#!/usr/bin/perl -T

use warnings;
use strict;
use File::Find;
use Memoize;
use File::Spec;

$ENV{PATH}="/usr/local/bin:/usr/bin:/bin";

BEGIN {
	$blosxom::version="is a proper perl module too much to ask?";
	do "/usr/bin/markdown";
}

my ($srcdir, $destdir, %links, %oldlinks, %oldpagemtime, %renderedfiles,
    %pagesources);
my $wiki_link_regexp=qr/\[\[([^\s]+)\]\]/;
my $wiki_file_regexp=qr/(^[-A-Za-z0-9_.:\/+]+$)/;
my $wiki_file_prune_regexp=qr!((^|/).svn/|\.\.|^\.|\/\.|\.html?$)!;
my $verbose=0;
my $wikiname="wiki";
my $default_pagetype=".mdwn";
my $cgi=0;
my $url="";
my $cgiurl="";
my $svn=1;

sub usage {
	die "usage: ikiwiki [options] source dest\n";
}

sub error ($) {
	if ($cgi) {
		print "Content-type: text/html\n\n";
		print "Error: @_\n";
		exit 1;
	}
	else {
		die @_;
	}
}

sub debug ($) {
	print "@_\n" if $verbose;
}

sub mtime ($) {
	my $page=shift;
	
	return (stat($page))[9];
}

sub possibly_foolish_untaint ($) {
	my $tainted=shift;
	my ($untainted)=$tainted=~/(.*)/;
	return $untainted;
}

sub basename {
	my $file=shift;

	$file=~s!.*/!!;
	return $file;
}

sub dirname {
	my $file=shift;

	$file=~s!/?[^/]+$!!;
	return $file;
}

sub pagetype ($) {
	my $page=shift;
	
	if ($page =~ /\.mdwn$/) {
		return ".mdwn";
	}
	else {
		return "unknown";
	}
}

sub pagename ($) {
	my $file=shift;

	my $type=pagetype($file);
	my $page=$file;
	$page=~s/\Q$type\E*$// unless $type eq 'unknown';
	return $page;
}

sub htmlpage ($) {
	my $page=shift;

	return $page.".html";
}

sub readfile ($) {
	my $file=shift;

	local $/=undef;
	open (IN, "$file") || error("failed to read $file: $!");
	my $ret=<IN>;
	close IN;
	return $ret;
}

sub writefile ($$) {
	my $file=shift;
	my $content=shift;

	my $dir=dirname($file);
	if (! -d $dir) {
		my $d="";
		foreach my $s (split(m!/+!, $dir)) {
			$d.="$s/";
			if (! -d $d) {
				mkdir($d) || error("failed to create directory $d: $!");
			}
		}
	}
	
	open (OUT, ">$file") || error("failed to write $file: $!");
	print OUT $content;
	close OUT;
}

sub findlinks {
	my $content=shift;

	my @links;
	while ($content =~ /$wiki_link_regexp/g) {
		push @links, lc($1);
	}
	return @links;
}

# Given a page and the text of a link on the page, determine which existing
# page that link best points to. Prefers pages under a subdirectory with
# the same name as the source page, failing that goes down the directory tree
# to the base looking for matching pages.
sub bestlink ($$) {
	my $page=shift;
	my $link=lc(shift);
	
	my $cwd=$page;
	do {
		my $l=$cwd;
		$l.="/" if length $l;
		$l.=$link;

		if (exists $links{$l}) {
			#debug("for $page, \"$link\", use $l");
			return $l;
		}
	} while $cwd=~s!/?[^/]+$!!;

	#print STDERR "warning: page $page, broken link: $link\n";
	return "";
}

sub isinlinableimage ($) {
	my $file=shift;
	
	$file=~/\.(png|gif|jpg|jpeg)$/;
}

sub htmllink ($$) {
	my $page=shift;
	my $link=shift;

	my $bestlink=bestlink($page, $link);

	return $link if $page eq $bestlink;
	
	# TODO BUG: %renderedfiles may not have it, if the linked to page
	# was also added and isn't yet rendered! Note that this bug is
	# masked by the bug mentioned below that makes all new files
	# be rendered twice.
	if (! grep { $_ eq $bestlink } values %renderedfiles) {
		$bestlink=htmlpage($bestlink);
	}
	if (! grep { $_ eq $bestlink } values %renderedfiles) {
		return "<a href=\"$cgiurl?do=create&page=$link&from=$page\">?</a>$link"
	}
	
	$bestlink=File::Spec->abs2rel($bestlink, dirname($page));
	
	if (isinlinableimage($bestlink)) {
		return "<img src=\"$bestlink\">";
	}
	return "<a href=\"$bestlink\">$link</a>";
}

sub linkify ($$) {
	my $content=shift;
	my $file=shift;

	$content =~ s/$wiki_link_regexp/htmllink(pagename($file), $1)/eg;
	
	return $content;
}

sub htmlize ($$) {
	my $type=shift;
	my $content=shift;
	
	if ($type eq '.mdwn') {
		return Markdown::Markdown($content);
	}
	else {
		error("htmlization of $type not supported");
	}
}

sub linkbacks ($$) {
	my $content=shift;
	my $page=shift;

	my @links;
	foreach my $p (keys %links) {
		next if bestlink($page, $p) eq $page;
		if (grep { length $_ && bestlink($p, $_) eq $page } @{$links{$p}}) {
			my $href=File::Spec->abs2rel(htmlpage($p), dirname($page));
			
			# Trim common dir prefixes from both pages.
			my $p_trimmed=$p;
			my $page_trimmed=$page;
			my $dir;
			1 while (($dir)=$page_trimmed=~m!^([^/]+/)!) &&
			        defined $dir &&
			        $p_trimmed=~s/^\Q$dir\E// &&
			        $page_trimmed=~s/^\Q$dir\E//;
				       
			push @links, "<a href=\"$href\">$p_trimmed</a>";
		}
	}

	$content.="<hr><p>Links: ".join(" ", sort @links)."</p>\n" if @links;
	return $content;
}

sub finalize ($$) {
	my $content=shift;
	my $page=shift;

	my $title=basename($page);
	$title=~s/_/ /g;
	
	my $pagelink="";
	my $path="";
	foreach my $dir (reverse split("/", $page)) {
		if (length($pagelink)) {
			$pagelink="<a href=\"$path$dir.html\">$dir</a>/ $pagelink";
		}
		else {
			$pagelink=$dir;
		}
		$path.="../";
	}
	$path=~s/\.\.\/$/index.html/;
	$pagelink="<a href=\"$path\">$wikiname</a>/ $pagelink";
	
	my @actions;
	if (length $cgiurl) {
		push @actions, "<a href=\"$cgiurl?do=edit&page=$page\">Edit</a>";
		push @actions, "<a href=\"$cgiurl?do=recentchanges\">RecentChanges</a>";
	}
	
	$content="<html>\n<head><title>$title</title></head>\n<body>\n".
	          "<h1>$pagelink</h1>\n".
		  "@actions\n<hr>\n".
	          $content.
	          "</body>\n</html>\n";
	
	return $content;
}

sub render ($) {
	my $file=shift;
	
	my $type=pagetype($file);
	my $content=readfile("$srcdir/$file");
	if ($type ne 'unknown') {
		my $page=pagename($file);
		
		$links{$page}=[findlinks($content)];
		
		$content=linkify($content, $file);
		$content=htmlize($type, $content);
		$content=linkbacks($content, $page);
		$content=finalize($content, $page);
		
		writefile("$destdir/".htmlpage($page), $content);
		$oldpagemtime{$page}=time;
		$renderedfiles{$page}=htmlpage($page);
	}
	else {
		$links{$file}=[];
		writefile("$destdir/$file", $content);
		$oldpagemtime{$file}=time;
		$renderedfiles{$file}=$file;
	}
}

sub loadindex () {
	open (IN, "$srcdir/.index") || return;
	while (<IN>) {
		$_=possibly_foolish_untaint($_);
		chomp;
		my ($mtime, $file, $rendered, @links)=split(' ', $_);
		my $page=pagename($file);
		$pagesources{$page}=$file;
		$oldpagemtime{$page}=$mtime;
		$oldlinks{$page}=[@links];
		$links{$page}=[@links];
		$renderedfiles{$page}=$rendered;
	}
	close IN;
}	

sub saveindex () {
	open (OUT, ">$srcdir/.index") || error("cannot write to .index: $!");
	foreach my $page (keys %oldpagemtime) {
	print OUT "$oldpagemtime{$page} $pagesources{$page} $renderedfiles{$page} ".
	          join(" ", @{$links{$page}})."\n"
			if $oldpagemtime{$page};
	}
	close OUT;
}

sub rcs_update () {
	if (-d "$srcdir/.svn") {
		if (system("svn", "update", "--quiet", $srcdir) != 0) {
			warn("svn update failed\n");
		}
	}
}

sub rcs_commit ($) {
	my $message=shift;

	if (-d "$srcdir/.svn") {
		if (system("svn", "commit", "--quiet", "-m",
		           possibly_foolish_untaint($message), $srcdir) != 0) {
			warn("svn commit failed\n");
		}
	}
}

sub rcs_add ($) {
	my $file=shift;

	if (-d "$srcdir/.svn") {
		my $parent=dirname($file);
		while (! -d "$srcdir/$parent/.svn") {
			$file=$parent;
			$parent=dirname($file);
		}
		
		if (system("svn", "add", "--quiet", "$srcdir/$file") != 0) {
			warn("svn add failed\n");
		}
	}
}

sub rcs_recentchanges ($) {
	my $num=shift;
	my @ret;
	
	eval {use Date::Parse};
	eval {use Time::Duration};
	
	if (-d "$srcdir/.svn") {
		my $info=`LANG=C svn info $srcdir`;
		my ($svn_url)=$info=~/^URL: (.*)$/m;

		my $div=qr/^--------------------+$/;
		my $infoline=qr/^r(\d+)\s+\|\s+([^\s]+)\s+\|\s+(\d+-\d+-\d+\s+\d+:\d+:\d+\s+[-+]?\d+).*/;
		my $state='start';
		my ($rev, $user, $when, @pages, $message);
		foreach (`LANG=C svn log -v '$svn_url'`) {
			chomp;
			print STDERR "state: $state ($_)\n";
			if ($state eq 'start' && /$div/) {
				$state='header';
			}
			elsif ($state eq 'header' && /$infoline/) {
				$rev=$1;
				$user=$2;
				$when=ago(time - str2time($3));
		    	}
			elsif ($state eq 'header' && /^\s+(.*)/) {
				push @pages, $1;
			}
			elsif ($state eq 'header' && /^$/) {
				$state='body';
			}
			elsif ($state eq 'body' && /$div/) {
				push @ret, {rev => $rev, user => $user,
					date => $when, message => $message,
					pages => [@pages]},
				print STDERR "have: @ret\n";
				return @ret if @ret >= $num;
				
				$state='header';
				$message=$rev=$user=$when=undef;
				@pages=();
			}
			elsif ($state eq 'body') {
				$message.="$_<br>\n";
			}
		}
	}

	return @ret;
}

sub prune ($) {
	my $file=shift;

	unlink($file);
	my $dir=dirname($file);
	while (rmdir($dir)) {
		$dir=dirname($dir);
	}
}

sub refresh () {
	# Find existing pages.
	my %exists;
	my @files;
	find({
		no_chdir => 1,
		wanted => sub {
			if (/$wiki_file_prune_regexp/) {
				$File::Find::prune=1;
			}
			elsif (! -d $_) {
				my ($f)=/$wiki_file_regexp/; # untaint
				if (! defined $f) {
					warn("skipping bad filename $_\n");
				}
				else {
					$f=~s/^\Q$srcdir\E\/?//;
					push @files, $f;
					$exists{pagename($f)}=1;
				}
			}
		},
	}, $srcdir);

	my %rendered;

	# check for added or removed pages
	my @add;
	foreach my $file (@files) {
		my $page=pagename($file);
		if (! $oldpagemtime{$page}) {
			debug("new page $page");
			push @add, $file;
			$links{$page}=[];
			$pagesources{$page}=$file;
		}
	}
	my @del;
	foreach my $page (keys %oldpagemtime) {
		if (! $exists{$page}) {
			debug("removing old page $page");
			push @del, $renderedfiles{$page};
			prune($destdir."/".$renderedfiles{$page});
			delete $renderedfiles{$page};
			$oldpagemtime{$page}=0;
			delete $pagesources{$page};
		}
	}
	
	# render any updated files
	foreach my $file (@files) {
		my $page=pagename($file);
		
		if (! exists $oldpagemtime{$page} ||
		    mtime("$srcdir/$file") > $oldpagemtime{$page}) {
		    	debug("rendering changed file $file");
			render($file);
			$rendered{$file}=1;
		}
	}
	
	# if any files were added or removed, check to see if each page
	# needs an update due to linking to them
	# TODO: inefficient; pages may get rendered above and again here;
	# problem is the bestlink may have changed and we won't know until
	# now
	if (@add || @del) {
FILE:		foreach my $file (@files) {
			my $page=pagename($file);
			foreach my $f (@add, @del) {
				my $p=pagename($f);
				foreach my $link (@{$links{$page}}) {
					if (bestlink($page, $link) eq $p) {
		   				debug("rendering $file, which links to $p");
						render($file);
						$rendered{$file}=1;
						next FILE;
					}
				}
			}
		}
	}

	# handle linkbacks; if a page has added/removed links, update the
	# pages it links to
	# TODO: inefficient; pages may get rendered above and again here;
	# problem is the linkbacks could be wrong in the first pass render
	# above
	if (%rendered) {
		my %linkchanged;
		foreach my $file (keys %rendered, @del) {
			my $page=pagename($file);
			if (exists $links{$page}) {
				foreach my $link (@{$links{$page}}) {
					$link=bestlink($page, $link);
					if (length $link &&
					    ! exists $oldlinks{$page} ||
					    ! grep { $_ eq $link } @{$oldlinks{$page}}) {
						$linkchanged{$link}=1;
					}
				}
			}
			if (exists $oldlinks{$page}) {
				foreach my $link (@{$oldlinks{$page}}) {
					$link=bestlink($page, $link);
					if (length $link &&
					    ! exists $links{$page} ||
					    ! grep { $_ eq $link } @{$links{$page}}) {
						$linkchanged{$link}=1;
					}
				}
			}
		}
		foreach my $link (keys %linkchanged) {
		    	my $linkfile=$pagesources{$link};
			if (defined $linkfile) {
				debug("rendering $linkfile, to update its linkbacks");
				render($linkfile);
			}
		}
	}
}

# Generates a C wrapper program for running ikiwiki in a specific way.
# The wrapper may be safely made suid.
sub gen_wrapper ($$) {
	my ($svn, $rebuild)=@_;

	eval {use Cwd 'abs_path'};
	$srcdir=abs_path($srcdir);
	$destdir=abs_path($destdir);
	my $this=abs_path($0);
	if (! -x $this) {
		error("$this doesn't seem to be executable");
	}

	my @params=($srcdir, $destdir, "--wikiname=$wikiname");
	push @params, "--verbose" if $verbose;
	push @params, "--rebuild" if $rebuild;
	push @params, "--nosvn" if !$svn;
	push @params, "--cgi" if $cgi;
	push @params, "--url=$url" if $url;
	push @params, "--cgiurl=$cgiurl" if $cgiurl;
	my $params=join(" ", @params);
	my $call='';
	foreach my $p ($this, $this, @params) {
		$call.=qq{"$p", };
	}
	$call.="NULL";
	
	my @envsave;
	push @envsave, qw{REMOTE_ADDR QUERY_STRING REQUEST_METHOD REQUEST_URI
	               CONTENT_TYPE CONTENT_LENGTH GATEWAY_INTERFACE} if $cgi;
	my $envsave="";
	foreach my $var (@envsave) {
		$envsave.=<<"EOF"
	if ((s=getenv("$var")))
		asprintf(&newenviron[i++], "%s=%s", "$var", s);
EOF
	}
	
	open(OUT, ">ikiwiki-wrap.c") || error("failed to write ikiwiki-wrap.c: $!");;
	print OUT <<"EOF";
/* A wrapper for ikiwiki, can be safely made suid. */
#define _GNU_SOURCE
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>

extern char **environ;

int main (int argc, char **argv) {
	/* Sanitize environment. */
	char *s;
	char *newenviron[$#envsave+3];
	int i=0;
$envsave
	newenviron[i++]="HOME=$ENV{HOME}";
	newenviron[i]=NULL;
	environ=newenviron;

	if (argc == 2 && strcmp(argv[1], "--params") == 0) {
		printf("$params\\n");
		exit(0);
	}
	
	execl($call);
	perror("failed to run $this");
	exit(1);
}
EOF
	close OUT;
	if (system("gcc", "ikiwiki-wrap.c", "-o", "ikiwiki-wrap") != 0) {
		error("failed to compile ikiwiki-wrap.c");
	}
	unlink("ikiwiki-wrap.c");
	print "successfully generated ikiwiki-wrap\n";
	exit 0;
}

sub cgi () {
	eval q{use CGI};
	my $q=CGI->new;

	my $do=$q->param('do');
	if (! defined $do || ! length $do) {
		error("\"do\" parameter missing");
	}
	
	if ($do eq 'recentchanges') {
		my $list="<ul>\n";
		foreach my $change (rcs_recentchanges(100)) {
			print STDERR "change: $change....\n";
			$list.="<li>";
			foreach my $page (@{$change->{pages}}) {
				$list.=$page." ";
			}
			$list.="<br>\n";
			$list.="changed ".$change->{when}." by ".
			       htmllink("index", $change->{user}).
			       ": <i>".$change->{message}."<br>\n";
			$list.="</li>\n";
		}
		$list.="</ul>\n";
		
		print $q->header,
		      $q->start_html("RecentChanges"),
		      $q->h1("<a href=\"$url\">$wikiname</a>/ RecentChanges"),
		      $list,
		      $q->end_form,
		      $q->end_html;
		return;
	}
	
	my ($page)=$q->param('page')=~/$wiki_file_regexp/;
	if (! defined $page || ! length $page || $page ne $q->param('page') ||
	    $page=~/$wiki_file_prune_regexp/ || $page=~/^\//) {
		error("bad page name");
	}
	$page=lc($page);
	
	my $action=$q->request_uri;
	$action=~s/\?.*//;
	
	if ($do eq 'create') {
		if (exists $pagesources{lc($page)}) {
			# hmm, someone else made the page in the meantime?
			print $q->redirect("$url/".htmlpage($page));
		}

		my @page_locs;
		my ($from)=$q->param('from')=~/$wiki_file_regexp/;
		if (! defined $from || ! length $from ||
		    $from ne $q->param('from') ||
		    $from=~/$wiki_file_prune_regexp/ || $from=~/^\//) {
			@page_locs=$page;
		}
		else {
			my $dir=$from."/";
			$dir=~s![^/]+/$!!;
			push @page_locs, $dir.$page;
			push @page_locs, "$from/$page";
			while (length $dir) {
				$dir=~s![^/]+/$!!;
				push @page_locs, $dir.$page;
			}
		}
		
		$q->param("do", "save");
		print $q->header,
		      $q->start_html("Creating $page"),
		      $q->start_h1("<a href=\"$url\">$wikiname</a>/ Creating $page"),
		      $q->start_form(-action => $action),
		      $q->hidden('do'),
		      "Select page location:",
		      $q->popup_menu('page', \@page_locs),
		      $q->textarea(-name => 'content',
		               -default => "",
		               -rows => 20,
		               -columns => 80),
		      $q->br,
		      "Optional comment about this change:",
		      $q->br,
		      $q->textfield(-name => "comments", -size => 80),
		      $q->br,
		      $q->submit("Save Page"),
		      $q->end_form,
		      $q->end_html;
	}
	elsif ($do eq 'edit') {
		my $content="";
		if (exists $pagesources{lc($page)}) {
			$content=readfile("$srcdir/$pagesources{lc($page)}");
			$content=~s/\n/\r\n/g;
		}
		$q->param("do", "save");
		print $q->header,
		      $q->start_html("Editing $page"),
		      $q->h1("<a href=\"$url\">$wikiname</a>/ Editing $page"),
		      $q->start_form(-action => $action),
		      $q->hidden('do'),
		      $q->hidden('page'),
		      $q->textarea(-name => 'content',
		               -default => $content,
		               -rows => 20,
		               -columns => 80),
		      $q->br,
		      "Optional comment about this change:",
		      $q->br,
		      $q->textfield(-name => "comments", -size => 80),
		      $q->br,
		      $q->submit("Save Page"),
		      $q->end_form,
		      $q->end_html;
	}
	elsif ($do eq 'save') {
		my $file=$page.$default_pagetype;
		my $newfile=1;
		if (exists $pagesources{lc($page)}) {
			$file=$pagesources{lc($page)};
			$newfile=0;
		}
		
		my $content=$q->param('content');
		$content=~s/\r\n/\n/g;
		$content=~s/\r/\n/g;
		writefile("$srcdir/$file", $content);
		
		my $message="web commit from $ENV{REMOTE_ADDR}";
		if (defined $q->param('comments')) {
			$message.=": ".$q->param('comments');
		}
		
		if ($svn) {
			if ($newfile) {
				rcs_add($file);
			}
			# presumably the commit will trigger an update
			# of the wiki
			rcs_commit($message);
		}
		else {
			refresh();
		}
		
		print $q->redirect("$url/".htmlpage($page));
	}
	else {
		error("unknown do parameter");
	}
}

my $rebuild=0;
my $wrapper=0;
if (grep /^-/, @ARGV) {
	eval {use Getopt::Long};
	GetOptions(
		"wikiname=s" => \$wikiname,
		"verbose|v" => \$verbose,
		"rebuild" => \$rebuild,
		"wrapper" => \$wrapper,
		"svn!" => \$svn,
		"cgi" => \$cgi,
		"url=s" => \$url,
		"cgiurl=s" => \$cgiurl,
	) || usage();
}
usage() unless @ARGV == 2;
($srcdir) = possibly_foolish_untaint(shift);
($destdir) = possibly_foolish_untaint(shift);

if ($cgi && ! length $url) {
	error("Must specify url to wiki with --url when using --cgi");
}

gen_wrapper($svn, $rebuild) if $wrapper;
memoize('pagename');
memoize('bestlink');
loadindex() unless $rebuild;
if ($cgi) {
	cgi();
}
else {
	rcs_update() if $svn;
	refresh();
	saveindex();
}