ublog

   1 #!/usr/bin/perl
   2
   3 #
   4 # A minimal distributed multiuser blogging software [well ... better: hack]
   5 # that is based on Git
   6 #
   7 # Copyright 2011 Daniel Borkmann <borkmann@gnumaniacs.org>
   8 # Subject to the GNU GPL, version 2.
   9 #
  10 # ublog via Git Benefits:
  11 #
  12 #     + Distributed for a better censorship resistance
  13 #     + Secure since all sites are static HTML
  14 #     + Fast since all sites are static HTML
  15 #     + Backuped, since DVCS is used
  16 #     + Easy usage and setup, single blog text file
  17 #     + Can be used with multiple authors
  18 #     + RSS per author, tag and full RSS
  19 #     + Automatic content creation triggered on Git push
  20 #
  21 # txt2html syntax: http://txt2html.sourceforge.net/
  22 # On Debian: apt-get install txt2html
  23 #
  24 # Usage, i.e.: ublog blog.txt /var/www/htdocs/
  25 #
  26
  27 use strict;
  28 use warnings;
  29
  30 use HTML::TextToHTML;
  31
  32 my $prognam = "ublog";
  33 my $version = "0.1";
  34
  35 my $line = 0;
  36 my $verbose = 0;
  37
  38 use constant {
  39         NONE => -1,
  40         HEADER => 0,
  41         FOOTER => 1,
  42         ABOUT => 2,
  43         SETTINGS => 3,
  44         AUTHOR => 4,
  45         ENTRY => 5,
  46 };
  47
  48 my @abbr = qw( Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec );
  49
  50 my $header_txt = "";
  51 my $footer_txt = "";
  52 my $about_txt = "";
  53
  54 my %authors;
  55 my %entries;
  56 my %settings;
  57 my %tagsh;
  58 my %got;
  59
  60 sub debug
  61 {
  62         my $info = shift;
  63         if ($verbose) {
  64                 print $info;
  65         }
  66 }
  67
  68 sub parse_settings
  69 {
  70         my $l = shift;
  71         $_ = $l;
  72         if (/^\s*title\s*=\s*(.*)\s*$/) {
  73                 $settings{title} = $1;
  74                 debug("Title: $settings{title}\n");
  75         } elsif (/^\s*entries_per_site\s*=\s*(\d+)\s*$/) {
  76                 $settings{entries} = $1;
  77                 debug("Entries/site: $settings{entries}\n");
  78         } elsif (/^\s*git_clone\s*=\s*(.*)\s*$/) {
  79                 $settings{clone} = $1;
  80                 debug("Clone URL: $settings{clone}\n");
  81         } elsif (/^\s*root_url\s*=\s*(.*)\s*$/) {
  82                 $settings{root} = $1;
  83                 debug("Root URL: $settings{root}\n");
  84         } else {
  85                 die "Syntax error in l.$line!\n";
  86         }
  87 }
  88
  89 sub check_settings
  90 {
  91         if (not $settings{title}) {
  92                 die "Syntax error! No blog title!\n";
  93         }
  94         if (not $settings{entries}) {
  95                 die "Syntax error! No entries/site defined!\n";
  96         }
  97         if (not $settings{clone}) {
  98                 die "Syntax error! No Git clone URL provided!\n";
  99         }
 100         if (not ($settings{clone} =~ /^git:\/\//)) {
 101                 die "Syntax error! False clone URL!\n";
 102         }
 103         if (not $settings{root}) {
 104                 die "Syntax error! No blog root URL provided!\n";
 105         }
 106         if (not ($settings{root} =~ /^http:\/\//)) {
 107                 die "Syntax error! False root URL!\n";
 108         }
 109 }
 110
 111 sub parse_author
 112 {
 113         my $author = shift;
 114         my $l = shift;
 115         $_ = $l;
 116         if (/^\s*name\s*=\s*(.*)\s*$/) {
 117                 $authors{$author}->{name} = $1;
 118                 debug("Name: ".$authors{$author}->{name}."\n");
 119         } elsif (/^\s*email\s*=\s*(.*)\s*$/) {
 120                 $authors{$author}->{email} = $1;
 121                 debug("E-Mail: ".$authors{$author}->{email}."\n");
 122         } elsif (/^\s*web\s*=\s*(.*)\s*$/) {
 123                 $authors{$author}->{web} = $1;
 124                 debug("Website: ".$authors{$author}->{web}."\n");
 125         } elsif (/^\s*show_nic\s*=\s*(0|1)\s*$/) {
 126                 $authors{$author}->{usenic} = $1;
 127                 debug("Use nic: ".$authors{$author}->{usenic}."\n");
 128         } else {
 129                 die "Syntax error in l.$line!\n";
 130         }
 131 }
 132
 133 sub check_author
 134 {
 135         my $author = shift;
 136         if (not $authors{$author}->{name}) {
 137                 die "Syntax error! No author name!\n";
 138         }
 139 }
 140
 141 sub parse_entry
 142 {
 143         my $time = shift;
 144         my $l = shift;
 145         $entries{$time}->{text} .= $l;
 146 }
 147
 148 sub check_entry
 149 {
 150         my $time = shift;
 151         my $nic = $entries{$time}->{author};
 152         my $good = 0;
 153         foreach (keys(%authors)) {
 154                 if ($_ eq $nic) {
 155                         $good = 1;
 156                 }
 157         }
 158         if (not $good) {
 159                 die "Syntax error! Wrong author given!\n";
 160         }
 161 }
 162
 163 sub check_sections
 164 {
 165         if (not $got{header}) {
 166                 die "Syntax error! No header section present!\n";
 167         }
 168         if (not $got{footer}) {
 169                 die "Syntax error! No footer section present!\n";
 170         }
 171         if (not $got{about}) {
 172                 die "Syntax error! No about section present!\n";
 173         }
 174         if (not $got{settings}) {
 175                 die "Syntax error! No settings section present!\n";
 176         }
 177         if (not $got{author}) {
 178                 die "Syntax error! No author section present!\n";
 179         }
 180         if (not $got{entry}) {
 181                 die "Syntax error! No entry section present!\n";
 182         }
 183 }
 184
 185 sub parse
 186 {
 187         my $blog = shift;
 188         my $state = NONE;
 189         my ($author, $time, @tags);
 190         open BLOG, "<", $blog, or die $!;
 191         while (<BLOG>) {
 192                 $line++;
 193                 next if (/^\s*#/ and $state == NONE);
 194                 next if (/^\s+$/ and $state == NONE);
 195                 if (/^\s*header\s*=\s*{\s*$/) {
 196                         die "Syntax error in l.$line!\n" if ($state != NONE);
 197                         $state = HEADER;
 198                         $got{header} = 1;
 199                         debug("Found header!\n");
 200                 } elsif (/^\s*footer\s*=\s*{\s*$/) {
 201                         die "Syntax error in l.$line!\n" if ($state != NONE);
 202                         $state = FOOTER;
 203                         $got{footer} = 1;
 204                         debug("Found footer!\n");
 205                 } elsif (/^\s*about\s*=\s*{\s*$/) {
 206                         die "Syntax error in l.$line!\n" if ($state != NONE);
 207                         $state = ABOUT;
 208                         $got{about} = 1;
 209                         debug("Found about!\n");
 210                 } elsif (/^\s*settings\s*=\s*{\s*$/) {
 211                         die "Syntax error in l.$line!\n" if ($state != NONE);
 212                         $state = SETTINGS;
 213                         $got{settings} = 1;
 214                         debug("Found settings!\n");
 215                 } elsif (/^\s*author\s+(\w+)\s*=\s*{\s*$/) {
 216                         die "Syntax error in l.$line!\n" if ($state != NONE);
 217                         $state = AUTHOR;
 218                         $got{author} = 1;
 219                         $author = $1;
 220                         debug("Found author '$author'!\n");
 221                 } elsif (/^\s*entry\s+(\d+)\s+(\w+)\s+\[([\w\s,]*)\]\s*=\s*{\s*$/) {
 222                         die "Syntax error in l.$line!\n" if ($state != NONE);
 223                         $state = ENTRY;
 224                         $got{entry} = 1;
 225                         $time = $1;
 226                         $author = $2;
 227                         @tags = split(/,\s*/, $3);
 228                         $entries{$time}->{author} = $author;
 229                         push @{$entries{$time}->{tags}}, @tags;
 230                         $entries{$time}->{text} = "";
 231                         foreach my $tag (@tags) {
 232                                 $tagsh{$tag}++;
 233                         }
 234                         debug("Found header with '$time', '$author', '@tags'!\n");
 235                 } elsif (/^\s*}\s*$/) {
 236                         die "Syntax error in l.$line!\n" if ($state == NONE);
 237                         $state = NONE;
 238                         debug("Found close!\n");
 239                 } elsif ($state != NONE) {
 240                         if ($state == HEADER) {
 241                                 $header_txt .= $_;
 242                         } elsif ($state == FOOTER) {
 243                                 $footer_txt .= $_;
 244                         } elsif ($state == ABOUT) {
 245                                 $about_txt .= $_;
 246                         } elsif ($state == SETTINGS) {
 247                                 parse_settings($_);
 248                         } elsif ($state == AUTHOR) {
 249                                 parse_author($author, $_);
 250                         } elsif ($state == ENTRY) {
 251                                 parse_entry($time, $_);
 252                         } else {
 253                                 die "Wrong state in l.$line!\n";
 254                         }
 255                 } else {
 256                         die "Syntax error in l.$line!\n";
 257                 }
 258         }
 259         close BLOG;
 260 }
 261
 262 sub check_content
 263 {
 264         check_sections();
 265         check_settings();
 266         foreach (keys(%authors)) {
 267                 check_author($_);
 268         }
 269         foreach (keys(%entries)) {
 270                 check_entry($_);
 271         }
 272 }
 273
 274 sub generate_index
 275 {
 276         my $folder = shift;
 277         my $conv = new HTML::TextToHTML();
 278         my $sites = int(scalar(keys(%entries)) / $settings{entries}) + 1;
 279         my $last = scalar(keys(%entries)) % $settings{entries};
 280         my @keys = sort {$b <=> $a} keys(%entries);
 281         my @ldate = localtime(0);
 282
 283         if ($last == 0) {
 284                 if ($sites > 1) {
 285                         $sites--;
 286                         $last = $settings{entries};
 287                 }
 288         }
 289
 290         for (my $i = 0; $i < $sites; $i++) {
 291                 my $output = "";
 292                 my ($file, $number);
 293
 294                 $output .= "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\">\n";
 295                 $output .= "<html><head>\n";
 296                 $output .= "<title>".$settings{title}."</title>\n";
 297                 $output .= "<link href=\"style.css\" rel=\"stylesheet\" type=\"text/css\">";
 298                 $output .= "</head><body><h2>".$settings{title}."</h2>\n";
 299                 $output .= $conv->process_chunk($header_txt);
 300                 $output .= "<a href=\"".$settings{root}."/about.html\">About</a>, ";
 301                 $output .= "<a href=\"".$settings{root}."/feed.xml\">RSS</a>, ";
 302                 $output .= "Git: <a href=\"".$settings{clone}."\">".$settings{clone}."</a>";
 303                 $output .= "<br>\n<ul>";
 304
 305                 if ($i == $sites - 1) {
 306                         $number = $last;
 307                 } else {
 308                         $number = $settings{entries};
 309                 }
 310                 for (my $j = 0; $j < $number; $j++) {
 311                         my $key = shift @keys;
 312                         my $pre = "[<a href=\"".$settings{root}."/$key.html\">p</a>, ";
 313                         my $author = $entries{$key}->{author};
 314                         my @cdate = localtime($key);
 315
 316                         if ($authors{$author}->{usenic}) {
 317                                 $pre .= "<a href=\"".$settings{root}."/a_$author.xml\">$author</a>";
 318                         } else {
 319                                 $pre .= "<a href=\"".$settings{root}."/a_$author.xml\">".$authors{$author}->{name}."</a>";
 320                         }
 321                         if ($authors{$author}->{email} or $authors{$author}->{web}) {
 322                                 my $elem = 0;
 323                                 $pre .= " (";
 324                                 if ($authors{$author}->{email}) {
 325                                         $pre .= "<a href=\"mailto:".$authors{$author}->{email}."\">e</a>";
 326                                         $elem++;
 327                                 }
 328                                 if ($authors{$author}->{web}) {
 329                                         $pre .= ", " if $elem > 0;
 330                                         $pre .= "<a href=\"".$authors{$author}->{web}."\">w</a>";
 331                                 }
 332                                 $pre .= ")";
 333                         }
 334                         if (scalar(@{$entries{$key}->{tags}}) > 0) {
 335                                 $pre .= ", tags: ";
 336                                 foreach my $tag (@{$entries{$key}->{tags}}) {
 337                                         $pre .= "<a href=\"".$settings{root}."/t_$tag.xml\">$tag</a> ";
 338                                 }
 339                         }
 340                         $pre .= "]";
 341
 342                         if ($ldate[3] != $cdate[3] || $ldate[4] != $cdate[4] ||
 343                             $ldate[5] != $cdate[5]) {
 344                                 @ldate = @cdate;
 345                                 $output .= "\n</ul>\n<h3>$abbr[$ldate[4]] $ldate[3], ".(1900 + $ldate[5])."</h3>\n<ul>";
 346                         }
 347
 348                         $output .= "\n<li>$pre".$conv->process_chunk($entries{$key}->{text}, is_fragment => 1)."</li>";
 349                 }
 350
 351                 if ($i == $sites - 1) {
 352                         $output .= "\n</ul>\n<hr>\n";
 353                 } else {
 354                         $output .= "</ul><a href=\"".$settings{root}."/index_".
 355                                    ($i + 1).".html\">Next</a><hr>\n";
 356                 }
 357                 $output .= $conv->process_chunk($footer_txt);
 358                 $output .= "</body></html>";
 359
 360                 if ($i == 0) {
 361                         $file = "index.html";
 362                 } else {
 363                         $file = "index_$i.html";
 364                 }
 365
 366                 open OUT, ">", "$folder/$file" or die $!;
 367                 print OUT $output;
 368                 close OUT;
 369         }
 370 }
 371
 372 sub generate_about
 373 {
 374         my $folder = shift;
 375         my $conv = new HTML::TextToHTML();
 376         my $output = "";
 377
 378         $output .= "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\">\n";
 379         $output .= "<html><head>\n";
 380         $output .= "<title>".$settings{title}."</title>\n";
 381         $output .= "<link href=\"style.css\" rel=\"stylesheet\" type=\"text/css\">";
 382         $output .= "</head><body><h2>".$settings{title}."</h2>\n";
 383         $output .= $conv->process_chunk($about_txt);
 384         $output .= "<a href=\"".$settings{root}."/index.html\">Blog</a>, ";
 385         $output .= "<a href=\"".$settings{root}."/feed.xml\">RSS</a>, ";
 386         $output .= "Git: <a href=\"".$settings{clone}."\">".$settings{clone}."</a>";
 387         $output .= "<br><hr>\n";
 388         $output .= $conv->process_chunk($footer_txt);
 389         $output .= "</body></html>";
 390
 391         open OUT, ">", "$folder/about.html" or die $!;
 392         print OUT $output;
 393         close OUT;
 394 }
 395
 396 sub generate_entries
 397 {
 398         my $folder = shift;
 399         my $conv = new HTML::TextToHTML();
 400
 401         foreach my $key (keys(%entries)) {
 402                 my $output = "";
 403                 my $author = $entries{$key}->{author};
 404                 my @ldate = localtime($key);
 405                 my $pre = "[<a href=\"".$settings{root}."/$key.html\">p</a>, ";
 406
 407                 if ($authors{$author}->{usenic}) {
 408                         $pre .= "<a href=\"".$settings{root}."/a_$author.xml\">$author</a>";
 409                 } else {
 410                         $pre .= "<a href=\"".$settings{root}."/a_$author.xml\">".$authors{$author}->{name}."</a>";
 411                 }
 412                 if ($authors{$author}->{email} or $authors{$author}->{web}) {
 413                         my $elem = 0;
 414                         $pre .= " (";
 415                         if ($authors{$author}->{email}) {
 416                                 $pre .= "<a href=\"mailto:".$authors{$author}->{email}."\">e</a>";
 417                                 $elem++;
 418                         }
 419                         if ($authors{$author}->{web}) {
 420                                 $pre .= ", " if $elem > 0;
 421                                 $pre .= "<a href=\"".$authors{$author}->{web}."\">w</a>";
 422                         }
 423                         $pre .= ")";
 424                 }
 425                 if (scalar(@{$entries{$key}->{tags}}) > 0) {
 426                         $pre .= ", tags: ";
 427                         foreach my $tag (@{$entries{$key}->{tags}}) {
 428                                 $pre .= "<a href=\"".$settings{root}."/t_$tag.xml\">$tag</a> ";
 429                         }
 430                 }
 431                 $pre .= "]";
 432
 433                 $output .= "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\">\n";
 434                 $output .= "<html><head>\n";
 435                 $output .= "<title>".$settings{title}."</title>\n";
 436                 $output .= "<link href=\"style.css\" rel=\"stylesheet\" type=\"text/css\">";
 437                 $output .= "</head><body><h2>".$settings{title}."</h2>\n";
 438                 $output .= $conv->process_chunk($header_txt);
 439                 $output .= "<a href=\"".$settings{root}."/about.html\">About</a>, ";
 440                 $output .= "<a href=\"".$settings{root}."/feed.xml\">RSS</a>, ";
 441                 $output .= "Git: <a href=\"".$settings{clone}."\">".$settings{clone}."</a>";
 442                 $output .= "<br>\n";
 443                 $output .= "<h3>$abbr[$ldate[4]] $ldate[3], ".(1900 + $ldate[5])."</h3>\n<ul>";
 444                 $output .= "\n<li>$pre".$conv->process_chunk($entries{$key}->{text}, is_fragment => 1)."</li>";
 445                 $output .= "\n</ul>\n<hr>\n";
 446                 $output .= $conv->process_chunk($footer_txt);
 447                 $output .= "</body></html>";
 448
 449                 open OUT, ">", "$folder/$key.html" or die $!;
 450                 print OUT $output;
 451                 close OUT;
 452         }
 453 }
 454
 455 sub generate_full_rss
 456 {
 457         my $folder = shift;
 458         my $output = "";
 459         my @keys = sort {$b <=> $a} keys(%entries);
 460
 461         $output .= "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 462         $output .= "<rss version=\"2.0\">\n";
 463         $output .= "<channel>\n";
 464         $output .= "<title>".$settings{title}."</title>\n";
 465         $output .= "<link>".$settings{root}."</link>\n";
 466         $output .= "<description>".$settings{title}.", all ramblings</description>\n";
 467         $output .= "<language>en</language>\n";
 468
 469         foreach my $key (@keys) {
 470                 $output .= "<item>\n";
 471                 $output .= "<title>".$entries{$key}->{text}."</title>\n";
 472                 $output .= "<link>".$settings{root}."/$key.html</link>\n";
 473                 $output .= "<guid>".$settings{root}."/$key.html</guid>\n";
 474                 $output .= "</item>\n";
 475         }
 476
 477         $output .= "</channel>\n";
 478         $output .= "</rss>\n";
 479
 480         open OUT, ">", "$folder/feed.xml" or die $!;
 481         print OUT $output;
 482         close OUT;
 483 }
 484
 485 sub generate_author_rss
 486 {
 487         my $folder = shift;
 488         my @keys = sort {$b <=> $a} keys(%entries);
 489         my %aitems;
 490
 491         foreach my $key (@keys) {
 492                 my $author = $entries{$key}->{author};
 493                 if (not $aitems{$author}) {
 494                         $aitems{$author} = "";
 495                 }
 496                 $aitems{$author} .= "<item>\n";
 497                 $aitems{$author} .= "<title>".$entries{$key}->{text}."</title>\n";
 498                 $aitems{$author} .= "<link>".$settings{root}."/$key.html</link>\n";
 499                 $aitems{$author} .= "<guid>".$settings{root}."/$key.html</guid>\n";
 500                 $aitems{$author} .= "</item>\n";
 501         }
 502
 503         foreach my $author (keys(%aitems)) {
 504                 my $output = "";
 505                 $output .= "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 506                 $output .= "<rss version=\"2.0\">\n";
 507                 $output .= "<channel>\n";
 508                 $output .= "<title>".$settings{title}."</title>\n";
 509                 $output .= "<link>".$settings{root}."</link>\n";
 510                 $output .= "<description>".$settings{title}.", $author\'s ramblings</description>\n";
 511                 $output .= "<language>en</language>\n";
 512                 $output .= $aitems{$author};
 513                 $output .= "</channel>\n";
 514                 $output .= "</rss>\n";
 515
 516                 open OUT, ">", "$folder/a_$author.xml" or die $!;
 517                 print OUT $output;
 518                 close OUT;
 519         }
 520 }
 521
 522 sub generate_tag_rss
 523 {
 524         my $folder = shift;
 525         my @keys = sort {$b <=> $a} keys(%entries);
 526         my %titems;
 527
 528         foreach my $key (@keys) {
 529                 my @tags = @{$entries{$key}->{tags}};
 530                 foreach my $tag (@tags) {
 531                         if (not $titems{$tag}) {
 532                                 $titems{$tag} = "";
 533                         }
 534                         $titems{$tag} .= "<item>\n";
 535                         $titems{$tag} .= "<title>".$entries{$key}->{text}."</title>\n";
 536                         $titems{$tag} .= "<link>".$settings{root}."/$key.html</link>\n";
 537                         $titems{$tag} .= "<guid>".$settings{root}."/$key.html</guid>\n";
 538                         $titems{$tag} .= "</item>\n";
 539                 }
 540         }
 541
 542         foreach my $tag (keys(%titems)) {
 543                 my $output = "";
 544                 $output .= "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
 545                 $output .= "<rss version=\"2.0\">\n";
 546                 $output .= "<channel>\n";
 547                 $output .= "<title>".$settings{title}."</title>\n";
 548                 $output .= "<link>".$settings{root}."</link>\n";
 549                 $output .= "<description>".$settings{title}.", $tag-tagged ramblings</description>\n";
 550                 $output .= "<language>en</language>\n";
 551                 $output .= $titems{$tag};
 552                 $output .= "</channel>\n";
 553                 $output .= "</rss>\n";
 554
 555                 open OUT, ">", "$folder/t_$tag.xml" or die $!;
 556                 print OUT $output;
 557                 close OUT;
 558         }
 559 }
 560
 561 sub generate_content
 562 {
 563         my $folder = shift;
 564         mkdir($folder, 0777);
 565         generate_index($folder);
 566         generate_about($folder);
 567         generate_entries($folder);
 568         # Yeah, sloppy and inefficient
 569         generate_full_rss($folder);
 570         generate_author_rss($folder);
 571         generate_tag_rss($folder);
 572 }
 573
 574 sub main
 575 {
 576         my $blog = shift;
 577         my $folder = shift;
 578         print "Parsing $blog blog ...\n";
 579         parse($blog);
 580         print "Checking ...\n";
 581         check_content();
 582         print "Generate html files in $folder ...\n";
 583         generate_content($folder);
 584         print "Done!\n";
 585 }
 586
 587 sub help
 588 {
 589         print "\n$prognam $version\n";
 590         print "http://gnumaniacs.org\n\n";
 591         print "Usage: $prognam <blog-text-file> <output-folder>\n\n";
 592         print "Please report bugs to <borkmann\@gnumanics.org>\n";
 593         print "Copyright (C) 2011 Daniel Borkmann <borkmann\@gnumanics.org>,\n";
 594         print "License: GNU GPL version 2\n";
 595         print "This is free software: you are free to change and redistribute it.\n";
 596         print "There is NO WARRANTY, to the extent permitted by law.\n\n";
 597
 598         exit;
 599 }
 600
 601 if ($#ARGV + 1 != 2) {
 602         help();
 603 }
 604
 605 main(@ARGV);