#! /bin/csh -f
# To fetch files saved in [IN] such as attachments and .html email bodies.
	# New attempt.  Stay with old design.  *.* is so fast that fancier
		# patterns aren't worthwhile except theoretically.
# Can now take 2 params 'bbc' and 3  --
	# -- which means get the 3rd last bbcnews email AND fetch it.

## These addresses are NOT urls BUT my unix file names.
set pickup = ~/public_html/pickup  # my name for my unix dir
set pickup = ~/public_html/pickup  # my name for my unix dir
set IN = ~/mail/IN
set ind = 1
if ("$1" == 'bbc' &&  "$2") set ind = "$2"

## Not sure if this is right or necessary
bbcextract "$ind" # Extracts latest bbc daily .html file
if ("$1" == 'bbc') exit

set suffixes = (ics mp4 mp3 zip pdf doc{,x} ppt{,x} ps xls{,x} csv gif jp{,e}g png\
	txt htm{,l} cw odp odt)
	## Have done case-indiff matching differently below.
		## But could do it like here.

set tab = '	'
set unapproved = ()
set ignMax = 0;
set msgignore = ({A,B,C,D,E,F,G,H,I,J}{A,B,C,D,E,F,G,H,I,J});
        # Creates (only) 100 slots in array; with CAPS so if you see the
             # contents, you'll immed. know something is wrong with the script.

cd $IN
	# Patterns with glob.  Finds all files in current dir
		## with extensions consisting of only 1 or 2 digits
set toignore = (*.{[1234567890],[1234567890][1234567890]})

# MAC FILES e.g. .doc .ps etc.
# webdir = pickup
echo ''
  # Cannot handle a '[' in the file name.  The glob fails and exits.
foreach f (*.*) # excludes no-extension; starting with dot; being a dir
	if (! -e "$f") break  # should abort silently if no matches at all
	#if ($f =~ .*) continue   Excluded by *.* pattern above
	
	if (-x "$f") then
		echo "'$f' is executable '$f'."
	endif

	set ftail = "$f:e"  # file extension
	if ("$ftail" == '') continue # This catches patterns above with no .
	if ("$ftail" == 'bak') continue # Expected, allowed; not to be moved
	if ("$ftail" == 'old') continue # allowed; not to be moved

	## Main section:  Only process files with extensions of 4 or less
	set l = (`echo -n $ftail | wc -m;`) # -n don't count LF as a char
			#if ("$ftail" =~ ???  ||  "$ftail" =~ ????)
	if ($l <= 4) then  # Look at all extensions of 4 chars or less.
		set lcftail = (` echo $ftail |  tr '[:upper:]' '[:lower:]' `)
			#edit ftail to lcftail

		set doit = 0
		foreach s ($suffixes) 
			# if approved suffix set bools to action it
			if ("$lcftail" == "$s") then
				set doit = 1
				break
			endif
		end
		set ignore = 0
		foreach fig ($toignore) 
			if ("$fig" == "$f") then
				set ignore = 1
				@ ignMax++;
			     set msg = "${tab}Ignoring suffix '$f:e' (in '$f')";
			     set msgignore[$ignMax] = "$msg";
			     break
			endif
		end

		if (! $doit  &&  ! $ignore) then
			set unapproved = ($unapproved .$ftail)
			continue
		endif
		if (! $doit) continue  # On to next file, w/o action.

		# ACTION
		echo "Moving '$f' to [$pickup]"
		chmod 644 "$f"  # so browser can see it
		mv "$f" $pickup
	else
		#echo  " File '$f' remains."
	endif
end


echo ''; echo '====================';
echo "Tacitly allowed and not fetched suffixes: ";
while ($ignMax)
        echo "$msgignore[1]";   shift msgignore;   @ ignMax--;
		# echo appends LF to each slot, when it is printed out.
end

echo ''; # Separate the two list-prints
echo "Unapproved and so not fetched suffixes:"
echo -n '	'; echo "$unapproved"
	#unset nonomatch
exit

ls -ad .*
rm .*.swp


NEAT TECH FEATURES IN THIS SCRIPT
1) Auto expansion of suffix patterns e.g. htm{,l} --> htm html
2) Avoid such expansion to deal with upper/lower case variants of suffixes, 
	by only testing lcase versions of what filenames turn up.
3) Collecting msgs of a given type into an array;
	and printing out together later on.
	Used indexed arrays for this.
4) Define var $tab with a tab in; to help formatting msgs;
	So each msg can be formattted with sps and tabs; and preserved as such.
5) Only need LF at end of each msg, so done on printout.