scan-adoc-refs

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5 use IO::File;
   6 use JSON;
   7
   8 use Data::Dumper;
   9
  10 my $environments = {
  11     default => 1,
  12     wiki => 1,
  13     manvolnum => 1,
  14     pvelogo => 0, # ignore
  15 };
  16
  17 my $fileinfo = {};
  18
  19 my $start_env = [];
  20 foreach my $e (keys %$environments) {
  21     push @$start_env, $e if $environments->{$e};
  22 }
  23
  24 my $env_stack = [$start_env];
  25 my $env_name_stack = [];
  26
  27 sub reset_environment_stack {
  28     $env_stack = [$start_env];
  29     $env_name_stack = [];
  30 }
  31
  32 sub push_environment {
  33     my ($env, $not) = @_;
  34
  35     die "undefined environment '$env'\n" if !defined($environments->{$env});
  36
  37     # FIXME: this seems wrong  (nested env?)?
  38     return if !$environments->{$env}; # do not track
  39
  40     if ($not) {
  41         my $new_env = [];
  42         foreach my $e (@{$env_stack->[-1]}) {
  43             if ($e ne $env) {
  44                 push @$new_env, $e;
  45             }
  46         }
  47         die "empty environment" if !scalar($new_env);
  48         push @$env_stack, $new_env;
  49     } else {
  50         push @$env_stack, [$env];
  51     }
  52
  53     push @$env_name_stack, $env;
  54 }
  55
  56 sub pop_environment {
  57     my ($env) = @_;
  58
  59     die "undefined environment '$env'\n" if !defined($environments->{$env});
  60
  61     return if !$environments->{$env}; # do not track
  62
  63     pop @$env_stack;
  64     my $res = pop @$env_name_stack;
  65
  66     die "environment missmatch ($res != $env)\n" if $res ne $env;
  67 }
  68
  69 sub register_include {
  70     my ($filename, $include_filename, $env_list) = @_;
  71
  72     return if $include_filename !~ m/\.adoc$/; # skip attributes.txt
  73
  74     foreach my $e (@$env_list) {
  75         $fileinfo->{include}->{$e}->{$filename}->{$include_filename} = 1;
  76     }
  77 }
  78
  79 sub register_blockid {
  80     my ($filename, $blockid, $reftext, $env_list) = @_;
  81
  82     foreach my $e (@$env_list) {
  83         my $fn = $fileinfo->{blockid}->{$e}->{$blockid};
  84         die "blockid '$blockid' already defined in $fn"
  85             if defined($fn);
  86         $fileinfo->{blockid}->{$e}->{$blockid} = $filename;
  87         $fileinfo->{reftext}->{$e}->{$blockid} = $reftext
  88             if defined($reftext);
  89     }
  90 }
  91
  92 sub register_title {
  93     my ($filename, $env, $doctype, $title, $blockid) = @_;
  94
  95     # fixme: what about other macros?
  96     $title =~ s/\{pve\}/Proxmox VE/g;
  97     $title =~ s!http://\S+\[(.*?)\]!$1!g;
  98
  99     # register document title (onyl once)
 100     if (!defined($fileinfo->{titles}->{$env}->{$filename})) {
 101
 102         $fileinfo->{titles}->{$env}->{$filename} = $title;
 103
 104         if (defined($doctype)) {
 105             $fileinfo->{doctype}->{$env}->{$filename} = $doctype;
 106         } else {
 107             die "unable to change title (no doctype)"
 108                 if !defined($fileinfo->{doctype}->{$env}->{$filename});
 109         }
 110     } elsif (!defined($doctype)) {
 111         # change title via :title: attribute
 112         $fileinfo->{titles}->{$env}->{$filename} = $title;
 113     }
 114
 115     if (defined($doctype) && ($env eq 'manvolnum') && ($doctype == 0)) {
 116         if ($title =~ m/.*\(([1-8])\)\s*$/) {
 117             $fileinfo->{mansection}->{$env}->{$filename} = $1;
 118         }
 119     }
 120
 121     if ($blockid) {
 122         die "internal error"
 123             if !defined($fileinfo->{blockid}->{$env}->{$blockid});
 124         $fileinfo->{reftitle}->{$env}->{$blockid} = $title;
 125     }
 126 }
 127
 128 sub scan_adoc_file {
 129     my ($filename) = @_;
 130
 131     reset_environment_stack();
 132
 133     # print "SCAN $filename\n";
 134
 135     my $fh = IO::File->new("$filename", "r") or
 136         die "unable to open file '$filename' - $!\n";
 137
 138     my $env_last_line = {};
 139     my $env_last_blockid = {};
 140
 141     while (defined (my $line = <$fh>)) {
 142         if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
 143             my ($not, $env, $text) = ($1, $2, $3);
 144             die "unsuported ifdef usage - implement me" if $text;
 145             push_environment($env, $not);
 146             next;
 147         } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
 148             my ($env, $text) = ($1, $2);
 149             die "unsuported ifdef usage - implement me" if $text;
 150             pop_environment($env);
 151             next;
 152         } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
 153             register_include($filename, $1, $env_stack->[-1]);
 154             next;
 155         }
 156
 157         # try to detect titles
 158         foreach my $e (@{$env_stack->[-1]}) {
 159             if ($line =~ m/^=====+$/) {
 160                 register_title($filename, $e, 0, $env_last_line->{$e},
 161                                $env_last_blockid->{$e});
 162             } elsif ($line =~ m/^-----+$/) {
 163                 register_title($filename, $e, 1, $env_last_line->{$e},
 164                                $env_last_blockid->{$e});
 165             } elsif ($line =~ m/^~~~~~+$/) {
 166                 register_title($filename, $e, 2, $env_last_line->{$e},
 167                                $env_last_blockid->{$e});
 168             } elsif ($line =~ m/^\^\^\^\^\^+$/) {
 169                 register_title($filename, $e, 3, $env_last_line->{$e},
 170                                $env_last_blockid->{$e});
 171             } elsif ($line =~ m/^= +(\S.*?)( +=)?$/) {
 172                 register_title($filename, $e, 0, $1, $env_last_blockid->{$e});
 173             } elsif ($line =~ m/^== +(\S.*?)( +==)?$/) {
 174                 register_title($filename, $e, 1, $1, $env_last_blockid->{$e});
 175             } elsif ($line =~ m/^=== +(\S.*?)( +===)?$/) {
 176                 register_title($filename, $e, 2, $1, $env_last_blockid->{$e});
 177             } elsif ($line =~ m/^==== +(\S.*?)( +====)?$/) {
 178                 register_title($filename, $e, 3, $1, $env_last_blockid->{$e});
 179             }
 180
 181             $env_last_line->{$e} = $line;
 182             chomp $env_last_line->{$e};
 183         }
 184
 185         if ($line =~ m/^:(\S+?):\s*(.*\S)?\s*$/) {
 186             my ($key, $value) = ($1, $2);
 187             if ($key eq 'pve-toplevel') {
 188
 189                 foreach my $e (@{$env_stack->[-1]}) {
 190                     my $title = $fileinfo->{titles}->{$e}->{$filename};
 191                     die "not title for toplevel file '$filename' (env=$e)\n"
 192                         if !defined($title);
 193                     $fileinfo->{toplevel}->{$e}->{$filename} = 1;
 194                 }
 195             } elsif ($key eq 'title') {
 196                 foreach my $e (@{$env_stack->[-1]}) {
 197                     register_title($filename, $e, undef, $value);
 198                 }
 199             }
 200         }
 201
 202         if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
 203             my $blockid = $1;
 204             die "implement me" if $blockid =~m/,/;
 205             my $reftext = '';
 206             register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
 207             foreach my $e (@{$env_stack->[-1]}) {
 208                 $env_last_blockid->{$e} = $blockid;
 209             }
 210         }
 211
 212         if ($line =~ m/^\s*$/) {
 213             foreach my $e (@{$env_stack->[-1]}) {
 214                 delete $env_last_blockid->{$e};
 215             }
 216         }
 217
 218         # fixme: "anchor:"
 219         # bibliography anchors
 220         if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
 221             my $blockid = $1;
 222             die "implement me" if $blockid =~m/,/;
 223             register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
 224         }
 225     }
 226 }
 227
 228 my $scanned_files = {};
 229 while (my $filename = shift) {
 230     next if $filename !~ m/\.adoc$/; # skip attributes.txt
 231     next if $scanned_files->{$filename};
 232
 233     scan_adoc_file($filename);
 234     $scanned_files->{$filename} = 1;
 235 }
 236
 237 sub resolve_link_target {
 238     my ($env, $filename) = @_;
 239
 240     my $include_hash = $fileinfo->{include}->{$env};
 241
 242     my $repeat = 1;
 243
 244     while ($repeat) {
 245         $repeat = 0;
 246         foreach my $fn (keys %$include_hash) {
 247             if ($include_hash->{$fn}->{$filename}) {
 248                 next if ($fn eq 'pve-admin-guide.adoc') &&
 249                     $fileinfo->{outfile}->{$env}->{$filename};
 250                 $filename = $fn;
 251                 $repeat = 1;
 252                 last;
 253             }
 254         }
 255     }
 256
 257     return $filename;
 258 }
 259
 260
 261 # try to generate output file mapping
 262 foreach my $e (@$start_env) {
 263     my $toplevel_hash = $fileinfo->{toplevel}->{$e};
 264     foreach my $fn (sort keys %$toplevel_hash) {
 265         my $mansection = $fileinfo->{mansection}->{manvolnum}->{$fn};
 266         if ($e eq 'wiki') {
 267             my $realfn = $fn;
 268             $realfn =~ s/\.adoc$//;
 269             if (defined($mansection) && ($mansection == 5)) {
 270                 $realfn .= ".$mansection";
 271             }
 272             $realfn = "$realfn-plain.html";
 273             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 274         } elsif ($e eq 'manvolnum') {
 275             my $realfn = $fn;
 276             $realfn =~ s/\.adoc$//;
 277             die "toplevel file '$fn' is not marked as manual page!" if !$mansection;
 278             $realfn .= ".$mansection";
 279             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 280         } elsif ($e eq 'default') {
 281             my $realfn = $fn;
 282             $realfn =~ s/\.adoc$//;
 283             if (defined($mansection) && ($mansection == 5)) {
 284                 $realfn .= ".$mansection";
 285                 $realfn = "$realfn.html";
 286             } else {
 287                 if (($fn ne 'pve-admin-guide.adoc') &&
 288                     $fileinfo->{doctype}->{$e}->{$fn} == 0) {
 289                     $realfn = "chapter-$realfn.html";
 290                 } else {
 291                     $realfn = "$realfn.html";
 292                 }
 293             }
 294             $fileinfo->{outfile}->{$e}->{$fn} = $realfn;
 295         }
 296     }
 297 }
 298
 299 # now resolve blockids
 300 foreach my $e (@$start_env) {
 301     my $blockid_hash = $fileinfo->{blockid}->{$e};
 302     foreach my $blockid (keys %$blockid_hash) {
 303         my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
 304         if ($e eq 'wiki') {
 305             my $title = $fileinfo->{titles}->{$e}->{$fn};
 306             $title =~ s/\s/_/g;
 307             die "found not title for '$fn' in env '$e'" if !$title;
 308             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
 309
 310             # we do not produce wiki pages for all content
 311             #my $realfn = $fileinfo->{outfile}->{$e}->{$fn};
 312             #warn "no output file mapping for '$fn' ($e)\n" if !$realfn;
 313
 314         } elsif ($e eq 'default') {
 315             my $realfn = $fileinfo->{outfile}->{$e}->{$fn} ||
 316                 die "no output file mapping for '$fn'\n";
 317             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/pve-docs/$realfn#$blockid";
 318         } elsif ($e eq 'manvolnum') {
 319             # we do not produce manpages for all content
 320             # my $realfn = $fileinfo->{outfile}->{$e}->{$fn} ||
 321             # warn "no output file mapping for '$fn'\n";
 322             $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
 323         }
 324     }
 325 }
 326
 327
 328 print to_json($fileinfo, { pretty => 1,  canonical => 1 } );