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) = @_;
  94
  95     # fixme: what about other macros?
  96     $title =~ s/\{pve\}/Proxmox VE/g;
  97     $title =~ s!http://\S+\[(.*?)\]!$1!g;
  98
  99     $fileinfo->{titles}->{$env}->{$filename} = $title;
 100
 101     if (defined($doctype)) {
 102         $fileinfo->{doctype}->{$env}->{$filename} = $doctype;
 103     } else {
 104         die "unable to change title (no doctype)"
 105             if !defined($fileinfo->{doctype}->{$env}->{$filename});
 106     }
 107 }
 108
 109 sub scan_adoc_file {
 110     my ($filename) = @_;
 111
 112     reset_environment_stack();
 113
 114     # print "SCAN $filename\n";
 115
 116     my $fh = IO::File->new("$filename", "r") or
 117         die "unable to open file '$filename' - $!\n";
 118
 119     my $env_last_line = {};
 120
 121     while (defined (my $line = <$fh>)) {
 122         if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
 123             my ($not, $env, $text) = ($1, $2, $3);
 124             die "unsuported ifdef usage - implement me" if $text;
 125             push_environment($env, $not);
 126             next;
 127         } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
 128             my ($env, $text) = ($1, $2);
 129             die "unsuported ifdef usage - implement me" if $text;
 130             pop_environment($env);
 131             next;
 132         } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
 133             register_include($filename, $1, $env_stack->[-1]);
 134             next;
 135         }
 136
 137         # try to detect titles
 138         foreach my $e (@{$env_stack->[-1]}) {
 139             my $title = $fileinfo->{titles}->{$e}->{$filename};
 140             next if defined($title);
 141
 142             if ($line =~ m/^=====+$/) {
 143                 register_title($filename, $e, 0, $env_last_line->{$e});
 144             } elsif ($line =~ m/^-----+$/) {
 145                 register_title($filename, $e, 1, $env_last_line->{$e});
 146             } elsif ($line =~ m/^= +(\S.*?)( +=)?$/) {
 147                 register_title($filename, $e, 0, $1);
 148             } elsif ($line =~ m/^== +(\S.*?)( +==)?$/) {
 149                 register_title($filename, $e, 1, $1);
 150             }
 151
 152             $env_last_line->{$e} = $line;
 153             chomp $env_last_line->{$e};
 154         }
 155
 156         if ($line =~ m/^:(\S+?):\s*(.*\S)?\s*$/) {
 157             my ($key, $value) = ($1, $2);
 158             if ($key eq 'pve-toplevel') {
 159
 160                 foreach my $e (@{$env_stack->[-1]}) {
 161                     my $title = $fileinfo->{titles}->{$e}->{$filename};
 162                     die "not title for toplevel file '$filename' (env=$e)\n"
 163                         if !defined($title);
 164                     $fileinfo->{toplevel}->{$e}->{$filename} = 1;
 165                 }
 166             } elsif ($key eq 'title') {
 167                 foreach my $e (@{$env_stack->[-1]}) {
 168                     register_title($filename, $e, undef, $value);
 169                 }
 170             }
 171         }
 172
 173         if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
 174             my $blockid = $1;
 175             die "implement me" if $blockid =~m/,/;
 176             my $reftext = '';
 177             register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
 178         }
 179         # fixme: "anchor:"
 180         # bibliography anchors
 181         if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
 182             my $blockid = $1;
 183             die "implement me" if $blockid =~m/,/;
 184             register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
 185         }
 186     }
 187 }
 188
 189 my $scanned_files = {};
 190 while (my $filename = shift) {
 191     next if $filename !~ m/\.adoc$/; # skip attributes.txt
 192     next if $scanned_files->{$filename};
 193
 194     scan_adoc_file($filename);
 195     $scanned_files->{$filename} = 1;
 196 }
 197
 198 sub resolve_link_target {
 199     my ($env, $filename) = @_;
 200
 201     my $include_hash = $fileinfo->{include}->{$env};
 202
 203     my $repeat = 1;
 204
 205     while ($repeat) {
 206         $repeat = 0;
 207         foreach my $fn (keys %$include_hash) {
 208             if ($include_hash->{$fn}->{$filename}) {
 209                 $filename = $fn;
 210                 $repeat = 1;
 211                 last;
 212             }
 213         }
 214     }
 215
 216     return $filename;
 217 }
 218
 219 # now resolve blockids
 220 foreach my $e (@$start_env) {
 221     my $blockid_hash = $fileinfo->{blockid}->{$e};
 222     foreach my $blockid (keys %$blockid_hash) {
 223         my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
 224         if ($e eq 'wiki') {
 225             my $title = $fileinfo->{titles}->{$e}->{$fn};
 226             $title =~ s/\s/_/g;
 227             die "found not title for '$fn' in env '$e'" if !$title;
 228             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
 229         } else {
 230             $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
 231         }
 232     }
 233 }
 234
 235
 236 print to_json($fileinfo, { pretty => 1 } );