scan-adoc-refs

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5 use Getopt::Long;
   6 use IO::File;
   7 use JSON;
   8
   9 use Data::Dumper;
  10
  11 my $generate_depends;
  12
  13 GetOptions("depends=s" => \$generate_depends) or
  14     die("Error in command line arguments\n");
  15
  16
  17 my $environments = {
  18     default => 1,
  19     wiki => 1,
  20     manvolnum => 1,
  21     pvelogo => 0, # ignore
  22 };
  23
  24 my $fileinfo = {};
  25
  26 my $start_env = [];
  27 foreach my $e (keys %$environments) {
  28     push @$start_env, $e if $environments->{$e};
  29 }
  30
  31 my $env_stack = [$start_env];
  32 my $env_name_stack = [];
  33
  34 sub reset_environment_stack {
  35     $env_stack = [$start_env];
  36     $env_name_stack = [];
  37 }
  38
  39 sub push_environment {
  40     my ($env, $not) = @_;
  41
  42     die "undefined environment '$env'\n" if !defined($environments->{$env});
  43
  44     # FIXME: this seems wrong  (nested env?)?
  45     return if !$environments->{$env}; # do not track
  46
  47     if ($not) {
  48         my $new_env = [];
  49         foreach my $e (@{$env_stack->[-1]}) {
  50             if ($e ne $env) {
  51                 push @$new_env, $e;
  52             }
  53         }
  54         die "empty environment" if !scalar($new_env);
  55         push @$env_stack, $new_env;
  56     } else {
  57         push @$env_stack, [$env];
  58     }
  59
  60     push @$env_name_stack, $env;
  61 }
  62
  63 sub pop_environment {
  64     my ($env) = @_;
  65
  66     die "undefined environment '$env'\n" if !defined($environments->{$env});
  67
  68     return if !$environments->{$env}; # do not track
  69
  70     pop @$env_stack;
  71     my $res = pop @$env_name_stack;
  72
  73     die "environment missmatch ($res != $env)\n" if $res ne $env;
  74 }
  75
  76 sub register_include {
  77     my ($filename, $include_filename, $env_list) = @_;
  78
  79     foreach my $e (@$env_list) {
  80         $fileinfo->{include}->{$e}->{$filename}->{$include_filename} = 1;
  81     }
  82 }
  83
  84 sub register_blockid {
  85     my ($filename, $blockid, $reftext, $env_list) = @_;
  86
  87     foreach my $e (@$env_list) {
  88         my $fn = $fileinfo->{blockid}->{$e}->{$blockid};
  89         die "blockid '$blockid' already defined in $fn"
  90             if defined($fn);
  91         $fileinfo->{blockid}->{$e}->{$blockid} = $filename;
  92         $fileinfo->{reftext}->{$e}->{$blockid} = $reftext
  93             if defined($reftext);
  94     }
  95 }
  96
  97 sub register_title {
  98     my ($filename, $env, $doctype, $title, $blockid) = @_;
  99
 100     # fixme: what about other macros?
 101     $title =~ s/\{pve\}/Proxmox VE/g;
 102     $title =~ s!http://\S+\[(.*?)\]!$1!g;
 103
 104     # register document title (onyl once)
 105     if (!defined($fileinfo->{titles}->{$env}->{$filename})) {
 106
 107         $fileinfo->{titles}->{$env}->{$filename} = $title;
 108
 109         if (defined($doctype)) {
 110             $fileinfo->{doctype}->{$env}->{$filename} = $doctype;
 111         } else {
 112             die "unable to change title (no doctype)"
 113                 if !defined($fileinfo->{doctype}->{$env}->{$filename});
 114         }
 115     } elsif (!defined($doctype)) {
 116         # change title via :title: attribute
 117         $fileinfo->{titles}->{$env}->{$filename} = $title;
 118     }
 119
 120     if (defined($doctype) && ($env eq 'manvolnum') && ($doctype == 0)) {
 121         if ($title =~ m/.*\(([1-8])\)\s*$/) {
 122             $fileinfo->{mansection}->{$env}->{$filename} = $1;
 123         }
 124     }
 125
 126     if ($blockid) {
 127         die "internal error"
 128             if !defined($fileinfo->{blockid}->{$env}->{$blockid});
 129         $fileinfo->{reftitle}->{$env}->{$blockid} = $title;
 130     }
 131 }
 132
 133 sub scan_adoc_file {
 134     my ($filename) = @_;
 135
 136     reset_environment_stack();
 137
 138     # print "SCAN $filename\n";
 139
 140     my $fh = IO::File->new("$filename", "r") or
 141         die "unable to open file '$filename' - $!\n";
 142
 143     my $env_last_line = {};
 144     my $env_last_blockid = {};
 145
 146     while (defined (my $line = <$fh>)) {
 147         if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
 148             my ($not, $env, $text) = ($1, $2, $3);
 149             die "unsuported ifdef usage - implement me" if $text;
 150             push_environment($env, $not);
 151             next;
 152         } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
 153             my ($env, $text) = ($1, $2);
 154             die "unsuported ifdef usage - implement me" if $text;
 155             pop_environment($env);
 156             next;
 157         } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
 158             register_include($filename, $1, $env_stack->[-1]);
 159             next;
 160         }
 161
 162         # try to detect titles
 163         foreach my $e (@{$env_stack->[-1]}) {
 164             if ($line =~ m/^=====+$/) {
 165                 register_title($filename, $e, 0, $env_last_line->{$e},
 166                                $env_last_blockid->{$e});
 167             } elsif ($line =~ m/^-----+$/) {
 168                 register_title($filename, $e, 1, $env_last_line->{$e},
 169                                $env_last_blockid->{$e});
 170             } elsif ($line =~ m/^~~~~~+$/) {
 171                 register_title($filename, $e, 2, $env_last_line->{$e},
 172                                $env_last_blockid->{$e});
 173             } elsif ($line =~ m/^\^\^\^\^\^+$/) {
 174                 register_title($filename, $e, 3, $env_last_line->{$e},
 175                                $env_last_blockid->{$e});
 176             } elsif ($line =~ m/^= +(\S.*?)( +=)?$/) {
 177                 register_title($filename, $e, 0, $1, $env_last_blockid->{$e});
 178             } elsif ($line =~ m/^== +(\S.*?)( +==)?$/) {
 179                 register_title($filename, $e, 1, $1, $env_last_blockid->{$e});
 180             } elsif ($line =~ m/^=== +(\S.*?)( +===)?$/) {
 181                 register_title($filename, $e, 2, $1, $env_last_blockid->{$e});
 182             } elsif ($line =~ m/^==== +(\S.*?)( +====)?$/) {
 183                 register_title($filename, $e, 3, $1, $env_last_blockid->{$e});
 184             }
 185
 186             $env_last_line->{$e} = $line;
 187             chomp $env_last_line->{$e};
 188         }
 189
 190         if ($line =~ m/^:(\S+?):\s*(.*\S)?\s*$/) {
 191             my ($key, $value) = ($1, $2);
 192             if ($key eq 'pve-toplevel') {
 193
 194                 foreach my $e (@{$env_stack->[-1]}) {
 195                     my $title = $fileinfo->{titles}->{$e}->{$filename};
 196                     die "not title for toplevel file '$filename' (env=$e)\n"
 197                         if !defined($title);
 198                     $fileinfo->{toplevel}->{$e}->{$filename} = 1;
 199                 }
 200             } elsif ($key eq 'title') {
 201                 foreach my $e (@{$env_stack->[-1]}) {
 202                     register_title($filename, $e, undef, $value);
 203                 }
 204             }
 205         }
 206
 207         if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
 208             my $blockid = $1;
 209             die "implement me" if $blockid =~m/,/;
 210             my $reftext = '';
 211             register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
 212             foreach my $e (@{$env_stack->[-1]}) {
 213                 $env_last_blockid->{$e} = $blockid;
 214             }
 215         }
 216
 217         if ($line =~ m/^\s*$/) {
 218             foreach my $e (@{$env_stack->[-1]}) {
 219                 delete $env_last_blockid->{$e};
 220             }
 221         }
 222
 223         # fixme: "anchor:"
 224         # bibliography anchors
 225         if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
 226             my $blockid = $1;
 227             die "implement me" if $blockid =~m/,/;
 228             register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
 229         }
 230     }
 231 }
 232
 233 my $scanned_files = {};
 234 while (my $filename = shift) {
 235     next if $filename !~ m/\.adoc$/; # skip attributes.txt
 236     next if $scanned_files->{$filename};
 237
 238     scan_adoc_file($filename);
 239     $scanned_files->{$filename} = 1;
 240 }
 241
 242 sub resolve_link_target {
 243     my ($env, $filename) = @_;
 244
 245     my $include_hash = $fileinfo->{include}->{$env};
 246
 247     my $repeat = 1;
 248
 249     while ($repeat) {
 250         $repeat = 0;
 251         foreach my $fn (keys %$include_hash) {
 252             if ($include_hash->{$fn}->{$filename}) {
 253                 next if ($fn eq 'pve-admin-guide.adoc') &&
 254                     $fileinfo->{outfile}->{$env}->{$filename};
 255                 $filename = $fn;
 256                 $repeat = 1;
 257                 last;
 258             }
 259         }
 260     }
 261
 262     return $filename;
 263 }
 264
 265
 266 # try to generate output file mapping
 267 foreach my $e (@$start_env) {
 268     my $toplevel_hash = $fileinfo->{toplevel}->{$e};
 269     foreach my $fn (sort keys %$toplevel_hash) {
 270         my $mansection = $fileinfo->{mansection}->{manvolnum}->{$fn};
 271         if ($e eq 'wiki') {
 272             my $realfn = $fn;
 273             $realfn =~ s/\.adoc$//;
 274             if (defined($mansection) && ($mansection == 5)) {
 275                 $realfn .= ".$mansection";
 276             }
 277             $realfn = "$realfn-plain.html";
 278             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 279         } elsif ($e eq 'manvolnum') {
 280             my $realfn = $fn;
 281             $realfn =~ s/\.adoc$//;
 282             die "toplevel file '$fn' is not marked as manual page!" if !$mansection;
 283             $realfn .= ".$mansection";
 284             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 285         } elsif ($e eq 'default') {
 286             my $realfn = $fn;
 287             $realfn =~ s/\.adoc$//;
 288             if (defined($mansection) && ($mansection == 5)) {
 289                 $realfn .= ".$mansection";
 290                 $realfn = "$realfn.html";
 291             } else {
 292                 if (($fn ne 'pve-admin-guide.adoc') &&
 293                     $fileinfo->{doctype}->{$e}->{$fn} == 0) {
 294                     $realfn = "chapter-$realfn.html";
 295                 } else {
 296                     $realfn = "$realfn.html";
 297                 }
 298             }
 299             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 300         }
 301     }
 302 }
 303
 304 # now resolve blockids
 305 foreach my $e (@$start_env) {
 306     my $blockid_hash = $fileinfo->{blockid}->{$e};
 307     foreach my $blockid (keys %$blockid_hash) {
 308         my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
 309         if ($e eq 'wiki') {
 310             my $title = $fileinfo->{titles}->{$e}->{$fn};
 311             $title =~ s/\s/_/g;
 312             die "found not title for '$fn' in env '$e'" if !$title;
 313             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
 314
 315             # we do not produce wiki pages for all content
 316             #my $realfn = $fileinfo->{outfile}->{$e}->{$fn};
 317             #warn "no output file mapping for '$fn' ($e)\n" if !$realfn;
 318
 319         } elsif ($e eq 'default') {
 320             my $realfn = $fileinfo->{outfile}->{$e}->{$fn} ||
 321                 die "no output file mapping for '$fn'\n";
 322             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/pve-docs/$realfn#$blockid";
 323         } elsif ($e eq 'manvolnum') {
 324             # we do not produce manpages for all content
 325             # my $realfn = $fileinfo->{outfile}->{$e}->{$fn} ||
 326             # warn "no output file mapping for '$fn'\n";
 327             $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
 328         }
 329     }
 330 }
 331
 332 if ($generate_depends) {
 333
 334     my $tmpfilename = "${generate_depends}.tmp";
 335     my $outfh = IO::File->new($tmpfilename, "w") ||
 336         die "unable to open temporary file '$tmpfilename'\n";
 337
 338     my $depends = {};
 339     foreach my $e (@$start_env) {
 340         my $env_data = $fileinfo->{include}->{$e};
 341         foreach my $fn (keys %$env_data) {
 342             foreach my $dep (keys %{$env_data->{$fn}}) {
 343                 $depends->{$fn}->{$dep} = 1;
 344             }
 345         }
 346     }
 347
 348     my $res = '';
 349
 350     my $depend_varname_hash = {};
 351
 352     foreach my $fn (sort keys %$depends) {
 353         my $basename = uc($fn);
 354         $basename =~s/\.adoc$//i;
 355         $basename =~s/[^A-Za-z0-9]/_/g;
 356
 357         my $var1text = "${basename}_ADOCSOURCES =";
 358         my $var2text = "${basename}_ADOCDEPENDS = \$\{${basename}_ADOCSOURCES\}";
 359         $depend_varname_hash->{$fn} = "${basename}_ADOCDEPENDS";
 360
 361         foreach my $dep (sort keys %{$depends->{$fn}}) {
 362             if ($dep =~ m/-(opts|synopsis).adoc$/) {
 363                 $var2text .= " \\\n\t$dep";
 364             } else {
 365                 $var1text .= " \\\n\t$dep";
 366             }
 367         }
 368         $res .= "$var1text\n\n";
 369         $res .= "$var2text\n\n";
 370     }
 371
 372     my $man_sources_hash = {};
 373     foreach my $sourcefile (keys %{$fileinfo->{outfile}->{manvolnum}}) {
 374         $man_sources_hash->{$sourcefile} = 1;
 375         my $ihash = $fileinfo->{include}->{manvolnum}->{$sourcefile};
 376         foreach my $include (keys %$ihash) {
 377             if ($include !~ m/-(opts|synopsis).adoc$/) {
 378                 $man_sources_hash->{$include} = 1;
 379             }
 380         }
 381     }
 382
 383     foreach my $e (@$start_env) {
 384         my $filelist = $fileinfo->{outfile}->{$e};
 385         foreach my $sourcefile (sort keys %$filelist) {
 386             my $varname = $depend_varname_hash->{$sourcefile};
 387             my $target = $filelist->{$sourcefile};
 388             $res .= "$target: \$\{$varname\}\n\n";
 389
 390             if ($e eq 'manvolnum') {
 391
 392                 $res .= "$target.html: \$\{$varname\}\n\n";
 393                 $res .= "$target-plain.html: \$\{$varname\}\n\n";
 394             }
 395         }
 396     }
 397
 398     my $varname = "MANUAL_SOURCES";
 399     $res .= "$varname =";
 400     foreach my $sourcefile (sort keys %$man_sources_hash) {
 401         $res .= " \\\n\t$sourcefile";
 402     }
 403     $res .= "\n\n";
 404
 405     $varname = "CHAPTER_LIST";
 406     $res .= "$varname =";
 407     my $filelist = $fileinfo->{outfile}->{default};
 408     foreach my $sourcefile (sort keys %$filelist) {
 409         my $target = $filelist->{$sourcefile};
 410         $res .= " \\\n\t$target";
 411     }
 412     $res .= "\n\n";
 413
 414     $varname = "MANUAL_PAGES";
 415     $res .= "$varname =";
 416     $filelist = $fileinfo->{outfile}->{manvolnum};
 417     foreach my $manpage (sort keys %$filelist) {
 418         my $target = $filelist->{$manpage};
 419         $res .= " \\\n\t$target";
 420     }
 421     $res .= "\n\n";
 422
 423     $varname = "WIKI_IMPORTS";
 424     $res .= "$varname =";
 425     $filelist = $fileinfo->{outfile}->{wiki};
 426     foreach my $sourcefile (sort keys %$filelist) {
 427         my $target = $filelist->{$sourcefile};
 428         $res .= " \\\n\t$target";
 429     }
 430     $res .= "\n\n";
 431
 432     print $outfh $res;
 433     close($outfh);
 434
 435     rename($tmpfilename, $generate_depends) ||
 436         die "rename failed - $!";
 437 }
 438
 439 print to_json($fileinfo, { pretty => 1,  canonical => 1 } );