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 scan_adoc_file {
  93     my ($filename) = @_;
  94
  95     reset_environment_stack();
  96
  97     # print "SCAN $filename\n";
  98
  99     my $fh = IO::File->new("$filename", "r") or
 100         die "unable to open file '$filename' - $!\n";
 101
 102     my $env_last_line = {};
 103
 104     while (defined (my $line = <$fh>)) {
 105         if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
 106             my ($not, $env, $text) = ($1, $2, $3);
 107             die "unsuported ifdef usage - implement me" if $text;
 108             push_environment($env, $not);
 109             next;
 110         } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
 111             my ($env, $text) = ($1, $2);
 112             die "unsuported ifdef usage - implement me" if $text;
 113             pop_environment($env);
 114             next;
 115         } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
 116             register_include($filename, $1, $env_stack->[-1]);
 117             next;
 118         }
 119
 120         # try to detect titles
 121         foreach my $e (@{$env_stack->[-1]}) {
 122             my $title = $fileinfo->{titles}->{$e}->{$filename};
 123             next if defined($title);
 124
 125             if (($line =~ m/^=====+/) || ($line =~ m/^-----+/)) {
 126                 $fileinfo->{titles}->{$e}->{$filename} = $env_last_line->{$e};
 127             }
 128             $env_last_line->{$e} = $line;
 129             chomp $env_last_line->{$e};
 130         }
 131
 132         # fixme: also scan <<>>
 133
 134         while ($line =~ m/xref:([^\s\[\]]+)\[([^\]]*)\]/g) {
 135             # print "$filename xref:$1 [$2]\n";
 136         }
 137
 138         if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
 139             my $blockid = $1;
 140             die "implement me" if $blockid =~m/,/;
 141             my $reftext = '';
 142             register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
 143         }
 144         # fixme: "anchor:"
 145         # bibliography anchors
 146         if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
 147             my $blockid = $1;
 148             die "implement me" if $blockid =~m/,/;
 149             register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
 150         }
 151     }
 152 }
 153
 154 my $scanned_files = {};
 155 while (my $filename = shift) {
 156     next if $filename !~ m/\.adoc$/; # skip attributes.txt
 157     next if $scanned_files->{$filename};
 158
 159     scan_adoc_file($filename);
 160     $scanned_files->{$filename} = 1;
 161 }
 162
 163 sub resolve_link_target {
 164     my ($env, $filename) = @_;
 165
 166     my $include_hash = $fileinfo->{include}->{$env};
 167
 168     my $repeat = 1;
 169
 170     while ($repeat) {
 171         $repeat = 0;
 172         foreach my $fn (keys %$include_hash) {
 173             if ($include_hash->{$fn}->{$filename}) {
 174                 $filename = $fn;
 175                 $repeat = 1;
 176                 last;
 177             }
 178         }
 179     }
 180
 181     return $filename;
 182 }
 183
 184 # now resolve blockids
 185 foreach my $e (@$start_env) {
 186     my $blockid_hash = $fileinfo->{blockid}->{$e};
 187     foreach my $blockid (keys %$blockid_hash) {
 188         my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
 189         if ($e eq 'wiki') {
 190             my $title = $fileinfo->{titles}->{$e}->{$fn};
 191             $title =~ s/\{pve\}/Proxmox VE/g;
 192             $title =~ s/\s/_/g;
 193             die "found not title for '$fn' in env '$e'" if !$title;
 194             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
 195         } else {
 196             $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
 197         }
 198     }
 199 }
 200
 201
 202 print to_json($fileinfo, { pretty => 1 } );