contrib/mw-to-git/git-mw.perl

   1 #!/usr/bin/perl
   2
   3 # Copyright (C) 2013
   4 #     Benoit Person <benoit.person@ensimag.imag.fr>
   5 #     Celestin Matte <celestin.matte@ensimag.imag.fr>
   6 # License: GPL v2 or later
   7
   8 # Set of tools for git repo with a mediawiki remote.
   9 # Documentation & bugtracker: https://github.com/moy/Git-Mediawiki/
  10
  11 use strict;
  12 use warnings;
  13
  14 use Getopt::Long;
  15 use URI::URL qw(url);
  16 use LWP::UserAgent;
  17 use HTML::TreeBuilder;
  18
  19 use Git;
  20 use MediaWiki::API;
  21 use Git::Mediawiki qw(clean_filename connect_maybe
  22                                         EMPTY HTTP_CODE_PAGE_NOT_FOUND);
  23
  24 # By default, use UTF-8 to communicate with Git and the user
  25 binmode STDERR, ':encoding(UTF-8)';
  26 binmode STDOUT, ':encoding(UTF-8)';
  27
  28 # Global parameters
  29 my $verbose = 0;
  30 sub v_print {
  31         if ($verbose) {
  32                 return print {*STDERR} @_;
  33         }
  34         return;
  35 }
  36
  37 # Preview parameters
  38 my $file_name = EMPTY;
  39 my $remote_name = EMPTY;
  40 my $preview_file_name = EMPTY;
  41 my $autoload = 0;
  42 sub file {
  43         $file_name = shift;
  44         return $file_name;
  45 }
  46
  47 my %commands = (
  48         'help' =>
  49                 [\&help, {}, \&help],
  50         'preview' =>
  51                 [\&preview, {
  52                         '<>' => \&file,
  53                         'output|o=s' => \$preview_file_name,
  54                         'remote|r=s' => \$remote_name,
  55                         'autoload|a' => \$autoload
  56                 }, \&preview_help]
  57 );
  58
  59 # Search for sub-command
  60 my $cmd = $commands{'help'};
  61 for (0..@ARGV-1) {
  62         if (defined $commands{$ARGV[$_]}) {
  63                 $cmd = $commands{$ARGV[$_]};
  64                 splice @ARGV, $_, 1;
  65                 last;
  66         }
  67 };
  68 GetOptions( %{$cmd->[1]},
  69         'help|h' => \&{$cmd->[2]},
  70         'verbose|v'  => \$verbose);
  71
  72 # Launch command
  73 &{$cmd->[0]};
  74
  75 ############################# Preview Functions ################################
  76
  77 sub preview_help {
  78         print {*STDOUT} <<'END';
  79 USAGE: git mw preview [--remote|-r <remote name>] [--autoload|-a]
  80                       [--output|-o <output filename>] [--verbose|-v]
  81                       <blob> | <filename>
  82
  83 DESCRIPTION:
  84 Preview is an utiliy to preview local content of a mediawiki repo as if it was
  85 pushed on the remote.
  86
  87 For that, preview searches for the remote name of the current branch's
  88 upstream if --remote is not set. If that remote is not found or if it
  89 is not a mediawiki, it lists all mediawiki remotes configured and asks
  90 you to replay your command with the --remote option set properly.
  91
  92 Then, it searches for a file named 'filename'. If it's not found in
  93 the current dir, it will assume it's a blob.
  94
  95 The content retrieved in the file (or in the blob) will then be parsed
  96 by the remote mediawiki and combined with a template retrieved from
  97 the mediawiki.
  98
  99 Finally, preview will save the HTML result in a file. and autoload it
 100 in your default web browser if the option --autoload is present.
 101
 102 OPTIONS:
 103     -r <remote name>, --remote <remote name>
 104         If the remote is a mediawiki, the template and the parse engine
 105         used for the preview will be those of that remote.
 106         If not, a list of valid remotes will be shown.
 107
 108     -a, --autoload
 109         Try to load the HTML output in a new tab (or new window) of your
 110         default web browser.
 111
 112     -o <output filename>, --output <output filename>
 113         Change the HTML output filename. Default filename is based on the
 114         input filename with its extension replaced by '.html'.
 115
 116     -v, --verbose
 117         Show more information on what's going on under the hood.
 118 END
 119         exit;
 120 }
 121
 122 sub preview {
 123         my $wiki;
 124         my ($remote_url, $wiki_page_name);
 125         my ($new_content, $template);
 126         my $file_content;
 127
 128         if ($file_name eq EMPTY) {
 129                 die "Missing file argument, see `git mw help`\n";
 130         }
 131
 132         v_print("### Selecting remote\n");
 133         if ($remote_name eq EMPTY) {
 134                 $remote_name = find_upstream_remote_name();
 135                 if ($remote_name) {
 136                         $remote_url = mediawiki_remote_url_maybe($remote_name);
 137                 }
 138
 139                 if (! $remote_url) {
 140                         my @valid_remotes = find_mediawiki_remotes();
 141
 142                         if ($#valid_remotes == 0) {
 143                                 print {*STDERR} "No mediawiki remote in this repo. \n";
 144                                 exit 1;
 145                         } else {
 146                                 my $remotes_list = join("\n\t", @valid_remotes);
 147                                 print {*STDERR} <<"MESSAGE";
 148 There are multiple mediawiki remotes, which of:
 149         ${remotes_list}
 150 do you want ? Use the -r option to specify the remote.
 151 MESSAGE
 152                         }
 153
 154                         exit 1;
 155                 }
 156         } else {
 157                 if (!is_valid_remote($remote_name)) {
 158                         die "${remote_name} is not a remote\n";
 159                 }
 160
 161                 $remote_url = mediawiki_remote_url_maybe($remote_name);
 162                 if (! $remote_url) {
 163                         die "${remote_name} is not a mediawiki remote\n";
 164                 }
 165         }
 166         v_print("selected remote:\n\tname: ${remote_name}\n\turl: ${remote_url}\n");
 167
 168         $wiki = connect_maybe($wiki, $remote_name, $remote_url);
 169
 170         # Read file content
 171         if (! -e $file_name) {
 172                 $file_content = git_cmd_try {
 173                         Git::command('cat-file', 'blob', $file_name); }
 174                         "%s failed w/ code %d";
 175
 176                 if ($file_name =~ /(.+):(.+)/) {
 177                         $file_name = $2;
 178                 }
 179         } else {
 180                 open my $read_fh, "<", $file_name
 181                         or die "could not open ${file_name}: $!\n";
 182                 $file_content = do { local $/ = undef; <$read_fh> };
 183                 close $read_fh
 184                         or die "unable to close: $!\n";
 185         }
 186
 187         v_print("### Retrieving template\n");
 188         ($wiki_page_name = clean_filename($file_name)) =~ s/\.[^.]+$//;
 189         $template = get_template($remote_url, $wiki_page_name);
 190
 191         v_print("### Parsing local content\n");
 192         $new_content = $wiki->api({
 193                 action => 'parse',
 194                 text => $file_content,
 195                 title => $wiki_page_name
 196         }, {
 197                 skip_encoding => 1
 198         }) or die "No response from remote mediawiki\n";
 199         $new_content = $new_content->{'parse'}->{'text'}->{'*'};
 200
 201         v_print("### Merging contents\n");
 202         if ($preview_file_name eq EMPTY) {
 203                 ($preview_file_name = $file_name) =~ s/\.[^.]+$/.html/;
 204         }
 205         open(my $save_fh, '>:encoding(UTF-8)', $preview_file_name)
 206                 or die "Could not open: $!\n";
 207         print {$save_fh} merge_contents($template, $new_content, $remote_url);
 208         close($save_fh)
 209                 or die "Could not close: $!\n";
 210
 211         v_print("### Results\n");
 212         if ($autoload) {
 213                 v_print("Launching browser w/ file: ${preview_file_name}");
 214                 system('git', 'web--browse', $preview_file_name);
 215         } else {
 216                 print {*STDERR} "Preview file saved as: ${preview_file_name}\n";
 217         }
 218
 219         exit;
 220 }
 221
 222 # uses global scope variable: $remote_name
 223 sub merge_contents {
 224         my $template = shift;
 225         my $content = shift;
 226         my $remote_url = shift;
 227         my ($content_tree, $html_tree, $mw_content_text);
 228         my $template_content_id = 'bodyContent';
 229
 230         $html_tree = HTML::TreeBuilder->new;
 231         $html_tree->parse($template);
 232
 233         $content_tree = HTML::TreeBuilder->new;
 234         $content_tree->parse($content);
 235
 236         $template_content_id = Git::config("remote.${remote_name}.mwIDcontent")
 237                 || $template_content_id;
 238         v_print("Using '${template_content_id}' as the content ID\n");
 239
 240         $mw_content_text = $html_tree->look_down('id', $template_content_id);
 241         if (!defined $mw_content_text) {
 242                 print {*STDERR} <<"CONFIG";
 243 Could not combine the new content with the template. You might want to
 244 configure `mediawiki.IDContent` in your config:
 245         git config --add remote.${remote_name}.mwIDcontent <id>
 246 and re-run the command afterward.
 247 CONFIG
 248                 exit 1;
 249         }
 250         $mw_content_text->delete_content();
 251         $mw_content_text->push_content($content_tree);
 252
 253         make_links_absolute($html_tree, $remote_url);
 254
 255         return $html_tree->as_HTML;
 256 }
 257
 258 sub make_links_absolute {
 259         my $html_tree = shift;
 260         my $remote_url = shift;
 261         for (@{ $html_tree->extract_links() }) {
 262                 my ($link, $element, $attr) = @{ $_ };
 263                 my $url = url($link)->canonical;
 264                 if ($url !~ /#/) {
 265                         $element->attr($attr, URI->new_abs($url, $remote_url));
 266                 }
 267         }
 268         return $html_tree;
 269 }
 270
 271 sub is_valid_remote {
 272         my $remote = shift;
 273         my @remotes = git_cmd_try {
 274                 Git::command('remote') }
 275                 "%s failed w/ code %d";
 276         my $found_remote = 0;
 277         foreach my $remote (@remotes) {
 278                 if ($remote eq $remote) {
 279                         $found_remote = 1;
 280                         last;
 281                 }
 282         }
 283         return $found_remote;
 284 }
 285
 286 sub find_mediawiki_remotes {
 287         my @remotes = git_cmd_try {
 288                 Git::command('remote'); }
 289                 "%s failed w/ code %d";
 290         my $remote_url;
 291         my @valid_remotes = ();
 292         foreach my $remote (@remotes) {
 293                 $remote_url = mediawiki_remote_url_maybe($remote);
 294                 if ($remote_url) {
 295                         push(@valid_remotes, $remote);
 296                 }
 297         }
 298         return @valid_remotes;
 299 }
 300
 301 sub find_upstream_remote_name {
 302         my $current_branch = git_cmd_try {
 303                 Git::command_oneline('symbolic-ref', '--short', 'HEAD') }
 304                 "%s failed w/ code %d";
 305         return Git::config("branch.${current_branch}.remote");
 306 }
 307
 308 sub mediawiki_remote_url_maybe {
 309         my $remote = shift;
 310
 311         # Find remote url
 312         my $remote_url = Git::config("remote.${remote}.url");
 313         if ($remote_url =~ s/mediawiki::(.*)/$1/) {
 314                 return url($remote_url)->canonical;
 315         }
 316
 317         return;
 318 }
 319
 320 sub get_template {
 321         my $url = shift;
 322         my $page_name = shift;
 323         my ($req, $res, $code, $url_after);
 324
 325         $req = LWP::UserAgent->new;
 326         if ($verbose) {
 327                 $req->show_progress(1);
 328         }
 329
 330         $res = $req->get("${url}/index.php?title=${page_name}");
 331         if (!$res->is_success) {
 332                 $code = $res->code;
 333                 $url_after = $res->request()->uri(); # resolve all redirections
 334                 if ($code == HTTP_CODE_PAGE_NOT_FOUND) {
 335                         if ($verbose) {
 336                                 print {*STDERR} <<"WARNING";
 337 Warning: Failed to retrieve '$page_name'. Create it on the mediawiki if you want
 338 all the links to work properly.
 339 Trying to use the mediawiki homepage as a fallback template ...
 340 WARNING
 341                         }
 342
 343                         # LWP automatically redirects GET request
 344                         $res = $req->get("${url}/index.php");
 345                         if (!$res->is_success) {
 346                                 $url_after = $res->request()->uri(); # resolve all redirections
 347                                 die "Failed to get homepage @ ${url_after} w/ code ${code}\n";
 348                         }
 349                 } else {
 350                         die "Failed to get '${page_name}' @ ${url_after} w/ code ${code}\n";
 351                 }
 352         }
 353
 354         return $res->decoded_content;
 355 }
 356
 357 ############################## Help Functions ##################################
 358
 359 sub help {
 360         print {*STDOUT} <<'END';
 361 usage: git mw <command> <args>
 362
 363 git mw commands are:
 364     help        Display help information about git mw
 365     preview     Parse and render local file into HTML
 366 END
 367         exit;
 368 }