--- mail2lj/mail2lj.pl	2007/08/12 19:54:44	1.1
+++ mail2lj/mail2lj.pl	2007/08/14 05:38:33	1.5
@@ -42,6 +42,8 @@
 # Original script seems to be distributed as freeware, so I stick to that
 # decision.  No warranty whatsoever, of course - use at your own risk ;-).
 #
+# Changes by Boris Veytsman - added --cut option
+#
 # ------------------------------------------------------------------------
 
 use	strict ;
@@ -115,6 +117,7 @@ my %tr = (
 
 # ------------------------------------------------------------------------ #
 # Changed by LG - added parsing of command line.
+# Changed by BV - added options cut
 # ------------------------------------------------------------------------ #
 my 	%Opt = (); 			# Main options go here
 my 	$opt_h ; 			# Help flag
@@ -123,6 +126,9 @@ my 	$opt_addfrom ; 			# Add the From fie
 my 	$opt_addfromh ; 		# Add the htmlized From to the post
 my 	$opt_keepspaces ; 		# HTML-encode multiple spaces in e-mail
 my 	@opt_taglist ; 			# command-line taglist first goes here
+my 	$opt_ljcut ;    		# Add lj-cut after line number N
+my 	$ljcut_delta = 5 ; 		# No lj-cut if less lines left after it
+my 	$opt_ljcut_text ; 		# A text for lj-cut.
 my 	$Parse = GetOptions( \%Opt,
 			'user|u=s',
 			'password|passwd|p=s',
@@ -133,6 +139,7 @@ my 	$Parse = GetOptions( \%Opt,
 			'prop_opt_backdated|backdated|back-dated|backdate|back-date|back!',
 			'subject|subj|s=s',
 			'taglist|tags|tag|t=s' => \@opt_taglist,  # Will tweak
+			'notaglist|notags|notag|not|no-taglist|no-tags|no-tag|no-t' => sub {undef @opt_taglist},
 			'usejournal|use-journal|use|journal|j=s',
 			'prop_current_mood|current_mood|mood=s',
 			'prop_current_music|current_music|music=s',
@@ -142,6 +149,8 @@ my 	$Parse = GetOptions( \%Opt,
 			'bounces|bounce|b=s' => \$opt_bounces,
 			'addfrom|add-from|from!' => \$opt_addfrom,
 			'addfromh|add-fromh|fromh!' => \$opt_addfromh,
+			'ljcut|lj-cut|cut|l=i'=>\$opt_ljcut,
+			'ljcut-text|lj-cut-text|cut-text|ljcuttext|cuttext=s'=>\$opt_ljcut_text,
 			'keep-spaces|keep-space|keepspaces|keepspace|spaces|space!' => \$opt_keepspaces,
 			'help|h' => \$opt_h,
 			   );
@@ -198,7 +207,13 @@ if ( exists $Opt{'comments'} ) {
 # with other parameters.
 $Opt{'prop_taglist'} = join( ", ", @opt_taglist )  if ( @opt_taglist ) ;
 
-# Convert all command line options to unicode.
+# Convert $opt_ljcut_text to UTF8.
+if ( defined $opt_ljcut_text ) {
+   $opt_ljcut_text = 
+	to_utf8({ -string => $opt_ljcut_text, -charset => $SystemCharset }) ;
+}
+
+# Convert all %Opt command line options to unicode.
 # Function href2utf8() uses a reference to input hash, so %Opt is
 # being modified "in-place".
 href2utf8( \%Opt, $SystemCharset) ;
@@ -215,7 +230,7 @@ my	$mp = new MIME::Parser() or die "new
 
 # Changed by LG - changed directory.
 # $mp->output_dir("$home/mimetmp") ;
-$mp->output_dir("/tmp/mimetmp-$ENV{user}") ;
+$mp->output_dir("/tmp/mimetmp-".$ENV{USER}) ;
 mkdir $mp->output_dir if not -d $mp->output_dir ; 	# Create it if missing
 
 # Get the whole mail.
@@ -514,7 +529,12 @@ sub	post_body2href {
 
 		# Changed by LG - added 'tags' option.  
 		} elsif ($var =~ /^tags?$/ || $var eq "taglist") {
-			$req_data->{prop_taglist} = $val;
+		    $req_data->{prop_taglist} = $val;
+
+		# Changed by LG - added 'notags' option.  Empty the preceding
+		# taglist if set to true, otherwise do nothing 
+		} elsif ($var =~ /^no-?tags?$/ || $var eq "no-?taglist") {
+		    $req_data->{prop_taglist} = "" if $val =~ /^\s*((on)|(yes))\s*$/i ;
 
 		# Anything else - just assign.
 		} else {
@@ -590,8 +610,18 @@ sub	post_me2req {
 
 	# Changed by LG - added options to add the 'From' field to the
 	# posted message.
+	# 
+	# NOTE: $from is already in UTF8.  Strictly speaking, everything
+	# 	that we add to it MUST ALSO BE IN UTF8 (i.e. you need to run
+	# 	a to_utf8() function on it).  But since all I'm adding is in
+	# 	ISO-8859-1 lower ASCII characters (which are guaranteed to
+	# 	have the same values in UTF8 as in plain ISO-8859-1), I'm
+	# 	cheating here and taking a shortcut. If you want to add
+	# 	something non-ASCII, you MUST convert it to UTF8 first!
+	# 	Be forewarned!
 	if ( $opt_addfrom ) {
 	   $hr->{event} = "From: $from" . "\n\n" . $hr->{event} ;
+	   $hr->{event} = $plain_from . $hr->{event} ;
 	} elsif ( $opt_addfromh ) {
 	   my $html_from = "<nobr><i><b>From:</b> $from</i></nobr>" ;
 	   $html_from =~ s/\@/[_\@_]/g ;
@@ -607,6 +637,36 @@ sub	post_me2req {
 	   $hr->{event} =~ s/\t/\&nbsp;\&nbsp;\&nbsp;\&nbsp;\&nbsp;\&nbsp;\&nbsp;\&nbsp;/g ;
 	   $hr->{event} =~ s/  / \&nbsp;/g ;
 	}
+	
+	#
+	# Change by BV - added the option to put lj-cut after '--cut XX' lines
+	#
+	# Tweaked by LG - only adding lj-cut if more than $ljcut_delta lines
+	# is left in the posting.
+	#
+	if ($opt_ljcut>0) {
+	    my $nlines = scalar( my @junk=split( /\n/, $hr->{event}, -1) ) - 1;
+	    my $start=0;
+	    for (my $i=0; $i<$opt_ljcut; $i++) {
+		$start=index($hr->{event},"\n",$start)+1;
+		if ($start == 0) {
+		    last;
+		}
+	    }
+	    # And insert the lj-cut if not too close to the end of the post.
+	    if ($start>0 ) {
+		if ( $nlines >= $opt_ljcut+$ljcut_delta ) {
+		   my $ljcut = ( $opt_ljcut_text =~ /^\s*$/ ) ?
+				'<lj-cut>' :
+				'<lj-cut text="' . $opt_ljcut_text . '">' ;
+		   substr($hr->{event}, $start,0) = $ljcut ;
+		} else {
+		   print STDERR "'--cut $opt_ljcut' requested, which is " .
+				"within $ljcut_delta of the total $nlines " .
+				"lines. Skipping lj-cut.\n" ;
+		}
+	    }
+	}
 
 	$req->content_type('application/x-www-form-urlencoded');
 	$req->content(href2string $hr) ;
@@ -860,6 +920,12 @@ Options:
                single or double quotes to protect from the shell.  Multiple
                '-t' options are allowed and taglists will be combined.
 
+--notaglist, --notags
+               Unsets all previously defined tags.  Thus, a call to 
+                  $shortname ... --tags X --tags Y ... --notags --tags Z
+               will yield a taglist consisting of just "Z".  This option is
+               rarely needed and added only for the sake of completeness.
+
 -d DATE, --date DATE
                Label posting with this date.  Date should be in LiveJournal's
                format: DD.MM.YYYY HH:mm.  If absent, current date/time is used.
@@ -923,6 +989,16 @@ Options:
                better preserved in the journal.  The option can be negated
                ('--nospaces').  Default is '--nospaces'.
 
+--ljcut NUM, --cut NUM, -l NUM
+               Inserts '<lj-cut>' after NUM lines of the post content.
+               If the resulting lj-cut happens to be within $ljcut_delta lines from
+               the end of the post, the cut will not be added.
+
+--ljcut-text TEXT, --cut-text TEXT, --cuttext TEXT
+               Text to use as lj-cut text parameter (in <lj-cut text="TEXT">).
+               If the text contains nothing but whitespace, it is ignored.
+               Remember to quote spaces and special characters from the shell.
+
 --charset CHARSET
                This option tells the script that all COMMAND LINE options are
                given in this charset.  Default is "$SystemCharset".
@@ -931,7 +1007,7 @@ Options:
                utf8).  It also has absolutely no effect on the in-the-body
                keywords (they are also governed by email's charset).  This
                option is meaningful ONLY for the text that you supply VIA
-               COMMAND LINE (e.g. '-s Subject').
+               COMMAND LINE (e.g. '-s Subject' or '--cuttext TEXT').
                
 -b xxx\@yyy, --bounces xxx\@yyy
                Normally, if errors occur during posting (e.g. wrong password),
@@ -959,6 +1035,7 @@ command line options), they should look
 	Security: private
 	Subject: Rzhevskij zhiv!
 	Tags:  Junk, Viva Rzhevskij!
+	Notags: yes 			# Clears all preceding tags
 	Formatted: on 			# Or equivalent "Autoformat: off"
 	Usejournal: gusary
 	Mood: okay