetworks.net