Using comment symbol for matching comment sections

Faki · 11-03-2021, 03:45 AM

I have the following bash function to extract sections between `## Mode: org` and `## # End of org`, using `#` as the comment character.

I would like to enable other comment characters instead of handling only `#`. Using a Character Class `[#;c!C]` could be a good plan.

I also want to allow a user-defined literal string for the texinfo comment identifier `@c`.

Code:

 
capture ()
 {
  local efile="$1"
  begorg='^[[:space:]]*## Mode: org$'
  endorg='^[[:space:]]*## # End of org$'
  awk -v bego="$begorg" -v endo="$endorg" \
    '$0 ~ bego { found=1; next } 
     $0 ~ endo { found=0; } 
     found { sub(/^[[:space:]]*#+[[:space:]]*/,""); print }' "$efile"
 }

pan64 · 11-03-2021, 03:50 AM

and what is your problem exactly?

MadeInGermany · 11-03-2021, 04:15 AM

A [ ] character set is an OR for single characters.
awk uses ERE that has
| for OR
and
( ) for grouping

Code:

([#;!]|@[Cc])

is one of # ; ! or @ followed by one of C c
The ( ) limit the scope of the |
in case something is appended (or prepended)

Faki · 11-03-2021, 04:26 AM

I would like to use a variable for character class. And also take care of C-Language comments //
Capital `C` is for Fortran Fixed Form files.

For single comment characters `[#;!C]` I would like the possibility any number of comment characters, hence `([#;!C])+`. But for other comment types (e.g. `//` for C-Language, and `@c` for Texinfo-Language) there will be no repeats, thus just
`[//|@c]`.

Code:

capture ()
{
 local efile="$1"
 
 local begorg endorg charcl

 i="1"
 if [ "$i" = "1" ]; then 
   charcl='^[[:space:]]*(#|;|!)+[[:space:]]*' 
 elif [ "$i" = "2" ]; then
   charcl='^[[:space:]]*(//|@c)[[:space:]]*' 
 fi 
 
 begorg="${charcl}"'Mode: org$'
 endorg="${charcl}"'# End of org$'
 
 awk -v ccls="$charcl" -v bego="$begorg" -v endo="$endorg" \
   '$0 ~ bego { found=1; next } 
    $0 ~ endo { found=0; } 
    found { sub(/ccls/,""); print }' "$efile"
}

pan64 · 11-03-2021, 05:28 AM

That's why you need to write more than one sub calls:

Code:

sub(/^[[:space:]]*#+[[:space:]]*/,"")
sub(/^[[:space:]]*//[[:space:]]*/,"")
... whatever ...

Or you can specify a complex regex too, but that would be the hard way....[

Code:

sub(/^[[:space:]]*<your regexp>[[:space:]]*/,"")

In that case probably I would rather use perl.

Faki · 11-03-2021, 06:21 AM

I agree. Have updated the function but have to recognise the language (whether `#`, `;`, `!`).

To avoid too complicated patterns, the if statement can be used to distinguish between single character comments (#;!) and double character comments (//|@c).

pan64 · 11-03-2021, 06:24 AM

you can specify the language and in a BEGIN block (awk) you can specify that regex depending on the language

Faki · 11-03-2021, 07:03 AM

It would like to figure out the language from the function itself automatically by checking the comment character from the mode line "Mode: org"

Or I can use the extension of the input file.

pan64 · 11-03-2021, 07:39 AM

so you need to implement that (in awk).

Code:

/<pattern>/ { set variable to A }
/other pattern/ { set variable to B }
... use this variable in sub

Faki · 11-03-2021, 09:50 AM

Have modified to check for any of the comment cases in a single regex.

Code:

capture ()
{
 local efile="$1"
 
 local charcl begorg endorg

 charcl='^[[:space:]]*([#;!]+|@c|\/\/)[[:space:]]*' 
 begrec="${charcl}"'Mode: org$'
 endrec="${charcl}"'# End of org$'
 
 awk -v ccls="$charcl" -v bego="$begorg" -v endo="$endorg" \
   '$0 ~ bego { found=1; next }
    $0 ~ endo { found=0; }
    found { sub(/ccls/,""); print }' "$efile"
}

But when I tried it, the comment characters are still displayed, rather than deleted.

Code:

awk: warning: escape sequence `\/' treated as plain `/'
 #  Assigns shell positional parameters or changes the values of shell
 #  options.  The -- option assigns the positional parameters to the
 #  arguments of {set}, even when some of them start with an option
 #  prefix `-'.
 ;  Assigns shell positional parameters or changes the values of shell
 ;  options.  The -- option assigns the positional parameters to the
 ;  arguments of {set}, even when some of them start with an option
 ;  prefix `-'.
 @c Assigns shell positional parameters or changes the values of shell
 @c options.  The -- option assigns the positional parameters to the
 @c arguments of {set}, even when some of them start with an option
 @c prefix `-'.

This was the input

Code:

 ## Mode: org
 #  Assigns shell positional parameters or changes the values of shell
 #  options.  The -- option assigns the positional parameters to the
 #  arguments of {set}, even when some of them start with an option
 #  prefix `-'.
 ## # End of org

 ;; Mode: org
 ;  Assigns shell positional parameters or changes the values of shell
 ;  options.  The -- option assigns the positional parameters to the
 ;  arguments of {set}, even when some of them start with an option
 ;  prefix `-'.
 ;; # End of org
 
 @c Mode: org
 @c Assigns shell positional parameters or changes the values of shell
 @c options.  The -- option assigns the positional parameters to the
 @c arguments of {set}, even when some of them start with an option
 @c prefix `-'.
 @c # End of org